MITCH JACOBY, C&EN CHICAGO

	RAMAN AFICIONADOS Wachs (left) and Stair, who use Raman spectroscopy in their work, organized a symposium devoted to its catalysis applications. PHOTO BY MITCH JACOBY

"In recent years, Raman spectroscopy in catalysis has taken off exponentially," observed Israel E. Wachs, a professor of chemical engineering at Lehigh University. One of the reasons for its popularity is that the method provides molecular-level information about catalysis, he explained. Another reason is that the technique can be used to probe catalytic systems while reactions are occurring. Such capabilities have long been sought in the catalysis field, Wachs remarked.

Other researchers share similar views. "People have realized that they need to study catalysts under operating conditions if they really want to understand how catalysts work," commented Peter C. Stair, a chemistry professor at Northwestern University. Raman spectroscopy has become one of the analytical methods of choice because it provides information about molecular species present on a catalyst and also gives structural information about the catalyst itself, Stair explained. "It's about the only technique that can provide--in a single experiment--information about the molecules undergoing reaction and the catalyst under reaction conditions," he stressed.

Given the ringing endorsements, it's no surprise that the popularity of Raman spectroscopy has surged since the early 1970s. Indeed, at a symposium entitled "Three Decades of Raman Spectroscopy of Heterogeneous Catalysts" at last month's American Chemical Society national meeting in New York City, Wachs presented a histogram showing that the number of Raman-based catalysis studies published annually mushroomed from a handful in the 1970s to nearly 300 in each of the past few years.

In addition to instrumentation improvements that have made Raman spectrometers more sensitive, quicker, and easier to use than ever before, newly developed Raman techniques have also played a part in expanding the method's role in catalysis research. As Stair noted, by bringing together Raman spectroscopists who use the technique in a variety of ways, the symposium provided "a snapshot that shows the current state of progress in the field."

Besides drawing attention to the widening use of this vibrational technique in heterogeneous catalysis, the symposium also marked the 75th anniversary of the discovery of the scattering effect, for which Chandrasekhara Venkata Raman won the Nobel Prize in Physics in 1930. The symposium was organized by Wachs and Stair and was cosponsored by the Division of Colloid & Surface Chemistry and by Jobin Yvon, a company based in Villeneuve d'Ascq, France, that manufactures Raman spectrometers and other instruments.

MANY INDUSTRIAL catalytic processes are based on chemical reactions that occur in the interior of zeolites, a family of porous aluminosilicate-type compounds. Yining Huang, a chemistry professor at the University of Western Ontario, London, Ontario, reported on studies of interactions between porous hosts and various types of molecules contained within the pores and interior channels of support materials. Huang noted that Fourier transform (FT) Raman spectroscopy is well suited to studying the behavior of guest molecules in zeolites because zeolites tend to produce weak Raman signals, which simplifies the job of detecting molecules adsorbed inside of them. In addition, by exciting vibrational transitions with near-IR laser light (1,064 nm), the FT Raman method reduces the fluorescence interference typically associated with zeolites.

Drawing on examples of 1,4-dichlorocyclohexane and other dichlorinated compounds, Huang showed that the vibrational method can be used to distinguish between various conformational isomers of adsorbed molecules even at room temperature. That ability is noteworthy, he explained, "because molecular conformation affects adsorption, diffusion, and the subsequent chemistry of adsorbed molecules."

The FT Raman method is also useful for interrogating chemical reactions and other types of processes that take place in zeolites. For example, Huang showed that the technique can be used to probe dimerization of Mn(CO)₅Br by detecting metal-metal bond formation. It also can distinguish between the three types of sites at which benzene adsorbs in zeolite ZSM-5.

	MAKING COKE Various types of carbon deposits (coke) on catalysts can be classified using UV Raman spectroscopy. Molecules such as pentacene and anthracene that have chainlike structures can be distinguished from those with sheetlike structures, such as coronene and pyrene.

SOME RESEARCHERS bend over backward to make sure their specimens are pristine and free of contaminants. But a little dirt doesn't bother Northwestern's Stair. "We want to look at systems that aren't clean, that have hydrocarbons and water and all the junk that we often have to live with when we work with catalysts," he announced. The reason that impurities don't bother Stair is that his research group probes catalysts using Raman spectroscopy in a somewhat unconventional manner.

