Proc. Nati. Acad. Sci. USA Vol. 74, No. 7, pp. 2639-2643, July 1977
Chemistry
Ultraviolet resonant Raman spectroscopy of nucleic acid components (AMP/excitation profile/ultraviolet Raman)
DANIEL C. BLAZEJ AND WARNER L. PETICOLAS Department of Chemistry, University of Oregon, Eugene, Oregon 97403
Communicated by V. Boekelheide, May 9, 1977
ABSTRACT The first resonant Raman excitation profile using UV as well as visible radiation is presented. Measurements of the intensity of the Raman spectrum of adenosine 5'-monophosphate as a function of the frequency of the incident laser Iight are presented in the range from 20 to 38 kK (1000 cm-'). The scattering intensity per molecule increases by about 105 as the laser is tuned from low to high frequencies. The Raman excitation profile has been calculated by using a simple form of the vibronic theory of Raman scattering. The theoretical curves are found to adequately fit the data using only the frequencies of the excited electronic states of AMP and their corresponding vibronic linewidths as adjustable parameters. The Raman bands at 1484 cm-l and 1583 cm-1 appear to obtain virtually all of their intensity from a weak electronic transition at 276 nm. The set of Raman bands in the range 1300 cm-1-1400 cm-l appear to derive at least part of their intensity from an electronic band whose 0-0 transition is in the 269-259 nm region although the possibility of some intensity arising from the vibronic mixing between these two electronic states cannot as yet be ruled out. A number of workers have recently obtained the resonant Raman spectra of heme proteins, visual pigments, and other molecules of biological interest that have absorption bands in the region 350-650 nm (for recent reviews see refs. 1-4). The limitation in this region is due to the fact that these are the wavelength limits of the argon and krypton gas lasers as well as typical dye lasers. However, many molecules of biological interest such as the aromatic amino acids and the nucleic acid bases have absorption bands that fall in the UV region. Consequently, we wish to present laser Raman data obtained with a dye laser that was frequency doubled into the UV region as the Raman excitation source. The intensity of the Raman spectrum of adenosine 5'-monophosphate as a function of the incident frequency of radiation from 514.5 to 268 nm (19.44-37.29 kK) is presented.
Experimental A schematic of the Raman spectrometer is shown in Fig. 1. The excitation source was a Chromatix CMX-4 pulsed dye laser containing ADP crystals as frequency doublers. The laser was operated with a 30 pulse per second repetition rate in which each pulse was about 1 ,usec long. Average power at the sample was typically 1-5 mW (30-150 microjoules per pulse). Three dyes from the Exciton Co. were used for this study: rhodamine 6G (305-289 nm), fluorol 7Ga (285-272 nm), and coumarin 522 (280-268 nm). The sample compartment and collection optics were somewhat modified from the usual Raman spectrometer although a 900 scattering geometry was maintained. To preserve achromaticity into the UV region, we have used an ellipsoidal mirror to focus the scattered light on the slits. The ellipsoidal mirror had some drawbacks in that it was difficult to align and the image of a finite source was rather distorted. However, because the scattered light emanated from a focused 2639
FIG. 1. Block diagram of Raman spectrometer.
laser beam 1 mm long and a few microns in diameter, the image quality was adequate for transfer into the monochromator. Both polarized and depolarized components of the scattered Raman light were collected. Samples were contained in 1 mm diameter quartz x-ray capillaries and supported in a brass block. The monochromator was a Spex 1400 with gratings blazed at 500 nm which were used in the second order. We assumed that the response of the monochromator at each different excitation wavelength was proportionally the same for the cacodylate band at 608 cm-1 and the AMP vibrations. The Raman light was detected by an ITT F4013 photomultiplier with a sapphire window and the resulting signal was analyzed with a Molectron LP 20 boxcar integrator. This instrument averaged a given number of pulse intensities and calculated a ratio for the sample relative to a reference standard. For most of the spectra either a 100 or 300 pulse average was used. A stock solution of 0.5 M sodium cacodylate, pH 7.1, was used to make all dilutions. As discussed below, the sodium cacodylate was both the buffer and the intensity standard for the Raman intensity measurements. Figs. 2 and 3 are representative spectra of AMP in the UV region at 295 and 267 nm. The higher noise levels at the lower wavelength were due primarily to the low output power of the laser, the pulse to pulse instability, and the more dilute nature of the solutions in the UV region. Even though there is a large relative enhancement of the Raman scattering in an absorption band there are losses due to self-absorption that must be compensated for with lower concentrations. The combination of low concentration and low average laser power (2 to 3 orders of magnitude less than the typical argon laser power levels) effectively results in a measured Raman signal that is still weak by conventional Raman standards. Nevertheless, we were able to measure peak intensities that were reproducible within + 15%. With digitization and data processing, the~e intensities should become much more noise free. On the whole, these resonant Raman spectra resembled those taken with visible laser radiation. However, it was very remarkable that no water bands appeared, as they were very prominent in the nonresonant spectrum.
2640
Proc. Natl. Acad. Sci. USA 74 (1977)
Chemistry: Blazej and Peticolas
1583
FIG. 2. Raman spectrum of 0.01 M AMP/0.25 M sodium cacodylate, pH 7.1. Excitation wavelength was 295 nm.
