Resonance Raman Spectroscopy - Analytical Chemistry (ACS

May 31, 2012 - Femtomolar Detection of Prostate-Specific Antigen: An Immunoassay Based on Surface-Enhanced Raman Scattering and Immunogold Labels...
0 downloads 0 Views 12MB Size
Resonance Raman Spectroscopy

Michael D. Morris David J. Wallan Department of Chemistry University of Michigan Ann Arbor, Mich. 48109

Resonance enhancement of Raman scattering occurs when t h e excitation wavelength corresponds to the wave­ length of a dipole-allowed electronic transition of a molecule. Vibrations responsible for the usually unresolved vibronic structure of the absorption band as well as totally symmetric vi­ brations are enhanced. T h e enhanced R a m a n bands may have intensities 10 2 -10 6 times greater t h a n normal Raman intensities. Consequently, res­ onance Raman spectra have low detec­ tion limits and are much simpler than normal Raman spectra, since only bands related to the "chromophore" are enhanced. T h e power of resonance enhancement is shown in Figure 1. T h e sample is 5 X 10~ 6 M cyanocobalamin and 0.05 M sodium nitrate aque­ ous solution. T h e spectrum is ob­ tained with 514.5-nm (Ar + laser) exci­ tation, a wavelength within the in­ tense lowest π — -π* transition of cyanocobalamin, b u t well removed from any nitrate absorption bands. T h u s , the strong ring vibration (1504 cm" 1 ) has about half the integrated intensity of the strongest nitrate band at 1055 c m - 1 , despite a concentration differ­ ence of four orders of magnitude.

Resonance R a m a n spectroscopy possesses many features t h a t make it attractive to the analytical chemist. These properties are outlined in Table I. However, problems of sample fluo­ rescence have limited its practical ap­ plications. In this REPORT we will out­ line the theory of resonance Raman spectroscopy and review the published analytical applications. We will also describe the most promising approach to fluorescence rejection, coherent res­ onance Raman spectroscopy, and summarize the current applications and the prospects for the future. Resonance R a m a n spectroscopy is the only high-resolution spectroscopic technique t h a t is routinely applicable to dilute aqueous solutions. For exam­ ple, detection limits of 2 Χ Ι Ο - 7 Μ for vitamin B12 have been obtained using only a simple photon-counting system (1). Recently, detection limits for βcarotene were pushed below 10~ 8 M, by use of a multipass cell (2). Struc­ tural studies are routinely carried out on solutions in the 1 0 - 4 - 1 0 ~ 5 M range. Resonance-Enhanced Raman Spectra T h e period of early exploration of applications of resonance R a m a n spectroscopy included vigorous debate about the origin of the effect and t h e form of its governing equations. Most of the controversies have been re­ solved, and several recent reviews dis­ cuss the theory clearly (3-7). T h e res­ onance R a m a n effect results from the promotion of an electron into an excit­ ed vibronic state, accompanied by im­ mediate relaxation into a vibrational level of the ground state. T h e process is not preceded by prior relaxation to

