Near-Infrared Excited Raman Optical Activity

LAURENCE A. NAFIE,* BRUCE E. BRINSON, XIAOLIN CAO,� DAVID A. RICE,OMAR M. RAHIM, RINA K. DUKOR, and NAOMI J. HALASDepartment of Chemistry, Syracuse University, Syracuse, New York 13244-4100 (L.A.N., X.C.); Critical Link, LLC, 6712 Brooklawn Pkwy,

Syracuse, New York 13211 (D.A.R., O.M.R.); Departments of Electrical Engineering and Chemistry, Rice University, Houston, Texas 77005(B.E.B., N.J.H.); and BioTools, Inc., 17546 Bee Line Hwy, Jupiter, Florida 33458 (R.K.D., L.A.N.)

Measurements of near-infrared scattered circular polarization Raman

optical activity (SCP-ROA) are presented using laser excitation at 780 nm

for samples of S-(�)-a-pinene and L-alanyl-L-alanine. These are the first

measurements of ROA outside the blue-to-green visible region between

488 and 532 nm. Comparison of Raman and ROA intensities measured

with excitation at 532 and 780 nm demonstrate that the expected

frequency to the fourth-power dependence for Raman scattering and the

corresponding fifth-power dependence for ROA are observed. It can be

concluded that, to within this frequency dependence, the same level of

efficiency of Raman and ROA measurements using commercial instru-

mentation with 532 nm excitation is maintained with the change to near-

infrared excitation at 780 nm.

Index Headings: Raman scattering; Raman optical activity; Near-infrared

laser excitation.

INTRODUCTION

Raman optical activity (ROA) was first observed experi-mentally over thirty years ago.1,2 These early measurementswere carried out using the incident circular polarization (ICP)modulation scheme, laser excitation at 488 nm with an argonion laser, and single-channel detection with a scanningmonochromator. A second form of ROA, originally namedthe degree of circularity,3,4 was observed experimentally forthe first time5 in 1988 and named scattered circular polarization(SCP) ROA. In the following year, two new forms of ROA, in-phase and out-of-phase dual circular polarization (DCPI andDCPII), were proposed6 and then subsequently measured.7,8 Inthe more than thirty years since the discovery of ROA, allmeasurements have used either 488 or 514 nm excitation froman argon ion laser or 532 nm excitation from a Nd:YVO4 laser.We report here the first measurements of ROA with laserradiation outside this blue-to-green region of the visiblespectrum. In particular, we present SCP-ROA spectra of thenatural terpene S-(-)-a-pinene and the peptide L-alanyl-L-alanine in H2O solution using near-infrared (NIR) laserexcitation at 780 nm. To gauge the quality of the NIR ROAspectra, we compared the new measurements to previousmeasurements of ROA using excitation at 532 nm, which hasbecome the standard laser wavelength for ROA measurements.

The motivation for the NIR ROA measurements is thewidespread success in recent years of NIR excited Ramanspectroscopy across the field of analytical spectroscopy forprocess monitoring and quality control. Excitation in the NIRregion dramatically reduces interfering fluorescence from mostsamples where this is a problem and makes the measurement ofa Raman spectrum a routine process that does not require

lengthy amounts of time for sample purification and prepara-tion. The achievement of NIR ROA would facilitate the use ofROA as a monitor of chirality in the chemical industry, wherechirality is rapidly gaining prominence and importance in theproducts produced, with the pharmaceutical and biopharma-ceutical industries being the leading examples.

EXPERIMENTAL

All SCP-ROA measurements were carried out on aChiralRAMANt ROA spectrometer from BioTools, Jupiter,FL. The design of this ROA spectrometer is based on onedeveloped by Werner Hug several years ago.9,10 The followingchanges were carried out to convert the ROA spectrometerfrom operation with laser excitation at 532 nm to operation at780 nm. A diode-pumped frequency-doubled Nd:YVO4 laserfrom Spectra Physics Lasers, Inc./Millennia IIs, 532 nm, wasreplaced with a NIR diode laser Sacher Lasertechnik GmbH/TIGER T780/PILOT 3000, 780 nm. All dynamic half-waveplates used to suppress polarization artifacts10 were changedfrom 532 to 780 nm operation. The Supernotch Plusholographic filter for Rayleigh line suppression at 532 nmfrom Kaiser Optical Systems, Inc., was changed to an HSPF780.0AR-2.0 Notch Filter at 780 nm from the same company.The notch filter has nearly a six-order-of-magnitude rejection at780 nm and a half-width at half-height of approximately 4 nm.The holographic transmission grating in a modified Holospecspectrograph from Kaiser Optical Systems, Inc., was changedfrom 532 nm near-Stokes to 780 nm near-Stokes spectralcoverage. The spectral resolution for excitation at 532 nm isapproximately 7 cm�1 for a fixed entrance slit determined bythe 70 micrometer diameter of the output optical fibers.9 Thenear-Stokes grating gives spectral coverage over the 1024 pixelcolumns of the charge-coupled device (CCD) of approximately2200 cm�1, and thus a resolution element of 7 cm�1