A major problem with working in the UV region was finding a suitable internal standard. None of the usual salts including phosphate, nitrate, chlorate, or sulfate yielded a measurable Raman signal even at 1 M concentrations. Sodium cacodylate has a very strong vibrational band at 608 cm-1 apparently due to some resonance enhancement from its absorption band at 53 kK. Peak height intensities were measured relative to the cacodylate 608-cm' line and standardized for the concentration by assuming that the Raman intensity varied linearly with the concentration. The AMP concentration ranged from 0.5 mM at 268 nm to 0.05 M at 514.5 nm. The Raman spectrum of AMP/cacodylate solutions was recorded at a number of different laser frequencies and peak height intensities were measured. Due to the large deviations in laser power and detection efficiencies over the frequency range of 18 kK it was necessary to use an internal standard. The 608 cm-1 band of cacodylate was very convenient because it was quite strong and did not interfere with the AMP spectrum. We assumed that the instrument response was linear over the range of a single scan which was less than 1000 cm-1 (less than 10 nm at 300 nm). Because the concentrations of both species were varied to obtain optimum spectra, a normalized relative intensity was calculated for the jth AMP band by
Rj(WL) =
Icac
[1]
= ij[caci/icadlAMP]
in which i is the measured intensity and we assumed scattering was linear with the concentration. We observed some measurable increase in the cacodylate band intensity with increasing laser frequency that was almost certainly due to a preresonance Raman effect. The measured intensity of the 608 cm-l cacodylate band was corrected for this preresonance effect by dividing the intensity by the energy denominator of the theoretical expression to be described below. Referencing each AMP Raman band intensity to the corrected 608
FIG. 3. Raman spectrum of 0.5 mM AMP/0.5 M sodium cacodylate, pH 7.1. Excitation wavelength was 267 nm.
standard yields intensity values that should be proportional to the absolute square of the Raman scattering tensor for the given AMP vibration. Each of these intensity ratios, Rj, was then calculated relative to its corresponding value with the 514.5 nm line of the argon laser. Thus, every Raman excitation curve started out at unity at 514.5 nm and increased with increasing laser frequency to 268 nm. This fixed the scale for each vibration and allowed a direct comparison of the gain in the Raman intensities with the increasing laser frequency. As we shall see, in going from the low to the high frequencies some Raman bands gained intensity considerably faster than others, a fact that allowed us to discriminate between the various excited electronic states from which these vibrations derived their intensity. Discussion of experimental results The vibronic theory of Raman scattering has been discussed by numerous -authors (1-16). In this theory, the Raman scattering tensor, apa, may be separated into a zeroth order term, first obtained by Tang and Albrecht (5) and called the A term, and additional B terms, C terms, etc. that contain vibronic mixing of two excited electronic states and may contain nonadiabatic corrections. However, because all of the vibrations described here possess a depolarization ratio less than 0.75, we will assume that they are all totally symmetric (i.e., in the plane of the adenine base) and use only the A term for the rigorous region and the A' term (sometimes called the trace B term) for the preresonance region (5, 6, 11). The A term is given by E Apa = ev
(I
efea(g&ieV) (evigi) mu MP,(gjjev) (evi gi)
Eev- Egi - hWL+ i rev
Ee -Egj + hWL + iFev/
[2]
in which ho.L is a quantum of laser light, hQ is a vibrational quantum, mge is the electronic transition dipole between ground and excited electronic states, polarized in direction p of the scattered light, (ev I gi ) is a Franck-Condon overlap integral between the vth vibrational level of the eth electronic state and the ith vibrational level of the ground state, Eev and Egi are the corresponding vibronic energies, and Fev is the corresponding damping. To use these equations, it is necessary to know EeO, the location of the 0-0 energy level of the eth state. If the exciting laser light falls below the resonance (i.e., absorption) region, then the A term must be replaced by an A' (often called trace B) equation in which there is no longer a summation over the vibrational levels of the resonant electronic state, the energy denominator contains E° (the vertical energy of the resonant state), and Fe is the line width of the whole electronic absorption band. The A' equation is given as
Chemistry: Blazej and Peticolas
c
1,000
sc
Proc. Natl. Acad. Sci. USA 74 (1977)
2641
E O. 10o00oo
,
X
100
.
10
C C
0
I
E
18 20 22 24 26 28 30 32 34 36 38
Frequency, kK FIG. 4. Corrected relative intensity (see text) of the Raman bands of AMP at 1338 cm-' and 1484 cm-' as a function of excitation frequency. The solid line is a theoretical line calculated by assuming the intensity came from an absorption band at 260 nm.