182 A · ANALYTICAL CHEMISTRY, VOL. 5 1 , NO. 2, FEBRUARY 1979

the lowest vibrational level of the ex­ cited state as in ordinary fluorescence. T h e distinction is shown in Figure 2. Consequently, the resonance R a m a n emission process is essentially instan­ taneous, and the resulting spectra con­ sist of narrow bands. For molecules in solution, electronic states are broadened by many closely spaced vi­ brational states. Excitation with ra­ diation anywhere within this continu­ um will give rise to the same reso­ nance R a m a n spectrum, with an in­ tensity proportional t o t h e absorption intensity. Spectra may be obtained using an excitation frequency j u s t below the absorption band. Such spec­ t r a display smaller resonance en­ hancement, typically less than tenfold, and are called preresonance R a m a n spectra. Not all of the normal Raman bands are equally enhanced. Only those vi­ brations t h a t exhibit a large change in equilibrium geometry upon elec­ tronic excitation will produce strongly resonance-enhanced Raman bands. In practical terms, this means t h a t two classes of vibrational modes will pro­ duce intense resonance-enhanced spectra. These are totally symmetric vibrations, and those nontotally sym­ metric vibrations t h a t vibronically couple two electronic states. T h e re­ sulting two classes of enhancement are called Α-term and B-term enhance­ ment, respectively. T h e y can be dis­ tinguished experimentally by their in­ tensity vs. excitation frequency de­ pendences (called excitation profiles) in the preresonance region. Note t h a t although resonance en­ hancement involves a true electronic excited state of a molecule, the vibra­ tional frequencies observed are the 0003-2700/79/0351-0182A$01.00/0 © 1979 American Chemical Society

Current Applications and Prospects

ground state frequencies of the molecule, as observed in infrared absorption or normal R a m a n spectroscopy. T h e reason is t h a t the resonance-enhanced scattering process starts and terminates in the ground electronic state of the molecule. Resonance R a m a n spectroscopy can therefore be considered a form of high-resolution vibronic spectroscopy. Since an electronic transition is often more or less localized in one p a r t of a complex molecule, the resonance R a m a n effect provides a way to selectively enhance the Raman bands due to vibrations of this chromophore. This selectivity is quite apparent in molecules such as the heme proteins, whose resonance Raman spectra have been extensively studied (8). It is easily shown t h a t the resonance R a m a n bands are due solely to vibrational modes of the tetrapyrrole chromophore. None of the bands associated with the protein is enhanced, and at the concentrations employed for bio-

Report

chemical studies ( 1 0 ~ 4 - 1 0 - 5 M ) , protein bands are too weak to be observed. T h e sensitivity of resonance R a m a n spectroscopy to only chromophore vibrational modes may be considered either a strength or a weakness. On the one hand, spectra are greatly simplified, and a series of molecules containing slightly different chromophores will give spectra t h a t are easily distinguished. On the other hand, if a series of molecules contains the same chromophore with, for example, different aliphatic side chains, the resonance Raman spectra will be nearly identical. T h e analytical advantages of resonance R a m a n spectroscopy were suggested as early as 1959 (9). At t h a t time, sample self-absorption and the lack of a multiline source of sufficient intensity limited resonance R a m a n spectroscopic research to a handful of specialists. T h e argon ion and krypton ion lasers, which became available in

Figure 1 . Resonance Raman spectrum of 5 X 10 6 /Wcyanocobalamin and 5 X 1 0 ~ 2 M nitrate ion in 0.1 M H C I . Excitation frequency: 19436 c m - 1 (514.5 nm)

Resonance Raman Spectroscopy Works well in aqueous solution Works well with solid samples Good detection limits, 1 0 ~ e - 1 0 - 8 M Good resolution, 10-20 c m - 1 (0.3-0.6 nm at 500 nm) Structure-sensitive spectra Simple spectra. Only "chromophore" vibrations contribute No special equipment needed. Spectra obtained on conventional Raman spectrometers Fast data acquisition. Scan rate 20-100 cm-1/min Multichannel detector sometimes usable

the late 1960's, were quickly adopted as the intense excitation sources needed for resonance Raman applications. Almost immediately, resonance Raman spectroscopy became a valuable tool for the study of such diverse biochemical systems as heme derivatives and the poly-ene visual pigments. Applications to the stud^ of inorganic molecules also quickly a p peared (10). Today commercially available dye lasers cover the wavelength range from about 300 nm into the near infrared. Frequency doubling extends this wavelength"range down to about 220 nm. T h u s , resonance R a m a n spectra can now be obtained over almost the entire range in which conventional electronic absorption spectra can be taken. Clearly, one should be able to substitute the high resolution of resonance R a m a n spectroscopy for some or most of the separation steps and interference removal reactions needed for low-resolution techniques such as fluorimetry or spectrophotometry. In fact, resonance R a m a n spectra of dyes and pesticides have been observed a t the micromolar level in such media as

