摘要:

Direct chiral resolution of aliphatic a-hydroxy acids using 2-hydroxypropyl-b-cyclodextrin in capillary electrophoresis Shuji Kodama,* Atsushi Yamamoto and Akinobu Matsunaga Toyama Institute of Health, 17-1 Nakataikoyama, Kosugi-machi, Toyama 939-0363, Japan. E-mail: lee07664@niftyserve.or.jp Received 21st September 1998, Accepted 17th November 1998 Chiral resolution of a-hydroxy acids (lactic acid, 2-hydroxybutyric acid, 2-hydroxy-3-methylbutyric acid, 2-hydroxyisocaproic acid) without derivatization was performed by capillary electrophoresis using 2-hydroxypropyl-b-cyclodextrin (2HP-b-CD).An increase in the bulkiness of the alkyl group in these acids enhanced the resolution. The association constants for complexes of 2HP-b-CD with the a-hydroxy acids were determined by spectrophotometric and electrophoretic methods. Good agreement was found between the two methods. An increase in the bulkiness of the alkyl group in these acids brought about an increase in the association constants.When cyclohexanol that was included in 2HP-b-CD was added to the background electrolyte, chiral separation of these compounds was completely prevented. These results suggest that a-hydroxy acids having a short alkyl chain, and also chiral compounds having aromatic rings, could be included in 2HP-b-CD to be chiroptically separated. Introduction According to recent reviews,1–7 cyclodextrins (CDs) and their derivatives have been widely applied in capillary electrophoresis (CE) for the separation of enantiomers of many compounds.It is well known that at least three-point simultaneous interactions are generally needed between a chiral selector and an enantiomer to cause physical separation, because optical isomers have a three-dimensional spatial character. In the inclusion complexation mechanism, the hydrophobic interaction between the cavity of the native or derivatized CDs and the hydrophobic part of the compounds, such as an aromatic ring, plays an important role in the stereoselective interaction.Inclusion complex formation has been studied by several methods, such as proton nuclear magnetic resonance spectroscopy, ultraviolet and visible spectroscopy, circular dichroism spectroscopy and thermoanalytical methods, and the equilibrium constants for the inclusion–complexation have been calculated.8–10 The migration time of analytes in CE was well characterized by Guttman et al.11 Wren and Rowe12 developed a theoretical model relating the mobility to the concentration of a CD selector.An extended model for the separation of enantiomers of weak electrolyte solutes considering both pH and concentration of chiral selector has been proposed by Rawjee et al.13,14 Recently, CE has been successfully used for the calculation of association constants using electrophoretic measurements.15–19 Rundlett and Armstrong20 showed the relationship between several standard linear plotting methods that could be used to calculate binding constants with CE and compared these constants calculated by CE with those determined by various methods, including spectroscopy, calorimetry and chromatography.Most compounds described in these reviews have aromatic rings. It has been reported that monoterpenes such as pinenes, camphene and limonene, which have no aromatic rings and are hydrophobic, are chiroptically separated by CE21 and also by gas chromatography22–24 and high-performance liquid chromatography25 using a cyclodextrin bonded stationary phase.Recently, we reported the direct chiral resolution of the hydrophilic pantothenic acid having no aromatic rings by CE using 2-hydroxypropyl-b-cyclodextrin (2HP-b-CD).26 However, it remains unclear whether pantothenic acid is included in the cavity of 2HP-b-CD or not. An interesting chiral recognition mechanism between b-CD and dinitrophenyl (DNP)-amino acids (valine and leucine) has been reported.8 According to this report, the DNP group forms stable inclusion complexes with the b-CD cavity, and the alkyl groups form a secondary inclusion complex with another b-CD cavity (in the case of DNP-l-amino acids) or are sterically repulsed by the hydroxyl groups at the edge of the cavity (in the case of DNP-d-amino acids).This may indicate that an alkyl group in the aliphatic acid is also included in the cavity of the CD. This raised the possibility that the direct chiral resolution of various organic acids having no aromatic rings could be accomplished with CE using CDs. In this work, the direct chiral resolution of the aliphatic ahydroxy acids using 2HP-b-CD in CE was studied.The association constants between 2HP-b-CD and a-hydroxy acids by using spectrophotometric and electrophoretic measurements were also studied. Experimental Reagents Heptakis(2,6-di-O-methyl)-b-cyclodextrin, 2-hydroxypropylb- cyclodextrin (average degree of substitution 7) and d-lactic acid lithium salt were obtained from Sigma (St.Louis, MO, USA). 2-Hydroxypropyl-a-cyclodextrin, 2-hydroxypropyl-gcyclodextrin and racemic 2-hydroxy-3-methylbutyric acid were obtained from Aldrich (Milwaukee, WI, USA). 2,3,6-Tri-Omethyl- b-cyclodextrin, dl-lactic acid lithium salt, dl-2-hydroxybutyric acid lithium salt, 0.05 mol l21 iodine solution, cyclohexanol and other chemicals (guaranteed grade) were Analyst, 1999, 124, 55–59 55purchased from Wako (Osaka, Japan).d- and dl-2-Hydroxyisocaproic acid were purchased from Tokyo Kasei (Tokyo, Japan). Apparatus Electrophoretic experiments were carried out using an HP3D capillary electrophoresis system (Hewlett-Packard, Palo Alto, CA, USA). Samples were injected by applying a pressure of 5 kPa for 4 s. The separations were performed in a poly(vinyl alcohol) (PVA) coated bubble cell capillary of 50 cm 3 50 mm id (Hewlett Packard). The capillary was kept at 15 °C. The analytes were detected at 200 nm.The power supply was operated in the constant-voltage mode, at 230 kV, and the substances migrated towards the positive pole. Spectrophotometric examinations were carried out using a Hitachi (Tokyo, Japan) U-2000 spectrophotometer. The sample compartment contained a 1 cm thick quartz cell that was kept at 25 °C. Buffer and sample preparation for CE The background electrolyte (BGE) in the electrophoretic experiments, unless stated otherwise, was 60 mm phosphate buffer (pH 6.0) containing 0–220 mm 2HP-b-CD and was filtered with a 0.22 mm filter before use.De-ionized water was prepared using a Yamato (Tokyo, Japan) WA-52G auto still. Stock standard solutions of 100 mm dl-lactic acids, dl- 2-hydroxybutyric acid, racemic 2-hydroxy-3-methylbutyric acid and dl-2-hydroxyisocaproic acid were separately prepared in de-ionized water, stored at 4 °C and diluted to 2 mm before use. Calculation of resolution Each electropherogram that was obtained was subtracted from a blank run and the resolution (Rs) of the enantiomer was calculated by using the following equation: Rs = 2[(t2 2 t1)/(w2 + w1)] where t is the migration time and w is the width of the peak at the baseline.Determination of association constants by UV/VIS spectrophotometry The association constant (Ka) for a 2HP-b-CD–a-hydroxy acid system was determined by spectrophotometric examination of the inhibitory effect of the a-hydroxy acid on the association of the CD with a light-absorbing substance as reported by Matsui and Mochida,9 except that