Nobel chemistry prize honors career in surface science

The Royal Swedish Academy of Sciences has awarded this year’s Nobel Prize in Chemistry to Gerhard Ertl, professor emeritus and former director of the physical chemistry department at the Fritz Haber Institute of the Max Planck Society in Berlin, Germany, “for his thorough studies of fundamental molecular processes at the gas–solid interface.” It’s the first time since 1999 that the prize has been awarded for work in physical chemistry, and the first since 1932 that it was awarded for the study of chemical reactions on surfaces.

Ertl was born in 1936 in Bad Cannstatt—now part of Stuttgart—in Germany. His early training in electro-chemistry, which deals with chemical processes at the solid–liquid interface, inspired him to start thinking about the chemistry of the solid–gas interface, a nearly unexplored field at the time. At the young age of 31, he became the director of the physical chemistry department at the University of Hanover, taking a similar position at Ludwig–Maximilians University in Munich in 1973 before moving to the Fritz Haber Institute in 1986. His long list of awards includes the 1998 Wolf Prize in Chemistry, which he shared with Gabor Somorjai of the University of California, Berkeley. ¹

A great deal of Ertl’s scientific insight involved finding the right tool—or, more usually, combination of tools—for the job. He was always on the lookout for new surface-probing methods, based on the various ways that photons and electrons interact with surface atoms; he invited experts on the techniques to come to his lab to teach him and his students how to use them. He built sophisticated ultrahigh-vacuum machines that incorporated several surface probes. And he often returned to important unresolved problems as new techniques became available.

Much of Ertl’s early surface-science work involved low-energy electron diffraction (LEED). The physics behind the technique dates back to Clinton Davisson and Lester Germer’s famous 1920s experiments, in which they found that slow-moving electrons fired at a crystalline surface were deflected in a diffraction pattern. The interest at the time was in the electron’s wave–particle duality; LEED’s use as a tool for surface characterization had to wait several more decades. The delay was due in part to the need for better vacuum technology, which was eventually developed in the semiconductor industry.

In the early 1960s, as soon as LEED systems became commercially available, Ertl applied for and obtained funding to buy one. “Before that time,” he says, “I had tried to build such an apparatus from glass in my laboratory, myself, but I was not very successful.” He recognized that the technique could be used to probe not only the structure of surfaces themselves but also the arrangements of adsorbed atoms and molecules. Since LEED was prone to error resulting from impurities on the surface, Ertl often combined it with other techniques to confirm his results. He always kept an eye toward applications to surface catalysis, the promotion of a chemical reaction by a surface that is not itself used up by the reaction. Surfaces act as catalysts by changing the structures or even breaking the bonds of adsorbed molecules. (See the articles by Ertl and Hans-Joachim Freund, Physics Today, January 1999, page 32 and by Somorjai and Jeong Park, October 2007, page 48 ).

Haber–Bosch process

Ertl’s work on adsorbate structures came to have great practical use in his characterization of the Haber–Bosch process for the synthesis of ammonia, NH₃, from nitrogen and hydrogen. Air is mostly N₂, but it’s tricky to get that nitrogen into a chemically reactive form that can be used to make explosives, fertilizer, or other products. The triple bond between two N atoms is one of the strongest of chemical bonds, and few chemical processes can break it.

Developed by Fritz Haber in 1909 and modified to a commercial scale by Carl Bosch and others in 1913, the Haber-Bosch process involves flowing heated N₂ and H₂ (the latter is easily derived from fossil methane) over a mixture of iron and other metals. The process was used for decades, and refined empirically, without anyone fully understanding how it worked. A lingering open question was whether the adsorbed N₂ molecules reacted with hydrogen directly or whether they first dissociated into two N atoms before reacting. Understanding that step was key to designing a way to make the reaction go faster.

In 1974 Ertl attended a conference in Switzerland at which Paul Emmett—a noted researcher on catalysis, and friend and brother-in-law of Linus Pauling—was honored for his 50 years of work on ammonia synthesis. Recalls Ertl, “At the end of the conference, Emmett gave a speech in which he concluded that despite the efforts over so many years, the mechanism is still not clear. That prompted me—I went back to my laboratory and said to my students, ‘Let’s try this and figure out what’s going on.’”

Nitrogen and hydrogen react at a detectable rate only at pressures that preclude the use of Ertl’s customary surface probes. But by 1981 Ertl had found a way to probe the high-pressure reaction through low-pressure surface studies. ² He noticed that at temperatures between 400 K and 600 K, adsorbed N atoms didn’t have enough energy to find partners and recombine into N₂ molecules that can desorb from the surface—but all the other molecules involved in the reaction were able to desorb. If he kept the temperature in that range, he could run the reaction at high pressure, pump the reaction chamber down to low pressure, and measure the concentration of N atoms left on the surface.