A'pa =
E
e
>
1000-
0-
(1IQIO)
mPeheemeg jfe)(E
(E°- hWL + hQ +
mgeheemeg
-
hWL + iFe) 1
[3] (Ee° + hCL-hQ + ire)(E° + hWL + ire)J This equation contains hee = (OEe/c9Q), the slope of the excited state potential energy curve evaluated at the equilibrium position of the ground state. The A' term is used to correct the measured cacodylate Raman intensities by assuming that Ee for cacodylate is 53,000 cm-1, and that re is negligible compared to E0 - hwL. Over the laser frequencies available, 1938,000 cmt1, this amounts to a calculated increase in the cacodylate intensity of about 15-fold. Thus, whereas the intensity of the AMP vibration relative to the cacodylate standard increases by 104, the total increase of the AMP Raman inten+
0
sities is 105 because the standard itself increases in intensity over this large range of excitation frequency. Fig. 4 shows the corrected relative intensity of the 1338 cm1 and 1484 cm-1 Raman lines of AMP as a function of the excitation laser frequency from 19.44 to 38 kK. The solid line was calculated from Eq. 3 by using the absorption maximum value at 260 nm as the Eeo term and by considering re as negligible, as was done in the only previous resonant Raman study of AMP (16). As may be seen, all of the experimental points fell to the left of the calculated curve. Furthermore, even very small values of re shifted the calculated curve to the right and increased the discrepancy. Attempts to use Eq. 3 in the resonance region failed presumably because the excited electronic states were too greatly shifted along the normal coordinates-for the A' term to be applicable (10). Consequently, we abandoned the A' term as an expression to accurately fit the excitation profile over the whole range and concentrated on the rigorous resonance region from 32 to 38 kK. Additonal calculations are underway to test the validity of the A' equation as the rigorous resonance region is approached. If the shift A4 in the excited state potential energy minimum along a normal coordinate Q is small, then, in the sum over the vibrational levels of the resonant absorption band, the v = 0, 1 terms, must dominate. This is easily seen because as Ae goes to zero, (vIj) = 6,v. Thus, the products (110)(010) and
l
100 32
33
34
l
I
35 36 37 Excitation frequency
38
FIG. 5. Detail of the excitation profile of the 1484 cm-' band of AMP. The solid line was calculated from the A term (Eq. 5) with We = 276 nm, FeO = 0.9 kK, and Fei = 0.6 kK.
(111) (1I0) are linear in A, whereas the products ( IJv) (vI0) are higher order in Ae for v 2 2 (9). Noting that (1gIe0) = -(leIgO) Ae/V2, we may write the A term as -
[EeO- hWL
+
-reo Eel-hWL
+
freil
X MgeMeg [4]
in which Aif indicates a summation only over 0,1 vibrational levels of the eth excited state. Because the intensity goes as A2, the intensity of the A01 term is given by its absolute square
22 M yMA2 2
e
2 +
(rel - FeO)2 1 [5] L[(EeO hOL )2 + reO][(Eel h(OL) e ~] in which Q is the circular vibrational frequency of the Raman r
-
-
active mode. Note that in this formula the Raman scattering intensity is proportional to the square of the displacement of the resonant excited state potential energy curve. For totally symmetric vibrations whose A4 values are small, this Eq. 5 may be more useful than the exact but more cumbersome equations used by others (9, 10). The major problem in trying to calculate a Raman excitation profile for AMP is the fact that the exact location of the electronic states is not known. Molecular orbital calculations (17) show that adenine substituted in the 9-position possesses two 7r-r* (in-plane) electronic transitions that lie in the 250- to 280-nm absorption manifold with its maximum absorption at 259 nm. Although the theoretical calculations place the two 7r-r* electronic states at 278 nm and 257 nm, there is no firm
Proc. Natl. Acad. Sci. USA 74 (1977)
Chemistry: Blazej and Peticolas
2642
Table 1. Calculated vibronic energies and bandwidths of AMP, with corresponding standard deviations
Ql (kK) Ee0, kK (nm) Feo (kK) Eei, kK (nm) Ie, (kK) SD 1.310 1.338 1.380 1.484 1.583
37.2 (269) 37.2 (269) 37.3 (268) 36.2 (276) 36.4 (275)
0.8 0.9 0.9 0.9 1.0
38.5 (260) 38.5 (259) 38.7 (259) 37.7 (265) 38.0 (263)
1.2 1.1 1.1 0.6 0.8
0.131 0.131 0.184 0.065 0.091
1.338 1.484
36.2 (276) 37.2 (269)
0.9 0.9
37.5 (266) 38.7 (259)
0.6 1.1
0.579 0.562
E o
10,000S
C.', C
experimental evidence for their exact location because the observed absorption band is broad and relatively structureless; such structure as exists in the AMP absorption spectrum shows shoulders at 276 nm and 267 nm as well as the maximum at 259 nm (18). Because all of the Raman-active vibrations and the two electronic states are all in-plane, it is not immediately possible to rule out vibronic coupling between the two in-plane electronic levels as a source of the Raman intensity even though all of the Raman bands possess depolarization ratios less than 0.75 and hence are formally totally symmetric. However, we have assumed in this preliminary work that all of the vibrations derive their intensity from the A term as approximated by Eq. 4.
Rather than trying to simply calculate the excitation profile as is customary, we attempted to use the experimentally measured excitation profile to obtain the exact position of the E0o bands for the two electronic levels. Thus, we fitted the experimental profiles for each of five vibrations of AMP with Eq. 5 using Eec, FreO, and reI as variable parameters and obtained lAnru-Iml
0
C C
E
a~1000-
0 ioo IOC
0
100
l
n 33
E
l
l
l
l
37 36 35 Excitation frequency
38
FIG. 7. The excitation profile of the Raman band of AMP at 1338 cm-'. The calculated curve is the A term with We = 269 nm, reo = 0.9 kK, andlrei = 1.1 kK.
the best value of these parameters for each vibration. Because the Raman intensities increased by several orders of magnitude over the incident frequency range, we chose to minimize the difference in the logarithms of the intensities in order to avoid giving too large a weighted value to the high intensity data. The standard deviation is thus given by S[ SD = n
0
34
(log Icalc-log Iexp.)2
112
n-1
10,00O LO
-
a CO 10001
e 32
I
I
33
34
I
I
I
36 37 35 Excitation frequency
I 38
FIG. 6. Detail of the excitation profile of the 1583 cm-' band of -AMP. The line was calculated from the A term (Eq. 5) with we = 275 nm, Feo = 1.0 kK, and rei = 0.8 kK.
