February 7,2000
Volume 78, Number 6
CENEAR 78 6 pp.41-50
ISSN 0009-2347

[Previous Story][Next Story]


Researchers hope that multidimensional IR and Raman techniques will complement the capabilities of multidimensional NMR

Stu Borman

C&EN Washington

Infrared spectroscopy and Raman spectroscopy—techniques used to determine the identity and structure of organic compounds by analyzing their molecular vibrations—normally yield one-dimensional absorption spectra plotted with respect to a changing frequency or time dimension.

For relatively simple molecules, these vibrational techniques work well, but for larger and more complex molecules they're much less useful. IR and Raman spectra of peptides and proteins, for example, generally consist of broad, overlapping absorptions whose components are largely indistinguishable and for the most part unidentifiable.

For inspiration in improving the ability of Raman and IR spectroscopy to handle larger molecules, vibrational spectroscopists can look to nuclear magnetic resonance spectroscopy, which has been revolutionized over the past quarter century by the development of multidimensional techniques. These techniques have transformed conventional NMR into a powerful tool capable of probing the dynamics of large biomolecules and determining the structure of proteins and other complex compounds at resolutions approaching those of X-ray crystallography. Essentially, multidimensional NMR takes what would otherwise be broad, overlapping NMR bands and extends them into two or more frequency or time dimensions, making it possible to view their component bands as individual discrete peaks.

Several research groups are now developing multidimensional vibrational techniques that will potentially complement multidimensional NMR methods. "In ordinary vibrational spectroscopy, you vary one parameter—say a frequency—and you plot a spectrum as a function of that frequency," explains chemistry professor Shaul Mukamel of the University of Rochester. "We call this one-dimensional spectroscopy because there is only one coordinate that you vary."

But with multidimensional spectroscopy, he says, "you come in with several light fields, and you can vary these frequencies simultaneously. You can then see the effects of one frequency on the other. These correlation plots . . . contain direct signatures of molecular structure and dynamics, in complete analogy with multidimensional NMR."

The basic idea is to use multiple IR beams to tease out some of the information hidden under those broad, featureless bands in conventional one-dimensional vibrational spectra of complex molecules. An advantage of multidimensional IR and Raman techniques is that they also can be used to study molecular dynamics, such as fast structural changes, on timescales faster than those accessible with NMR.

NMR spins are relatively easy to model, making it simple to take NMR spectra and convert them into data on molecular structure and dynamics. But vibrational absorptions are much more complex. "So the technology needs to be developed," Mukamel says. "It's not a straightforward copying from NMR." Nevertheless, "when you look at the theoretical description of NMR and vibrational spectroscopy, they're very similar in terms of the basic concepts." Therefore, researchers remain hopeful about the potential of multidimensional vibrational techniques.

"We are really on the road to developing, for vibrations, the ultrafast equivalent of pulsed NMR," says chemistry professor Michael D. Fayer of Stanford University. "This is going to be a very important field in chemistry, biology, materials science, and physics. In 10 years there will be commercial machines, like NMR machines, that will do a variety of IR pulse sequences for spectral enhancements and time-domain measurements."

Sidebar: Two-dimensional Raman spectrum shows coupling between intermolecular nuclear motions

The concept of studying vibrations using multidimensional spectroscopy is relatively new. Several studies on picosecond Raman echo spectroscopy—a technique that helped lead to current forms of multidimensional vibrational spectroscopy—were carried out in the 1980s and '90s. But a theory of multidimensional Raman spectroscopy that first pointed out the analogy with NMR was only proposed in 1993, by Mukamel and his colleague Yoshitaka Tanimura.

"Some of the basic ideas of multidimensional spectroscopy were first described theoretically, and Mukamel has really been an important stimulator in that direction," chemistry professor Robin M. Hochstrasser of the University of Pennsylvania says.

Mukamel is organizing a symposium at the American Physical Society's March Meeting 2000 in Minneapolis that will cover recent progress in multidimensional Raman and IR techniques and their connection to NMR and other techniques. "The contributions of Shaul Mukamel to the theoretical description of these types of experiments cannot be overstated," says Andrei Tokmakoff , assistant professor of chemistry at Massachusetts Institute of Technology.

