Identification of Stimulant Drugs by Surface-Enhanced Raman Spectrometry on Colloidal Silver

Vibrational Spectroscopy, 2 (1991) 145-154 145 Elsevier Science Publishers B.V., Amsterdam

Identification of stimulant drugs by surface-enhanced Raman spectrometry on colloidal silver

A. RupCrez, R. Montes and J.J. Lasema * Department of Physical Chemistry and Department of Analytical Chemistry, Faculty of Sciences, Uniuersity of Ma’laga, 29071 Mrilaga (Spain) (Received 2nd February 1991)

Abstract

The surface-enhanced Raman (SER) spectra of stimulant drugs, including mefenorex, pentylenetetrazole, L- amphetamine and pemoline, were obtained on colloidal silver. Silver colloids are prepared in a single step, at room temperature, by chemical reduction of Ag+ with sodium tetrahydroborate. Spectra are recorded using drug concentrations at the pg ml-’ level. Individual drugs can be identified by characteristic vibrational bands. The SER spectrum of human urine and that of human urine spiked with mixtures of stimulant drugs are reported.

Keywords: Raman spectrometry; Amphetamine; Mefenorex; Pemoline; Pentylenetetrazole; Pharmaceuticals; Sports doping; Stimulant drugs

The increase over the last 30 years in the use In recent years, a number of papers have been of drugs in sport (doping) has led the Intema- published demonstrating the detectability of low tional Olympic Committee (IOC) to establish a molecule concentrations (< lop4 M) by adsorp- list of banned substances and to introduce formal tion on rough metal surfaces and subsequent drug testing. The doping classification includes Raman scattering [surface-enhanced Raman scat- stimulants as one of the five regulated categories. tering (SERS)]. Raman spectra of substances at Stimulants include various types of drugs which trace concentrations are readily obtained [6-l 11, increase alertness, competitiveness and hostility even in the presence of a strong luminescence and reduce fatigue. Stimulants also produce a background. Luminescence from adsorbed species false sense of ability and may cause a loss of is generally quenched by radiationless energy judgement, leading to accidents within sports. transfer to the metal surface [12-141. The first These include irregular heart rate, which may enhancement of Raman signals by adsorption on lead to cardiac arrest, the development of para- a rough metal surface was observed in 1974 [15], noid psychosis, a rise in body temperature, which the SER spectrum of pyridine on electrochemi- can induce dehydration in warm weather, and the tally roughened silver electrodes being obtained. masking of fatigue, which can lead to circulatory Since then, different metals have been reported collapse, heart exhaustion and stroke. to be SER-active, including silver, gold, copper The determination of stimulants in urine (the and, to a lesser extent, platinum, nickel, indium sample of choice in doping control) is accom- and others. Diverse roughening procedures have plished using a variety of techniques including gas been also investigated for the production of ac- chromatography [l], liquid chromatography [2,3] tive surfaces 116,171. Excitation with red wave- and gas chromatography mass spectrometry [4,5]. lengths is a requirement for large enhancement

0924-2031/91/$03.50 0 1991 - Elsevier Science Publishers B.V. All rights reserved 146 A.RUPkREZETAL. factors using gold and copper as substrates [18,191. mineralized water was used throughout. Mefeno- However, most visible wavelengths can be used rex, pentylenetetrazole, L-amphetamine and pe- with silver. For chemical characterization, col- moline were purchased from Sigma. Stock aque- loidal silver seems to be the most amenable sub- ous solutions (100 pg ml-‘) contained about 5% strate in terms of production, storage and sample methanol to help dissolve the drugs. Silver hydro- handling, although difficulties with the repro- sols were prepared at room temperature in test ducibility of SERS intensities have been reported tubes by adding 1 ml of 1.0 x 10e3 M AgNO, to La. 3 ml of 2.0 x 10e3 M NaBH,. Hydrosols so pre- A few papers [21,22] on the applicability of the pared showed an absorption maximum at 380 nm SERS technique to biological samples are avail- with no other spectral features in the long-wave- able. This whole demonstrates the analytical use- length side of this band. Each sample was pre- fulness of SERS for the identification of drugs in pared by adding 0.1 ml of the drug solution to 3 human urine samples using a silver hydrosol as a ml of the silver colloid at room temperature. substrate. The SER spectrum of urine and those After mixing, the resulting test solution was of urine spiked with four stimulants were ob- transferred into the liquid cell and placed in the tained. The procedure is easy to perform and the sample chamber of the spectrometer. The SER drugs can be readily identified by characteristic spectra of urine were obtained by spiking 0.1 ml bands in the SER spectrum of spiked urine. of unfiltered urine with 0.1 ml of drug solution and adding the resulting sample to 3 ml of silver colloid at room temperature. Urine samples were EXPERIMENTAL taken from healthy human subjects.