in which n is the number of data points. The optimum results for each of five vibrations of AMP are given in the top part of Table 1. The remarkably good agreement between different sets of data indicates that there are two excited states of AMP with their 0-0 transitions occurring at 269 nm and 276 nm. The absorption maximum occurring at 259 nm is presumably dominated by 0-1 transitions of the 1310 to 1380 cm- vibrations and the 0-2 transitions of the 1484 cm-1 and 1583 cmvibrations. As may be seen in Figs. 5 and 6, the agreement between theory and experiment is excellent for the 1484 and 1583 cm-1 vibrations. Furthermore, our analysis confirms the suggestion of Tsuboi et al. (19) that these vibrations derive their intensity from an excited state at 276 nm. Fig. 7 shows the excitation profile for the 1338 cm-1 vibration that is representative of all three modes in the 1340 cm-1 region. The theoretical fit is not as good as in Figs. 4 and 5. Below 34 kK and around 36 kK, there are significant deviations that may exceed the experimental error and may give rise to the larger SD. Consequently, it is possible that vibronic mixing by means
Chemistry: Blazej and Peticolas of a B term occurs between the 276 nm state and a higher electronic state in the range 259-269 nm. The bottom section of Table 1 shows the SD values of two of the vibrations fitted with the parameters of the opposite absorption band. It is quite clear that the calculated SD is affected by the variations of the position of the electronic absorption levels. The rather large values of F', necessary to minimize the error between calculated and observed points in Figs. 5-7 may be justified by the broad, almost structureless absorption and fluorescence band (18) of adenine and its derivatives even in a neon matrix at 4 K (20) and by its extremely weak fluorescence. The excited states of adenine are heavily quenched by nonradiative processes and, in addition, there must be some inhomogeneous broadening probably due to many different types of hydrogen bonding of the adenine with the solvent. Finally, the analysis presented here that was obtained from a least squares minimization of the Raman excitation curve is in remarkable agreement with the small amount of structure observed in the electronic absorption band (18) of AMP. Thus, the tentative interpretation obtained from the Raman data is that the shoulder at 276 nm is the 0-0 transition of the first electronic state, the shoulder at 267 nm is due to an overlap of the 0-1 transition of the first electronic state with the 0-0 of the second, and the maximum at 259 nm is an overlap of the 0-2 and 0-1 levels of the first and second electronic states, respectively. However, standard deviation analysis of the excitation curves for the 1310-1380 cm-1 vibrations is not sufficiently good so that vibronic mixing by means of a B term between the 276 nm state and a second state at a higher frequency (say at 259 nm) cannot as yet be firmly ruled out. We wish to acknowledge the support of the U.S. Public Health Service (Grant GM 15547) and the National Science Foundation (Grant GB 29709).
The costs of publication of this article were defrayed in part by the payment of page charges from funds made available to support the research which is the subject of the article. This article must therefore
Proc. Natl. Acad. Sci. USA 74 (1977)
2643
be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact. 1. Spiro, T. C. (1974) Accounts of Chemical Research, 7, 339344. 2. Behringer, J. (1975) Mol. Spectrosc. 3, 163-280. 3. Bernstein, H. J. (1976) Mol. Spectrosc., in press. 4. Johnson, B. B. & Peticolas, W. L. (1976) Annu. Rev. Phys. Chem.
27,465-491. 5. Tang, J. & Albrecht, A. C. (1970) Raman Spectrosc., 2, 33-68. 6. Peticolas, W. L., Nafie, L., Stein, P. & Fanconi, B. (1970) J. Chem. Phys. 52, 1576-1584. 7. Friedman, J. M. & Hochstrasser, R. M. (1973) Chem. Phys. 1, 457-467. 8. Warshel, A. & Karplus, N. (1972) J. Am. Chem. Soc. 94, 5612-5624. 9. Inagaki, F., Tasumi, M. & Miyazawa, T. (1974) J. Mol. Spectrosc. 50,286-303. 10. Garozzo, M. & Galluzzi, F. (1976) J. Chem. Phys. 64, 15761584. 11. Stein, P., Miskowski, V., Woodruff, W. H., Griffin, J. P., Werner, K. G., Gaber, J. P. & Spiro, T. C. (1976) J. Chem. Phys. 64, 2159-2167. 12. Nafie, L. A., Pastor, R. W., Dabrowiak, J. C. & Woodruff, W. H. (1976) J. Am. Chem. Soc., 98,8007-8014. 13. Mingardi, M. & Siebrand, W. (1975) J. Chem. Phys. 62, 1074-1085. 14. Shelnutt, J. A., O'Shea, D. C., Yu, N. T., Cheung, L. D. & Felton, R. M. (1976) J. Chem. Phys. 64, 1156-1165. 15. Johnson, F., Nafie, L. & Peticolas, W. L. (1977) Chem. Phys. 19, 303-311. 16. Pezolet, M., Yu, T. J. & Peticolas, W. L. (1975) J. Raman Spectrosc. 3, 55-64. 17. Tanaka, M. & Nagakwa, S. (1966) Theor. Chim. Acta 6, 320332. 18. Gueron, M., Eisenger, J. & Lamola, A. A. (1974) Basic Principles of Nucleic Acid Chemistry (Academic Press, New York), pp. 311-342. 19. Tsuboi, M., Hirakawa, A. Y., Nishimira, Y. & Harada, I. (1974)
J. Raman Spectrosc. 2, 609-621. 20. Tomlinson, B. L. (1968) Ph.D. Dissertation, University of Cali-
fornia.