iodine was used as the lightabsorbing substance.A 2 ml volume of 30 mm phosphate buffer (pH 7.0) containing various concentrations of a-hydroxy acid was added to 2 ml of 30 mm phosphate buffer (pH 7.0) containing 0.08 mm iodine and 2 mm 2HP-b-CD and their absorption spectra (250–450 nm) were measured. The value was calculated according to the following equations: [I] = (DI° 2 DI)/De [C] = Kd([I]0 2 [I])/[I] Ka = {[C]0 2 [C] 2 ([I]0 2 [I])}/ [C]{[H]0 2 [C]0 + [C] + ([I]0 2 [I])} where DI° is the difference in absorbance between free and complexed iodine, DI is the change in absorbance of an iodine solution with the addition of 2HP-b-CD and a-hydroxy acid, De is the difference in the molar absorptivities for free and complexed iodine, Kd is the dissociation constant for the 2HP-b- CD–iodine complex as reported by Li and Purdy,8 [C]0, [I]0 and [H]0 are the total concentrations of 2HP-b-CD, iodine and ahydroxy acid, respectively, and [C] and [I] are the equilibrium concentrations for 2HP-b-CD and iodine, respectively. The absorption spectra were measured at five concentrations of ahydroxy acid.Results and discussion Chiral separation of a-hydroxy acids Lactic acid (LA), 2-hydroxybutyric acid (HBA), 2-hydroxy- 3-methylbutyric acid (HMBA) and 2-hydroxyisocaproic acid (HICA) were separated by CE using a BGE containing separately 90 or 200 mm a-CD, g-CD, 2,6-di-O-methyl-b-CD, 2,3,6-tri-O-methyl-b-CD, 2HP-a-CD, 2HP-b-CD or 2HP-g-CD (Table 1).A PVA-coated capillary, in which the electroosmotic flow was almost completely suppressed, was used to avoid the longer migration time with a fused-silica capillary. In the case of a-CD, g-CD and methylated b-CDs, the above analysis was carried out at a concentration of only 90 mm owing to their low solubilities. Of these CDs, 2HP-b-CD was found to be most effective for the resolution of all the a-hydroxy acids.This was in close agreement with the results of other studies in which 2HP-b-CD was suitable for the resolution of the enantiomers of a series of a-hydroxy acids having an aromatic ring, such as mandelic acid,27,28 and for that of pantothenic acid,26 having no aromatic rings. Fig. 1 shows the effect of 2HP-b-CD concentration on the resolution and migration time of these acids. Since the carboxyl group of each a-hydroxy acid is dissociated in the BGE at pH 6 and the electroosmotic flow is almost completely suppressed by using a PVA-coated capillary, the analytes migrate electrophoretically to the anode.Since all of the a-hydroxy acids have the same negative charge, the acids of smaller molecular size migrated faster in the absence of 2HP-b-CD. For both LA and HICA, the l-isomers were found to move faster than the disomers. This might indicate that the d-isomers formed a stronger diastereomer complex with 2HP-b-CD than the lisomers, because neutral 2HP-b-CD hardly migrates at all.For both HBA and HMBA, it is unknown whether the l-isomers moved faster than the d-isomers, because their d- or l-isomers are not commercially available. It was found that the resolution and migration time of all the a-hydroxy acids increased with increasing amount of 2HP-b-CD in the range of the 2HP-b-CD concentrations used in Fig. 1. Wren and Rowe12 developed a theoretical model relating mobility to the concentration of a CD selector.Also, it has been suggested that maximum resolution can be expected at the optimum CD concentration, Copt = (KAKB)21/2, where KA and KB are equilibrium constants. Table 1 Resolution of enantiomers of four a-hydroxy acids using different cyclodextrins as chiral selectors Resolution (RS) Concen- Cyclodextrin tration/mm LA HBA HMBA HICA a-CD 90 NSa 0.72 0.68 0.82 g-CD 90 NS NS NS 0.47 2,6-Di-O-methyl-b-CD 90 NS NS 1.03 NS 2,3,6-Tri-O-methyl-b-CD 90 NS NS NS 0.57 2HP-a-CD 90 NS NS 0.67 0.68 200 0.79 1.02 1.40 1.26 2HP-b-CD 90 NS 1.14 1.62 1.96 200 1.02 1.91 2.00 3.00 2HP-g-CD 90 NS NS NS 1.17 200 NS NS 1.45 3.62 a NS, not separated. 56 Analyst, 1999, 124, 55–59Therefore, it could be suggested that the 2HP-b-CD concentrations used were still below the optimum concentrations. Also, it was considered that the increased migration times resulted both from complexation and from the increased viscosity of the buffer due to the high cyclodextrin concentration.It was also found that an increase in the bulkiness of the alkyl group in these acids lowered the amount of 2HP-b-CD required for chiral separation. The effect of capillary temperature on the resolution and the migration time of a-hydroxy acids was studied. It was found that a lower capillary temperature caused increases in both the resolution and the migration time of all the a-hydroxy acids. This result was the same as those observed with other chiral compounds.6,26,29–31 According to Heuermann and Blaschke,31 the increase in the Rs value with a decrease in temperature might be explained by a decrease in rotational and/or vibrational energy, increasing the fixing of the enantiomers inside or at the rim of CD and thus increasing the enantioselectivity.Spectrophotometric determination of association constants In order to establish whether these racemic a-hydroxy acids were included in the cavity of 2HP-b-CD or not, the association constants Ka for the inclusion–complexation were measured by spectrophotometry as reported by Matsui and Mochida,9 except that iodine was used as a light-absorbing substance.In the presence of 2HP-b-CD, the absorption maxima of iodine at 287 and 351 nm were shifted to longer wavelengths (290 and 356 nm, respectively), and the absorption intensity was significantly enhanced. As a result, the dissociation constant Kd for the 2HP-b-CD–iodine complex calculated from the absorbance at 290 nm was 1.13 mm and was identical with that calculated from the absorbance at 356 nm.The Ka value was determined by using the absorbance at 356 nm, because the absorbance of the complex at 356 nm was not affected by an increase of the ionic strength to 0.6, but the absorbance at 290 nm was slightly increased. Fig. 2 shows that the addition of dl-HICA to the 2HP-b-CD–iodine system resulted in a decrease in absorbance, indicating that part of the added dl-HICA is included by 2HPb- CD to expel the complexed iodine to the bulk solution.The calculated Ka value of cyclohexanol used as a control was 0.198 ± 0.009 l mmol21, which was lower than the value (0.501 l mmol21) reported for the association of cyclohexanol with b- CD.9 This difference may result from the effect of the substituted group of b-CD and also the fact that the association constant for propranolol binding to b-CD17 is twice that for binding to methyl-b-CD.32 The Ka value of dl-mandelic acid, used as another control having an aromatic ring, was 0.0088 ± 0.0004 l mmol21, which was higher than the values (0.0028 l mmol21 for the d-isomer and 0.0024 l mmol21 for the lisomer) reported for the association of mandelic acid with g- CD.16 It could be suggested that this difference in the values resulted from the difference of the cavity size in these CDs to include the acid.The calculated Ka values of dl-LA, dl-HBA, racemic HMBA and dl-HICA with 2HP-b-CD were 0.0004 ± 0.0001, 0.0044 ± 0.0002, 0.0095 ± 0.0005 and 0.0144 ± 0.0007 l mmol21, respectively.Although these values are 10–100 times lower than those of hydrophobic compounds having