Instrumentation The Raman system has been described in a RESULTS AND DISCUSSION previous paper [231. Basically, it consisted of a helium-neon laser (Siemens, Model LGK 76268) The SER spectra of mefenorex, pentylenete- tuned at 632.8 nm. The power at the sample was trazole, L-amphetamine and pemoline on silver typically 30 mW. Spectra were obtained with a colloid are shown in Fig. 1. Table 1 summarizes 0.22-m double-grating spectrometer (Spex Indus- the vibrational bands of these substances and tries, Model 1680B). The Raman scattering was shows selective spectral features corresponding to collected at right-angles and detected with a ther- the presence of different heteroatoms and sub- moelectrically cooled photomultiplier tube stituent groups on the ring systems. The peak (Hamamatsu, Model R928) and a photon count- positions of the conventional Raman (CR) spec- ing system (Stanford Research, Model SR 400). tra of the pure drugs (liquid state for L- Data acquisition, storage, processing and plotting amphetamine and solid state for the others) also were controlled by an IBM AT compatible micro- are listed in Table 1. computer. All spectra reported represent single The SER spectrum of mefenorex shows two scans and are provided with spectral smoothing medium bands at 990 and 651 cm-i and a weak of 11 or 17 points (Savitzky-Golay algorithm). mode at 1580 cm-‘, which appears in the CR The time constant of the detection electronics spectrum at 1598 cm-‘. The band at 651 cm-‘, was 0.3 s and each spectrum consisted of 968 data also appearing in the CR spectrum, can be at- points. Frequencies were accurate to within 3 tributed to the interaction of bending mode of cm-’ for strong, sharp bands. The spectrometer the carbon atoms in positions 1 and 4 with the resolution was generally set to 14 cm-‘. stretching of the bond connecting the substituent to the ring [24]. The C-X distance increases as Chemicals and procedure the distance between C-l and C-4 decreases. The All chemicals were of analytical-reagent grade band at 990 cm-’ corresponds to in-plane ring or equivalent, and were used as received. De- deformation. 147 SERS OF STIMULANT DRUGS

- CH,CH(CH,)NH(CH,),CI / Q-

a00 ,200 Raman shift (cm-‘)

Raman shift (cm-‘)

L-omphetamlne

1 I 800 lma Isc4 Roman shift (cm-‘) Fig. 1. Surface-enhanced Raman spectra of stimulant drugs on colloidal silver. Drug concentration, 25 fig ml-’ on the measured sample. 148 A. RUPl?REZ ET AL.

The SER spectrum of pentylenetetrazole shows appears at 1580 cm-‘, probably due to the pres- a medium band at 1029 cm-’ and weak bands at ence of the C=N bond. Pemoline also exhibits a 674, 950, 1095, 1130, 1150 and 1584 cm-‘. Sev- medium band at 658 cm-’ and a strong band at eral other weak vibrational modes occur between 830 cm-‘. 1150 and 1550 cm-r and a medium band at 1640 The CR spectrum of mefenorex (Fig. 2) shows cm-‘. a strong band at 828 cm-’ which appear in the The SER spectrum of L-amphetamine shows SER spectrum as a weak band at 820 cm-‘. The the characteristic vibrational modes of the ben- medium band at 1379 cm- ’ could be due to zene ring at 990 and 1012 cm-’ and two modes at symmetric CH, deformation. Also, it exhibits 1112 and 1174 cm-‘, which may be due to the medium bands at 651, 1165 and 1316 cm-‘. Fig- primary amine group with a tertiary a-carbon ure 2c shows the CR spectrum of pentylenetetra- atom. A medium band at 1645 cm- ’ may be due zole, with the most prominent peaks at 1650 and to NH, bending. 661 cm-‘. Several weak bands occur in the spec- The SER spectrum of pemoline exhibits abun- tral zone examined. The weak SER band at 1385 dant spectral features. Medium bands at 1014 cm-’ appears in the CR spectrum at 1370 cm-‘. and 1047 cm-’ are observed. The medium band L-amphetamine (Fig. 2) shows a strong peak at at 1170 cm-’ and the very strong peak at 1640 990 cm-’ with a shoulder at 1012 cm-’ and weak cm-’ may be due to C-N stretching of the amine bands at 456, 493 and 821 cm-‘. These bands group [25] and NH, bending, respectively. The appear as weak vibrational modes in the SER weak band at 1745 cm-’ may be attributed to spectrum at similar Raman shifts. The other bands carbonyl group stretching. Another strong band in the CR spectrum also appear in the SER

