Surface-Enhanced Raman Effect

6XUIDFHHQKDQFHG5DPDQHIIHFW 7KRPDVYRQ)RHUVWHU &LWDWLRQ3K\V7RGD\  GRL 9LHZRQOLQHKWWSG[GRLRUJ 9LHZ7DEOHRI&RQWHQWVKWWSZZZSK\VLFVWRGD\RUJUHVRXUFH3+72$'YL 3XEOLVKHGE\WKH$,33XEOLVKLQJ//&

$GGLWLRQDOUHVRXUFHVIRU3K\VLFV7RGD\ +RPHSDJHKWWSZZZSK\VLFVWRGD\RUJ ,QIRUPDWLRQKWWSZZZSK\VLFVWRGD\RUJDERXWBXV 'DLO\(GLWLRQKWWSZZZSK\VLFVWRGD\RUJGDLO\BHGLWLRQ

Downloaded 04 Oct 2013 to 129.105.14.67. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms

ing the derivative of impedance: They applied an audio-frequency signal to a ring at the helium surface to vary the electron density and synchronously detected the impedance of this modulation frequency. Dips in this variation in impedance at four frequencies indicate four resonances, all of which disappear above a temperature of 0.457 K when the density is 4.4 X 108 cm" 2 . These resonances are strong evidence for the presence of a crystalline phase below this temperature. In an analysis of the foregoing experiment, Daniel Fisher and Philip M. Platzman (Bell Labs) and Halperin 4 calculated the positions of the resonances one would expect from coupled plasmon-ripplon modes excited by a triangular crystalline lattice, 0.51 microns on a side. Their analysis agrees well with the measured positions and observed behavior of the resonances for several densities and temperatures. The Grimes-Adams experiment relates indirectly to the theory of melting by Kosterlitz and Thouless. The electron crystal melts when the ratio T (inversely proportional to temperature) becomes less than a critical value determined in the experiment to be 137 ± 15. A Monte Carlo calculation5 by Robert Gann, Sudip Chakravarty and Geoffrey Chester (Cornell University) places this value in the range 110 to 140. Based on the Kosterlitz-Thouless theory, Thouless made an estimate that was too low because he used a value for the shear modulus at zero degrees. Rudolf Morf of Harvard recently simulated 6 by computer the temperature behavior of the shear modulus and arrived at a corrected value for T consistent with the Grimes-Adams experiment. Morf also showed that a more detailed version of the Kosterlitz-Thouless ideas due to Halperin, Nelson and Young gave a temperature-dependent shear modulus that agreed with his simulation. Other activity. Two-dimensional electron systems can also be studied in metal-oxide-semiconductor devices, with the important differences being that the electron densities are much higher and quantum-mechanical effects become important. Several MOS experiments have demonstrated unusual behavior, and a recent Bell Labs experiment has offered perhaps the clearest evidence to date for collective behavior of electrons below a certain temperature. (Because electron systems in MOS devices are more complicated than those of electrons on helium, one cannot claim the direct observation of a solid phase.) Barbara A. Wilson, S. James Allen Jr and Daniel C. Tsui of Bell Labs saw" a narrowing and shifting of the position of the cyclotron resonance at low enough electron densities (less than 10 11 per cm 2 ) and low enough temperatures (below 4.2 deg K) that they are in the extreme quantum limit, with only the lowest Landau level populated. The results may 18

require the assumption of some type of collective electron behavior such as charge density waves. Other experimenters are examining a wide variety of types of two-dimensional structures. They include gases weakly adsorbed or condensed on surfaces such as graphite, liquid crystals, various intercalated compounds, soap bubbles and lipid films on water. Evidence for ordering can come from any behavior sensitive to coherence among neighbors, such as x-ray or neutron scattering. The observation of the transition from an ordered to disordered state is difficult because the transition is expected to be a very smooth second-order phase transition. Thus most thermodynamic parameters are expected to undergo only subtle changes during the transition. A final class of experiments consists of computer calculations, based on either molecular dynamics or on Monte Carlo techniques. Such experiments have some interesting ideas to test, such as those of Halperin and Nelson. These theorists have predicted that the melting described by Kosterlitz and Thouless proceeds through an intermediate phase. The low-temperature phase is one in which there is both long-range order of bond orientation and quasi long-range translational order, that is, correlations that decay as an inverse fractional power of the distance. The intermediate phase, which Halperin