Chemistry
Ultraviolet resonant Raman spectroscopy of nucleic acid components (AMP/excitation profile/ultraviolet Raman)
DANIEL C. BLAZEJ AND WARNER L. PETICOLAS Department of Chemistry, University of Oregon, Eugene, Oregon 97403
Communicated by V. Boekelheide, May 9, 1977
ABSTRACT The first resonant Raman excitation profile using UV as well as visible radiation is presented. Measurements of the intensity of the Raman spectrum of adenosine 5'-monophosphate as a function of the frequency of the incident laser Iight are presented in the range from 20 to 38 kK (1000 cm-'). The scattering intensity per molecule increases by about 105 as the laser is tuned from low to high frequencies. The Raman excitation profile has been calculated by using a simple form of the vibronic theory of Raman scattering. The theoretical curves are found to adequately fit the data using only the frequencies of the excited electronic states of AMP and their corresponding vibronic linewidths as adjustable parameters. The Raman bands at 1484 cm-l and 1583 cm-1 appear to obtain virtually all of their intensity from a weak electronic transition at 276 nm. The set of Raman bands in the range 1300 cm-1-1400 cm-l appear to derive at least part of their intensity from an electronic band whose 0-0 transition is in the 269-259 nm region although the possibility of some intensity arising from the vibronic mixing between these two electronic states cannot as yet be ruled out. A number of workers have recently obtained the resonant Raman spectra of heme proteins, visual pigments, and other molecules of biological interest that have absorption bands in the region 350-650 nm (for recent reviews see refs. 1-4). The limitation in this region is due to the fact that these are the wavelength limits of the argon and krypton gas lasers as well as typical dye lasers. However, many molecules of biological interest such as the aromatic amino acids and the nucleic acid bases have absorption bands that fall in the UV region. Consequently, we wish to present laser Raman data obtained with a dye laser that was frequency doubled into the UV region as the Raman excitation source. The intensity of the Raman spectrum of adenosine 5'-monophosphate as a function of the incident frequency of radiation from 514.5 to 268 nm (19.44-37.29 kK) is presented.
Experimental A schematic of the Raman spectrometer is shown in Fig. 1. The excitation source was a Chromatix CMX-4 pulsed dye laser containing ADP crystals as frequency doublers. The laser was operated with a 30 pulse per second repetition rate in which each pulse was about 1 ,usec long. Average power at the sample was typically 1-5 mW (30-150 microjoules per pulse). Three dyes from the Exciton Co. were used for this study: rhodamine 6G (305-289 nm), fluorol 7Ga (285-272 nm), and coumarin 522 (280-268 nm). The sample compartment and collection optics were somewhat modified from the usual Raman spectrometer although a 900 scattering geometry was maintained. To preserve achromaticity into the UV region, we have used an ellipsoidal mirror to focus the scattered light on the slits. The ellipsoidal mirror had some drawbacks in that it was difficult to align and the image of a finite source was rather distorted. However, because the scattered light emanated from a focused 2639
FIG. 1. Block diagram of Raman spectrometer.
laser beam 1 mm long and a few microns in diameter, the image quality was adequate for transfer into the monochromator. Both polarized and depolarized components of the scattered Raman light were collected. Samples were contained in 1 mm diameter quartz x-ray capillaries and supported in a brass block. The monochromator was a Spex 1400 with gratings blazed at 500 nm which were used in the second order. We assumed that the response of the monochromator at each different excitation wavelength was proportionally the same for the cacodylate band at 608 cm-1 and the AMP vibrations. The Raman light was detected by an ITT F4013 photomultiplier with a sapphire window and the resulting signal was analyzed with a Molectron LP 20 boxcar integrator. This instrument averaged a given number of pulse intensities and calculated a ratio for the sample relative to a reference standard. For most of the spectra either a 100 or 300 pulse average was used. A stock solution of 0.5 M sodium cacodylate, pH 7.1, was used to make all dilutions. As discussed below, the sodium cacodylate was both the buffer and the intensity standard for the Raman intensity measurements. Figs. 2 and 3 are representative spectra of AMP in the UV region at 295 and 267 nm. The higher noise levels at the lower wavelength were due primarily to the low output power of the laser, the pulse to pulse instability, and the more dilute nature of the solutions in the UV region. Even though there is a large relative enhancement of the Raman scattering in an absorption band there are losses due to self-absorption that must be compensated for with lower concentrations. The combination of low concentration and low average laser power (2 to 3 orders of magnitude less than the typical argon laser power levels) effectively results in a measured Raman signal that is still weak by conventional Raman standards. Nevertheless, we were able to measure peak intensities that were reproducible within + 15%. With digitization and data processing, the~e intensities should become much more noise free. On the whole, these resonant Raman spectra resembled those taken with visible laser radiation. However, it was very remarkable that no water bands appeared, as they were very prominent in the nonresonant spectrum.
2640
Proc. Natl. Acad. Sci. USA 74 (1977)
Chemistry: Blazej and Peticolas
1583
FIG. 2. Raman spectrum of 0.01 M AMP/0.25 M sodium cacodylate, pH 7.1. Excitation wavelength was 295 nm.
