32
Anal. Chem. 1983, 55,32-34
PETRIN, 1607-17-6; 5-ISMN, 16051-77-7; 2-MNG, 620-12-2; PEDN, 1607-01-8.
LITERATURE CITED (1) Prlstera, F.; Halik, M.; Castelli, A.; Fredericks, W. Anal. Chem. 1980, 32, 495. (2) Whitnack, G. C.; Mayfield, M. M.; Gantz, E. S. C. Anal. Chem. 1985, 37, 899. (3) Di Carlo, F. J.; Crew, M. C.; Skiow, N. J.; Couthinho. C. B.; Nonkin, P.; Simon, F.; Bernstein, A. J. Pharmacol. Exp. Ther. 1978, 153, 254. (4) Di Carlo, F. J.; Crew, M. C.; Haynes, L. J.; Meigar, M. D.; Gaia, R. L. Biochem. Pharmacol. 1988, 17, 2179. (5) Reed, D. E.; Akester, J. M.; Prather, J. F.; Tuckosh, J. R.; McCurdy, D. H.; Yeh, C. J. Pharmacol. Exp. Ther. 1977, 202, 32. (6) Crew, M. C.;Gala, R. L.; Hayes, L. J.; Di Carlo, F. J. Blochem. Parmacol. 1971, 20, 3077. (7) Crew, M. C.; Meigar, M. D.; Di Carlo, F. J. J. Pharmacol. Exp. Ther. 1975, 192, 218. (8) Roseel, M. T.; Bogaert, M. G. J. Chromafogr. 1974, 64, 364. (9) Davldson, I. W. F.; Di Carlo, F. J.; Szabo, E. I. J. Chromafogr. W71, 57, 345. (IO) Assinder, D. F.; Chasseaud, L. F.; Taylor, T. J. Pharm. Scl. 1977, 66,
(11) Yap, P. S.K.; McNiff, E. F.; Fung, H. L. J. Pharm. Sci. 1978, 67, 583. (12) Armstrona, J. A.: Marks, G. S.: Armstrona. P. W. Mol. Pharmacol. lg80, 16; 112. (13) Chandler, C. D.; Gibson, G. R.; Bolleter, W. T. J. Chromafogr. 1974, 100. . - - . 155.
(14) Chin, D. A.; Prue, D. 0.; Mlcheluccl, J.; Kho, T.; Warner, C. R . J. Pharm. Sci. 1977, 66, 1143. (15) Baake, D. M.; Carter, J. E.; Amann, A. H. J. Pharm. Sci. 1977, 66, 1143. (16) Crouthamel, W. G.; Dorsch, 8. J. Pharm. Sci. 197% 66, 481. (17) Hannemann, R. E.; Erb, R. J.; Stoltman, W. P.; Bronson, E. C.; WilIlams, E. J.; Long, R. A.; Hull, J. H.; Starbuck. R. R. Clin. Pharmacol.
Ther. 1981, 35.(16) Lafleur, A. L.; Morrlseau, B. D. Anal. Chem. 1980, 52, 1313. (19) Spanggord, R. J.; Keck, R. G. J. Pharm. Sci. 1980, 69, 444. (20) Fine, D. H.; Rufeh, F.; Lieb, D.; Rounbehler, D. P. Anal. Chem. 1975, 47. 1188. (21) Doyle, E.; Chasseaud, L. F.; Taylor, T. Blopharm. Drug Dispos. 1980, 1 , 141. (22) Malbica, J. 0.; Monson, K.; Nellson, K.; Sprlssler, R. J. Pharm. Scl. 1977, 66, 385.
RECEIVED for review June 24,1982. Accepted October 4,1982.
775.
Determination of Phencyclidine and Phenobarbital in Complex Mixtures by Fourier-Transformed Infrared Photoacoustic Spectroscopy M. G. Rockiey,* M. Woodard,
H. H. Richardson, D. M. Davis, N.
Purdle, and J. M. Bowen
Department of Chemistry, Oklahoma State lJniversi@, Stlllwater, Oklahoma 74078
Fourier-transformed infrared photoacoustlc spectroscopy has been used to quantitate the amounts of controlled substances (phencyclidine and phenobarbital) in substrates such as lactose and parsley. Quantltatlon to 1% accuracy is demonstrated, although saturation effects might be exhibited in m e of the spectra.
