Rapid Indirect Determination of Very Low Levels of Cocaine by Tandem On-line Continuous Separation and Inductively Coupled Plasma Atomic Emission Spectrometric Detection Journal of Analytical Atomic Spectrometry I I ALBERTO MENENDEZ GARCIA ENRIQUE SANCHEZ URIA AND ALFREDO SANZ-MEDEL* Department of Physical and Analytical Chemistry Faculty of Chemistry University of Oviedo Oviedo Spain A very sensitive precise and automated method is proposed for the indirect determination of cocaine. The method is based on selective extraction into chloroform of the protonated alkaloid as an ion pair with Dragendorff reagent; BiH is generated directly from the organic phase and the volatile species is brought to an ICP-AES system for bismuth detection. Both separation steps extraction and gas-liquid separation are accomplished in a continuous manner with on-line final ICP- AES detection.The optimum chemical continuous flow and instrumental variables were determined. The proposed method allows the determination of cocaine over a wide range of concentration (20 ng ml-'-100 pg ml-') and the precision at a concentration of 0.5 pg ml-' was 3%. The method was applied to the determination of cocaine in confiscated samples. Keywords Cocaine; alkaloids; liquid-liquid extraction; continuous pow; hydride generation; inductively coupled plasma atomic emission spectrometry Cocaine a psychotropic drug is methylbenzoylecgonine a derivative of ecgonine which in turn is a carboxylic acid derivative of tropine. Cocaine has been used for medical and therapeutic purposes mainly as a local anaesthetic in ophthal- mology.Its euphorizing action on the nervous central system and the facility for its intake however have led to the spread of its consumption as a drug of abuse. This increasing consump- tion of cocaine necessitates the development of analytical methods in order to achieve adequate control of this drug in consignments of drugs or in biological samples e.g. urine to measure the consumption of cocaine by drug addicts and to control the therapy of drug addiction cure programmes as it is known that metabolites of cocaine such as benzoylecgonine and small amounts of the untransformed drug are excreted in the urine in 24 h; in some cases valuable information can be obtained from the determination of cocaine in hair. Immunoanalysis has been claimed to be the ideal technique for screening the water-soluble metabolite benzoylecgonine in urine samples.For this purpose there are commercial kits for the rapid quantification of cocaine in biological fluids. However positive results obtained by immunoassays must be confirmed by using non-immunological techniques. Gas chromatography is a useful technique for the determi- nation of cocaine (or its metabolites) but prior extraction of the drug'- is usually required for biological samples. HPLC usually in the reverse-phase mode has also been used for cocaine determination with non-polar Spherisorb C or pBondpack c18 column^.^-^ Capillary electroph~resis,~ supercritical fluid chromatographys and electrochemical methodsg have also been proposed for cocaine determination. * To whom correspondence should be addressed.However GC-MS is the technique most frequently used for the determination of cocaine and its metabolite^.'^-'^ Methods based on the formation of ion pairs extractable into organic solvents between the protonated alkaloid and a negative- ly charged metal complex e.g. CO(SCN),~- Ni(SCN),2- or BiI,- have also been applied for cocaine determination. Detection can be carried out by spectr~photometry'~-'~ of the organic phase. Such an extraction scheme has been used for indirect determination using AAS. In one method,17 protonated cocaine (Coc. H)' is continuously extracted into 1,2-dichloromethane with BiI,- providing the Bi in the organic solvent. Further the cocaine ion pair may be precipitated as the metal complex and re-dissolved in a flow injection system for final determination of Bi by AAS."