TABLE 1 Observed surface-enhanced Raman shifts (SERS, cm-‘) on colloidal silver, conventional Raman shifts (CRS, cm-‘) and relative peak intensities (in parentheses, on a semiquantitative scale from 1 to 10) of stimulant drugs

L-Amphetamine Pemoline Pentylenetetrazole Mefenorex

SERS CRS SERS CRS SERS CRS SERS CRS 453 (1) 456 (2) 530 (1) 573 (1) 595 (1) 628 (1) 651(4) 651 (4) 491(l) 493 (2) 619 (1) 646 (4) 674 (3) 661(5) 745 (2) 747 (2) 821(l) 670 (3) 658 (4) 670 (1) 749 (1) 779 (1) 820 (1) 828 (6) 910 (1) 742 (4) 774 (2) 833 (7) 826 (1) 826 (2) 990 (5) 917 (3) 959 (1) 821 (6) 830 (7) 911 (3) 904 (2) 1030 (2) 1174 (1) 1165 (4) 990 (2) 990 (10) 895 (3) 955 (3) 950 (2) 1086 (3) 1360 (1) 1316 (4) 1012 (2) 1012 (5) 920 (3) 1097 (1) 1029 (5) 1172 (2) 1580 (2) 1379 (5) 1112 (2) 1174 (1) 1014 (5) 1170 (5) 1095 (3) 1306 (2) 1640 (7) 1521 (1) 1174 (3) 1204 (3) 1047 (4) 1187 (1) 1130 (2) 1370 (2) 1598 (9) 1257 (1) 1291(l) 1153 (5) 1221 (1) 1150 (3) 1446 (3) 1646 (10) 1310 (1) 1372 (1) 1170 (5) 1306 (3) 1178 (1) 1598 (3) 1412 (1) 1449 (2) 1190 (2) 1318 (4) 1235 (1) 1650 (9) 1470 (1) 1606 (4) 1225 (1) 1375 (6) 1288 (2) 1590 (2) 1650 (7) 1280 (2) 1385 (4) 1360 (1) 1645 (6) 1306 (2) 1513 (4) 1385 (2) 1350 (5) 1595 (8) 1492 (2) 1372 (2) 1650 (10) 1520 (2) 1458 (1) 1584 (1) 1519 (1) 1640 (4) 1580 (8) 1640 (10) 1745 (3) SERS OF STIMULANT DRUGS 149

spectrum as very weak vibrational modes. The Raman spectrum of pemoline (Fig. 2) indicates that in the frequency region where C=C and C=N ring stretching vibration are expected [26], a very strong peak at about 1595 cm-’ with a shoulder at 1580 cm-i appears. The band at about 1650 cm-’ may correspond to NH, bending. The bands shown in the typical frequency region of the benzene ring are smaller in intensity than in the SER spectrum. Other medium or strong bands are observed at 646, 833, 1170, 1306, 1318, 1375, 1385 cm-’ and 1513 cm-‘. In general, the band widths (FWHM) in the solid-state spectra are smaller than those in the SER spectra. These results demonstrate that each compound can be identified by its characteristic vibrational bands. However, whereas SER is able to provide such information at the pg ml-’ level, CR spectrometry needs much higher concentrations. This illustrates the main advantage of SERS, i.e., fingerprinting capability at trace concentration levels. It can be established from Table 1 that stimulants such as pemoline with a large number of functional groups tend to provide an SER spectrum that shows a larger number of spectral features than those of drugs with simpler structures such as mefenorex. This is due to the fact that the adsorbate-induced aggregation of the colloid and the consequent surface-enhancement effect depend on the chemical structure of the adsorbate [271. As a result, in a given sample where one of several drugs may be present, the SERS identification of a drug with a large number of active groups should be more feasible than that of a drug with a simpler structure. Regarding the relative detection power of the SERS technique, a similar trend can be established, with pemoline showing the largest SERS signals of the four drugs tested.