aromatic rings with cyclodextrins as reported previously,15,17–20 it could be suggested that these a-hydroxy acids were apparently included in the cavity of 2HP-b-CD. The CD cavity possesses a hydrophobic character and exhibits an affinity for the hydrophobic group of the guest compound.Hence it was found that an increase in the bulkiness of the alkyl group in these acids enhanced the association constants. Calculation of association constants using CE Fig. 1 shows that an increase in the amount of 2HP-b-CD increased the resolution and the migration time of the ahydroxy acids. The resolution increases as the difference in migration time between two enantiomers increases and as the width of peak at the baseline decreases. However, the width of the peak is more difficult to describe mathematically, because it is affected by band broadening due to diffusion and other factors.12 We propose defining the differential time selectivity coefficient, SDT, as the ratio of the average migration time of a pair of enantiomers to the difference in migration time: SDT = (TB 2 TA )/[(TA + TB)/2] (1) Fig. 1 Effect of 2HP-b-CD concentration on the enantiomeric resolution and migration time of a-hydroxy acids. Racemic a-hydroxy acids (2 mm) were separated by CE.The BGE was composed of various concentrations of 2HP-b-CD containing 60 mm phosphate buffer (pH 6.0). 2, LA; ½, HBA; “, HMBA; and 8, HICA. Fig. 2 Effect of dl-HICA addition on the absorption spectra of the 2HP-b- CD–iodine complex. A, 0.04 mm iodine; B, A + 1 mm 2HP-b-CD; C, B + 30 mm dl-HICA; D, B + 75 mm dl-HICA; and E, B + 150 mm dl- HICA. Analyst, 1999, 124, 55–59 57where TA and TB are migration times of enantiomers A and B.The buffer viscosity will affect the mobility of all the ionic species and hence the resultant current. It has been reported that the relative viscosity determined from measuring the current agreed well with that determined by direct measurement.12 According to that report, it was found that the current values for the buffer without 2HP-b-CD and with 200 mm 2HP-b-CD were 45.2 and 19.3 mA to give a ratio of 2.34. Goodall and coworkers17, 33 reported that observed mobilities should be converted into corrected mobilities before calculation of association constants. However, an advantage using SDT is to be able to cancel out the factors such as viscosity and diffusion.Expressing the migration time (TA) in terms of effective mobility of enantiomer (mA) , the operating voltage (V), the total capillary length (L) and the capillary length to the detector (l), yields TA = lL/mAV (2) As a PVA-coated capillary was used, we assumed that electroosmotic flow was negligible.Wren and Rowe12 described a separation model that incorporated equations for mobility and differential mobility. Mobility was defined as mA = (mF + mCKA[C])/(1 + KA[C]) (3) where mF is the electrophoretic mobility of both enantiomers A and B in the free state, mC is the electrophoretic mobility of both enantiomers A and B in the complexed state, [C] is the concentration of 2HP-b-CD and KA and KB are formation constants for inclusion complexes. Substitution of eqns.(2) and (3) into eqn. (1) and rearrangement yield SDT = 2 (mF 2 mC)(KB 2 KA)[C]/ {2mF + (mF + mC)(KA + KB)[C] + 2mCKAKB[C]2} (4) The double reciprocal expression of eqn. (4) yields eqn. (5), which shows the relationship between the 1/SDT and 1/[C]: 1/SDT = mF/(mF2mC) (KB2KA) [C] + (mF + mC) (KA + KB)/ 2 (mF 2 mC) (KB 2 KA) + mC KA KB [C]/(mF 2 mC) (KB 2 KA) (5) The molecular masses of the complexes are 12–18 times higher than those of the free acids, whereas both the free acids and the complexes have the same charge. Therefore, it is thought that the effective mobility of free a-hydroxy acids is much higher than that of complexes.When mF > > mC is postulated for the sake of simplicity, we obtain 1/SDT = 1/(KB 2 KA) [C] + (KA + KB)/2 (KB 2 KA) (6) That is, when 1/SDT is plotted against 1/[C], the slope is 1/(KB 2 KA) and the intercept on the ordinate is (KA + KB)/2 (KB 2 KA). As shown in Fig. 3, a straight line (r > 0.99) was obtained by plotting 1/SDT versus 1/[C]. It was found that the differences between the KB and KA values of these four a-hydroxy acids were very small and that KB and KA values calculated by the CE method were similar to the Ka values determined by spectrophotometry (Table 2).Therefore, it could be suggested that this calculation method, using the CE system with a PVAcoated capillary, was useful for the determination of the association constant. Inhibition of chiral resolution of dl-a-hydroxy acids by cyclohexanol Since cyclohexanol has a high association constant with 2HP-b- CD, as mentioned above, it was hoped that cyclohexanol would act as a competitor between the guest and host molecules.It has been reported that when cyclohexanol was added to the BGE, a higher concentration of b-CD was required for maximum resolution of tioconazole enantiomers.17 In order to confirm the inclusion of the a-hydroxy acids in the cavity of 2HP-b-CD, the effect of cyclohexanol on the resolution was studied by CE (Fig. 4). It was found that the addition of cyclohexanol to the BGE completely prevented the resolution of the a-hydroxy acids and resulted in a decrease in their migration times. It could be suggested that cyclohexanol inhibited the formation of the inclusion complex of the a-hydroxy acids with 2HP-b-CD and that the free enantiomers of each analyte migrated with the same mobility. Therefore, this supported the result obtained by using spectrophotometry that the a-hydroxy acids could be included in the cavity of 2HP-b-CD.Fig. 3 Relationship between 1/SDT and 1/[C]. 2, LA; ½, HBA; “ HMBA; and 8, HICA. Table 2 Association constants between a-hydroxy acids and 2HP-b-CD a-Hydroxy acid KA a l mmol21 KB a l mmol21 Ka b l mmol21 LA 0.00038 0.00042 0.0004 HBA 0.00288 0.00304 0.0044 HMBA 0.0137 0.0144 0.0095 HICA 0.0225 0.0238 0.0144 a KA and KB calculated by electrophoretic experiments are association constants of each enantiomer of the a-hydroxy acids for inclusion complexes with 2HP-b-CD.b Ka measured by spectrophotometry is the association constant between racemic a-hydroxy acids and 2HP-b-CD. Fig. 4 Inhibition effect of cyclohexanol on the enantiomeric resolution of a-hydroxy acids. Racemic a-hydroxy acids (2 mm) were separated by CE using the following BGEs. A, 60 mm 2HP-b-CD containing 60 mm phosphate buffer (pH 6.0); B, A + 180 mm cyclohexanol; C, 200 mm 2HP-b- CD containing 60 mm phosphate buffer (pH 6.0); D, C + 200 mm cyclohexanol.Peaks: 1 = LA; 2 = HBA; 3 = HMBA; and 4 = HICA . l and d are indicated l- and d-isomers, respectively. 