ANALYTICAL CHEMISTRY, VOL. 5 1 , NO. 2, FEBRUARY 1979 · 183 A

Vm

"ex

Δι/ Γ

Resonance Raman

Fluorescence

Figure 2. Energy diagram showing difference between resonance Raman emission and fluorescence Relaxation to ground vibrational state, which precedes fluorescent emission, depicted by undulating arrow

river water and cherry soda with little or no matrix interference (11-13). T h e combination of resonance R a m a n spectroscopy and some simple "chromophore developing" chemistry could result in many practical analyti­ cal methods. T h e catecholamines are an example of a system of closely re­ lated compounds amenable to this technique. T h e aminochromes, which are formed by oxidation of catechol­ amines with air or one of several com­ mon oxidizing agents, are useful for this purpose (14-16). Many molecules, such as heme derivatives, vitamin B12, poly-ene visual pigments and antibiot­ ics, have suitable absorption spectra and are easily studied with no chemi­ cal transformations at all. Resonance R a m a n spectroscopy is also being used to investigate tempo­ ral changes in chemical systems. Sev­ eral groups have used this method t o examine generated species at electro­ chemical diffusion layers (17,18). Spectra have been obtained using multichannel detectors, such as vidicons (19), and 7-ns time resolution has been observed (20). Vibrational spectra of solids are readily obtained using resonance R a m a n spectroscopy. Recently, t h e technique has been employed to probe the olefin adsorption on a zeolite sub­ strate (21). T h e chemical transforma­ tions of rose bengal adsorbed on a ZnO sinter electrode have also been investi­ gated (22). Resonance R a m a n spectroscopy is now established as a useful tool for a

wide variety of problems. Nonetheless, analytical chemists have not often em­ ployed it, largely because many poten­ tially interesting samples are fluo­ rescent. Fluorescence appears as a broad band background signal and may be so intense as to completely ob­ scure R a m a n signals. Urine and serum, for example, are sufficiently fluorescent t h a t resonance R a m a n spectroscopy in these common matri­ ces is generally difficult or impossible. T h e same problem may occur with paint samples, colored fabrics, or pharmaceutical preparations. Sample fluorescence has long plagued Raman spectroscopists, and many have developed ad hoc tech­ niques for dealing with t h e problem. Extensive sample purification or de­ struction of fluorescent impurities in a laser beam is sometimes helpful. However, if the fluorescent culprit is the species of interest, these tech­ niques are useless, and even addition of quenching agents is usually not suf­ ficient. Recently, time resolution of Raman signals from the much slower fluorescence emission has been pro­ posed (23-27). However, fluorescence rejection is generally only ten to one hundredfold, an insufficient factor for many analytical applications. In some cases, time resolution actually lowers signal-to-noise ratios (27).

Coherent Raman Spectroscopy Recently, coherent R a m a n spectros­ copy (28-31 ) has been proposed as a method for obtaining R a m a n spectra