*19 In this work the concept of tandem on-line continuous separation based on two continuous separation steps carried out on-line with ICP-AES detection previously proposed in our laboratory,20 was used for cocaine determination. Dragendorff reagent21 was selected for generating the ion pair (C0c-H)' Bi14- which was extracted in chloroform in a continuous mode; BiH was directly generated on-line from the organic phase and continuously introduced into the torch of the ICP for AES measurement of bismuth.EXPERIMENTAL Reagents All chemicals were of analytical-reagent grade unless stated otherwise. Alkaloids. A stock standard solution of cocaine (Sigma-Aldrich Quimica Madrid Spain) was prepared in ethanol at a concen- tration of 1000 pg ml-' and stored at 0-4 "C in PVC container.Ethanol solutions of the following alkaloids were used for the interference study heroin morphine codeine methadone atro- pine lidocaine quinine atropine and strichnine. Dragendor- reagent. A standard solution of Bi( NO) 5H20 (Merck Darmstadt Germany) was prepared by dissolving 5.0 g of the salt in 10 ml of HNO and diluting with distilled water to 100 ml. A 50 g amount of KI (Merck) was dissolved in 100ml of distilled water and Dragendorff reagent was obtained by mixing 5ml of the Bi"' solution 62.5ml of the iodide solution and 487.5ml of distilled water. The concen- tration of tetraiodobismuthate(n1) ( Bi14-) in this solution was 9.3 x lo- mol 1-' (pH 2.0).Solutions of this reagent of lower concentration were prepared by dilution of suitable volumes in doubly distilled water. Sodium tetrahydroborate(@ (NaBH,) solution 1.5% m/v DMF. The solution was prepared daily using NaBH obtained from Merck. Journal of Analytical Atomic Spectrometry August 1996 Vol. 1 1 (561 -565) 561Apparatus All glass material was cleaned with dilute nitric acid (1 +9) and rinsed with ultra-pure water (Milli-Q; Millipore Molsheim France) before use. A Perkin-Elmer (Norwalk CT USA) ICP/5000 spec- trometer equipped with a Perkin-Elmer Data System 10s laboratory computer was used for ICP measurements using the 223.06 nm bismuth emission line. A Minipuls-2 multi-channel pump from Gilson (Worthington OH USA) and an HP4 peristaltic pump from Scharlau Science (Villiers le Bel France) were used for pumping the reagent feeding the tandem on-line continuous separation device.A continuous phase separator made of PTFE with 3 pm pore diameter membranes (Millipore FSLW 02500 Type FS) was employed. Circular membranes (2.5 cm diameter) were used for continuous liquid-liquid extraction of the ion pair of protonated cocaine and tetraiodobismuthate(II1). The general operating mode of the tandem on-line separation device and its coupling to the ICP torch can be seen in Fig. 1. A description of its operation mode for this application is given below. General Procedure The device used in our tandem on-line system allowing for ion-pair formation and its simultaneous and continuous extrac- tion into CHC13 and the direct generation of BiH from this solvent is depicted in Fig.1. First a 2.9 ml min-' flow of cocaine (or blank) in ethanolic medium at pH 1.0 (adjusted with HN03) merges in a T-piece with a 0.5 ml min-' flow of lo- moll-' Dragendorff reagent. The resulting flow reaches the mixing coil and the formation of the (Coc-H)' BiI,- ion pair takes place. Then this aqueous combined flow (3.40 ml min-') merges with a 0.70 ml min-' flow of CHC& at the solvent segmenter where the continuous extraction of the ion pair starts to take place and continues along the extraction coil. The PTFE 'phase separator' leads the aqueous phase continuously to waste and the chloroform phase containing the analyte is merged with NaBH and glacial acetic acid with direct formation of BiH,.This volatile com- pound generated in the continuous mode when the organic phase mixes with a 0.80 ml min-' flow of 1.5% m/v