As Huang noted, interference from fluorescence is a common problem nagging researchers who use Raman scattering to study catalysts. Trace quantities of metals in zeolites left over from synthesis and sample preparation techniques can lead to sample fluorescence that buries the Raman signals. The same is true of carbon layers (coke) that accumulate and adhere to catalyst surfaces following hydrocarbon processing. But several years ago, Stair and Can Li, who currently is the director of the State Key Laboratory of Catalysis at the Dalian Institute of Chemical Physics, in China, used UV laser light--instead of light in the visible range--to excite Raman transitions. They discovered that not only does the fluorescence interference disappear, but the method yields sharp and intense spectra of coke.

"If you try to measure Raman spectra in the region where fluorescence occurs, then the inherently weak Raman signal is overwhelmed by fluorescence," Stair explained. "But when the excitation is done with UV light, then the Raman spectra span a narrow wavelength range that does not overlap with the fluorescence signals."

Underscoring that point, Stair presented data showing that when a "gunked-up" deactivated rhodium catalyst was probed with green light (514 nm), the Raman spectrum was featureless. But when the very same spent catalyst was excited with 257-nm UV light, the resulting spectrum exhibited sharp bands that arise from ring-breathing motions of polynuclear aromatic compounds in the coke and other features that correspond to C2H bending motions.

With the ability to examine coked catalysts, the UV Raman method could pave the way to understanding mechanisms of coke formation. Such studies could lead to more durable catalysts that resist deactivation from coke buildup or to more efficient catalytic processes in which some coke is required to initiate hydrocarbon conversion reactions.

But the method isn't without its disadvantages, Stair acknowledged. The relatively high energy UV light causes undesirable sample heating and photochemistry, which can destroy molecules adsorbed on the catalyst. Although UV-mediated decomposition processes persist even when the laser output is attenuated, Stair and graduate student Yek Tann Chua figured out how to sidestep the problem by devising a type of chemical reactor that churns and agitates a powdered catalyst such that fresh material is continuously delivered to the tiny spot probed by the laser. "We're still destroying the sample photochemically," Stair noted, but the effect is diluted by the catalyst motion, and the Raman signal is derived primarily from fresh catalyst.

Armed with the UV Raman method, Stair and coworkers examined the nature of coke formed during methanol-to-hydrocarbons conversion in the acidic form of zeolite ZSM-5 (also known as H-MFI). The process, which has been studied widely because of its significance to petrochemistry, was developed some 25 years ago by scientists at Mobil for converting methanol to olefins and other hydrocarbons for gasoline formulations.

The Northwestern group ran the reaction at elevated pressure and at various temperatures in such a way that volatile molecules exit from the zeolite cavities, leaving behind just the heavy hydrocarbons, which ultimately form coke. Stair reported that under those conditions his group detected polyolefins, cyclopentadienyl species, and polyaromatic hydrocarbons [J. Catal., 213, 39 (2003)]. Those observations support an acid-catalyzed reaction mechanism--proposed previously by scientists at the University of Karslruhe, in Germany--in which cyclopentadienyl species serve as key intermediates in the formation of polyaromatic hydrocarbons.

Just recently, the group also detected cyclopentenyl carbenium ions, which have been fingered as key intermediates in methanol-to-olefins chemistry on the basis of nuclear magnetic resonance spectroscopy studies conducted by James F. Haw, a chemistry professor at the University of Southern California.

SHOWING A SERIES of reference UV Raman spectra, Stair proposed that in addition to distinguishing between polyolefin, polyaromatic, and graphitic forms of coke, the technique can be used to further classify polyaromatic types of coke. Specifically, by comparing the intensities of two vibrational band regions (near 1,400 and 1,600 cm^-1), Stair showed that in compounds such as phenanthrene and pentacene that are characterized by 1-D chainlike structures, the bands have nearly equal intensity. In contrast, compounds with 2-D sheetlike structures, such as coronene and pyrene, exhibit very unequal band intensities. On that basis, Stair concluded that the coke formed in typical methanol-to-hydrocarbons chemistry in zeolites consists primarily of 1-D polyaromatics.

Only a fraction of the vibrational bands observed in the UV Raman experiments have been assigned so far, Stair pointed out. "Part of the job now is to look at enough model compounds so that we can begin to make connections between all of the spectral features we see and the nature of the hydrocarbons left on the catalyst."