of fluorescent samples. T h e fluores­ cence rejection of these techniques is indeed spectacular. An early report presented spectra of samples deliber­ ately spiked with laser dyes (32), and the spectra of dyes themselves have been reported. Spectra of many other highly fluorescent molecules have since been observed. Coherent Raman spectroscopy appears t o provide t h e most promising approach to analytical R a m a n spectroscopy. T h e term coherent Raman spec­ troscopy has come to include several closely related techniques. These in­ clude coherent anti-Stokes R a m a n spectroscopy, CARS (32-34), Ramaninduced Kerr effect spectroscopy, R I K E S (35, 36), and inverse R a m a n spectroscopy, 1RS (37, 38). Several other techniques suggested recently are all closely related to these three. All the techniques depend on the in­ teraction of two intense, generally pulsed, laser beams with a sample. When the frequencies of the lasers dif­ fer by a Raman-active frequency, then one can observe coherent emission (CARS), absorption out of one beam (1RS) or a change in the polarization state of one of t h e beams (RIKES). Although CARS, and with it fairly broad interest in coherent R a m a n spectroscopy, is only 4 or 5 years old, the advantages and problems of coher­ ent R a m a n spectroscopy are already becoming clear. T h e advantages are summarized in Table II. For our pur­ poses, the excellent fluorescence rejec­ tion is t h e most important. Some of the problems are merely those of a new technology. At present, one must assemble one's own instru­ ment, which usually lacks the conve­ nience features of commercial devices. More serious is t h e presence in CARS of a nonresonant background signal, which limits its sensitivity. At this writing, detection limits are at least

Table II. Advantages of Coherent Raman Spectroscopy High conversion efficiency Strong signals in most cases Low average incident power required Short spectral acquisition time—even single pulse Nearly 100% collection of signal Good spatial rejection of fluorescence

Good spectral rejection of fluorescence Résolution determined by laser Only low-resolution single or double monochromator needed Interference filter adequate in some cases

ANALYTICAL CHEMISTRY, VOL. 5 1 , NO. 2, FEBRUARY 1979 · 185 A

two orders of magnitude higher for CARS than for conventional R a m a n spectroscopy under equivalent condi­ tions. Refinements in instrumentation could improve detection limits some­ what. Newly emerging coherent tech­ niques, to which we shall return later, may soon provide the sensitivity need­ ed to exploit the full potential of reso­ nance Raman spectroscopy. T h e CARS apparatus used in our laboratory is similar to systems used by most investigators for CARS of so­ lutions. A 1-MW nitrogen laser pumps two dye lasers. T h e dye laser, which generates the beam 002, has a motordriven grating, allowing spectral scan­ ning. T h e other dye laser is manually tunable only. For reasons of economy, reliability, and ease of operation, the two transversely p u m p e d dye lasers are driven by a single nitrogen laser and form the most common CARS laser system. Virtually any sort of sample cell can be used for CARS. We have used spec­ trophotometer cells of 1- or 2-mm p a t h length. Melting point capillaries are useful, both because they are inex­ pensive and because they act as very short focal length lenses and help in aligning the system (39). Capillaries can also serve as flow cells for CARS. Nitrogen laser-pumped dye lasers have very short pulses, typically 5-10 ns long, depending on design. More­ over, the repetition rate is low, typi­ cally 10-25 pulses per second. Because of this short pulse length and low duty cycle, CARS data must be processed through a gated integrator. For strong signals the detector can be a P I N diode. More frequently, it is necessary t o use a photomultiplier. T h e common 1P28 is suitable. Because CARS sig­ nals show large pulse-to-pulse varia­ tions, it is useful to ratio to a reference signal. Typically, ωχ is monitored, but a cell filled with solvent can be used to generate a reference background signal. T h e gated integrator itself is a weak link in the coherent R a m a n experi­ mental system. T h e device is linear to no more t h a n 0.1% and operates over about three orders of magnitude of signal intensity. These linearity and dynamic range constrictions limit the effectiveness of background subtrac­ tion, averaging over many pulses or digital filtering. T h e presence of the nonresonant background signal is a fundamental limitation of CARS and is the origin of the poor sensitivity of this tech­ nique. CARS nonresonant background signals relative to water are shown for selected solvents in Table III. T h e key factor is t h a t the nonresonant back­ ground signal for aromatic solvents is about an order of magnitude below the signals for most other solvent sys-

Table III. Relative CARS Background Intensities a Solvent

Background

Water (reference) D20 Methanol Ethanol Chloroform Carbon tetrachloride Benzene Toluene m-Xylene Benzyl chloride

1.0 1.0 1.0 1.8 2.6 2.6 0.09 0.09 0.09 0.12

" Reprinted with permission from ref. 40. Copyright 1976 National Academy of Sciences.

terns. Therefore, lower detection lim­ its have been reported for samples in benzene t h a n for those in other com­ mon solvents. Overcoming the nonres­ onant background is the object of in­ tensive research. We will summarize progress in this area a t the end of this review.