Multidimensional Raman experiments have been carried out by several groups. These include professor of chemical physics Koos Duppen's group at the University of Groningen, the Netherlands; chemistry professor Graham R. Fleming of the University of California, Berkeley, and coworkers, including Tokmakoff; the collaborative team of Keisuke Tominaga of the department of chemistry at Kobe University, Japan, and materials science professor Keitaro Yoshihara of the Japan Advanced Institute of Science & Technology, Hokuriku; chemistry professor John C. Wright and coworkers at the University of Wisconsin, Madison; and professor of chemistry and chemical biology Andreas C. Albrecht and coworkers at Cornell University.

The Duppen, Fleming-Tokmakoff, and Tominaga-Yoshihara groups, for example, have all attempted to use 2-D Raman spectroscopy to study intermolecular nuclear motions and structural dynamics in liquid CS2. In what Fleming calls "a brilliant piece of detection," his postdoc David A. Blank (who will be an assistant professor of chemistry at the University of Minnesota, Minneapolis, this fall) discovered recently "that all previous work, including our own, was totally compromised by what we called third-order cascades." Fleming and coworkers believe that the spectra acquired up to that point actually contained no structural information beyond that already available in ordinary one-dimensional Raman spectra.

However, Tominaga and Yoshihara disagree that they were observing a cascading signal instead of a true 2-D signal. "This issue has not been settled yet and is still in controversy," Tominaga says. And Mukamel notes that the role of cascade processes in multidimensional IR was first discussed by Wright.

In any case, Fleming, Blank, and associate professor Minhaeng Cho of the Theoretical Physical Chemistry Lab at Korea University, Seoul, believe they clarified the theoretical relationship between the cascades and true multidimensional signals, and Fleming's group then used that information to design a new geometry for the experiment. "We have now, we believe, generated a true two-dimensional . . . Raman signal for the first time," Fleming says. "We are just about to send a paper to the Journal of Chemical Physics."

Sidebar: Learning the relationship between spectrum and structure

Possible long-range applications of multidimensional Raman spectroscopy, Fleming says, include structure determinations analogous to those currently carried out by 2-D NMR, studies of complex liquids on molecular size scales, and studies of vibrational interactions and coupling in liquids, glasses, and proteins at unprecedented levels of precision.

In the area of multidimensional IR (as opposed to Raman) spectroscopy, four groups have been particularly active. Hochstrasser's group has focused its efforts on obtaining structures of peptides and small proteins with time-domain multidimensional IR techniques, in analogy with multidimensional NMR. Fayer's group has been experimenting with time-domain vibrational echo spectroscopy to enhance spectral features. Wright and coworkers have developed a frequency-domain multidimensional IR technique for potential analytical chemistry applications. And in the theoretical area, Mukamel's group has been working on developing models and algorithms to "invert" multidimensional IR spectra to extract the underlying structural and dynamics information.

In time-domain multidimensional IR and Raman techniques, one hits a sample with short IR pulses and plots a spectrum as a function of varying time differences between the pulses. In frequency-domain studies, on the other hand, the sample is hit with continuous-wave IR frequencies instead of discrete pulses of radiation, and the frequencies are varied directly to obtain an absorption spectrum.

Time-domain and frequency-domain methods "are mathematically identical," Mukamel says. "But in practice, for certain applications one or the other is to be preferred. If you want to follow a fast process, like solvent fluctuations around a solute or the dynamics of hydrogen bonding or protein folding, or if you want to initiate a process and then follow it in real time, then you have to do it with a time-domain technique"—analogous to the way Nobel Prize-winning chemistry professor Ahmed H. Zewail of California Institute of Technology uses femtosecond spectroscopy to obtain information about the dynamics of chemical reactions. "For other applications, such as simplifying a complex spectrum by suppressing signals from selected components, frequency-domain multidimensional spectroscopy may be appropriate, or even preferred," Mukamel says.

In research by Hochstrasser's group, the objective has been to determine structure and the way structure changes with time. "The aim is not to copy or replace NMR, because I doubt very much if that's possible," Hochstrasser says. "But what is possible is to determine structure with much higher time resolution, compared with NMR."

NMR is "a wonderful technique," he says. "I don't believe that two-dimensional IR will ever be a serious competitor to two-dimensional NMR for the determination of large structures. However, to look at structural changes of peptides, protein fragments, or other small biological molecules on a faster than NMR timescale—this should be very interesting indeed. It's the timescale that's important."

Relaxation, diffusion, and other dynamic processes occurring on nanosecond timescales are already accessible by NMR, Hochstrasser says. "But in fact, generally speaking, if you have any complexity and you have structures that are interchanging on timescales much faster than milliseconds to microseconds, then the signals that tell us about coupling between spins get averaged out in NMR."