Every conceivable analytical technique has been applied to the difficult problem of drug analysis (1). Because drugs are minor constituents in rather complex mixtures, a primary requirement in almost every analysis is to first effectively separate the mixture and, in so doing, concentrate the drug so that it has a concentration which exceeds the limits of detection of the method to be used. Qualitative identification is readily accomplished by the many chromatographic procedures either alone or in tandem with mass spectrometry or, in some cases, by spot tests. Quantitation is less readily accomplished and frequently requires compound derivatization and repeated instrument calibration. Innovative techniques which can simplify the analytical protocol either by requiring less stringent separation needs or by eliminating subsequent sample prepreparation steps are valuable in that time, a precious commodity in clinical and criminalistic laboratories, is saved. A recent innovation, for example, is the application of circular dichroism spectropolarimetry (2) to the direct analysis of drugs. A rapidly emerging technique which has the potential to expedite analysis is photoacoustic spectroscopy (PAS) (3,4). Its use in the ultraviolet spectral range was described in the early literature and, more recently, reports have appeared for the application of PAS in the mid-infrared range by both dispersive (5-9) and Fourier-transformed methods (FTIR-
PAS) (10-13). The advantages of FTIR-PAS over dispersive methods are well delineated in these articles. They include speed of analysis, high incident power (useful for PAS studies), and frequency multiplexing. The disadvantages of FTIR-PAS are that the sampling depth varies with frequency, there is some difficulty in knowing exactly what to use as a power correcting reference spectrum, and the thermal and optical properties of the samples control the frequency dependence of the PAS signal strength. These difficulties have been discussed in some detail by Krishnan (14) and by Royce et al. (15). These advantages notwithstanding, it was felt that for routine quantitative and qualitative analysis the currently used methods of FTIR-PAS analysis would prove to be a significantly useful tool in the modern forensic analysis laboratory by virtue of the speed of analysis, use of commercially available components, and other features described below. In this paper we describe the direct quantitative determination of phenobarbital and phencyclidine (PCP) using FTIR-PAS. Sample mixtures, in powdered solid form, were all in-house preparations and were tested for both the qualitative and quantitative capabilities of the method. EXPERIMENTAL SECTION The samples were prepared gravimetriclly and made somewhat homogeneous in composition by shaking for 5 min in a Wiggle-bug. Reference FTIR-PAS spectra of the pure components of the various mixtures were stored on disk for subsequent subtraction as will be explained below. PCP was phencyclidine hydrochloride obtained from NIDA via Research Triangle Institute. Parsley was from a local grocery store. This mixture was coground. Phenobarbital was from Sigma. Lactose was from Aldrich. The procedure for analysis of the FTIR-PAS spectra has been described elsewhere. There were essentially no new features with the exception of some minor modifications to th cell design. The cell used for these experiments incorporated a Bruel and Kjaer
0003-2700/83/0355-0032$0 1.50/0 0 1982 American Chemical Soclety
ANALYTICAL CHEMISTRY, VOL. 55, NO. 1, JANUARY 1983 6E
+
52.5
I
45
Table I. Analysis of Phenobarbital‘ sample % calcd 1 2 3 4 5
u1
37.5
e
u2 u2 u3 u3 u4 u5
z
i” 22.5
8.4 19.2 26.7 40.8 36.8 12.0 10.8 10.0
11.3 12.8 12.2 36.8
33
% given
7.5 19.8 27.1 40.7
37.1 12.3 9.9 9.9 12.3 12.3 10.7
39.9
a Calculated (by FTIR-PAS) and given weight percentages of phenobarbital in lactose. U1-U5 are “unknowns” with typical precisions of independent measurements of the same sample as indicated.
15
7.5
WAVWHXRS
Flgure 1. Top spectrum Is that of ca. 10 mg of pure Phenobarbital with 2000 scans at 8 cm-’ resolution, corrected by reference to a carbon-black spectrum. Bottom spectrum Is the residual spectrum from 2000 scans at 8 CKI-‘ resolution of 12.1% by weight phenobarbital in lactose after subtracting the spectra of the pure components in a 0.12 to 0.88 ratio of phenobarbital to lactose.