Resonance-Enhanced Coherent Raman Spectra Coherent anti-Stokes R a m a n spec­ t r a can be resonance enhanced under the same conditions as spontaneous (conventional) R a m a n spectra (40, 41). T h e ωχ (pump) frequency is made coincidental with an electronic ab­ sorption maximum, and W2 (probe) is scanned to generate the spectrum. Ex­ citation profiles have been measured and found to conform to theory quite well (42). Resonance inverse R a m a n spectra have also been reported (43), b u t very little work has been done in this area.

T h e spectra of laser dyes them­ selves have been obtained and demon­ strate the power of coherent resonance Raman spectroscopy. These molecules fluoresce with high q u a n t u m efficien­ cy, approaching unity in the case of perylene, and conventional resonance R a m a n spectra of these molecules cannot be obtained (44). Biochemical applications of reso­ nance-enhanced CARS are now begin­ ning to appear. Recently, resonance CARS spectra of several flavins, in­ cluding FAD, glucose oxidase, and ri­ boflavin binding protein, have been reported (45, 46). Partial band assign­ ments and some preliminary structur­ al correlations have been presented. T h e resonance-enhanced CARS spec­ t r u m of adriamycin has been used to reinterpret the electronic spectrum of t h a t molecule (47), an important anti­ tumor drug. Flavins and adriamycin are highly fluorescent, and their reso­ nance R a m a n spectra cannot be ob­ tained by conventional means. Resonance CARS spectra are par­ ticularly susceptible to distortion by interference between the R a m a n sig­ nal of the solute and the nonresonant background of the solvent. Since these spectra are obtained with solute con­ centrations of 10~ 3 M or lower, there is an enormous excess of solvent, and negative and dispersive peaks are common. Figure 3 shows part of an ad­ riamycin spectrum in which the R a m a n bands appear dispersive. T h e midpoints between the maximum and minimum emission correspond to the positions of the conventional R a m a n bands. As ωχ is moved away from the origin of the electronic transition, the CARS bands are gradually trans­ formed to negative peaks (47). T h e same transition from positive bands

Figure 3. Resonance CARS spectrum of 3 X 10~3 Madriamycin in pH 6 buffer ω, = 20 000 c m - 1 (500 nm), origin of electronic transition. At this concentration, bands appear disper­ sive. Arrows show midpoint of each band, which corresponds well to bands observed in preresonance Raman spectrum of molecule

186 A · ANALYTICAL CHEMISTRY, VOL. 5 1 , NO. 2, FEBRUARY

1979

A FULLY AUTOMATIC MICROPROCESSOR CONTROLLED X-RAY SPECTROMETER

DIANO HAS IT !

DIANO now offers completely unattended instrument control at a price even lower than competitors' outdated manual systems! The secret — a powerful microprocessor coupled to a new torque drive system. A universal bi-directional interface also allows easy communication with most available I/O devices. This permits fuller utilization of existing laboratory computers and software. In addition, DIANO offers a full range of Digital Equipment Corporation's PDP-11 computers with DIANO's ACUPAK™ Fortran language software — the most extensive and advanced package of x-ray fluorescence software available. See the automatic spectrometer with the highest Performance/Cost Ratio in history at Cleveland or let us have one of our local sales representatives contact you with full details.