Mhure spectra: aggregation kinetics and molec- ular adsorption The SER spectra of three binary mixtures of pemoline, mefenorex and pentylenetetrazole were obtained on silver hydrosols. Most vibrational bands appearing in the mixtures correlate fairly $a I I I ‘* Ramd~shift (c’%l) lwo well with the vibrational features shown by the Fig. 2. Conventional Raman spectra of stimulant drugs. isolated compounds. It is known [ll] that the 1.50 A. RUPkREZ ET AL. relative intensities of SER features arising from The SER spectra of a mixture of mefenorex two components in a mixture depend not only on and pemoline obtained 5 and 30 min after sample their solution concentrations, but also on their preparation are shown in Fig. 3. It is apparent relative adsorptivities. As similar concentrations that pemoline is readily adsorbed on the silver of the drugs were used, the extent of the aggrega- colloidal particles, causing a displacement of tion process depended on the chemical structure mefenorex from the surface. The only spectral of the adsorbate [27]. For the isolated com- feature of mefenorex occurs at early aggregation pounds, the aggregation and SERS intensity in- at 745 cm- ‘, this band being shifted to 830 cm-’ creased in the order L-amphetamine < mefenorex (a pemoline vibrational mode) in the 30-min spec- < pentylenetetrazole < pemoline. In the spectra trum. Ho other mefenorex bands can be recog- of mixtures, described below, the order of ad- nized. However, as colloid aggregation pro- sorption observed followed the same pattern. gresses, pemoline vibrational modes at 830, 1014,

p.lTl. 5 min. ‘\

J - 4 3ma hift (cm-‘)

014

30 min.

I 1200 1600 Raman shift (cm-‘)

Fig. 3. SER spectra of mixtures of mefenorex and pemoline obtained 5 and 30 min after mixing colloidal silver and drug solution. Drug concentration, 25 pg ml-‘. SERS OF STIMULANT DRUGS 151

1153,1225 and 1640 cm-’ are enhanced. Shifts in compared with the orientations of the pure com- other SERS bands as aggregation proceeds may pounds [l 11. indicate adsorbate reorganization at the silver The SER spectrum for mixtures of pemoline surface. A new band centred at about 1485 cm-’ and pentylenetetrazole at early aggregation stages appears at early aggregation (5 min), as shown in (Fig. 4) shows characteristic bands corresponding Fig. 3. This band disappears as the aggregation to both drugs. Pentylenetetrazole is recognized by proceeds. This band is probably the result of a weak, reproducible bands at 674, 749, 1095 and combination of effects that may include interac- 1584 cm-‘, whereas pemoline is recognized by a tions between molecules and differences in their medium-intensity band at about 1700 cm-‘. This orientations at the surface when co-adsorbed, band, presumably due to carbonyl stretching, ap-

5 min.

I I 1 400 600 1200 1600 Roman shift (cm-‘)

60 min

!Ii75 1590

1 I 1 So0 1200 1600 Roman shift (cm-‘)

Fig. 4. SER spectra of mixtures of pentylenetetrazole and pemoline obtained 5 and 60 min after mixing colloidal silver and drug solution. Drug concentration, 25 pg ml -l. The inset represents an enlargement of the 1575-1590 cm-’ section of the spectrum. 152 A. RlJPl?REZ ET AL. pears at a frequency slightly shorter than that in the spectrum of isolated pemoline. The band at (0) 1014 cm-’ of pemoline and the band at 1029 cm-’ of pentylenetetrazole are not well resolved and appear as a strong band centred at 1015 cm-‘. After 1 h, the intensity of pemoline bands at 774 and 830 cm-’ increased (Fig. 41, and any bands of pentylenetetrazole were obscured by enhancement of pemoline; for instance, the pentylenetetrazole band at 1029 cm-’ now appears as a shoulder in the band at 1014 cm-’ of pemoline. However, the band at 1584 cm-’ corresponding to the C=N ring stretching mode of Ih I I I pentylenetetrazole is reproducible and can be 400 600 1200 1600 easily identified (inset in Fig. 4). The carbonyl Raman shift (cm-‘) stretching of pemoline now appears at about 1745 cm-‘. For mefenorex-pentylenetetrazole mixtures only weak bands appear at the beginning of the aggregation process. After 20 min, only the mefenorex band at 990 cm-’ appears. These results confirm those obtained for the isolated drugs, i.e., pemoline, more abundant in functional groups, was adsorbed on the silver surface more readily than the other drugs. When the drug has fewer SER-active functional groups, e.g., mefenorex, it is displaced from the surface by pemoline as hydrosol aggregation proceeds, as is apparent by the manifestation of new bands be- longing to pemoline.