A major problem with working in the UV region was finding a suitable internal standard. None of the usual salts including phosphate, nitrate, chlorate, or sulfate yielded a measurable Raman signal even at 1 M concentrations. Sodium cacodylate has a very strong vibrational band at 608 cm-1 apparently due to some resonance enhancement from its absorption band at 53 kK. Peak height intensities were measured relative to the cacodylate 608-cm' line and standardized for the concentration by assuming that the Raman intensity varied linearly with the concentration. The AMP concentration ranged from 0.5 mM at 268 nm to 0.05 M at 514.5 nm. The Raman spectrum of AMP/cacodylate solutions was recorded at a number of different laser frequencies and peak height intensities were measured. Due to the large deviations in laser power and detection efficiencies over the frequency range of 18 kK it was necessary to use an internal standard. The 608 cm-1 band of cacodylate was very convenient because it was quite strong and did not interfere with the AMP spectrum. We assumed that the instrument response was linear over the range of a single scan which was less than 1000 cm-1 (less than 10 nm at 300 nm). Because the concentrations of both species were varied to obtain optimum spectra, a normalized relative intensity was calculated for the jth AMP band by
Rj(WL) =
Icac
[1]
= ij[caci/icadlAMP]
in which i is the measured intensity and we assumed scattering was linear with the concentration. We observed some measurable increase in the cacodylate band intensity with increasing laser frequency that was almost certainly due to a preresonance Raman effect. The measured intensity of the 608 cm-l cacodylate band was corrected for this preresonance effect by dividing the intensity by the energy denominator of the theoretical expression to be described below. Referencing each AMP Raman band intensity to the corrected 608
FIG. 3. Raman spectrum of 0.5 mM AMP/0.5 M sodium cacodylate, pH 7.1. Excitation wavelength was 267 nm.
standard yields intensity values that should be proportional to the absolute square of the Raman scattering tensor for the given AMP vibration. Each of these intensity ratios, Rj, was then calculated relative to its corresponding value with the 514.5 nm line of the argon laser. Thus, every Raman excitation curve started out at unity at 514.5 nm and increased with increasing laser frequency to 268 nm. This fixed the scale for each vibration and allowed a direct comparison of the gain in the Raman intensities with the increasing laser frequency. As we shall see, in going from the low to the high frequencies some Raman bands gained intensity considerably faster than others, a fact that allowed us to discriminate between the various excited electronic states from which these vibrations derived their intensity. Discussion of experimental results The vibronic theory of Raman scattering has been discussed by numerous -authors (1-16). In this theory, the Raman scattering tensor, apa, may be separated into a zeroth order term, first obtained by Tang and Albrecht (5) and called the A term, and additional B terms, C terms, etc. that contain vibronic mixing of two excited electronic states and may contain nonadiabatic corrections. However, because all of the vibrations described here possess a depolarization ratio less than 0.75, we will assume that they are all totally symmetric (i.e., in the plane of the adenine base) and use only the A term for the rigorous region and the A' term (sometimes called the trace B term) for the preresonance region (5, 6, 11). The A term is given by E Apa = ev
(I
efea(g&ieV) (evigi) mu MP,(gjjev) (evi gi)
Eev- Egi - hWL+ i rev
Ee -Egj + hWL + iFev/
[2]
in which ho.L is a quantum of laser light, hQ is a vibrational quantum, mge is the electronic transition dipole between ground and excited electronic states, polarized in direction p of the scattered light, (ev I gi ) is a Franck-Condon overlap integral between the vth vibrational level of the eth electronic state and the ith vibrational level of the ground state, Eev and Egi are the corresponding vibronic energies, and Fev is the corresponding damping. To use these equations, it is necessary to know EeO, the location of the 0-0 energy level of the eth state. If the exciting laser light falls below the resonance (i.e., absorption) region, then the A term must be replaced by an A' (often called trace B) equation in which there is no longer a summation over the vibrational levels of the resonant electronic state, the energy denominator contains E° (the vertical energy of the resonant state), and Fe is the line width of the whole electronic absorption band. The A' equation is given as
Chemistry: Blazej and Peticolas
c
1,000
sc
Proc. Natl. Acad. Sci. USA 74 (1977)
2641
E O. 10o00oo
,
X
100
.
10
C C
0
I
E
18 20 22 24 26 28 30 32 34 36 38
Frequency, kK FIG. 4. Corrected relative intensity (see text) of the Raman bands of AMP at 1338 cm-' and 1484 cm-' as a function of excitation frequency. The solid line is a theoretical line calculated by assuming the intensity came from an absorption band at 260 nm.