NaBH in DMF solution and a 1.0ml min-' flow of glacial acetic acid is transferred to the ICP torch by a 0.3 1 min-' continuous stream of argon and the liquid and gaseous phases are separated in the gas-liquid separator (see Fig. 1). The BiH3 reaches the plasma where Bi is monitored at the 223.06nm emission line. The GC determination of cocaine in real samples was car- ried out in an external laboratory following a standardized procedure using a phenylsilicone Ultrados column (HP 10091B-012) of 12 m x 0.33 pm id.22 RESULTS AND DISCUSSION Choice of the Organic Solvent for Extraction and the Emission Line The solvents used in our previous work usually well tolerated by the ICP (without disturbance of the plasma stability) include xylene and IBMK.20323*24 Unfortunately the cocaine ion-pair extraction is accompanied by Bi14H complex extrac- tion and therefore it is unsuitable for indirect determination via Bi detection.For this extraction we verified that only low- polarity organic solvents such as chloroform or dichloro- methane,'7.21.25 were able to secure the selective extraction of the ion pair formed between cocaine and Dragendorff reagent. Chloroform provided an intensity-to-background ratio slightly higher than dichloromethane for Bi and superior plasma stability in the continuous system and was therefore selected for the extraction. In order to select the emission line for Bi measurement the four most sensitive lines recommended in the literature were tried. The highest signal-to-background ratio was observed in our system at the 223.06 nm emission line of bismuth.Optimization of the ICP-AES Parameters The three most critical instrumental parameters affecting the emission signals in ICP-AES (rf power carrier gas flow rate and viewing height) were optimized first in our instrument using a simplex optimization method utilizing the minimum 'background equivalent concentration' as the optimization criterion. The optimum experimental values found for these parameters are summarized in the top part of Table 1. Optimization of Chemical Parameters for the Extraction The concentration of BiI,- reagent and pH are the two main chemical parameters to be optimized in order to establish the best experimental conditions for the continuous extraction of the (Coc-H)' BiI,- ion pair into chloroform.For the optimiz- Fig. 1 and ICP Schematic diagram of the continuous liquid-liquid CocH + Bi14- extraction coupled on-line with a hydride generator gas-liquid separator 562 Journal of Analytical Atomic Spectrometry August 1996 Vol. 1 1ation of both parameters a 270 cm extraction coil and a 100 cm mixing coil (Fig. 1) along with a reference 10 pg ml-' cocaine solution in ethanol were used. The effect of BiI,- concentration on the continuous ion-pair extraction was tested in the concen- tration range 10-5-10-3 mol 1-' fixing the extraction pH at 2.0. As can be seen in Fig. 2 the concentration of BiI,- is critical for the efficient extraction of the ion pair for concen- trations of BiI,- higher than 2 x loF4 moll-' the extraction efficiency decreased slightly.A lo- moll-' Bi1,- solution was finally selected for subsequent experiments. The influence of pH on the extraction was then tested in the range 1-4. The results obtained are plotted in Fig. 3. As can be seen pH had a critical effect on the continuous extraction efficiency. Maximum ion-pair extraction was achieved at pH 1 decreasing dramatically at higher pH. These results agree well with the chemistry expected as an acidic medium will be necessary for the protonation of the NH groups of the cocaine molecule (pK= 8.5); in this medium cocaine forms (Coc*H)+ and an ion pair with BiI,-. For pH values higher than 1.0 the efficiency of extraction decreases because deprotonation takes place.On the other hand the effective concentration of BiI,- is probably lower (bismuth precipitated as a basic salt at pH 3.5 or higher). Consequently