corresponds to approximately three columns of fibers. For780 nm excitation, the near-Stokes grating is changed toprovide similar spectral coverage in wavenumbers, and as aresult the grating dispersion of wavelengths in nm per pixelacross the CCD is reduced. As a result, even though a givenresolution of wavelengths from the fiber-optics entrance slit hasa lower number of wavenumbers for 780 nm excitationcompared to 532 nm excitation, the resolution does notincrease for a fixed input slit width. Consequently, the spectralresolution for 780 nm excitation remains approximately thesame as that for 532 nm excitation at 7 cm�1.

The CCD camera used for 532 nm excitation, the CCDspfrom BioTools/Critical Link, was equipped with a visible anti-reflection (AR)-coated, back-illuminated, 1024 3 255 pixelchip (E2V 3011BI from E2V Technologies, Inc.) operating inthe temperature range�30 to�40 8C. Lower temperatures havelittle effect on ROA measurements because very high Raman

Received 7 May 2007; accepted 10 July 2007.* Author to whom correspondence should be sent. E-mail: lnafie@syr.edu.

� Current address: Department of Global Cellular & Analytical Resources,Amgen Inc. Thousand Oaks, CA 91320.

Volume 61, Number 10, 2007 APPLIED SPECTROSCOPY 11030003-7028/07/6110-1103$2.00/0

� 2007 Society for Applied Spectroscopy

count levels (typically greater than 109 counts per spectralpoint) obscure the low dark counts of the CCD chip employed.For NIR excitation at 780 nm, the same CCD detector chip wasfound to give near-optimum performance. The options of deepdepletion added too high a level of dark current for theoperating temperature range used, and the NIR AR-coatingadded only a few percent to the quantum efficiency over thespectral range of the NIR Raman spectrum.

Samples of S-(�)-a-pinene and the peptide L-alanyl-L-alanine were obtained from Aldrich-Sigma and used withoutfurther purification.

RESULTS AND DISCUSSION

In Fig. 1, we present Raman and ROA spectra of S-(�)-a-pinene with the conditions indicated in the figure caption. Thegeneral appearance and quality of the two sets of Raman andROA spectra are quite comparable. Both Raman spectra werecollected to approximately the same level of total counts withpeak maxima near 3 3 109. There is an approximate advantageof 4.5 to 5.5 for the lower vibrational frequencies with 532 nmexcitation versus 780 nm excitation associated with thefrequency-to-the-fourth power of Raman scattering. This canbe seen qualitatively by comparing the laser power depen-dences and total exposure times required to obtain approxi-mately the same number of Raman counts for the two spectra.Other differences can be seen in the two Raman spectra. Thenotch filter used at 780 nm is not as narrow as the one used at532 nm. The half-width point on the Stokes side at 784 nmoccurs at approximately 70 cm�1, and the effect of the filter isnot reduced to a non-filtered level until 790 nm orapproximately 160 cm�1. The first Raman band at approxi-mately 130 cm�1, while visible, is noticeably reduced in sizerelative to nearby bands compared to its relative intensity in the532 nm Raman spectrum. Another difference is a generaldiminishment of the Raman spectrum at 780 nm toward highervibrational frequencies due to the fall-off in the sensitivity of

the CCD as wavelengths further into the near-infrared regionare measured.

Comparing the ROA spectra, we note that the ROAintensities at 780 nm are smaller relative to the Ramanintensities than their counterparts at 532 nm. The reason is dueto a wavelength dependence of all forms of optical activitywhereby molecules are less able to discriminate betweencircular polarization states as the wavelength increases. Forcircularly polarized radiation, one wavelength corresponds toone 360-degree rotation of the electric field vector about thepropagation direction. This is the main reason why vibrationalCD is smaller relative to its parent IR absorption by roughly anorder of magnitude compared to the intensity of electronic CDrelative to parent ultraviolet–visible absorption.3,4 As a result,ROA intensity exhibits a frequency-to-the-fifth power depen-dence on the laser excitation frequency.