Time-domain multidimensional IR experiments, on the other hand, have an intrinsic time resolution in the picosecond range. "This means if we look, for example, at a peptide in water and the peptide has a variety of conformations, some of which or all of which exist on longer than picosecond timescales, then we can see all these conformations, instead of just the average conformation" that one would see with multidimensional NMR, Hochstrasser notes. "So you learn about the nature of the energy surface of the peptide, and one could only obtain this kind of information previously by doing theoretical calculations."

Sidebar: DOVE-FWM has analytical applications

The researchers began their efforts in this area two years ago when Hochstrasser and coworkers Peter Hamm and Manho Lim developed a multidimensional IR version of an NMR experiment called double resonance, a technique in which two resonances (vibrations in the IR case) are excited consecutively [ J. Phys. Chem. B, 102, 6123 (1998) ]. "We particularly focused on the amide vibrators," says Hochstrasser. "Any protein has a lot of amide IR transitions, and they're all coupled to one another in a manner that depends on the structure of the protein. The idea was to be able to determine the magnitudes of the couplings in these networks and then to work back to try to get a structure."

The group attempted to use the technique to determine the structure of a model pentapeptide synthesized by biochemistry and biophysics professor William F. DeGrado of the University of Pennsylvania School of Medicine. The peptide's structure was well known by NMR and by X-ray diffraction. Hochstrasser and coworkers did not succeed in calculating the peptide's structure from the 2-D IR spectrum they obtained. But they did have some success working in the opposite direction. Using nonlinear optical theory, they calculated from first principles what the 2-D IR spectrum should have looked like based on the known NMR and X-ray structure data, and they achieved a decent match between the resulting theoretical spectrum and the one they had obtained experimentally.

In a recent study, Hochstrasser and coworkers also developed a multidimensional IR technique analogous to COSY (correlative spectroscopy), an NMR technique that detects coupling interactions between different atoms in a molecule [J. Chem. Phys., 112, 1907 (2000)]. In the multidimensional IR version of COSY, IR cross-peaks represent interactions between different vibrational modes in a molecule. Hochstrasser and coworkers use isotopic substitution to modify the vibrational coupling between various parts of a molecule as a means of obtaining structural information about it.

Potentially, one can work back from such 2-D IR spectra to determine a structure. That hasn't been done yet, "but that is what my objective is," Hochstrasser says, "and perhaps we will be able to do that before the end of the year." In comparing experimental results with calculations in the studies carried out by his group so far, he says, "The agreement has really been very good when you look at it in detail—which is one of the reasons why we feel that this whole methodology is so promising."

Studies by Cho and coworkers suggest that 2-D IR and 2-D Raman spectroscopy also can be used to obtain structural information similar to that provided by NOESY—nuclear Overhauser effect NMR spectroscopy, a technique that plays an essential role in NMR-based protein structure determinations. These 2-D vibrational techniques "can be directly used to extract crucial information about the through-space connectivity of two molecules or two subgroups in a protein," Cho says.

There's significant promise "for being able to go further with these technologies as more people get involved with them and more methodologies are invented," Hochstrasser says. "These are still very early days, but the field is very exciting because it really seems that there's a very good possibility of being able to look at the dynamics of structural change on timescales that are not accessible to NMR."

In a notable example of the development of new 2-D vibrational methodologies, Fayer and coworkers have done "a lot of beautiful work using IR echo spectroscopy that contributes to a knowledge of vibrational coherence," Hochstrasser says. "This is basic research on the coherent interaction of vibrators with IR pulses, and it's very important." A coherent technique is one in which a series of pulses or waves bear a fixed phase relationship to each other.

Graduate student Kent Meyer (left) and Wright discuss DOVE-FWM instrument. [Photo by Jerrold J. Jacobsen ]

Fayer explains that he and his coworkers use "a type of 2-D IR spectroscopy and theory that is on the road to NMR-type tricks." With vibrational echo spectroscopy, he says, "we can suppress the strong background in an IR spectrum and emphasize one peak over another."

Vibrational echo spectroscopy is the multidimensional IR analog of still another NMR technique—spin echo spectroscopy. In the 2-D vibrational version of the technique, a sequence of IR pulses causes the sample to generate an additional pulse of IR radiation, the echo. This is the ultrafast IR equivalent of the NMR spin echo, in which a sequence of radiofrequency pulses causes the sample to produce an additional radiofrequency pulse, the spin echo.