A'pa =
E
e
>
1000-
0-
(1IQIO)
mPeheemeg jfe)(E
(E°- hWL + hQ +
mgeheemeg
-
hWL + iFe) 1
[3] (Ee° + hCL-hQ + ire)(E° + hWL + ire)J This equation contains hee = (OEe/c9Q), the slope of the excited state potential energy curve evaluated at the equilibrium position of the ground state. The A' term is used to correct the measured cacodylate Raman intensities by assuming that Ee for cacodylate is 53,000 cm-1, and that re is negligible compared to E0 - hwL. Over the laser frequencies available, 1938,000 cmt1, this amounts to a calculated increase in the cacodylate intensity of about 15-fold. Thus, whereas the intensity of the AMP vibration relative to the cacodylate standard increases by 104, the total increase of the AMP Raman inten+
0
sities is 105 because the standard itself increases in intensity over this large range of excitation frequency. Fig. 4 shows the corrected relative intensity of the 1338 cm1 and 1484 cm-1 Raman lines of AMP as a function of the excitation laser frequency from 19.44 to 38 kK. The solid line was calculated from Eq. 3 by using the absorption maximum value at 260 nm as the Eeo term and by considering re as negligible, as was done in the only previous resonant Raman study of AMP (16). As may be seen, all of the experimental points fell to the left of the calculated curve. Furthermore, even very small values of re shifted the calculated curve to the right and increased the discrepancy. Attempts to use Eq. 3 in the resonance region failed presumably because the excited electronic states were too greatly shifted along the normal coordinates-for the A' term to be applicable (10). Consequently, we abandoned the A' term as an expression to accurately fit the excitation profile over the whole range and concentrated on the rigorous resonance region from 32 to 38 kK. Additonal calculations are underway to test the validity of the A' equation as the rigorous resonance region is approached. If the shift A4 in the excited state potential energy minimum along a normal coordinate Q is small, then, in the sum over the vibrational levels of the resonant absorption band, the v = 0, 1 terms, must dominate. This is easily seen because as Ae goes to zero, (vIj) = 6,v. Thus, the products (110)(010) and
l
100 32
33
34
l
I
35 36 37 Excitation frequency
38
FIG. 5. Detail of the excitation profile of the 1484 cm-' band of AMP. The solid line was calculated from the A term (Eq. 5) with We = 276 nm, FeO = 0.9 kK, and Fei = 0.6 kK.
(111) (1I0) are linear in A, whereas the products ( IJv) (vI0) are higher order in Ae for v 2 2 (9). Noting that (1gIe0) = -(leIgO) Ae/V2, we may write the A term as -
[EeO- hWL
+
-reo Eel-hWL
+
freil
X MgeMeg [4]
in which Aif indicates a summation only over 0,1 vibrational levels of the eth excited state. Because the intensity goes as A2, the intensity of the A01 term is given by its absolute square
22 M yMA2 2
e
2 +
(rel - FeO)2 1 [5] L[(EeO hOL )2 + reO][(Eel h(OL) e ~] in which Q is the circular vibrational frequency of the Raman r
-
-
active mode. Note that in this formula the Raman scattering intensity is proportional to the square of the displacement of the resonant excited state potential energy curve. For totally symmetric vibrations whose A4 values are small, this Eq. 5 may be more useful than the exact but more cumbersome equations used by others (9, 10). The major problem in trying to calculate a Raman excitation profile for AMP is the fact that the exact location of the electronic states is not known. Molecular orbital calculations (17) show that adenine substituted in the 9-position possesses two 7r-r* (in-plane) electronic transitions that lie in the 250- to 280-nm absorption manifold with its maximum absorption at 259 nm. Although the theoretical calculations place the two 7r-r* electronic states at 278 nm and 257 nm, there is no firm
Proc. Natl. Acad. Sci. USA 74 (1977)
Chemistry: Blazej and Peticolas
2642
Table 1. Calculated vibronic energies and bandwidths of AMP, with corresponding standard deviations
Ql (kK) Ee0, kK (nm) Feo (kK) Eei, kK (nm) Ie, (kK) SD 1.310 1.338 1.380 1.484 1.583
37.2 (269) 37.2 (269) 37.3 (268) 36.2 (276) 36.4 (275)
0.8 0.9 0.9 0.9 1.0
38.5 (260) 38.5 (259) 38.7 (259) 37.7 (265) 38.0 (263)
1.2 1.1 1.1 0.6 0.8
0.131 0.131 0.184 0.065 0.091
1.338 1.484
36.2 (276) 37.2 (269)
0.9 0.9
37.5 (266) 38.7 (259)
0.6 1.1
0.579 0.562
E o
10,000S
C.', C
experimental evidence for their exact location because the observed absorption band is broad and relatively structureless; such structure as exists in the AMP absorption spectrum shows shoulders at 276 nm and 267 nm as well as the maximum at 259 nm (18). Because all of the Raman-active vibrations and the two electronic states are all in-plane, it is not immediately possible to rule out vibronic coupling between the two in-plane electronic levels as a source of the Raman intensity even though all of the Raman bands possess depolarization ratios less than 0.75 and hence are formally totally symmetric. However, we have assumed in this preliminary work that all of the vibrations derive their intensity from the A term as approximated by Eq. 4.
Rather than trying to simply calculate the excitation profile as is customary, we attempted to use the experimentally measured excitation profile to obtain the exact position of the E0o bands for the two electronic levels. Thus, we fitted the experimental profiles for each of five vibrations of AMP with Eq. 5 using Eec, FreO, and reI as variable parameters and obtained lAnru-Iml
0
C C
E
a~1000-
0 ioo IOC
0
100
l
n 33
E
l
l
l
l
37 36 35 Excitation frequency
38
FIG. 7. The excitation profile of the Raman band of AMP at 1338 cm-'. The calculated curve is the A term with We = 269 nm, reo = 0.9 kK, andlrei = 1.1 kK.
the best value of these parameters for each vibration. Because the Raman intensities increased by several orders of magnitude over the incident frequency range, we chose to minimize the difference in the logarithms of the intensities in order to avoid giving too large a weighted value to the high intensity data. The standard deviation is thus given by S[ SD = n
0
34
(log Icalc-log Iexp.)2
112
n-1
10,00O LO
-
a CO 10001
e 32
I
I
33
34
I
I
I
36 37 35 Excitation frequency
I 38
FIG. 6. Detail of the excitation profile of the 1583 cm-' band of -AMP. The line was calculated from the A term (Eq. 5) with we = 275 nm, Feo = 1.0 kK, and rei = 0.8 kK.