In order to provide some quantitative measure to the two setsof Raman and ROA spectra, their relative intensities arecompared in Table I. Here, the Raman spectra in Fig. 1 arescaled to the number of counts that would be obtained at fourpoints in the spectra for 500 minutes of total exposure timewith 200 mW of power in each experiment. The total exposuretime of 500 minutes was a convenient level of scaling forcomparison between 532 nm and 780 nm excited ROA/Ramanspectra. The relatively short scan times shown in the figureswere chosen to highlight the differences in the signal qualitybetween the two excitation sources. At 500 minutes exposuretime for the ROA spectra, virtually no noise would bediscernable.

In Table I, first the Raman intensities are compared and thenthe ROA intensities are compared. For the ROA data, anadditional comparison is provided, namely the ratio of theROA-to-Raman ratio for the two excitation frequencies. Thisratio eliminates differences in the responses of the instrumentsas a function of Raman frequency for the two sets of data. Allof these ratios can then be compared to the ratios one calculatespurely on the basis of the ideal frequency-to-the-fourth

FIG. 1. Visible and near-infrared Raman and ROA spectra of S-(�)-a-pinene. Visible spectra were excited at 532 nm with 0.20 W laser power at the sample, 1664detector scans, 0.147 seconds illumination time per scan, and 4 minutes 4 seconds total exposure time. Near-infrared spectra were excited at 780 nm with 0.50 Wlaser power at the sample, 1184 detector scans, 0.515 seconds illumination time per scan, and 10 minutes 9 seconds total exposure time. The spectral resolution isapproximately 7 cm�1 for excitation at both 532 nm and 780 nm.

1104 Volume 61, Number 10, 2007

(Raman) and frequency-to-the-fifth (ROA) dependencies.Because these theoretical ratios are based on intensities andnot Raman photon counts, equal intensities at 532 nm and 780nm would require 780/532 (or 1.47) times more counts at 780nm. Each photon carries only 532/780 times the energy at 780nm versus 532 nm, and hence, in terms of Raman countsproduced by the CCD, the frequency-to-the-fourth power

dependence becomes a frequency-to-the-third power countdependence, and for ROA, the frequency-to-the-fifth powerdependence becomes a frequency-to-the-fourth power countdependence. The theoretical ratios for ROA-to-Raman ratiosare also provided, which is simply the first-order dependenceon the spectral frequency. From these results, we see that thereare some variations away, but not far away, from what oneexpects on the basis of frequency dependence alone. Incomparing the Raman spectra, differences in instrumentalresponses and efficiencies are important factors. In particularwe know that approximately half of the reduction in responseto higher wavenumbers arises from lower transmission, andtherefore high reflection percentages, associated with theobjective lens and the window of the CCD detector. Inaddition, there is a reduction in efficiency of the CCD responseat longer wavelengths that is more significant for 780 nmexcitation than for 532 nm excitation. Another interestingpossibility is that some pre-resonance effects may contribute todeviations from expected signal levels. Some vibrationalmodes may experience the onset of pre-resonance enhancementat 532 nm more than other modes. This effect, if present, isprobably quite small, because the Raman and ROA spectrafrom band to band have very similar relative intensities.

In Fig. 2 we show the 532 nm and 780 nm Raman and ROAspectra of L-alanyl-L-alanine in aqueous solution. Thesespectra provide a second example of the performances ofRaman and ROA spectral measurements with 532 nm and 780nm excitation. In this comparison, the total exposure timeswere the same in both cases, and somewhat higher laser powerwas used for the 532 nm excitation, 1.0 W versus 0.5 W. Theresult was nearly-an-order-of-magnitude-higher Raman countswith 532 nm excitation compared with 780 nm excitation. Dueto the lower total Raman counts, the ROA spectra measured at780 nm have noticeably higher noise levels. On the other hand,it can be seen that the Raman spectrum with 532 nm excitationis riding on a relatively higher fluorescence background. Forthis particular sample, background fluorescence is not a

TABLE I. Comparison of Raman and ROA intensities for a-pinenemeasured with 532 nm and 780 nm excitation.

Comparison of Raman intensities for 500 minutes of collection at 0.2 W,counts 3 109

Frequency Visible Near-infrared Ratio

380 cm�1 50 12 4.2780 cm�1 100 18 5.5850 cm�1 130 21 6.2

1430 cm�1 190 20 9.5

Comparison of ROA intensities for 500 minutes of collection at 0.2 W,counts 3 106

Frequency Visible Near-infrared Ratio ROA/Raman ratio

380 cm�1 35 4.5 7.8 1.9780 cm�1 90 7.5 12 2.2850 cm�1 38 3.2 11.8 1.9

1430 cm�1 100 6.5 15.4 1.6

Calculated frequency to third (Raman) and frequency to the fourth (ROA)dependencea

Frequency Raman ROA ROA/Raman ratio

380 cm�1 3.2 4.8 1.48780 cm�1 3.4 5.0 1.50850 cm�1 3.4 5.1 1.50

1430 cm�1 3.6 5.4 1.53

a The frequency-to-the-fourth and frequency-to-the-fifth power dependences ofRaman and ROA intensities, respectively, become frequency-to-the-third andfrequency-to-the-fourth power dependences on Raman and ROA photoncounts.