a pH value of 1.0 (adjusted with HNO,) was finally selected for extraction. Another chemical parameter to be optimized is the NaBH concentration in DMF (necessary for BiH formation directly in the chloroform phase). For its optimization a 10 pg ml-1 cocaine solution to feed the tandem on-line system was used and glacial acetic acid was selected to give an acidic medium for hydride generation from the organic phase. Concentrations of NaBH in DMF in the range 0.5-4% m/v were tested. The ICP-AES signal observed (Fig. 4) increased with increasing concentration of NaBH up to 1 YO. From 1 YO to 2% a plateau was reached and decreasing values of the net emission intensity of bismuth were observed at higher concentrarions probably Fig.2 Effect efficiency (I,) on-line system 0 2 4 6 8 1 0 [BiI;]/mol I-' x 10" of BiI,- concentration on the continuous extraction of 10pgml-' of cocaine in CHCI by the tandem 800 c v) -8 650 E 600 .- 550 0 1 2 3 4 5 PH Fig.3 efficiency (I,) into chloroform of 10 pg ml-' of cocaine Effect of pH on the tandem on-line continuous extraction 200 f 0 1 2 3 4 [NaBH,] in DMF (%) Fig. 4 Effect of NaBH concentration (in DMF) on the continuous generation of BiH by tandem on-line continuous separation with final ICP-AES Bi determination because larger amounts of excess hydrogen are produced that dilute the Bi and at the same time decrease the residence time of the analyte in the plasma leading to poor stability of the plasma.In consequence a 1.5% m/v NaBH concentration was selected for the general procedure. Optimization of the Continuous Flow Parameters Continuous flow parameters to be optimized for the automatic indirect determination of cocaine in our continuous tandem on-line device were the lengths of both the extraction and mixing coils the aqueous flow rate (qw) and the organic flow The lengths of the mixing and extraction coils (Fig. 1) were optimized using a 10 pg ml-' cocaine solution. First using a 270 cm length of the extraction coil the influence of the mixing coil length on the extraction was tested. This parameter seems not to be critical for the ion-pair extraction because the same analytical signal was obtained with coils between 30 and 100 cm long.The influence of the extraction coil length between 50 and 300 cm on the Bi ICP-AES signal was then examined. The results obtained (Fig. 5) show that the analytical signal increases with increasing length of the coil up to 200cm and for longer coils a plateau is reached. An extraction coil of 200cm is long enough for efficient ion-pair extraction so a safer 250cm long extraction coil was selected for the general procedure. The diameter of the coils was less critical as we have already verified in other similar experiments carried out with our tandem on-line system in previous s t ~ d i e s ; ~ ' ? ~ ~ a 0.7 mm id coil was selected. The influence of the total aqueous flow rate qw (sample + reagent solution flow rates) was optimized using a flow rate of chloroform of 0.70mlmin-' for the continuous extraction of a 10 pg ml-1 solution of cocaine at the previously optimized pH value and concentration of Dragendorff reagent.rate (40). 5 0 100 150 200 250 300 Coil lengthkm Fig. 5 of cocaine Effect of the length of the extraction coil on the determination Journal of Analytical Atomic Spectrometry August 1996 Vol. 11 563The sample flow rate was varied between 0.3 and 3.0 ml min-'. With higher total aqueous flow rates a higher mass of cocaine is extracted and a higher analytical signal was observed. However when the sample flow rate was > 3.0 ml min-' the precision and the efficiency of extraction worsened. A 2.90 ml min- sample flow rate (RSD = 0.7%) was finally selected for subsequent experiments.Using this flow rate of 2.90mlmin-' the effect of the chloroform flow rate qo on the extraction was investigated. Values of qo ranging between 0.1 and 1.0 ml min-' were tested and the results