If the reactive nitrogen species were the N₂ molecules, some N₂ molecules would still dissociate before they had a chance to react. The resulting N atoms would be stuck—they would have no way of leaving the surface through reaction or desorption—so the concentration of N atoms would be large regardless of the H₂ pressure used in the reaction. But Ertl found that the N-atom concentration dropped sharply as he ran the reaction at higher and higher H₂ pressures. In a conclusion he described as “unequivocal,” he had found that hydrogen was reacting with N atoms, not N₂ molecules. In other experiments, Ertl elucidated the later steps of the mechanism (see figure 1) by looking at the reverse reaction, starting from ammonia and going backwards. ³

Carbon monoxide oxidation

One of the most fascinating systems Ertl studied is the surface-catalyzed oxidation of carbon monoxide. Like the Haber–Bosch process, CO oxidation has a practical use: CO is often the product of incomplete combustion of carbon-based fuels—in car engines, for example. It’s desirable to convert as much as possible of the poisonous CO to CO₂ by reacting it with oxygen. That reaction is one of the functions of catalytic converters, widely used in automobiles since 1975. Several metals can catalyze the reaction, but platinum exhibits some of the most interesting chemistry and physics. (CO oxidation on the surface of another metal, ruthenium, is discussed in Physics Today, October 1999, page 19 .)

In principle, the reaction itself is fairly simple. Molecules of CO and O₂ adsorb on the surface, and each O₂ molecule dissociates into two adsorbed O atoms. An adsorbed CO molecule combines with an adsorbed O atom to form a molecule of CO₂, which immediately leaves the surface. But around 1970 researchers noticed some strange behavior in platinum-catalyzed CO oxidation: The reaction rate oscillated in time. In 1982 Ertl reproduced the oscillations under idealized conditions: a clean, well-defined single-crystal surface under low pressure. ⁴ He found that the Pt(100) surface made the reaction rate oscillate, but the Pt(111) surface did not.

Ertl’s proposed explanation ⁵ for the oscillations had to do with surface reconstruction, the rearrangement of the atoms on a surface into a structure different from that of the bulk. Many surfaces reconstruct as a way of reducing the strain caused by the reduced coordination of the surface atoms—but an adsorbate covering enough of the surface can undo the reconstruction and restore the bulk structure. In the Pt(100) case, either CO or O can remove the reconstruction, but if they’re present together, they react, so in practice a bulk-structure region of the surface is either CO covered or oxygen covered. That makes a total of three structures the Pt(100) surface can adopt: the CO-covered bulk configuration, the oxygen-covered bulk configuration, and the reconstructed configuration. The Pt(111) surface does not reconstruct at all in the 300–600 K temperature range under which the reaction is carried out, which ultimately explains the lack of oscillations on that surface.

The reconstructed surface exhibits a low reactivity because the sticking coefficient for oxygen, a measure of the adsorption probability, is very low. CO can easily adsorb onto the reconstructed surface, but O₂ cannot, so CO accumulates until there is enough of it on the surface to undo the reconstruction. The CO-covered bulk-structure surface also has a low reactivity, but for a different reason. An adsorbed O₂ molecule dissociates immediately, so it requires two adjacent unoccupied sites on the surface; when CO covers so much of the surface, those empty pairs are hard to come by. Occasionally, though—perhaps by chance, or perhaps, as Ertl speculated, due to a surface defect—two adjacent sites do open up, and an O₂ molecule can adsorb. The molecule’s two O atoms quickly react with nearby CO molecules, freeing a larger cluster of sites, paving the way for even more O₂ adsorption, and eventually forming an oxygen-covered domain.

The oxygen-covered region is highly reactive because a CO molecule requires only a single isolated surface site to adsorb and quickly react. Keeping the O₂ pressure greater than the CO pressure ensures that O remains the dominant species on the surface, for a while. If, in the course of the reaction, a patch of the surface is cleared of most of its adsorbates, it reconstructs, and O₂ adsorption is inhibited once again. The reconstructed region grows as CO molecules diffuse into the surrounding oxygen-covered areas, and the cycle begins again.

Ertl realized that if he wanted to fully understand the reaction, he needed to measure the surface variations not only in time but also in space. The bulk-structure CO-rich regions could be distinguished from the oxygen-rich regions and the reconstructed regions through LEED or through the photoelectric effect. The work function of the oxygen-covered surface is greater than that of the CO-covered surface, which in turn is greater than the work function of the mostly clean, reconstructed surface. From those experiments Ertl observed the structures moving across the surface in waves (see figure 2(a)).