58 Analyst, 1999, 124, 55–59Conclusion Direct chiral resolution of aliphatic a-hydroxy acids was performed by CE using 2HP-b-CD. Using two methods, spectrophotometry and CE, we showed that these a-hydroxy acids could be included in the cavity of 2HP-b-CD. It is well known that at least three-point simultaneous interactions are generally needed between a chiral selector and an enantiomer to cause physical separation. It can be assumed that one of the three-point interactions is the inclusion of an alkyl group in the a-hydroxy acid with 2HP-b-CD.As the hydrophobic interaction with the cavity alone is not sufficient to permit the separation of enantiomers,5 it is suggested that two other interactions between the hydroxyl and carboxyl groups on the asymmetric center of a-hydroxy acids and the secondary and/or primary hydroxypropyl groups of the CD are responsible for chiral recognition. We propose defining the differential time selectivity coefficient, SDT, as the ratio of the average migration time of a pair of enantiomers to the difference in migration time.By calculating SDT as a function of 2HP-b-CD concentration, the association constants of the a-hydroxy acids with 2HP-b-CD can be obtained, and they show satisfactory agreement with those obtained by spectrophotometry. There have been many reports on the direct chiral resolution of various compounds with one or more aromatic rings by CE using CDs and their derivatives. However, using the present CE method, we were able to achieve the chiral separation of ahydroxy acids having no aromatic rings, and without having to derivatize the acids.Therefore, the chiral resolution of various organic compounds having no aromatic rings may become possible with CE using native and derivatized CDs.References 1 J. Snopek, I. Jelinek, and E. Smolkova-Keulemansova, J. Chromatogr., 1992, 609, 1. 2 S. Terabe, K. Otsuka and H. Nishi, J. Chromatogr., 1994, 666, 295. 3 M. Novotny, H. Soini and M. Stefansson, Anal. Chem., 1994, 66, 646A. 4 H. Nishi and S. Terabe, J. Chromatogr. A, 1995, 694, 245. 5 S. Fanali, J. Chromatogr. A, 1996, 735, 77. 6 S. Fanali, J. Chromatogr. A, 1997, 792, 227. 7 I. S. Lurie, J. Chromatogr. A, 1997, 792, 297. 8 S. Li and W. C. Purdy, Anal. Chem., 1992, 64, 1405. 9 Y. Matsui and K. Mochida, Bull. Chem. Soc. Jpn., 1979, 52, 2808. 10 E. A. Lewis and L. D. Hansen, J. Chem. Soc., Perkin Trans. 2, 1973, 2081. 11 A. Guttman, A. Paulus, A. S. Cohen, N. Brinberg and B. L. Karger, J. Chromatogr., 1988, 448, 41. 12 S. A. C. Wren and R. C. Rowe, J. Chromatogr., 1992, 603, 235. 13 Y. Y. Rawjee, D. U. Staerk and G. Vigh, J. Chromatogr. A., 1993, 635, 291. 14 Y. Y. Rawjee and G. Vigh, Anal. Chem., 1994, 66, 619. 15 S. A. C. Wren, J. Chromatogr., 1993, 636, 57. 16 I. E. Valko, H. A. H. Billiet, J. Frank and K. Ch. A. M. Luyben, Chromatographia, 1994, 38, 730. 17 S. G. Penn, G. Liu, E. T. Bergstrom, D. M. Goodall and J. S. Loran, J. Chromatogr., A, 1994, 680, 147. 18 K.-H. Gahm and A. M. Stalcup, Anal. Chem., 1995, 67, 19. 19 S. Piperaki, S. G. Penn and D. M. Goodall, J. Chromatogr. A, 1995, 700, 59. 20 K. L. Rundlett and D. W. Armstrong, J. Chromatogr. A, 1996, 721, 173. 21 K.-H. Gahm, L. W. Chang and D. W. Armstrong, J. Chromatogr. A, 1997, 759, 149. 22 A. Steinborn, R. Reinhardt, W. Engewald, K. Wyssuwa and K. Schulze, J. Chromatogr. A, 1995, 697, 485. 23 K. L. Rundlett and D. W. Armstrong, J. Chromatogr. A, 1996, 721, 173. 24 B. E. Kim, K. P. Lee, K. S. Park, S. H. Lee and J. H. Park, Chromatographia, 1997, 46, 145. 25 D. W. Armstrong and J. Zukowski, J. Chromatogr. A., 1994, 666, 445. 26 S. Kodama, A. Yamamoto and A. Matsunaga, J. Chromatogr. A, 1998, 811, 269. 27 A. Nardi, A. Eliseev, P. Bocek and S. Fanali, J. Chromatogr., 1993, 638, 247. 28 I. E. Valko, H. A. H. Billiet, J. Frank, and K. Ch. A. M. Luyben, J. Chromatogr. A, 1994, 678, 139. 29 A. Guttman and N. Cooke, J. Chromatogr. A, 1994, 680, 157. 30 K. D. Altria, D. M. Goodall and M. M. Rogan, Chromatographia, 1992, 34, 19. 31 M. Heuermann and G. Blaschke, J. Chromatogr., 1993, 678, 267. 32 S. A. C. Wren, Electrophoresis, 1995, 16, 2127. 33 L. Knights, G. Liu, A. Ruddick and D. M. Goodall, J. Phys. Chem., 1995, 99, 3875. Paper 8/07351A Analyst, 1999, 124, 55–59 59

ISSN:0003-2654    

DOI:10.1039/a807351a

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Simultaneous determination of cis- and trans-resveratrol in wines by capillary zone electrophoresis J. J. Berzas Nevado,* A. M. Contento Salcedo and G. Casta�neda Pe�nalvo Department of Analytical Chemistry and Foods Technology, University of Castilla-La Mancha, 13071 Ciudad Real, Spain Received 16th September 1998, Accepted 16th November 1998 A capillary zone electrophoresis method for determining cis- and trans-resveratrol isomers is proposed. Optimal conditions for the quantitative separation were investigated.A background electrolyte solution consisting of 40 mM borate buffer adjusted to pH 9.5, hydrodynamic injection and 5 kV of separation voltage were used. Good linearity and precision were obtained for the two isomers. Detection limits of 0.06 mg L21 for trans-resveratrol and 0.08 mg L21 for cis-reveratrol were obtained. The developed method is rapid and sensitive and it has been applied to determine cis- and trans-resveratrol in several red wines.The samples were purified and enriched by passing them through a preconditioned C18 cartridge and eluting the isomers with acetonitrile–water (3 + 7). Introduction The phytoalexin trans-3,5,4A-trihydroxystilbene (resveratrol) was first reported in the skins of grapes1 and later in wines.2–10 Research on the resveratrol content in wines has been due to the interest in the prevention of cancer and heart disease by ingestion of chemical agents that reduce the risk of carcinogenesis. 11–13 Resveratrol inhibits cellular events associated with tumour initiation, promotion and progression. The compound also functions as an antimutagen and has anti-fungal properties. A number of investigations on the resveratrol concentrations of commercial wines have been conducted. HPLC techniques are the most commonly used procedures3,4,7,9,14–17 but gas chromatographic,2,18,19 GC-MS6,20,21 and capillary zone electrophoresis (CZE)22 techniques have also been proposed. Resveratrol exists in two isomeric forms (Fig. 1) that can be present in variable amounts in commercial wines. The trans isomer is transformed to the cis form under UV light.3 The physiological activity of the cis form has not been studied previously, so it is important to distinguish it from the trans isomer and to quantify each separately. Methods to assay cis-resveratrol are less numerous, though more recently various HPLC techniques also became available. 19,23–25 The problem with this isomer is the lack of a suitable standard.The only published procedure to synthesize resveratrol yielded the stable trans isomer as the main product6,26,27 and only this isomer is commercially available at the present time. We have applied similar procedures to those reported previously24,25 to overcome this limitation and to propose the first CZE method to assay both isomers simultaneously. The determination of cis- and trans-resveratrol in wine generally requires the use of extraction and preconcentration techniques prior to CZE, in order to simplify the electropherograms.This is because those compounds are present in wine at very low concentrations and the matrix of the wines is highly complex. Firstly, liquid–liquid extraction procedures have been proposed for sample preparation.28,29 More recently, solid phase extraction has been applied to separate trans-resveratrol and other phenolic compounds13,22,24,25 using mainly solvents such as methanol and ethyl acetate to elute these analytes from C18 columns.In this paper, we propose a rapid method for CZE to determine cis- and trans-resveratrol. We used, prior to the electrophoretic separation, a C18 cartridge for the extraction and preconcentration