in which n is the number of data points. The optimum results for each of five vibrations of AMP are given in the top part of Table 1. The remarkably good agreement between different sets of data indicates that there are two excited states of AMP with their 0-0 transitions occurring at 269 nm and 276 nm. The absorption maximum occurring at 259 nm is presumably dominated by 0-1 transitions of the 1310 to 1380 cm- vibrations and the 0-2 transitions of the 1484 cm-1 and 1583 cmvibrations. As may be seen in Figs. 5 and 6, the agreement between theory and experiment is excellent for the 1484 and 1583 cm-1 vibrations. Furthermore, our analysis confirms the suggestion of Tsuboi et al. (19) that these vibrations derive their intensity from an excited state at 276 nm. Fig. 7 shows the excitation profile for the 1338 cm-1 vibration that is representative of all three modes in the 1340 cm-1 region. The theoretical fit is not as good as in Figs. 4 and 5. Below 34 kK and around 36 kK, there are significant deviations that may exceed the experimental error and may give rise to the larger SD. Consequently, it is possible that vibronic mixing by means
Chemistry: Blazej and Peticolas of a B term occurs between the 276 nm state and a higher electronic state in the range 259-269 nm. The bottom section of Table 1 shows the SD values of two of the vibrations fitted with the parameters of the opposite absorption band. It is quite clear that the calculated SD is affected by the variations of the position of the electronic absorption levels. The rather large values of F', necessary to minimize the error between calculated and observed points in Figs. 5-7 may be justified by the broad, almost structureless absorption and fluorescence band (18) of adenine and its derivatives even in a neon matrix at 4 K (20) and by its extremely weak fluorescence. The excited states of adenine are heavily quenched by nonradiative processes and, in addition, there must be some inhomogeneous broadening probably due to many different types of hydrogen bonding of the adenine with the solvent. Finally, the analysis presented here that was obtained from a least squares minimization of the Raman excitation curve is in remarkable agreement with the small amount of structure observed in the electronic absorption band (18) of AMP. Thus, the tentative interpretation obtained from the Raman data is that the shoulder at 276 nm is the 0-0 transition of the first electronic state, the shoulder at 267 nm is due to an overlap of the 0-1 transition of the first electronic state with the 0-0 of the second, and the maximum at 259 nm is an overlap of the 0-2 and 0-1 levels of the first and second electronic states, respectively. However, standard deviation analysis of the excitation curves for the 1310-1380 cm-1 vibrations is not sufficiently good so that vibronic mixing by means of a B term between the 276 nm state and a second state at a higher frequency (say at 259 nm) cannot as yet be firmly ruled out. We wish to acknowledge the support of the U.S. Public Health Service (Grant GM 15547) and the National Science Foundation (Grant GB 29709).
The costs of publication of this article were defrayed in part by the payment of page charges from funds made available to support the research which is the subject of the article. This article must therefore
Proc. Natl. Acad. Sci. USA 74 (1977)
2643
be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact. 1. Spiro, T. C. (1974) Accounts of Chemical Research, 7, 339344. 2. Behringer, J. (1975) Mol. Spectrosc. 3, 163-280. 3. Bernstein, H. J. (1976) Mol. Spectrosc., in press. 4. Johnson, B. B. & Peticolas, W. L. (1976) Annu. Rev. Phys. Chem.
27,465-491. 5. Tang, J. & Albrecht, A. C. (1970) Raman Spectrosc., 2, 33-68. 6. Peticolas, W. L., Nafie, L., Stein, P. & Fanconi, B. (1970) J. Chem. Phys. 52, 1576-1584. 7. Friedman, J. M. & Hochstrasser, R. M. (1973) Chem. Phys. 1, 457-467. 8. Warshel, A. & Karplus, N. (1972) J. Am. Chem. Soc. 94, 5612-5624. 9. Inagaki, F., Tasumi, M. & Miyazawa, T. (1974) J. Mol. Spectrosc. 50,286-303. 10. Garozzo, M. & Galluzzi, F. (1976) J. Chem. Phys. 64, 15761584. 11. Stein, P., Miskowski, V., Woodruff, W. H., Griffin, J. P., Werner, K. G., Gaber, J. P. & Spiro, T. C. (1976) J. Chem. Phys. 64, 2159-2167. 12. Nafie, L. A., Pastor, R. W., Dabrowiak, J. C. & Woodruff, W. H. (1976) J. Am. Chem. Soc., 98,8007-8014. 13. Mingardi, M. & Siebrand, W. (1975) J. Chem. Phys. 62, 1074-1085. 14. Shelnutt, J. A., O'Shea, D. C., Yu, N. T., Cheung, L. D. & Felton, R. M. (1976) J. Chem. Phys. 64, 1156-1165. 15. Johnson, F., Nafie, L. & Peticolas, W. L. (1977) Chem. Phys. 19, 303-311. 16. Pezolet, M., Yu, T. J. & Peticolas, W. L. (1975) J. Raman Spectrosc. 3, 55-64. 17. Tanaka, M. & Nagakwa, S. (1966) Theor. Chim. Acta 6, 320332. 18. Gueron, M., Eisenger, J. & Lamola, A. A. (1974) Basic Principles of Nucleic Acid Chemistry (Academic Press, New York), pp. 311-342. 19. Tsuboi, M., Hirakawa, A. Y., Nishimira, Y. & Harada, I. (1974)
J. Raman Spectrosc. 2, 609-621. 20. Tomlinson, B. L. (1968) Ph.D. Dissertation, University of Cali-
fornia.