SER spectra in urine Urine is a complex mixture with abundant Raman shift (cm-‘) suspended material. Its SER spectrum is shown Fig. 5. (a) SER spectra of human urine obtained 5 and 30 min in Fig. 5. The top and bottom curves in Fig. 5a after mixing colloidal silver and urine sample. (b) SER spec- correspond to the spectra obtained 5 and 30 min trum of the 30 min-urine sample with an enlarged ordinate scale. after the mixing of urine with the sol, respectively. As observed, the freshly prepared sample is characterized by a strong signal background, this compound is one of the main components of which may be due to elastic scattering from urine urine. To confirm this point, the SER spectrum suspended material or to luminescence from urine of urea was obtained (Fig. 6). Although the urea components. Some Raman bands are also de- concentration used for this spectrum may differ tected. After 30 min, the signal background has considerably from the value in urine, the spec- decreased substantially, allowing the identifica- trum shows a band at about 1020 cm-‘. This tion of SERS features. Figure 5b corresponds to band has been also reported in the bulk Raman the 30-min spectrum with an enlarged ordinate spectrum of urea [28]. A band centred at 1450 scale for easier spectral analysis. Superimposed cm-’ is characteristic of urine at early aggrega- on the background, the spectrum shows a strong tion stages (Fig. 5a). As observed, no bands in band at 1020 cm-’ which could be due to urea as this region appear in Fig. 5b. The band appearing SERS OF STIMULANT DRUGS 153 at 1635 cm-’ may be due to NH, bending of urea or other primary amine compounds present (0) in urine. Also, weak bands appear at 825, 968, 1087, 1253, 1360, 1580 and 1740 cm-‘. To show the possible applicability of SER to doping analysis, urine was spiked with the drugs at 25 pg ml-‘. The concentrations used are in the concentration range expected in a therapeutic sample, but much lower than the concentrations expected under the usual overdosage utilized for doping purposes, When urine spiked with pemoline and pentylenetetrazole was added to colloidal silver, weak, reproducible bands of both drugs were observed in the SER spectrum (Fig.

7a). Figure 7b shows the spectrum of blank urine Roman shift (cm-‘) for comparison purposes. The spectrum of the mixture shows the characteristic background of urine poorly aggregated in colloids (as shown in Fig. 7b), with the broad band centred at 1450 (b) cm-‘. Pentylenetetrazole is identified by a medium band at 1029 cm-’ with a shoulder at 1047 cm-’ corresponding to pemoline. Pentylenetetrazole is confirmed by a weak band at 1130 cm- ‘. However, pemoline is again domi- nant and it is recognized by bands at 830, 1170

I 1 1 I 800 1200 1600 Roman shift (cm-‘) Fig. 7. (a) SER spectrum of human urine spiked with 2.5 pg ml-’ each of pentylenetetrazole and pemoline obtained 5 min after mixing colloidal silver and the sample; (b) SER spectrum of a urine blank after 5 min.

and 1745 cm-‘. It is interesting that spiked urine samples did not spontaneously develop into further aggregation stages as with samples (isolated drugs or their mixtures) not involving urine. Con- sequently, spectral changes taking place as aggregation proceeds cannot be observed. It could be I I I # 400 800 1200 1800 argued that the simultaneous adsorption of urine Roman shift (cm-‘) electrolytes and drug adsorbates on the silver