DIANO CORPORATION 8 COMMONWEALTH AVENUE VVOBURN, MASSACHUSETTS 01801 U.S.A. TEL.(617)-935-4310 TELEX 94-9306 CIRCLE 46 ON READER SERVICE CARD 188 A ·

ANALYTICAL CHEMISTRY, VOL. 5 1 , NO. 2, FEBRUARY

1979

to dispersive to negative can also be observed at fixed p u m p wavelength, simply by diluting the sample. Although resonance-enhanced CARS was first reported only 2 years ago, the technique is already proving valuable for mechanistic and structural studies. Because of the nonresonant background signal, detection limits in aqueous solution are seldom below 1 0 - 5 M and can be pushed to 10~ 7 M in benzene solution in exceptional cases. These high detection limits preclude widespread use of resonance CARS for trace analysis. Development of a background-free or a t least low background coherent R a m a n spectroscopic method is an area of active research. Optical cancellation of nonresonant signals and techniques free of nonresonant signals are being explored in several labs. Raman induced Kerr effect spectroscopy (RIKES) can be used to cancel the background optically (35, 36). This technique exploits birefringence induced in a sample when the two incident laser beams differ in frequency by a R a m a n active frequency. A major experimental problem with R I K E S is t h a t it requires the use of linearly polarized light with an extinction ratio of 10 5 or better. Ordinary Glan laser prisms are not capable of such high extinction ratios, and expensive polarizing optics are required. An alternative approach called " C W stimulated R a m a n " (48-50) has been demonstrated. It is essentially inverse R a m a n or R a m a n gain spectroscopy. T h e absorption from a laser beam or the energy gained by a laser beam is measured under inverse R a m a n conditions. T h e gain or absorption from an argon ion laser is measured. T h e second tunable laser beam is chopped, and a lock-in amplifier is used to measure only the component of the argon beam synchronous with the chopped beam. By this technique very small induced intensity changes can be measured. Another inverse Raman technique uses the induced absorption generated on an argon laser beam by a pulsed dye laser. T h e high peak power available from a pulsed dye laser means t h a t the observed signal is much larger than t h a t found by the CW technique (51). Signal-to-noise ratios are thus higher. R a m a n gain and inverse R a m a n measurements require absorption of energy by the sample and therefore are intrinsically free of any nonreson a n t background. Measurement of R a m a n spectra by means of these phenomena should allow generation of undistorted spectra down to very low concentrations. T h e Kerr effect and the inverse R a m a n / R a m a n gain techniques are

Peak separation you've never seen before.

in their infancy. Both look quite promising. One or both of these vari­ ants of coherent R a m a n spectroscopy may give analytical chemists the means to obtain Raman spectra of flu­ orescent materials at trace concentra­ tions within the next year or two.

Acknowledgment T h e authors thank G. Patrick Ritz and Jeanne Haushalter who have been extensively involved in the develop­ ment and applications of our coherent R a m a n system.

References

* INSTEAD OF THIS

THIS

New from this X-ray fluorescence analyzer.

A new and unique pulse processor gives our SPECTRACE™ analyzer guaranteed resolution of 150 eV —the best ever achieved in an energy-dispersive X-ray fluorescence analyzer. This resolu­ tion, coupled with peak shifts no greater than 3 eV at maximum count rate, lets you detect light elements more reliably and quantify them more accurately. The spectrum at left above shows an actual peak separation using the new pulse processor. That at right, with overlapping peaks, was made using a conventional processor in the same analyzer. SPECTRACE 440 can make auto­ matic unattended analyses of up to

40 samples —solids, liquids, powders or deposits on filter paper. Makes qualitative and quantitative analyses of all elements from Na to Pu. Han­ dles samples from 1 mm in diameter to a foot high. Has superb sensitivity, repeatability and long-term stability. Our applications laboratory and engineering group will work with you to help solve your analytical problem. That way you know you'll get the results you want after your SPECTRACE is delivered. Over 50 SPECTRACE analyzers are now in use. And we'll give you the name of every single user— because we know every one will give you a favorable report. One thing they'll report is the service they get: personal instruction, an applications school, software school—and follow-up visits just to make sure you're getting the most out of your SPECTRACE. Send for our highly informative brochure now. Contact United Scientific Corporation, Analytical Instrument Division, Dept. C, 1400 Stierlin Road, RO. Box 1389, Mt. View, California 94042. Phone (415) 969-9400.