and Nelson call the "hexatic phase," is characterized by quasi long-range bond order but only short-range (exponential decay) translational order. In the liquid, the long-range orientational order disappears completely, due to dissociation of disclination pairs. (In a triangular solid, a disclination would be a lattice site with five or seven nearest neighbors in this otherwise six-fold coordinate structure.) Such ideas are still being investigated. Some open questions concern the effects of a substrate and of thickness on the system's "two-dimensional" behavior. The difficulty of such theoretical questions and the corresponding difficulty of experiments promise to keep this field a challenging one. BGL References 1. P. C. Hohenberg, Phys. Rev. 158,383 (1967). N. D. Mermin, H. Wagner, Phys. Rev. Lett. 17,1133(1966). 2. B. I. Halperin, D. R. Nelson, Phys. Rev. Lett. 41, 121, 519E (1978). A. P. Young, Phys. Rev. B19,1855 (1979). 3. C. C. Grimes, G. Adams, Phys. Rev. Lett. 42, 795(1979). 4. D. S. Fisher, B. I. Halperin, P. M. Platzman, Phys. Rev. Lett. 42, 798 (1979). 5. R. C. Gann, S. Chakravarty, G. V. Chester, Phys. Rev. B20, 326(1979). 6. R. H. Morf, Phys. Rev. Lett, 43, 931 (1979). 7. B. A. Wilson, S. J. Allen Jr, D. C. Tsui, Phys. Rev. Lett. 44, 479 (1980).

Surface-enhanced Raman effect A recently discovered and as yet incompletely understood effect promises to become an excellent tool for investigating molecules adsorbed onto metal surfaces. The effect consists of a spectacular enhancement—by factors of up to around 10B—of Raman scattering by monolayers of molecules adsorbed onto microscopically rough metal surfaces (rough on a scale of 500-1000 A). One of the most exciting prospects is that the effect will become a useful analytical tool for studying catalysis and other processes that take place on surfaces. As Elias Burstein, one of the early investigators in the field, put it, we are just learning how to put microscopic amplifiers onto metal surfaces. The Raman effect is, effectively, the inelastic scattering of photons (usually in the visible range) by molecules: Part of the incident energy is ultimately converted to a molecular excitation (vibration, for example); the remainder usually leaves as a photon with a reduced frequency. (This photon is "Stokes" radiation; upward shifts of the frequency are also possible—"anti-Stokes" radiation—but rarely seen in these experiments.) The spectrum of Raman frequency shifts is characteristic of the

molecule and its surroundings. Until recently it was believed that the only way to increase the intensity of the Raman lines is to increase the number of scattering molecules or the intensity of the incident light; in some cases, a resonant interaction between adsorbed molecules and their substrate could also lead to enhancements. Raman studies are typically performed with laser beams focused to 5 X 10"3 cm2; for ordinary Raman scattering one usually has something like 1015 scattering molecules in the beam cross section. A monolayer on a smooth surface contains on the order of 1012 molecules in that area, so that the scattered radiation would not be intense enough to be seen. Consequently it was assumed that it would be impossible to detect the Raman lines in scattering by atomic monolayers. But in 1974 Martin Fleischmann and his colleagues at the University of Southampton, England reported1 observing Raman lines from pyridine molecules adsorbed (from solution) onto a silver electrode that had been roughened electrochemically. At first it was not clear that this was something special; investigators supposed that the surface area had increased because of the roughening, so that an ad-

PHYSICS TODAY / APRIL 1980

oaded 04 Oct 2013 to 129.105.14.67. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us