ONE WAY TO BUMP UP ordinarily weak Raman signals is to take advantage of the surface-enhanced Raman effect. Years ago, researchers discovered that when Raman scattering occurs from molecules adsorbed on roughened metal surfaces, the signal can be enhanced by a factor of roughly 1 million relative to the scattering caused by the molecules in the gas or liquid phases. Christopher T. Williams, an assistant professor of chemical engineering at the University of South Carolina, Columbia, uses surface-enhanced Raman spectroscopy (SERS) to study catalytic systems.

Stereoselective chemical reactions generally fall in the domain of solution-phase (homogeneous) catalysis. But in a few cases, such as the hydrogenation of -ketoesters studied by Williams, surface-mediated reactions can proceed with a high degree of enantioselectivity (C&EN, March 25, 2002, page 43). In the best known example of that type of reaction, ethyl pyruvate is converted to ethyl lactate over an alumina-supported platinum catalyst that's treated with cinchonidine, a chiral alkaloid. Cinchonidine functions as a chiral template (also referred to as a modifier) that steers the reaction toward (R)-ethyl lactate with an enantiomeric excess of roughly 95%.

Williams noted that several observations regarding that stereoselective reaction have been made from kinetic studies. For example, it's known that enantiopure products are formed with the greatest selectivity when cinchonidine is present in a narrow concentration range (roughly one molecule per 10 platinum atoms). It's also known that the modifier itself is hydrogenated during reaction and that activity and enantioselectivity drop off quickly when the reaction is carried out above roughly 50 °C. But the reasons for the observed behavior aren't well understood, he said.

To gain insight into the reaction mechanism, Williams, postdoctoral associate Wei Chu, and graduate student Rene J. LeBlanc prepared test specimens and probed their vibrational transitions using SERS. The samples were made by electrodepositing thin films of platinum onto roughened gold electrodes and then submerging the electrodes in ethanol solutions of cinchonidine.

From the Raman spectra, which were interpreted with the help of quantum mechanical calculations, the South Carolina group found that cinchonidine adsorbs on the catalyst surface through -bonding interactions between its quinoline unit (fused-ring portion) and platinum [Catal. Commun., 3, 547 (2002)].

The spectra also reveal the effect of cinchonidine concentration on the molecule's surface orientation. By comparing the relative intensities of spectral bands corresponding to an in-plane quinoline C-C stretching motion with an out-of-plane C-H wagging motion, Williams and coworkers deduced that as the modifier concentration is increased, the tilt angle between the quinoline ring and the surface also increases--meaning the molecule "stands up" on the surface. These findings, which are consistent with surface studies conducted by other researchers, indicate that as the surface gets crowded, cinchonidine deviates from its optimal orientation, which lowers enantioselectivity.

In related experiments, the South Carolina team recorded dramatic increases in the cinchonidine SERS signal and a change in the modifier's surface orientation when hydrogen was bubbled into the solution. Williams explained that under those conditions, the vinyl function on the quinuclidine moiety is hydrogenated, thereby converting the modifier to dihydrocinchonidine. He added that the reaction causes the quinoline ring of the adsorbed molecule to adopt a more parallel orientation to the surface, which, in turn, enhances the Raman signal.

Switching to the effects of temperature, Williams reported that raising the temperature above 50 °C causes the modifier to desorb from platinum and alters its surface orientation. Given cinchonidine's finicky reaction preferences, it's not surprising therefore that enantioselectivity falls when conditions get too hot.

	HANDY MODIFIER Cinchonidine functions as a chiral modifier that steers reactions toward high enantioselectivity.

ONE AREA OF CURRENT concentrated effort in catalysis research focuses on devising experimental methods that can probe catalytic reactions in multiple ways simultaneously as reactions occur under conventional reaction conditions. In some of the newest experimental setups, for example, Raman spectroscopy and other spectroscopy methods probe catalyst surfaces rapidly and simultaneously, while other analytical techniques are used to monitor gas-phase reaction products. By probing catalysts in action, the newly named "operando" (Latin for working) spectroscopy approach aims to uncover structure-activity relationships and other pertinent information that can lead to improved catalysts and catalytic processes.

T. A. (Xander) Nijhuis, a postdoc at Utrecht University, in the Netherlands, who works with chemistry professor Bert M. Weckhuysen, demonstrated the effectiveness of the operando approach by presenting results of recent time-resolved experiments in which Raman spectroscopy, UV-Vis spectroscopy, gas chromatography (GC), and mass spectrometry were used simultaneously to study propane dehydrogenation to propene over alumina-supported chromium oxide catalysts. Catalytic processes of that type are used to produce millions of tons of olefins annually.