"We did the first experiments that used an IR coherent pulse sequence to bring out spectral features," Fayer says, when the group used vibrational echo spectroscopy to obtain the vibrational spectrum of the CO ligand in myoglobin-CO [J. Chem. Phys., 109, 5455 (1998)]. The technique made it possible to isolate the CO spectrum by discriminating against background vibrational signals arising from protein and solvent absorptions.

Wright's group, like Fayer's team, is using 2-D IR to greatly simplify vibrational spectra. But Wright and coworkers use frequency-domain multidimensional IR techniques instead of the time-domain multidimensional methods used in the Hochstrasser and Fayer studies, and Wright's group is focusing on analytical chemistry applications. An important advantage of working in the frequency domain (instead of the time domain) is that "you can access a broader range of frequencies than you can from the femtosecond bandwidth of a pulse," Hochstrasser says, "but in principle the interactions with the system are the same."

Wright and postdoc Wei Zhao (now assistant professor of chemistry at the University of Arkansas, Little Rock) have developed a frequency-domain multidimensional technique that enhances spectroscopic selectivity for one species in a mixture, eliminates interfering absorptions, and emphasizes cross-peaks that probe intra- and intermolecular interactions [ J. Am. Chem. Soc., 121, 10994 (1999) ; Phys. Rev. Lett., 84, 1411 (2000)]. "They have demonstrated very beautifully that they can resolve transitions that are unresolved with one-dimensional techniques," Mukamel says. "It's a big step."

Wright says the new method, called doubly vibrationally enhanced four-wave mixing (DOVE-FWM), is the culmination of work that started in his group in 1978. The research was carried out with funding assistance from the National Science Foundation's Division of Chemistry.

In DOVE-FWM, three lasers—two tunable IR lasers and a fixed-frequency visible laser—are used to acquire a 2-D vibrational spectrum. "The three lasers are focused into a sample, and nonlinear interactions result in the production of new frequencies," Wright explains. "The intensities of these new frequencies are monitored as a function of the two IR frequencies, and the intensity is enhanced when the laser frequencies match vibrational resonances in the sample."

The key idea, Wright says, "is that DOVE methods create vibrational coherences. These are analogous to the coherences that you find in NMR methods like COSY, and they therefore provide a basis for a similar type of 2-D spectroscopy."

In their JACS paper, Wright and Zhao demonstrated the technique's spectral selectivity capabilities on acetonitrile mixtures. In a seven-component mixture, acetonitrile's absorption bands were buried under interfering absorptions from other components. But in a DOVE-FWM spectrum of the mixture, a single acetonitrile absorption—the compound's 3,164 cm-1 combination band—was all that appeared in the spectrum.

In addition to its ability to selectively enhance individual components in sample mixtures, DOVE-FWM is capable of narrowing spectral features, thereby increasing spectral resolution. "We believe that this approach will be important for analytical, biochemical, and environmental measurements because the method is selective to modes that are associated with intra- and intermolecular interactions," Wright says.

In general, multidimensional IR techniques are easier to apply than multidimensional Raman techniques "because they involve fewer pulses," Hochstrasser says. But in principle, multidimensional IR and Raman methods "should be complementary once the methodologies are worked out," he says.

Tokmakoff notes that multidimensional Raman and IR experiments are best at addressing very different types of chemical problems. "Technology limits the practical use of the IR experiments to higher frequency vibrations," he says, whereas "the Raman methods are excellent for lower frequency modes—although technology does not limit them from use at any vibrational frequency. IR methods are best for studying dilute solutes and therefore are the method of choice for looking at proteins and peptides. Raman methods currently are being used to study neat liquids to understand collective molecular structure and dynamics."

And according to Duppen, "The intrinsic time resolution of 2-D Raman is, in principle, even higher than that of the 2-D IR methods," making 2-D Raman much more useful than 2-D IR for following molecular motions in real time.

However, researchers can also use 2-D IR/Raman spectroscopy—a combination of the two methods that enables them to take advantage of the best aspects of both. "You want to be able to tailor the experiment to your problem," Tokmakoff says. "If you realize that 2-D vibrational spectroscopy can be done either in the time or frequency domain and with either IR or Raman, you have an enormous number of potential 2-D experiments to use for solving a wide variety of chemical problems."

[Previous Story][Next Story]

Chemical & Engineering News
Copyright © 2000 American Chemical Society