Surface for enhanced Raman scattering. This silver surface, prepared electrochemically by John Bergmann, produces an enhancement of roughly 5 X 104. The lumps are about 0.1 micron across; the area shown is about 2.7 X 1.6 microns. (Scanning electron micrograph from Bell Labs.)

sorbed monolayer puts many more molecules into the beam. In early 1976 Richard Van Duyne and David Jeanmaire at Northwestern University observed the effect2 and recognized that the enhancement by a factor of 106 that they saw could not be explained merely by the increased surface area: The enhancement is a new effect. (Because it was so puzzling a phenomenon, the paper reporting it was delayed for more than a year, as referees asked for confirmation and clarification.) Independently M. G. Albrecht and J. A. Creighton (University of Kent, Canterbury, England) made the same observation3 in early 1977. Many other groups all over the world have since become involved in these investigations. The field has even merited sessions at recent conferences, such as the Second USA-USSR Symposium on Light-Scattering in Solids in New York last May, and the Midwinter Solid-State Research Conference in Laguna Beach in January. There are a few review papers, such as those by Van Duyne,4 by Elias Burstein, C. Y. Chen and S. Lundquist,5 and by Thomas Furtak and Jesus ReyesCorona;6 but, as Van Duyne said, review articles are hard to write in a field changing as quickly as this one. Other systems. Soon after the initial observations, other workers observed similar enhancements for cyanide ions, carbon monoxide and more than seventy other molecules, organic and inorganic. At first all experiments were carried out in the same general way as the initial observations: One prepares silver electrodes electrochemically and than lets the molecules be adsorbed onto the electrode from a solution. To observe the effect one scatters light off the electrode while it is in the solution. Several groups have observed enhanced scattering by monolayers on other metals,

chiefly copper and gold. James Tsang and John Kirtley at the IBM Watson Research Center, and Burstein and his colleagues and students at the University of Pennsylvania made the first of these observations; subsequently the effects were seen by Furtak and his group at the Ames Lab at Iowa State, Van Duyne's group at Northwestern and Bruno Pettinger at the Fritz Haber Institute in Berlin. While the majority of the early work was done with immersed electrodes, prepared electrochemically, other systems were also being investigated. Albrecht and Creighton, for example, used colloidal solutions. Burstein's group evaporated silver on top of molecular films deposited on glass; in these circumstances the silver formed islands. Tsang, Kirtley and their collaborators deposited first a molecular monolayer and then silver onto the oxidized surface of aluminum films. Because other interactions—with the solvent, for example—may complicate the picture, the wet electrochemical systems are not very "clean" from a physicist's point of view, Miles Klein of the University of Illinois (Urbana) told us. Klein, together with his student Thomas Wood (now at Bell Labs, Holmdel), therefore have been studying the effect in uacuo. Since then, Jack Rowe, Charles Shank, Cherry Murray and Dirk Zwemer at Bell Labs (Murray Hill), Richard Colton, James Murday and R. R. Smardzewski at the Naval Research Lab, Eugene Bradley and his collaborators at the University of Kentucky, as well as other groups have also been using ultra-high vacuum systems to study the enhancement. Results. The exact nature of the effect is not yet clear. Van Duyne pointed out that for a while there was no consensus on anything—even the existence of the effect. But with the accumulation of experimental results, investigators have