See SPECTRACE at the Pittsburg Conference, Booths 1702 & 1704

UNITED SCIENTIFIC ANALYTICAL INSTRUMENTS DIVISION CIRCLE 212 O N READER SERVICE CARD

190 A · ANALYTICAL CHEMISTRY, VOL. 5 1 , NO. 2, FEBRUARY 1979

(1) C.-W. Tsai and M. D. Morris, Anal. Chim. Acta, 76,193 (1975). (2) L. C. Hoskins and U. Alexander, Anal. Chem., 49, 695 (1977). (3) J. Behringer, in "Molecular Spectros­ copy, Specialist Periodic Reports," R. F. Barrow, D. A. Long, and D. J. Millen, Eds., The Chemical Society, pp 170-3, London, England, 1974. (4) J. Behringer, ibid., Vol. 3, pp 163-280, 1975. (5) B. B. Johnson and W. L. Peticolas, Ann. Rev. Phys. Chem., 27, 465 (1977). (6) T. G. Spiro and P. Stein, ibid., 28,501 (1977). (7) A. Warshel, Ann. Rev. Biophys. Bioeng., 6, 273 (1977). (8) T. G. Spiro, in "Vibrational Spectra and Structure," J. R. Durig, Ed., pp 101-20, Elsevier, Amsterdam, The Neth­ erlands, 1976. (9) J. Brandmiiller, Z. Anal. Chem., 170, 29 (1959). (10) R.J.H. Clark, in "Advances in InfraRed and Raman Spectroscopy," R.J.H. Clark and R. E. Hester, Eds., Vol. 1, pp 143-72, Heyden, London, England, 1975. (11) C. W. Brown and P. F. Lynch, J. Food Sci., 41,1231 (1976). (12) L. VanHaverbeke, P. F. Lynch, and C. W. Brown, Anal. Chem., 50, 315 (1978). (13) R. J. Thibeau, L. VanHaverbeke, and C. W. Brown, Appl. Spectrosc, 32, 98 (1978). (14) M. D. Morris, Anal. Chem., 47, 2453 (1975). (15) M. S. Rahaman and M. D. Morris, Talanta, 23, 65 (1976). (16) M. D. Morris, Anal. Lett., 9, 469 (1976). (17) D. L. Jeanmaire, M. R. Suchanski, and R. P. Van Duyne, J. Am. Chem. Soc, 97,1699 (1975). (18) J. L. Anderson and J. R. Kincaid, Appl. Spectrosc, 32, 356 (1978). (19) W. H. Woodruff and G. H. Atkinson, Anal. Chem., 48,186 (1976). (20) W. H. Woodruff and S. Farquharson, ibid., 50,1389 (1978). (21) P. J. Trotter, J. Phys. Chem., 82, 2396 (1978). (22) H. Yamada, Y. Amamiya, and H. Tsubomura, Chem. Phys. Lett., 56, 591 (1978). (23) P. P. Yaney, J. Opt. Soc. Am., 62, 1297 (1972). (24) R. P. Van Duyne, D. L. Jeanmaire, and D. F. Shriver, Anal. Chem., 46, 213 (1974). (25) F. E. Lytle and M. S. Kelsey, ibid., ρ 855. (26) J. M. Harris, R. W. Chrisman, F. E. Lytle, and R. S. Tobias, ibid., 48,1937 (1976). (27) S. Burgess and I. W. Shepherd, J. Phys., E, 10,617 (1977). (28) A. B. Harvey, Anal. Chem., 50,905A (1978).