FIG. 2. Visible and near-infrared Raman and ROA spectra of a 2 M solution of L-alanyl-L-alanine in H2O with 10 minutes total exposure time. Visible spectra wereexcited at 532 nm with 1.00 W laser power at the sample and 1.1 seconds illumination time per scan. Near-infrared spectra were excited at 780 nm with 0.50 W laserpower at the sample and 2.5 seconds illumination time per scan.

APPLIED SPECTROSCOPY 1105

problem, but for more complex biological molecules, such aslarger peptides or proteins, fluorescence can be a majorproblem. Here, excitation at 780 nm becomes an advantageeven though longer collection times are in general required.

Besides fluorescence suppression and potential routineindustrial applications of ROA, we are interested in usingNIR ROA as a tool in the search for surface-enhanced Ramanscattering (SERS) ROA (SEROA) for which SERS substrates,such as NIR-tuned nanoshells,11 offer many attractiveadvantages. We are currently pursuing this line of investigationusing NIR excited ROA instrumentation.

CONCLUSION

We have presented the first examples of ROA spectracollected with NIR laser excitation. These are also the firstROA spectra recorded with laser excitation outside the blue-to-green spectral region between 488 and 532 nm. Comparisonsof the Raman and ROA spectra of neat S-(�)-a-pinene and L-alanyl-L-alanine in H2O demonstrate that the major differencesin normalized intensities for these two excitation regions can beattributed to the expected frequency-to-the-fourth powerdependence for the Raman spectra and a frequency-to-the-fifthpower dependence for the ROA spectra. Relatively smalldeviations from this ideal behavior can be attributed todifferences in spectral responses associated with the notchfilters, reflection surfaces, and CCD detectors. In the case of L-alanyl-L-alanine, noticeably lower levels of fluorescence,relative to Raman intensities, confirms a principal objectivein moving the laser excitation from the visible to the NIRregion. We conclude that changing the laser excitationfrequency from 532 nm to 780 nm meets all of the anticipated

objectives and that the level of ROA instrumental efficiencyavailable with 532 nm excitation has been successfullytransferred to ROA measurement with laser excitation at 780nm.

The availability of NIR excitation ROA opens opportunitiesfor ROA to be used as a diagnostic for routine quality controland process monitoring of chiral or bio-pharmaceuticalmolecules where currently only NIR absorption or NIR excitedRaman are used. Another potential application is themeasurement of SEROA where, for example, some SERSsubstrates can be tuned to NIR resonance enhancement.11

ACKNOWLEDGMENTS

This work was supported by STTR grants from the Air Force Office ofScientific Research to BioTools, Inc. and Syracuse University and toNanospectra Biosciences, Inc. and Rice University.

1. L. D. Barron, M. P. Bogaard, and A. D. Buckingham, J. Am. Chem. Soc.95, 603 (1973).

2. W. Hug, S. Kint, G. F. Bailey, and J. R. Scherer, J. Am. Chem. Soc. 97,5589 (1975).

3. P. W. Atkins and L. D. Barron, Mol. Phys. 16, 453 (1969).

4. L. D. Barron, Molecular Light Scattering and Optical Activity (CambridgeUniversity Press, Cambridge, 2004), 2nd ed.

5. K. M. Spencer, T. B. Freedman, and L. A. Nafie, Chem. Phys. Lett. 149,367 (1988).

6. L. A. Nafie and T. B. Freedman, Chem. Phys. Lett. 154, 260 (1989).

7. D. Che, L. Hecht, and L. A. Nafie, Chem. Phys. Lett. 180, 182 (1991).

8. G.-S. Yu and L. A. Nafie, Chem. Phys. Lett. 222, 403 (1994).

9. W. Hug and G. Hangartner, J. Raman Spectrosc. 30, 841 (1999).

10. W. Hug, Appl. Spectrosc. 57, 1 (2003).

11. J. B. Jackson and N. B. Halas, Proc. Natl. Acad. Sci. U.S.A. 101, 17930(2004).

1106 Volume 61, Number 10, 2007