of these components in wine. The proposed method is simple and faster than by the described HPLC methods previously used for determining cis- and transresveratrol. Experimental Reagents The organic solvents were HPLC grade (methanol and acetonitrile, Sharlau, Barcelona, Spain).Milli-Q water (Millipore, Watford, Herts., UK) was used throughout the study. Trans-resveratrol was purchased from Sigma (St. Louis, MO, USA). A stock standard solution of 250 mg L21 was prepared in acetonitrile and stored in the dark. Cis-resveratrol was prepared from the trans isomer by UV-irradiation.24 Working standard solutions were prepared by diluting the stock standard solution with purified water or wine.A 100 mM sodium tetraborate stock buffer solution was used as background electrolyte. A solution of 0.1 M sodium hydroxide was used for conditioning the capillary. The extraction of trans- and cis-resveratrol was performed in a reverse phase cartridge C18 (Waters Sep-pak, Milford, MA, USA). The cartridge was conditioned before use by means of 5 ml of methanol followed by 5 ml of buffer solution, pH = 7.00 (potasium phosphate monobasic–sodium phosphate dibasic). Fig. 1 Structures of cis- and trans-resveratrol.Analyst, 1999, 124, 61–66 61Apparatus A Beckman (Fullerton, CA, USA) P/ACE System 5510 capillary electrophoresis equipped with a diode-array UV/VIS detector and controlled by a Dell DIMENSION P133V with P/ ACE station software was used. The separation was made using a fused silica capillary (25 cm 3 75 mm id) maintained in a cartridge with a detection window of 100 3 800 mm. Operating conditions The capillary was conditioned prior to its first use by flushing with 0.1 M NaOH for 10 min then with water for 5 min and finally with the electrolyte solution for 5 min.The running buffer was 40 mM sodium tetraborate (pH = 9.5) with a voltage of 5 kV, average current of 40 mA and temperature of 25 °C. Samples were injected by hydrodynamic injection for 5 s. Electropherograms were recorded at 320 nm. The capillary was flushed between two separations with 0.1 M NaOH (1 min), ultrapure water (1 min) and fresh buffer (2 min).Duplicate injections of the solution were performed and average peak areas corrected (area/migration time) were used for the quantitation. Sample preparation Wine (25 ml) was poured into a beaker and neutralized to pH 7.00 by means of NaOH. The cartridge was then slowly loaded with the neutralized wine. After, the cartridge was washed with 2 ml of water, 2 ml of buffer solution, pH 7.00, and 4 ml acetonitrile–water (1 + 9). Cis- and trans-resveratrol were then eluted with 4 ml acetonitrile–water (3 + 7).The cartridge was then washed with 5 ml of methanol and 5 ml of buffer solution. Results and discussion Optimization of the electrophoretic procedure Effect of electrolyte pH. The pH of the running electrolyte had a significant impact on the ionization of the acidic silanols of the capillary wall and on the electrophoretic mobilities of the isomers studied. A pH in the range 7.0–10.7 was chosen for method development to separate cis- and trans-resveratrol by CZE.In order to determine the optimum pH value, a set of electrolytes at several pH values were tested. The influence of the electrolyte pH on the migration times of the studied isomers is depicted in Fig. 2. A pH of 9.5 was selected as optimum in order to minimize analysis times with a good resolution between peaks. Effect of ionic strength of electrolyte. The optimum ionic strength of the electrolyte must be a balance between an acceptably low current to minimize the noise and a good peak efficiency.The effect of the concentration of buffer solution from 10 to 50 mM on the migration time of the isomers is shown in Fig. 3. As can be seen, when the concentration of buffer increases the migration times of cis- and trans-resveratrol also increase. A concentration of 40 mM of buffer was selected to maintain good peak shape and low current in order to minimize the noise and baseline aberrations. Effect of voltage applied. The electroosmotic flow-rate (EOF) and the velocity of migrating isomers are proportional to the applied voltage used for separation.Application of a high voltage reduces analysis times, but may lead to significant losses of resolution and peak efficiencies, because excessive heating occurs within the capillary. The choice of operating voltage should be optimized in conjunction with the choice of electrolyte concentration, capillaimensions and temperature to produce an acceptable current level.The influence of the applied voltage upon the migration times is shown in Fig. 4. As expected when the voltage increases from 3 to 10 kV, the migration times of both isomers decrease as well Fig. 2 Influence of electrolyte pH. Operating conditions: variable pH buffers (40 mM), 5 kV and 25 °C. Fig. 3 Influence of buffer molarity. Operating conditions: variable buffer molarity (pH = 9.5), 5 kV and 25 °C. Fig. 4 Influence of voltage. Operating conditions: 40 mM borate buffer, pH = 9.5, variable voltage and 25 °C. 62 Analyst, 1999, 124, 61–66as the resolution between peaks. Therefore, the earlier value of 5 kV was employed in all the studies because good resolution between peaks was obtained and with higher values high currents were observed. Optimization of the washing step. It is important to maintain a consistent EOF from run to run since any variation results in poor migration time precision. Sample components can become adsorbed onto the capillary surface and change the effective charge on the wall, resulting in a change in EOF.To prevent difficulties owing to adsorption and to ensure a consistent EOF, the capillary is flushed between injections with a dilute NaOH solution that strips the top surface of the capillary wall. A 0.1 M NaOH solution was used to rinse the capillary beween injections. Samples were analysed following varying rinsing times from 1 to 5 min. Good precision in the migration times were observed in this interval of time.A time of 1 min was selected as suitable to obtain repeatable migration times (ten sequential injections). After the rinsing with 0.1 M NaOH, a 2 min rinsing with the electrolyte was chosen for the preparation of the capillary before performing the sample injection. From these studies, the following electrophoretic conditions were selected: electrolyte, 40 mM Na2B4O7, pH 9.5; voltage, 5 kV; capillary, fused-silica (25 cm 3 75 mm id); injection, hydrodynamic, 5 s; temperature, 25 °C; detection signal, 320 nm.Fig. 5 shows the electropherogram corresponding to the standards of cis- and trans-resveratrol. From this electropherogram, it can be ascertained that the selected electrophoretic procedure is excellent for the separation of both isomers. As it can be seen in this figure, the determination of cis- and transresveratrol in wines is faster by the proposed method in this work (CE) than LC (typical analysis times of 6.5 min in CE and at least 10 min in LC13,23–25).Solid-phase extraction in a reversed phase cartridge First, the method was applied to the analysis of wines which had not been subjected to any special treatment, but due to the presence of a large quantity of various interferent compounds and the low concentration of resveratrol, it was necessary to extract the compounds of interest