(29) W. M. Toiles, J. W. Nibler, J. R. McDonald, and A. B. Harvey, Appl. Spectrosc, 31, 253 (1977). (30) M. D. Levenson, Phys. Today, 30 (5), 44 (1977). (31) H. C. Andersen and B. S. Hudson, in "Molecular Spectroscopy, Specialist Periodic Reports," R. F. Barrow, D. A. Long, and D. J. Millen, Eds., Vol. 5, The Chemical Society, London, England, 1977. (32) R. F. Begley, A. B. Harvey, R. L. Byer, and B. S. Hudson, J. Chem. Phys., 61, 2466 (1974). (33) P. R. Régnier and J.P.E. Taran, Appl. Phys. Lett., 23, 240 (1973). (34) R. F. Begley, A. B. Harvey, and R. L. Byer, ibid., 25, 387 (1974). (35) D. Heiman, R. W. Hellworth, M. D. Levenson, and G. Martin, Phys. Rev. Lett., 36,189 (1976). (36) M. D. Levenson and J. J. Song, J. Opt. Soc. Am., 66, 641 (1976). (37) W. J. Jones and B. P. Stoicheff, Phys. Rev. Lett., 13,657(1964). (38) E. S. Yeung, J. Mol. Spectrosc, 53, 379 (1974). (39) L. Β. Rogers, J. D. Stuart, L. P. Goss, Τ. Β. Malloy, Jr., and L. A. Carreira, Anal. Chem., 49, 959 (1977). (40) J. Nestor, T. G. Spiro, and G. Klauminzer, Proc. Nat. Acad. Sci., 73, 3329 (1976). (41) Β. J. Hudson, W. Hetherington, S. Kramer, I. Chalsay, and G. K. Klauminzer, ibid., ρ 3798. (42) L. A. Carreira, T. C. Maguire, and T. B. Malloy, J. Chem. Phys., 66, 2621 (1977). (43) S. H. Lin, E. S. Reid, and C. J. Tredwell, Chem. Phys. Lett., 29, 389 (1974). (44) L. A. Carreira, T. C. Maguire, and T. B. Malloy, Jr., 32nd Symp. on Molecular

Spectroscopy, Paper No. WE7, Ohio State University, Columbus, Ohio, June 1977. (45) P. B. Dutta, J. R. Nestor, and T. G. Spiro, Proc. Nat. Acad. Sci., 74, 4146 (1977). (46) P. B. Dutta, J. R. Nestor, and T. G. Spiro, Biochem. Biophys. Res. Commun., 83, 209 (1978). (47) G. P. Ritz and M. D. Morris, J. Raman Spectrosc., submitted for publi­ cation.

(48) A. Owyoung, Opt. Commun., 22, 323 (1977). (49) A. Owyoung and E. D. Jones, Opt. Lett., 1,152 (1977). (50) A. Owyoung, IEEE J. Quantum Elec­ tron., QE-14,192 (1978). (51) M. D. Morris, D. J. Wallan, G. P. Ritz, and J. P. Haushalter, Anal. Chem., 50, 1796 (1978). Financial support by the National Institutes of Health (Grant GM22604).

αβ@Μη^ΒΗΒΒηΒΐηΒϊβίΙίΕ93Β

Michael Morris (right) is associate professor of chemistry a t the University of Michigan. His research interests are coherent R a m a n spectroscopy and other applications of laser spectroscopy. David Wallan (left) is a graduate s t u d e n t in chemistry at the University of Michigan. He holds a full-year ACS Division of Analytical Chemistry Fellowship, sponsored by the Procter & Gamble Co.

The laboratory performers from Houston Atlas. ilfrThe hallmark of the Houston Atlas performers is the achievement of superior accuracy, reliability, and effortless precision performance . . . without the need for skilled technicians. Plus - application flexi­ bilities that deliver even more performance for your dollar. ^ Whether you work with gases or liquids or solids, in low ppb or high %, Houston Atlas has a "can-do" performer for you.