4175 0.5-in. foil electret microphone with a 2642 preamplifier and a 2610 battery operated power supply, all of these components being available off the shelf from Bruel and Kjaer. The preamplifier was sealed with rubber cement to prevent acoustic noise pickup from the room. ‘The spectra were all ratioed against a carbon black standard. Since this work was completed, it has emerged in studies by others (16) that carbon black is not such a good reference for FT1.12-PAS spectral measurements because it has several spectral features which are not intense but nevertheless present. Because the cell constructed here indicated the presence of no acoustic resonances over the range of audio modulation frequencies present in the FTIR beam, the spectra would just as easily have been corrected for source power variations with wavelength by ratioing to the standard DTGS detector response. An excellent description of the problems associated with spectral correction of FTIR-PAS spectra has been recently published by Royce et a]. (15). Therefore, besides band intensity problems associated with specular reflectance and saturation, the pure component spectra reported here have some band intensity errors which may be as much as 5% of the total band intensity. However, while this means that the spectra reported here cannot be used for study of absolute band intensities, they can be used for quantitative purposes because the same reference carbon-black spectrum was used to correct all spectra measured. As a result, the relative errors in using the spectral subtraction methods outlined below prove to be negligible and have not been considered further for the purposes of this work. Two-component mixtures were made with the compositions of these mixtures being made known to the individual running the FTIR-PAS spectra. In this way a so-called calibration line was inferred. The calibration line was prepared by subtracting the spectra of the individual pure components from the spectra of the mixture and repeating the process until the flattest possible base line was obtained, as shown by Figure 1for the case of 12.1% by weight phenobarbital in lactose. Following the experimental derivation of a calibration line or curve (for all cases studied here the set of points was well correlated with a straight line) a few samples (‘blinds”) were analyzed in which the concentration was known to the preparer but not to the analyst. Those points are represented by a “U”prefix in Tables I and 11.
RESULTS AND DISCUSSION An illustration of some typic1 spectra is given in Figure 1 for the case of a 1290 by weight phenobarbital in lactose
Table 11. Analysis of Phencyclidinea sample % calcd 1
4.0
2 3 5 6
9.6 7.2 3.4 12.2 13.8 8.8 14.5 5.8 14.3
u1 u2 u3 u4 u5
% given
4.1 9.7 7.1
3.3 12.2 14.9 6.0
17.9 3.5 13.1
a Calculated (by FTIR-PAS) and given weight percentages of PCP on parsley. Typical sample weights were about 1 0 mg.
system. The top spectrum is that of 10 mg of pure phenobarbital. The bottom spectrum is that which remains after 0.12 x the pure lactose spectrum and 0.88 x the pure phenobarbital spectrum are subtracted from the spectrum of the 12.1% by weight phenobarbital in lactose mixture. It can be seen that the base line is essentially flat with the notable exception of the band a t ca. 1700 cm-l. The flatness of this base line was estimated by eye and was not further analyzed by any computer algorithm, although with the purchase of suitable software this could also have been done. After several trials it was determined that the band a t 1700 cm-l which remains outstanding after spectral subtraction of the component spectra is not quantitatively related to the amount of phenobarbital in the mixture. While the exact cause of this outstanding feature has not been determined here, it is hypothesized that it could be due to a number of different phenomena. It could be that the pure component spectrum of phenobarbital was measured with a sample which had a slightly different grain size distribution than that present in the mixtures. In such a case, the amount of reflected light might be different at that particular band. That is, the phenomenon would correspond to some sensitivity of the photoacoustic spectrum to the amount of reflected light, especially for the case of intense transitions where the amount of scattered or reflected light will be greatest. Another possible cause for this phenomenon which also could be explained on the basis of variations in the distribution of grain sizes is the occurrence of saturation (14,16). Since the presence of saturation in the generation of the photoacoustic signal is dependent on the ratio of the thermal to the optical thickness of the sample, the appearance of this anomaly in band intensity could be associated with differing amounts of signal saturation present in the pure and mixed samples for absorption into this particular band. As to whether or not either
34
ANALYTICAL CHEMISTRY, VOL. 55,
NO. 1, JANUARY 1983