to obtain a cleaner electropherogram. C-18 cartridges were used to extract cis- and transresveratrol.Variables such as organic solvent, proportion and volume of organic solvent : water ratio in order to elute the analytes free from interferences were studied. A cleaner electropherogram was obtained when acetonitrile– water (3 + 7) was utilized to desorb the analytes; previously, the cartridge charged with the wine sample was washed with 2 mL of pH 7.0 buffer solution and 4 mL of acetonitrile–water (1 + 9) in order to minimize the interferences. Finally cis- and transresveratrol were eluted with 4 mL acetonitrile–water (3 + 7).This volume of eluent was found to be enough to elute quantitatively the analytes at the concentration levels present in the wine samples. Fig. 6 shows a representative electropherogram resulting from the analysis of a red wine. As can be seen, good resolution between the interferent compounds of wine and the compounds analyzed was obtained. Perfomance of the method Stability of the solutions. Although this test is often considered as part of the ruggedness of the procedure, it should be carried out at the beginning of the procedure validation because it determines the validity of the data of the other tests. At the beginning, a stock standard solution of transresveratrol was obtained by dissolving appropriate amounts of this compound in 96% ethanol (v/v).This solution was protected from light and stored at 4 °C. The stability was established by preparing fresh solutions daily.The stock solution was diluted with water by a factor of 20 in a calibrated flask and carrying out a CZE separation step at 25 °C using a potential of 5 kV, 40 mM borate, pH 9.5 and 5 s of injection time. Analyses were repeated every day for 3 d and then at 7 and 14 d. It could be seen that, for 3 d, the trans-resveratrol stock solution showed unchanging electropherograms but, for a week, retention time of resveratrol increased considerably. In order to confirm this fact, the UV absorption spectra were obtained at the apex peaks.The overlay spectra are presented in Fig. 7. Different spectra were obtained as might have been expected. A similar study was carried out for resveratrol in acetonitrile. In this solvent, the response factors of standard solutions were found to be unchanged for at least up to 10 d. Less than a 0.2% concentration difference was found between the solutions freshly prepared and those aged for 10 d. Linearity. The calibration graphs for trans-resveratrol were produced from results obtained by injecting standard solutions in the range 0.5–20 mg L21.For cis-resveratrol calibration, aliquots of the trans isomer stock solution were diluted in water to cover the range 2–30 mg L21. A portion of each standard was irradiated for 30 min at Fig. 5 Electropherogram of a standard mixture of cis- and transresveratrol. Operating conditions: 40 mM borate buffer, pH = 9.5, 5 kV and 25 °C. Fig. 6 Electropherogram of a sample of red wine.Operating conditions: 40 mM borate buffer, pH = 9.5, 5 kV and 25 °C. Analyst, 1999, 124, 61–66 63254 nm. Whereas the non-irradiated standard only contained the single peak of trans-resveratrol, this peak was diminished upon irradiation and was preceded by an earlier peak which was shown to be cis-resveratrol in an amount identical to the decrease in the trans-isomer (Fig. 8). Values for the cisresveratrol standards were therefore assigned on the basis of the decrease in trans-resveratrol following irradiation. Each point of the calibration graph corresponded to the mean value obtained from three independent area measurements.The corresponding regression equation and other characteristic parameters for the determination of both isomers are show in Table 1. The regression line passed through the origin with tcalc = 0.35 for cis-resveratrol and 1.93 for trans-resveratrol (P > 0.05). Precision. The precision of the proposed method for determining cis- and trans-resveratrol is expressed in terms of relative standard deviation (RSD).In order to test the precision of the electrophoretic procedure, eight injections of a standard of 4 and 5 mg L21 of cis- and trans-resveratrol, respectively, were carried out sequentially. This operation was repeated over 3 d. The precision of the migration time and peak area corrected were excellent with RSD (%) values (n = 24) of 0.25 and 0.8 for migration time and 1.3 and 1.8 for peak area corrected for trans- and cis-resveratrol respectively.To evaluate the extraction method, a wine sample spiked with 2 mg L21 of trans- and cis-reveratrol were analysed independently six times. This analysis was repeated over two days. The average of the recoveries from the spiked wine was 104.17 and 108.60 with RSD (%) of 1.95 and 6.50 for trans- and cisresveratrol respectively. Furthermore, six replicate analyses were performed on a wine with cis- and trans-resveratrol, the procedure was repeated over two days.The average content of trans- and cis-resveratrol was 1.60 ± 0.06 mg L21 and 1.07 ± 0.06 mg L21, respectively, with the confidence intervals evaluated at P = 0.05. These values are in agreement with the extrapolated values found using the method of the standard additions. Recovery. In order to test the accuracy of the proposed method, several aliquots of irradiated trans-resveratrol standards were added into a wine that had none present.These samples were analysed using the extraction and electrophoretic procedure described in this work. Good results were obtained, as can be seen in Table 2. In all the cases triplicate analyses were made. Limits of detection (LD) and quantitation (LQ). The LD and LQ were calculated by measuring six blanks, using the maximun sensitivity allowed by the system and calculating the standard deviation of this response. LD was estimated by multiplying the standard deviation by a factor of three.The LQ was defined as ten times the standard deviation. The LD and LQ obtained considering a concentration factor of 6.25 for each isomer, are summarized in Table 1. The LQ was subsequently validated separately by the analysis of six standards prepared at 0.25 mg L21 for trans-resveratrol and at 0.30 mg L21 for the cis-isomer. Analysis of wine samples To demonstrate the usefulness of the extraction and CZE methods developed, several wines produced in Castilla-La Mancha (Spain) were analyzed. Fig. 7 Absorption spectra of trans-resveratrol (8 mg L21) (A) freshly prepared in ethanol and (B) after seven days prepared in ethanol. Fig. 8 (A) Electropherogram of a standard of trans-resveratrol; (B) electropherogram of a standard of trans-resveratrol irradiated for 30 min. Table 1 Statistical parameters cis-Resveratrol trans-Resveratrol Equation Y = (280.84 ± 230) + (1296.91 ± 53)X Y = (2847 ± 440) + (1750.79 ± 36)X r 0.9967 0.9991 LD/mg L21 0.080 0.064 LQ/mg L21 0.30 0.25 64 Analyst, 1999, 124, 61–66The use of a photodiode detector allowed us to confirm the identity of the peak not only by its migration times, but also by the overlay of the UV-VIS spectra with a standard.The techniques studied for validating the peak purity corresponding to cis- and trans-resveratrol in the analysed wines were: normalising and comparing spectra from several peak sections; and absorbance at two wavelengths.Both techniques demonstrate