obtained are shown in Fig.6. The Bi signal increased with increase in flow rate up to a maximum at 0.75 ml min-' probably as a consequence of the improvement in the efficiency of extraction due to a larger contact surface between the two phases. For qo values >0.75 ml min-' it seems that a lower preconcentration factor prevails as the analytical signal began to decrease. The precision worsened for both high and low values of qo. In consequence a compro- mise value of qo of 0.70 ml min-' (RSD =0.8%) was selected for the general procedure.This value provides a preconcen- tration factor in the organic phase of 4. The optimum values found for the instrumental flow and chemical parameters involved are all summarized in Table 1. Analytical Performance Characteristics Using the 3s IUPAC criterion the LOD attained using the experimental conditions given in Table 1 with the proposed tandem on-line ICP-AES system determination was found to 1200 ..& 1100 E 1000 -g 800 .- s C .- 900 C 0 5 700 600 0 0.2 0.4 0.6 0.8 1 FIOW rate/mt min-' Fig.6 Effect of organic phase flow-rate (qo) on the tandem on-line continuous extraction of 10 pg ml- ' cocaine solution with lo- moll-' BiI,- solution at pH 1.0 (sample flow rate 2.90 ml min-'). The precision of the intensity of the signals estimated by the standard deviation is given for each measurement in bar [numbers below each bar RSD (YO)] Table 1 Optimum parameters (instrumental flow and chemical) for the indirect determination of cocaine by tandem on-line continuous separation and ICP-AES final determination (A= 223.06 nm) Plasma ICP- Rf forward power Carrier gas flow Viewing height (above coil) Continuous pow- Mixing coil length Extraction coil length Internal diameter of coil Sample flow rate Chloroform flow rate Glacial acetic acid flow rate NaBH solution flow rate Bi14- concentration pH of sample Organic phase NaBH concentration Chemical- 1350 W 0.4 1 min-' 11 mm 50 cm 250 cm 0.7 mm 2.90 ml min- 0.70 ml min-' 1.00 ml min-' 0.80 ml min-' be 2 ng ml-'.The precision attained at a cocaine level of 0.5 pg ml-' was 3% and the linear analytical range extended from 10 x LOD to 100 pg ml-'.The sampling frequency was about 12 samples per hour. Interference Studies Only common alkaloids containing NR groups in their mol- ecule that could form ion pairs with BiI,- were investigated as potential interferents. The maximum permissible concen- trations of the alkaloids studied as interferences were estab- lished by determining 0.5 pg ml-' of cocaine in ethanolic solution in the presence of increasing amounts of each of the interferents following the general procedure. The tolerated limits of alkaloids potentially interfering with the proposed determination of cocaine are given in Table 2. As can be seen from the recoveries obtained in each case the method is sensitive and relatively selective because a 20-fold excess of quinine and heroin and a 65-fold excess of papaverine are tolerated.Application to Real Samples In order to validate the analytical capability of the proposed method the determination of cocaine in four confiscated samples was carried out. The samples were dissolved in ethanol and filtered through Sep-Pak cartridges into 50ml flasks. These ethanolic solutions were analysed for cocaine in our laboratory using the proposed method and by GC using a standardized procedure22 in an external laboratory. Comparative results are given in Table 3. Good agreement was obtained. CONCLUSIONS The tandem on-line continuous separation device using the Bi1,- as counter ion provides a clear separation of cocaine into chloroform as an ion pair and the indirect determination of this alkaloid by ICP-AES coupled on-line as a detector for bismuth was achieved.The proposed automatic method allows the indirect determination of this drug at pgl-' levels in the presence of other alkaloids and can be used as an alternative method to more conventional