Three is better than two according to Berkeley Lab researchers who've debunked a longstanding assumption concerning how hydrogen molecules break apart and adhere to a catalyst surface.

Using scanning tunneling microscopy, the team determined that groups of three or more vacancies on a palladium surface are required to facilitate a process called dissociative hydrogen adsorption, in which a hydrogen molecule separates into two hydrogen atoms, each of which then occupy a vacancy.

	[Click for larger image]
	Under the scanning tunneling microscope, vacancies appear as red bumps in the image of a palladium surface. The two three-vacancy clusters, one located near the top and one on the right, can facilitate hydrogen adsorption.

Their research, reported in the April 17 issue of Nature, counters the widely held belief that only a pair of vacancies is needed to trigger hydrogen adsorption. The discovery may prompt scientists to rethink the underlying principles that drive many industrially important catalytic processes, such as those used in gasoline reformulation and hydrogen fuel cells.

"This paper is a call to theorists," says Miquel Salmeron, a physicist with Berkeley Lab's Materials Sciences Division. "It reveals that we don't fully understand catalysis, even in simple systems."

Catalyst surfaces supply chemical reactions with atoms. In a hydrogen fuel cell, for example, a metal surface adsorbs hydrogen atoms from a gas such as methane and supplies these atoms to an electrochemical reaction that generates electrical power. But although adsorption is a cornerstone of hundreds of catalytic processes, it is only partially understood. Researchers know that in order for a molecule to break apart and adsorb onto a catalyst surface, it must find a home among the surface's constantly fluctuating mix of atoms. If the molecule can't find this home, or active site, it bounces off.

Researchers also know that in many cases these active sites are composed of atom-free vacancies. Called atomic adsorption sites, these vacancies jitterbug across a catalyst surface, sometimes moving alone, sometimes forming short-lived clusters of two, three, or more. Every so often, this random dance brings together enough vacancies to create a home for a hydrogen molecule — a phenomenon at the heart of how catalyst surfaces work, and also where scientists' understanding becomes hazy. For years, they believed an active site needs only as many vacancies as the number of atoms in a molecule. To adsorb a hydrogen molecule's two atoms, only two vacancies must momentarily pair up. This assumption is repeated in many textbooks, and it makes sense — one vacancy for each atom — but it's wrong.

The truth didn't reveal itself until Salmeron and colleagues trained a scanning tunneling microscope onto a catalyst surface made of palladium. The microscope detects individual atoms, or the absence of atoms, using a probe that tapers down to a point only a single atom across. This atom-tipped probe skims across a surface, always maintaining the same current between it and the surface atoms. If the probe moves over an atom with an electron cloud that facilitates electronic conduction, it rises up. If it moves over an atom with an electron cloud that does not favor conduction, such as a hydrogen atom, the probe drops down. These irregularities are fed into a computer that produces an atom-by-atom contour map of the surface.

[Click for larger image]
Hydrogen atoms don't favor conduction, so they appear much smaller than vacancies under a scanning tunneling microscope.

Salmeron's team used this special microscope to observe the formation of sites that adsorb hydrogen. They cooled a palladium surface to an extremely low temperature, which slows its atomic movement, and exposed it to hydrogen molecules. They then watched what happened. Like a bird's eye view of people mish-mashing their way through a crowded train station, the vacant atomic adsorption sites skitter across the surface. When two vacancies bump together, creating the condition scientists long believed is ripe for hydrogen adsorption, nothing happens. The two vacancies merely dance together for a fraction of a second, then part ways.

"Although the pair of vacancies are constantly bombarded with hydrogen molecules, they never adsorb hydrogen atoms," Salmeron says. "This proves that two vacancies don't work."

But when three vacancies cluster together, two vacancies in the group quickly accept a hydrogen atom. This sequence of images, from a cluster of three vacancies to two adsorbed hydrogen atoms and one remaining vacancy, marks the first time the formation and function of an active site has been observed. The same holds true when the microscope captures four vacancies clustered together: hydrogen atoms soon fill two vacancies, leaving two remaining vacancies. Based on these images, the Berkeley Lab team concludes hydrogen adsorption requires active sites containing three or more vacancies. Why this is so, however, remains unanswered.

"We haven't solved a mystery. We've only unveiled a mystery," Salmeron says. "Two vacancies don't work, three or more are needed. Now it's up to the theoreticians to explain why."

Yet even as it poses questions, the team's research may enable scientists to more quickly refine catalyst systems. For example, a better understanding of the catalyst surface used in gasoline reformulation may lead to streamlined reformulation processes. In most cases, however, this pursuit of ever more efficient catalysts is guided by intuition and trial and error — a steady but slow process. Experimentally derived information, such as the three-or-more-vacancies rule, is a rare luxury that allows scientists to predict rather than guess how catalyst surfaces work, which could expedite their improvement.

"Our research helps take the guesswork out of making better catalysts," Salmeron says.

In addition to Salmeron, Toshi Mitsui, Mark Rose, Evgueni Fomin, and Frank Ogletree of the Materials Sciences Division contributed to the research.

Additional information

• "Dissociative hydrogen adsorption on palladium requires aggregates of three or more vacancies," by T. Mitsui, M. K. Rose, E. Fomin, D. F. Ogletree, and M. Salmeron, Nature, 17 April 2003