the purity of the obtained peak in all the cases. The results obtained from wine samples are given in Table 3. All determinations were carried out in triplicate. In order to evaluate the possible matrix effect, the method of standard addition was used for the determinations of these isomers in wines. In both cases the application of the t-test for the slopes of the calibration graphs showed no significant statistical differences. Consequently there is no evidence of systematic error affecting the determination of cis- and trans-resveratrol in wine by the proposed method.The concentrations found by using this method are shown in Table 3 and, as can be seen, they coincide with those obtained without standard addition by the proposed method. Conclusion In this work, a method is described for the extraction and determination of trans- and cis-resveratrol in wine by CZE. Although trans-resveratrol has been previously determined in wine by CZE, this is the first report on the determination of cisresveratrol concentrations in wine by CZE.It could be concluded that CZE can be an alternative to traditional existing methods for the determination of trans- and cis-resveratrol in wine. The proposed method is faster than those previously proposed for the determination of both isomers. The results obtained concerning linearity, recovery, precision and sensitivity were highly satisfactory and comparable to those obtained by the proposed methods in the literature. The developed method allows the determination of cis- and trans-resveratrol at low levels with detection limit of 0.08 mg L21 for the cis-isomer and 0.06 mg L21 for trans-resveratrol.References 1 L. L. Creasy and J. J. Coffee, J. Am. Soc. Hortic. Sci., 1988, 113, 230. 2 P. Jeandet, R. Bessis and B. Gantheron, Am. J. Enol. Vitic., 1991,42, 41. 3 E. H. Siemann and L. L. Creasy, Am. J. Enol. Vitic., 1992, 43, 49. 4 R. M. Lamuela-Raventós and A. L. Waterhouse, J. Agric. Food Chem., 1993, 41, 521. 5 F. Mattivi, Z. Lebensm. Unters. Forsch., 1993, 196, 522. 6 P. Jeandet, R. Bessis, B. F. Maune and M. Sbaghi, J. Wine Res., 1993, 4, 79. 7 J. P. Roggero and P. Archie, Sci. Aliments, 1994, 14, 99. 8 K. D. McMurtrey, J. Minn, K. Pobenz and T. P. Schultz, J. Agric. Food Chem., 1994, 42, 2077. 9 R. Pezet, V. Pont and P. J. Cuenat, J. Chromatogr. A, 1994, 663, 191. 10 D. M. Goldberg, J. Yan, E. Ng, E.P. Diamandis, A. Karumanchiri, G. Soleas and A. L. Waterhouse, Am. J. Enol. Vitic., 1995, 41, 50. 11 M. B. Sporn and D. L. Newton, Fed. Proc., 1979, 38, 2528. 12 M. Jang, L. Cai, G. O. Udeani, K. V. Slowing, C. E. Thomas, C. W. W. Beecher, H. H. S. Fong, N. R. Farnsworth, A. D. Kinghorn, R. G. Mehta, R. C. Moon and J. M. Pezzuto, Science, 1997, 275, 218. 13 D. M. Goldberg, E. Tsang, A. Karumanchiri, E. P. Diamandis, G. Soleas and E. Ng, Anal. Chem., 1996, 68, 1688. 14 M.Fregoni, L. Bavaresco, M. Petegolli, M. Trevisan and C. Ghebbioni, Vignevini, 1994, 21, 33. 15 F. Mattivi, Riv. Vitic. Enol., 1993, 41, 37. 16 F. Mattivi and G. Nicolini, L’Enotecnico, 1993, 29, 81. 17 J. P. Roggero, P. Archier and S. Coen, Sci. Aliment., 1992, 12, 37. 18 R. Barlass, M. Miller and T. J. Douglas, Am. J. Enol. Vitic., 1987, 38, 65. 19 P. Langcake and W. V. McCarthy, Vitis, 1979, 18, 244. 20 D. M. Goldberg, A. Karumanchiri, E. P. Diamandis, J. Yan, G. L. Soleas and A.L. Waterhouse, Technical Abstracts-American Society for Enology and Viticulture, Sacramento Community Convention Center, Sacramento, CA, June 22–25, 1993. 21 G. J. Soleas, D. M. Golberg, E. P. Diamandis, A. Karumanchiri, E. Ng and J. Yan, Technical Abstracts-American Society for Enology and Table 2 Recoverya trans-Resveratrol cis-Resveratrol Added/mg L21 Found/mg L21 Recovery (%) Added/mg L21 Found/mg L21 Recovery (%) Sample 1 0.9 0.95 ± 0.07 106.1 ± 9.5 0.2 0.23 ± 0.01 115.0 ± 4 Sample 2 1.5 1.63 ± 0.16 108.0 ± 12.9 0.4 0.34 ± 0.05 89.5 ± 4.8 Sample 3 2.0 1.95 ± 0.22 97.5 ± 10.9 0.7 0.8 ± 0.02 111.1 ± 2.2 Sample 4 2.5 2.66 ± 0.10 106.6 ± 4.2 0.96 1.06 ± 0.02 110.3 ± 2.1 a Results are mean ± s (n = 3).Table 3 Analysis of wines Standard addition/mg L21 Direct measurement/mg L21 Wine trans-Resveratrol cis-Resveratrol trans-Resveratrol cis-Resveratrol Rias bajas 0.40 0.29 0.43 ± 0.07 0.32 ± 0.04 Se�norio Llanos 1.46 0.98 1.60 ± 0.06 1.07 ± 0.06 Estola 2.18 —b 2.05 ± 0.07 — Vi�na Albali 0.6 0.53 0.65 ± 0.07 0.49 ± 0.02 D. Eugenio — — — — Vi�na Cuerva — — — — Yuntero — — — — Tomillar 0.73 0.25 0.74 ± 0.07 0.3 ± 0.02 Cason Histórico — — — — Campo Bello — — — — Pata Negra 0.88 0.11 0.91 ± 0.02 0.12 ± 0.02 Caserio Vigón — — — — a Mean value ± s (n = 3). b Not detected. Analyst, 1999, 124, 61–66 65Viticulture. Sacramento Community Convention Center, Sacramento, CA, June 22–25, 1993. 22 L. Arce, M. T. Tena, A. Rios and M. Valcárcel, Anal. Chim. Acta, 1998, 359, 27. 23 R. M. Lamuela-Raventós, A. I. Romero-Perez, A. L. Waterhouse and M. C. de la Torre-Boronat, J. Agric. Food Chem., 1995, 43, 281. 24 D. M. Goldberg, E. Ng, A. Karumanchiri, E. P. Diamandis and G. Soleas, J. Chromatog. A, 1995, 708, 89. 25 A. Gonzalo and P. Vidal, Alimentacion, 1995, 14(8), 67. 26 M. Moreno-Manas and R. Pleixats, Anal. Quim., 1985, 81, 157. 27 D. M. Goldberg, J. Yan, E. P. Diamandis and G. J. Soleas, Clin. Biochem., 1993, 26, 126. 28 C. Garcia-Viquera and P. Bridle, Food Chem., 1995, 54, 349. 29 G. Cartoni, F. Coccioli and A. Jasionowska, J. Chromatogr. A, 1995, 709, 209. Paper 8/07226D 66 Analyst, 1999, 124, 61&nd

ISSN:0003-2654    

DOI:10.1039/a807226d

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Determination of polyaromatic hydrocarbons and some related compounds in industrial waste oils by GPC-HPLC-UV C. Ner�ýn* and C. Dome�no Department of Analytical Chemistry, Centro Polit�ecnico Superior, University of Zaragoza, Maria de Luna 3, 50015 Zaragoza, Spain Received 29th September 1998, Accepted 3 November 1998 The determination of polyaromatic hydrocarbons (PAHs) in industrial waste oils was studied. The optimized procedure consisted in using low pressure gel permeation chromatography to separate the fraction containing PAHs and PCBs from the matrix.Both groups of contaminants were further fractionated on a small column of 3% deactivated alumina, using hexane to elute the PCB fraction and then hexane–dichloromethane (95 + 5) to elute PAHs. The final analysis was carried out by HPLC on a 2(1-pyrenyl)ethyldimethylsilica column with hexane as mobile phase in the isocratic mode and with UV detection at three different wavelengths, to quantify 16 priority PAHs.Recoveries were between 89 and 99%. Four different industrial waste oil samples were analyzed and the results obtained are discussed. Introduction Industrial waste oils are classified as toxic residues owing to the presence of toxic chemicals such as heavy metals, PCBs and PAHs, among others. Althougth the PAHs are not included in the legislation for waste oils, either in the USA or in Europe, their study is generally recommended since waste oils can be used in different combustion processes.1 As the volume of these residues is increasing, some alternatives have been proposed which include the recycling or the combustion approaches.However, the succes of any proposal depends critically on the composition of the waste oils. In this context, analytical methodology for the