GC-ECD. The latter technique and HPLC determinations are laborious as they involve Table 2 Interferences in the determination of cocaine by the proposed method (analyte concentration 0.5 pg ml-' i= 223.06 nm) Interferent Heroin Morphine Codeine Quinine Papaverine Atropine Strychnine Methadone ~~ Maximum amount tolerated/ pg ml-' 10 15 18 10 34 15 20 20 Recovery 99 104 102 97 101 98 100 98 (%I Table 3 Cocaine determination in confiscated samples by routine GC-ECD and by the proposed method Cocaine ("/.) moll-' 1 .o Chloroform 1.5% m/v in DMF Sample no.1 2 3 4 This work 30.75 & 1.5 30.20 f 1 .O 26.13 f 0.7 27.24k0.5 GC-ECD 30.70 k0.7 30.18+ 1.0 26.19k0.9 27.31 k0.4 564 Journal of Analytical Atomic Spectrometry August 1996 Vol. 1 1several time-consuming extractions and purifications of the extracts prior to injection into the chromatograph. Other alternatives such as electrochemical or immunoassay tech- niques lack the desired selectivity. Moreover for more sensitive determinations by GC further preconcentration steps involving evaporation of the organic extracts obtained are needed leading to even longer analysis times. In contrast using the continuous on-line ICP-AES procedure recommended here comparatively low levels of cocaine (pg 1-I) can be monitored requiring only 5 min per sample.Indirect methodologies proposed using AAS d e t e ~ t i o n ' ~ - ~ ~ and the same extraction principle provided detection limits in the range 0.2-2.5 pg ml-' of cocaine. Financial support from the DGICyT (Project PB 94-1331) during this work is gratefully acknowledged. The provision of confiscated cocaine samples by the Laboratory of Conse- jeria de Sanidad del Principado de Asturias and Gas Chromatographic cocaine determination realized in the last laboratory by J. M. Cabeza is also deeply appreciated. REFERENCES 1 LeBelle M. J. Dawson B. Lauriault G. and Savard C. Analyst 1991 116 1063. 2 Okeke C. C. Wynn J. E. and Patrick D. S. Chromatographia 1994 38 52.3 Moore J. M. Casale J. F. Klein R. F. X. Copper D. A. and Lydon J. J. Chromatogr. A 1994 659 163. 4 Schwartz R. S. and David K. O. Anal. Chem. 1985 57 1362. 5 Jatlow P. and Nadim H. Clin. Chem. 1990 36 1436. 6 Sukbuntering J. Walters A. Chow H. H. and Mayersohn M. J. Pharm. Sci. 1995 84 799. 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Tagliaro F. Poiesi C. Aiello R. Dorizzi R. Ghielmi S. and Maringo M. J. Chromatogr. 1993 638 303. Crowther J. B. and Henion J. D. Anal. Chem. 1985 57 2711. Fernandez Abedul M. T. Barreira J. R. Costa A. and Tuiih P. Electroanalysis 1991 3 409. Okeke C. C. Wynn J. E. and Patrick K. S. Chromatographia 1994 38 52. Corburt M. R. and Koves E. M. J . Forensic Sci. Int. 1994 39 136. Kintz P. and Mangin P. Forensic Sci. Int. 1995 73 93. de laTorre R. Ortufio J. Gonzalez M. L. Farre M. Cami J. and Segura J. J. Pharm. Biomed. Anal. 1995 13 305. Chichuev Yu. A. Farmatsiya (Moscow) 1984 33 70. Hernandez A. Gutierrez P. and Thomas J. Farmaco Ed. Prat. 1986 41 300. Mirzaeva Kh. A. and Ivanova N. I. Zh. Anal. Khim. 1984 39 1691. Eisman M. Gallego M. and Valcarcel M. Anal. Chem. 1992 64 1509. Eisman M. Gallego M. and Valcarcel M. J. Anal. At. Spectrom. 1993 8 1117. Eisman M. Gallego M. and Valcarcel M. J. Pharm. Biomed. Anal. 1994 12 179. MenCndez Garcia A. Sanchez Uria J. E. and Sanz Medel A. Anal. Chim. Acta 1990 234 133. Nerin C. Garnica A. and Cacho J. Anal. Chem. 1986,58 2617. Leach H. and Ramsey J. D. in Clarke's Isolation and IdentiJcation ofDrugs ed. Moffat A. C. Pharmaceutical Press London 1986 Menendez Garcia A. Perez Rodriguez M. C. Sanchez Uria J. E. and Sanz Medel A. Fresenius' J . Anal. Chem. 1995,353,128. Menkndez Garcia A. Fernandez Sinchez M. L. Sanchez Uria J. E. and Sanz Medel A. Mikrochim. Acta 1996 122 157. Travnikoff B. Anal. Chem. 1983 55 795. pp. 178-201. Paper 6/02054B Received March 25 1996 Accepted May 28 1996 Journal of Analytical Atomic Spectrometry August 1996 Vol. 11 565