arrived at some points of consensus: • The effect exists; it appears for many different kinds of molecules. • The roughness and the scale of roughness of the surface are important. • The strength of the effect depends on the optical properties of the substrate. • The enhanced Raman lines are accompanied by a broad-spectrum background. No one has yet seen the effect on a smooth surface, Klein told us. In fact, Klein said, Wood and he observed the effect from carbon monoxide adsorbed onto silver surfaces prepared by evaporating silver onto a cold smooth surface. For that system, the Raman effect is enhanced by a factor of 10s over what one would expect for a monolayer of CO molecules. Annealing the system—even only at room temperature—removes the enhancement. To determine the effect of roughness, Tsang and Kirtley deposited their aluminum films on calcium fluoride, whose roughness they could control; Rowe, Shank, Murray and Zwemer deposited silver iodide on silver, exposed it to light and developed it, leaving silver beads on the surface. In both cases, the scale of the surface roughness has a clear effect on the enhancement. The intensity of Raman lines for isolated molecules depends on the exciting frequency, increasing as w4, as does Rayleigh scattering. In the case of the enhanced Raman effect the behavior is not so simple: The a>4 dependence is modified by factors that depend on the rest of the system. In recent experiments involving pyridine on copper, Van Duyne's group measured Raman intensities as a function of the exciting frequency. They found that the Raman enhancement closely paralleled the reflectivity of copper: the larger the reflectivity, the larger the enhancement. Tsang and Kirtley had earlier reported analogous results for silver, and Colton's group at NRL has found comparable results for gold. The broad background was seen even in the first experiments. Its origin is not clear. Jonathan Heritage (Bell Labs, Holmdel) suggested to us that the enhancement is a kind of fluorescence phenomenon. He has investigated the enhancement by observing stimulated Raman scattering from cyanide on silver—the amplification of a second (weaker) laser beam tuned to one of the Raman lines produced by the exciting laser. He found no amplification when the second laser was tuned between the Raman lines, indicating that the background is not associated with the Raman effect. But this interpretation is not accepted by everyone. The effect appears not to require chemical bonding between the molecules and the substrate, Klein told us. This is suggested, for example, by the fact that carbon monoxide on silver, which produces one of the strongest enhancements,

PHYSICS TODAY / APRIL 1980 19 oaded 04 Oct 2013 to 129.105.14.67. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us

has the weakest chemisorption of the systems for which enhancement is seen. Furthermore, Rowe, Shank, Murray and Zwemer were able to determine, using Auger spectroscopy and low-energy electron diffraction, the thickness of the adsorbed layers in their experiments. They found that at least some lines are enhanced by effects that persist out to more than five molecular layers (about 30 A). Models. The difficulty of interpreting the data from the complex electrolyte systems and the rapid rate of accumulating experimental results has made it difficult to construct models to explain the effect. As Joseph Birman, a theorist at City College in New York told us, it was difficult to see what is general for all systems and what is specific. But any model or class of models will have to be able to produce a factor of 10e enhancement as a sort of yardstick. Burstein had pointed out that the geometry of the surface must be paramount. Somehow, the roughness produces an enhanced local electromagnetic field; the adsorbed molecules respond to that enhanced field. In one model, proposed by Van Duyne and his collaborators, the image dipole that is induced in the metal by the polarization of the adsorbed molecule greatly enhances the polarizability of the molecule. Although it is a classical—and more or less heuristic—model, Van Duyne said, it is a useful guide for suggesting experiments. Horia Metiu and Shlomo Efrima (University of California, Santa Barbara) independently proposed a more elaborate model of this type. Another class of models is based on the resonant Raman effect. Metiu and Efrima, for example, have also suggested that an interaction between the molecule and the substrate causes a shifting and broadening of molecular levels so as to make possible an electronic transition at the exciting frequency. Burstein and his coworkers have proposed that the enhancement is due to coupling between the molecular excitation and excited electron-hole pairs in the metal. The surface roughness plays the role of an antenna, strengthening the interaction with the radiation field. Many groups are now working on models that involve coupling of light to surface plasmons (transverse collective electron excitations) via the surface roughness. Among these are Kirtley, S. S. Jha and Tsang; Peter Wolff (MIT), Philip Platzman and Samuel McCall (Bell Labs); Ting-Kuo Lee (Institute for Theoretical Physics, Santa Barbara) and Birman; and Martin Moskovits (University of Toronto). Most investigators appear to agree with Van Duyne, who believes that no one mechanism is likely to produce the entire enhancement, and that the same causes do not act with the same strength in all systems. Applications. Surface chemists would 20