Pattern formation

In 1986 Ertl observed kinetic oscillations in the CO-oxidation reaction on the Pt(110) surface but found that they differed in many ways from those on the Pt(100) surface. ⁶ Because the Pt(100) structures propagated across the surface by diffusion, there wasn’t a high degree of correlation among distant parts of the surface, so the oscillations never formed quite the same shape twice. The Pt(110) oscillations, on the other hand, were regular and reproducible. The Pt(100) oscillations had periods of several minutes, but the Pt(110) oscillations took just a few seconds. The changing structures on the Pt(100) surface could be tracked by LEED, but LEED studies of the reaction on the Pt(110) surface initially got nowhere. And whereas the Pt(100) system could be made to oscillate over wide ranges of temperature, O₂ pressure, and CO pressure, a small change in any of those three parameters was enough to knock the Pt(110) system out of the oscillatory regime.

The main cause of the differing behavior is the extent to which O₂ adsorption is suppressed on the reconstructed surface. On the reconstructed Pt(100) surface, the O₂ sticking probability is so small that it can be ignored completely. But in the Pt(110) system, the O₂ sticking probabilities on the bulk-structure and reconstructed surfaces differ by less than a factor of 2. Since the likelihood of O₂ adsorption still decreases overall with increasing CO coverage, there’s a much smaller range of conditions under which the greater sticking probability on the bulk-structure surface has any effect. As a result, the Pt(110) system is much more sensitive than the Pt(100) system to the reactant pressures; that sensitivity can explain why the Pt(110) reaction rate oscillates so rapidly.

As the system cycles through the states of varying reactivity, the reactant pressures change. In the Pt(100) system, the changes in pressure (on the order of 1%) aren’t enough to noticeably affect the reaction, but in the much more sensitive Pt(110) system, they are. And since pressure waves in the gas phase propagate much more quickly than do diffusion waves of adsorbed molecules, a change in one region of the surface has an almost immediate effect on the reaction in other regions. As a result, even distant points on the surface can behave in a highly correlated way, and the reaction rate oscillates quickly, regularly, and reproducibly.

With a pressure-mediated mechanism postulated, ⁷ Ertl set out to image the Pt(110) surface during the reaction. He turned to photoemission electron microscopy (PEEM), a technique that measures the spatial and temporal variation in the surface work function. The photoelectrons from the entire surface are passed through a cathode lens to create images in real time, with much better time resolution than is possible using techniques that involve scanning the surface—just what Ertl needed to study the fast oscillations on the Pt(110) surface.

Originating in the early days of electron microscopy in the 1930s, PEEM later fell out of favor as transmission electron microscopy and scanning electron microscopy proved more suitable for most problems of interest. In the late 1980s, the Fritz Haber Institute’s Wilfried Engel developed a PEEM apparatus, optimized for studying surface reactions such as CO oxidation, that could easily be added onto Ertl’s ultrahigh-vacuum apparatus.

The results were surprising. Instead of the slow diffusion waves and irregular shapes on the Pt(100) surface, Ertl saw spirals (see figure 2(b)), bull’s-eye patterns, synchronized standing waves, and rapidly changing chaotic shapes, depending on the reaction parameters. ⁸ More recent work by Ertl and his colleagues has involved controlling the reaction in the chaotic regime by creating a feedback loop between the integrated PEEM intensity and the CO pressure ⁹ (see also Physics Today, July 2001, page 18 ).

Handling the pressure

Like many other surface techniques, PEEM works only at low pressure. But kinetic studies revealed evidence of qualitative differences in the CO reaction at low pressure and at high pressure: At low pressure, the oxygen atoms stay on the surface of the Pt crystal, but at high pressure some of them penetrate beneath the surface to form a Pt oxide. And the reaction rate at low pressure is slow enough that the heat generated by the exothermic reaction dissipates quickly enough that the temperature remains constant, but at high pressure significant temperature differences build up between regions of low and high reactivity, and those temperature differences in turn affect the reaction rate.

Imaging the surface under pressures higher than about 10⁻³ torr requires a technique that probes the surface with photons only. One suitable technique is ellipsomicroscopy, which measures spatial variations in the change in polarization of light reflected from the surface. The CO-covered and oxygen-covered surfaces change the polarization in different ways, due to their unequal dielectric properties. Harm Hinrich Rotermund, then head of the Fritz Haber Institute’s surface imaging group and now at Dalhousie University in Canada, worked on the technique, with Ertl’s guidance. Recalls Rotermund, “The day we saw our first spirals with ellipsomicroscopy he left an important visitor in the hands of his secretaries and rushed downstairs to the lab to witness our final success. His curiosity in science has always been his first priority.”