Nijhuis explained that in industrial plants, propane dehydrogenation causes the catalyst to undergo reduction. At the same time, coke accumulates on the catalyst and deactivates it. The reduced catalyst then becomes oxidized and reactivated by burning the coke in air. The process cycles every 20 to 30 minutes.

Using the combination of time-resolved techniques, the Utrecht group was able to correlate the catalyst activity (based on product composition) with the catalyst oxidation state (from UV-Vis spectroscopy) during dehydrogenation and regeneration phases of the catalytic cycle. In addition, using the Raman spectroscopy data, they could monitor coke accumulation on the catalyst and the corresponding disappearance of Cr₂O₃ spectral features as the coke layer occluded the catalyst surface [Phys. Chem. Chem. Phys, 5, 436 (2003)].

With their ability to continuously monitor the amount of coke on the catalyst and relate the coke coverage to the catalyst's performance, researchers will be able to adjust the cycle times dynamically to optimize the performance of the system, Nijhuis asserted. He added that, in an industrial setting, information of that type could be used to tweak the process parameters to maximize the product output and extend the life of the catalyst.

AT THE INSTITUTE of Catalysis & Petrochemistry at the Spanish Council for Scientific Research, in Madrid, Miguel A. Bañares also takes advantage of the operando approach--a term that he coined--to relate information about a catalyst's changing surface structure to its activity and selectivity. In Bañares' experimental setup, Raman spectra and GC data are measured simultaneously.

In one study, Bañares examined propane ammoxidation to acrylonitrile using alumina-supported Sb-V-O catalysts. Based on analysis of Raman spectral features, he reported that oxides of antimony and vanadium interact on the catalyst surface, forming SbVO₄ and Sb₂O₄. The data indicate that both phases of the catalyst are directly involved in the reaction's rate-determining step, he noted. In contrast, vanadia species on alumina are not involved in the ammoxidation reaction [Chem. Commun., 2002, 1292].

In related work carried out with chemical engineers at Delft University of Technology, in the Netherlands, Bañares investigated oxidative dehydrogenation of propane on vanadium oxide catalysts supported on alumina. Using Raman spectroscopy and GC in tandem, Bañares observed that under steady-state conditions, the catalyst remains in an oxidized state that's free of carbonaceous deposits. But in the absence of oxygen, propane reacts with the catalyst to form two types of cokelike deposits: a graphitic form and a more aliphatic form.

In a symposium presentation that was a little off the beaten path, Miodrag (Mickey) Micic, an applications scientist at Veeco Instruments, Santa Barbara, Calif., reported on simulation studies aimed at finding optimal conditions for Raman imaging on extremely fine length scales. The ultimate goal of the research effort is to develop a type of scanning probe microscope that can measure surface-enhanced Raman signals and generate chemically specific topographic maps with low-nanoscale resolution. The study, which Micic conducted while serving as a postdoc with H. Peter Lu at Pacific Northwest National Laboratory, focused on so-called apertureless near-field scanning optical microscopy.

One of the key findings reported by Micic is that not only is tip sharpness important to achieving the desired resolution, so is the ability to maintain control of the tip-to-sample (vertical or z) distance in the subnanometer range--a tall order, but not impossible to achieve, he said. Fine control is required because changes in the z-distance cause extremely strong fluctuations in signal strength. The simulations also indicated that under the proper conditions, a single metal nanoparticle buried under a lipid layer can be imaged. That result suggests that it should be possible to label membrane proteins with metal nanoparticles and selectively image them [J. Phys. Chem. B, 107, 1574 (2003)].

ALTHOUGH BOATLOADS of research papers and a few books have been written in the past few years on investigations of heterogeneous catalysis via Raman spectroscopy, University of Warwick physicists D. P. Woodruff and T. A. Delchar chose not to include the topic in their classic text "Modern Techniques of Surface Science." In the book, the authors commented that "Raman spectroscopy as a tool for studying surfaces and molecule-surface combinations is likely to become important if the enhancement effect can be extended to a wider range of examples." They also wrote that due to the "limited development" of the field so far (as of the 1980s and 1990s when the book was written and revised), it didn't seem appropriate to discuss the subject in detail.

Judging by the flurry of Raman-based catalysis research projects going on today, perhaps the technique's time has come at last.

Chemical & Engineering News
Copyright © 2003 American Chemical Society