of these two reasons is correct requires the analysis of a series of samples in which the grain size distribution of the pure and mixed samples are tightly specified. Even then it is not clear that the relative surface areas for the two samples would be the same. Nevertheless, it might be that the amplitude of this anomalous feature would decrease with decreasing grain size in which case at least one of these two proposed origins might apply. A third reason may be associated with photometric nonlinearity for intense bands. Finally, it is possible that Kubelka-Munk corrections, which are necessary to correct band intensities for reflectivity variations, might prove useful in distinguishing saturation from reflection problems. Despite these difficulties, the relative compositions of the various mixtures were obtained by spectral subtraction procedures. Adequate subtraction was inferred by assessing the flatness of the subtracted spectral base line in regions outside of those containing intense features. Table I contains the results obtained for the analysis of phenobarbital and lactose. In all cases for the phenobarbital and lactose system the measured weight percentages are accurate to within 1% and have a precision of better than 1%as shown by multiple but independent analyses of the same unknowns. For the two measurements of the same unknown, two independent samples were obtained from the container of sample provided. The PCP on parsley system was analyzed in similar fashion and the data are described in Table 11. The accuracy of the measured results here is not quite as good as for the case of phenobarbital and lactose. That is to be expected, however, since the samples of PCP on parsley cannot be made nearly as homogeneous as the phenobarbital and lactose system. As a result, sampling errors become more significant. The presence of two strong bands centered at 1600 cm-l and 1100 cm-l in the spectrum of parsley could not be substracted totally from the spectra of the mixtures, indicating the presence of the saturation and reflectivity problems also apparent in the phenobarbital-lactose system. The parsley was also incompletely dried. This resulted in the introduction of significant quantities of water vapor into the PAS cell, despite the presence of molecular sieves in the sample compartment. it is clear that the quantitative determination and qualitative analysis of these intractable drug mixtures are feasible by FTIR-PAS over the range of concentrations studied here. However, it is also apparent that, because of saturation and reflectance effects, technical improvements in the methodology
(an example might be careful spectral reflectance corrections) must be introduced before very complex multicomponent solid samples can be analyzed with ease. These phenomena manifest themselves in the difficulty associated with quantitation but will be less important in qualitation. In conclusion, the data presented here demonstrate the feasibility of rapid quantitative analysis of difficult samples (e.g., PCP on parsley) commonly encountered in the forensic laboratory. It is to be stressed that the samples are not adulterated in any way. It is therefore clear that this technique has qualitative and quantitative analytical capabilities. In addition to finding use in the forensic laboratory, it may find use wherever quantitation of solid mixtures is required such as in the analysis of food additives or pharmaceuticals and pesticides on grains.
ACKNOWLEDGMENT We thank the National Institute for Drug Abuse and the Research Triangle Institute for providing the standard samples. Registry No. phencyclidine, 77-10-1; phenobarbital, 50-06-6; lactose, 63-42-3. LITERATURE CITED Clarke, E. G. C. "Isolation and Identification of Drugs"; The Pharmaceutical Press: London, 1978; Vol. 1. Bowen, J. M.; Crone, T. A., Head, V. L.;McMorrow, H. A,; Kennedy, P. K.; Purdie, N. J. Forenslc Sci. 1981, 26 ( 4 ) , 664-670. Rosencwaig, A. "Photoacoustlcs and Photoacoustic Spectroscopy"; Wley-Interscience: New York, 1980; p 309. Pao, Y. H. "Optoacoustic Spectroscopy and Detection"; Academic: New York, 1977; p 244. Low, M. J. D.; Parodi, G. A. Spectrosc. Len. 1978, 7 1 , 581. Low, M. J. D.; Parodi, G. A. Infrared Phys. 1980, 20, 333. Low, M. J. D.; Parodi, G. A. Appl. Spectrosc. 1980, 34, 76. Low, M. J. D.; Parodi, G. A. J. Mol. Struct. 1980, 67, 119. Low, M. J. D.; Parodl, G. A. Spectrosc. Len. 1980, 73, 663. Rockley, M. G. Chem. Phys. Lett. 1979, 68, 455. Vldrlne, D. W. Appl. Spectrosc. 1980, 34, 313. Laufer, G.;Huneke, J. T.; Royce, 8. S. H.; Teng, Y. C. Appl. Phys. Lett. 1980, 3 7 , 617. Rockley, M. G.;Devlin, J. P. Appl. Specfrosc. 1980, 34, 407. Krlshnan, K. Appl. Spectrosc. 1981, 35, 549. Teng, Y. C.; Royce, 0. S. H. Appl. Opt. 1982, 27, 77. Risernan, S.M.; Eyring, E. M. Spectrosc. Lett. 1981, 74 (3), 163-185.
RECEIVEDfor review July 21,1982. Accepted October 4,1982. Supported in part by ARO Contract No. DAAG 29 81 KO09 and in part by NSF Grant No. CHE-7909388.