Fig. 6. SER spectrum of urea on colloidal silver. Concentra- surface results in the formation of an ionic layer, tion, 100 pg ml-l. preventing further collisions of colloidal particles 1.54 A.RUPkREZETAL. and crystal growth. Evidence for this stopped REFERENCES aggregation is found by comparison of Fig. 7a Y. Matsuki, K. Fukuhara, T. Ito, M. Otto, N. O’Hara, T. with Fig. 4. As observed, the spectrum of the Yui and T. Nambara, J. Chromatogr., 188 (1980) 177. mixture in urine (Fig. 7a) shows no sharp spectral C. Stanley, P. Mitchell and C.M. Kaye, Analyst, 110 (1985) features, in contrast to those obtained for mix- 83. K. Shimada, M. Tanaka, T. Nambara, Y. Imai, K. Abe and tures of the same drugs in the absence of urine K. Yashinaga, J. Chromatogr., 239 (1982) 585. (Fig. 4). As relatively large particles are needed E.L. Ghisalberti, in E.R. Garret and J. Mirtz (Eds.), Drug for the enhancement effect to be obtained with Fate and Metabolism. Methods and Thechniques, Dekker, red excitation (i.e., 632.8 nm as used here), no New York, 1979, p. 1. strong SERS peaks could be observed under the 5 J. Settlage and H. Jaeger, J. Chromatogr. Sci., 22 (1984) 192. conditions of stopped aggregation. However, it 6 E. Gantner, D. Steinert and J. Reinhardt, Anal. Chem., 57 could be still possible to observe the full enhance- (1985) 1658. ment effect with the small silver particles formed 7 J.J. Laserna, A.D. Campiglia and J.D. Winefordner, Anal. using a shorter wavelength laser (not available). Chim. Acta, 208 (1988) 21. In conclusion, the results demonstrate that the 8 A. Berthod, J.J. Laserna and J.D. Winefordner, Appl. Spectrosc., 41 (1987) 1137. stimulant drugs pemoline, mefenorex, L- 9 A. M. Alak and T. Vo-Dinh, Anal. Chim. Acta, 206 (1988) amphetamine and pentylenetetrazole are active 333. in SERS. The activity depends on the functional 10 J.J. Lasema, A. Berthod and J.D. Winefordner, Talanta, groups present in the compounds. In mixtures, 34 (1987) 745. the drugs are adsorbed selectively on the metal 11 J.J. Laserna, A.D. Campiglia and J.D. Winefordner, Anal. Chem., 61 (1989) 1697. surface, depending on their chemical structures. 12 D.J. Jeanmaire and R.P. Van Duyne, Electroanal. Chem., Urine provides a defined SER spectrum. Silver 84 (1977) 1. hydrosols can be developed as a practical sub- 13 S. Ohshima, T. Kajiwara, M. Hiramoto, K. Hashimoto and strate for the SERS identification of stimulants in T. Sakata, J. Phys. Chem., 90 (1986) 4474. urine samples. The technique fails in that the 14 A.C. Pineda and D. Ronis, J. Chem. Phys., 83 (1985) 5330. 15 M. Fleischmann, P.J. Hendra and A. McQuillan, Chem. optimum colloid aggregation needed for large Phys. Lett., 26 (1974) 163. Raman signals is difficult to achieve on a routine 16 J.J. Laserna, W.S. Sutherland and J.D. Winefordner, Anal. basis. SERS will not compete in the near future Chim. Acta, 237 (1990) 439. with other well established techniques for drug 17 A. Berthod, J.J. Laserna and J.D. Winefordner, J. Pharm. testing purposes. However, the identification of Biomed. Anal., 6 (1988) 599. 18 M. Kerker, Pure Appl. Chem., 53 (1981) 2083. characteristic vibrational modes in urine speci- 19 J.A. Creighton, in R.K. Chang and T.E. Furtak (Eds.), mens would be helpful in confirmatory analyses. Surface Enhanced Raman Scattering, Plenum, New York, 1982, p. 35. Financial support for this project was provided 20 J.J. Laserna A. Berthod and J.D. Winefordner, Mi- by the Direction General de Investigation crochem. J., 38 (1988) 125. 21 S. Nie, C.G. Castillo, K.L. Bergbauer, J.F.R. Kuck, Jr., Cientifica y TCcnica (Ministerio de Education y I.R. Nabiev and N.T. Yu, Appl. Spectrosc., 44 (1990) 571. Ciencia) and the Direction General de Universi- 22 I.R. Nabiev, R.G. Efremov and G.D. Chumanov, Sov. dades e Investigation (Consejeria de Education y Phys. Usp., 31 (1988) 241. Ciencia, Junta de Andalucia), Spain. 23 J.J. Laserna and R. Montes, Spectrosc. Int., 3 (1991) 32. 24 N.B. Colthup, H.D. Lawrence and S.E. Wiberley, Intro- duction to Infrared and Raman Spectroscopy, Academic, New York, 1975. 25 J.E. Stewart, J. Chem. Phys., 30 (1959) 1259. 26 M. Moskovits and J.S. Suh, J. Phys, Chem., 88 (1984) 5526. 27 R. Montes and J.J. Laserna, Analyst, 115 (1990) 1601. 28 B. Schrader and W. Meier, Raman/IR Atlas of Organic Compounds, Verlag Chemie, Weinheim, 1978.