clearly be delighted with an analytic technique that allows one to measure surface properties in complex systems, under "messy" (real world) conditions, without disturbing other processes, at high temperatures and with the ability to give one dynamical and high-speed time-resolved information. It would be useful, for example, for investigating catalysts, corroding surfaces, batteries and other systems. The enhanced Raman effect shows promise of being just such a tool, once it is understood. Most investigators in the field are interested in more general questions of spectroscopy and surface science, and Heritage suggested that the enhancement may be most useful in helping us sharpen our experimental skills to the point that we can observe Raman

spectra from catalysts or other surfaces in cases without enhancement. —TVF References 1. M. Fleischmann, P. J. Hendra, A. J. McQuillan, Chem. Phys. Lett. 26, 123 (1974). 2. D. L. Jeanmaire, R. P. Van Duyne, J. Electroanal. Chem. 84,1 (1977). 3. M. G. Albrecht, J. A. Creighton, J. Am Chem. Soc. 99, 5215(1977). 4. R. P. Van Duyne, in Chemical and Biological Applications of Lasers 4, C. B. Moore, ed., Academic, New York (1979). 5. E. Burstein, C. Y. Chen, S. Lundquist, in Light Scattering in Solids, J. L. Birman, H. Z. Cummins, K. K. Rebane, eds., Plenum, New York (1979). 6. T. E. Furtak, J. Reyes-Corona, Surf. Sci. (in press).

Langchow plans heavy-ion facility The Institute of Modern Physics in Langchow is one of three nuclear-physics institutes in the People's Republic of China and the only one I saw during my recent visit (PHYSICS TODAY, March 1980, page 32). The Institute has ambitous plans for a new heavy-ion facility whose first phase, scheduled to operate in 1985, is expected to produce heavy ions from carbon to xenon—with 50 MeV/ nucleon for low Z and 6 MeV/nucleon for high Z. The second phase would produce light ions with 100 MeV/nucleon and all ions up to uranium with 10 MeV/nucleon. The Institute, formally founded in 1963, inherited a Soviet-built cyclotron that was assembled in Langchow. Between 1963 and 1973 the cyclotron was entirely devoted to the measuring of fast-neutron cross sections and studying reactions of very light nuclei, presumably for military applications. Since 1973 the Institute has primarily been doing heavy-ion research, converting the old cyclotron so that it could accelerate carbon, nitrogen and oxygen. The old cyclotron, which has a 1.5meter-diameter pole face, is being converted to a 1.7-meter-diameter sectorfocusing cyclotron, scheduled for completion in 1982. This sector-focusing cyclotron will then be used as an injector for a separated-sector cyclotron, the first phase of the heavy-ion facility. During a lab tour, I saw the 1.7-meter pole pieces and a '/(-scale model magnet being tested. In the full-scale version, the field will be 16 kG, and the device will have three sectors. Construction of a large new building to house the separated-sector cyclotron and experimental areas is underway. The injection radius is to be 0.9 or 1 meter, and the extraction radius will be 3.2 meters; the value for K is 450. The Institute's director, Yang Cheng-chung, told me he

hopes that Phase 1, the separated-sector cyclotron, will operate in 1985, but the timetable depends on various factories in China—one might machine the magnet, another the coils, another the rf generators. Phase 2 of the Langchow heavy-ion facility would have a 20-MV tandem electrostatic accelerator injecting into the separated-sector cyclotron, thus extending the mass range from about xenon to uranium. Yang feels it is too early to predict when Phase 2 would be completed, but R&D on the tandem is being vigorously pursued now at the Institute of Modern Physics. The Langchow institute is cooperating with the Institute of Nuclear Research in Shanghai, which will have a 6-MV tandem to be built by the

Section ol an accelerating tube at the Institute of Modern Physics in Langchow is made of AI2O3 with titanium electrodes.

PHYSICS TODAY / APRIL 1980

oaded 04 Oct 2013 to 129.105.14.67. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us