determination of PCBs and PAHs is necessary. Previous work has been carried out with heavy metals2 and PCBs,3 but very few data exist on PAHs in waste oils, which was the main objective of this work.The determination of organic chemicals in waste oils is not an easy task, mainly owing to the lipidic and oily matrix, which has to be eliminated without modifying the analytes, PAHs in this case. Common methods for this step are saponification with alcoholic KOH, the use of sulfuric acid,4 gel permeation chromatography5–7 and the use of solid adsorbents. Saponification is only appropriate for very stable chemicals. Gel permeation is non-destructive, but it is often not efficient enough8 and usually requires a second clean-up step. The use of solid adsorbents such as silica, Florisil and alumina9–11 placed in columns (solid phase extraction) is not always appropriate for eliminating high amounts of lipids.In this work, a systematic study of several procedures applied to the determination of some priority PAHs established by the EPA12 and naphthalene and biphenyl in industrial waste oils was carried out. Experimental Apparatus A Hewlett-Packard (Palo Alto, CA, USA) HPLC 1050 Series chromatograph equipped with both autosampler and UV detector was used.The analytical column used was a Cosmosil packed with 2-(1-pyrenyl)ethyldimethylsilica (5 PYE), size 250 3 4.6 mm id. Hexane was used as the mobile phase at a flow rate of 1 ml min21. A Model 991 diode array detector coupled to the HPLC system was used to identify the PAHs in waste oils. Gel permeation chromatography (GPC) was carried out with an HPLC pump [Kontron (Zurich, Switzerland) LC-Pump T- 414] connected to a manual injector [Rheodyne (Cotati, CA, USA) Model 756], with an injection loop of 2 ml, a glass column of 40 cm 3 2 cm id filled with BioBeads SX3 and a Perkin-Elmer (Norwalk, CT, USA) Lambda 3 UV spectrophotometer equipped with a flow cell of 5 ml for detection.The detector response from GPC was monitored with a Perkin- Elmer Model 561 recorder. As the eluent, cyclohexane– dichloromethane (70 + 30) was used at a flow rate of 1 ml min21.A stainless steel filter equipped with an internal grid as support, appropriate fittings and a Millipore (Bedford, MA, USA) membrane filter of 25 mm diameter and 0.45 mm pore size was placed between the injector and the column, to prevent damage to the gel by particles carried by the oil. This filter was changed before each injection. All transfer lines were of PTFE capillary tubing. Reagents Naphthalene, biphenyl, dibenzofuran, acenaphthylene, fluorene, phenanthrene, anthracene, pyrene, fluoranthene, benz[a]- anthracene, chrysene, benzo[k]fluoranthene, benzo[a]pyrene, dibenz[a,h]anthracene, indene and benzo[ghi]perylene were supplied as certified standards by the Environmental Protection Agency, USA. Hexane, cyclohexane and dichloromethane (DCM) (residue analysis grade) were supplied by Merck (Darmstadt, Germany).Individual solutions of each compound were prepared in hexane. Joint standard solutions were obtained gravimetrically from each independent solution and appropriate dilution with hexane.All the standards were gravimetrically controlled. Samples Four different samples of waste oils were taken for analysis. All of them were supplied by an authorized manager of industrial Analyst, 1999, 124, 67–70 67toxic residues. These waste oils were classified according to their origin in automotive, hydraulic, machine and cutting oils. Elimination of lipids A sulfuric acid–silica impregnated column was prepared as follows.The silica was first activated at 550 °C in an oven for 5 h. Then, 50% m/m of 98% sulfuric acid was added to the silica and the mixture was shaken for 30 min to homogenize it. A 40 3 2 cm id glass column was carefully filled with this mixture. The GPC column was prepared as follows. A 50 g amount of BioBeads SX3, (a spherical porous styrene–divinylbenzene copolymer with 3% cross-linking), was placed in a 500 ml flask and mixed with 100 ml of the GPC elution solvent [cyclohexane –dichloromethane (70 + 30 v/v)] and the gel was left to stand for 24 h at 4 °C.The fully swollen gel was then de-gassed by applying a vacuum to the flask, before filling the column with the slurry. Elution solvent was pumped through the column at a flow rate of 1 ml min21 for 1.5 h prior to use. Quantification Calibration solutions from 100 to 900 ng g21 were prepared gravimetrically in hexane and 25 ml of each solution were injected into the 5 PYE HPLC column at a 1 ml min21 flow rate with hexane.In this way, a calibration plot for each individual compound was obtained for quantification. Signals were collected and integrated at 250, 320 and 376 nm, according to the procedure reported previously.13 This system allows the individual quantification of all the compounds under study based on the different absorption ability of each one at the mentioned wavelength. At 250 nm the peak obtained at 3.9 min corresponded to the sum of dibenzofuran and acenaphthylene.The peak at 4.33 min was the sum of phenanthrene and anthracene and the peak at 4.83 min was the sum of pyrene and fluoranthene. The signal obtained at 376 nm corresponded to anthracene (4.33 min) and fluoranthene (4.83 min) when working at 320 nm. The peak obtained at 3.90 min corresponded to acenaphtylene. The same procedure was applied to the clean extracts of waste oils. Detection limits were calculated by injecting successively diluted standard solutions into the HPLC system until the signal was no longer detected.The values obtained ranged from 0.2 ng for biphenyl to 1.4 ng for chrysene. Separation of PAHs A 5 PYE HPLC column was used for separation of individual compounds using hexane as the mobile phase at a flow rate of 1 ml min21, in the isocratic mode, according to the previous optimization.13 Results and discussion Elimination of the matrix As mentioned above, the matrix has to be eliminated without altering the composition of the analytes.This elimination can be carried out by chemical treatment using, for example, sulfuric acid or KOH, or by physico-chemical treatment through a size exclusion process. Chemical treatment has been succesfully applied to the elimination of lipids for the determination of PCBs. As, in principle, PAHs are stable enough to persist through a saponification step, this was the firsedure attempted. However, when a solution of 16 PAHs was added to the sulfuric acid–silica impregnated column and then eluted with hexane– dichloromethane (75 + 25) at 1 ml min21, the 50 ml fraction collected and analyzed following the recommended procedure showed only two peaks, corresponding to naphthalene and biphenyl.Consequently, these results demonstrate that this procedure cannot be applied to this sample. A non-destructive method such as GPC was studied. PAHs were recovered in the fraction from 35 to 75 ml when a flow rate of 1 ml min21 of cyclohexane–dichloromethane (70 + 30) was used as the mobile phase and the lipid fraction appeared in the first 35 ml.However, when the waste oil was injected into the GPC column, the bed became gray owing to the small particles carried by the oil and a new column was required. To avoid this problem, a removable membrane filter of 0.45 mm was placed between the injector and the GPC column to remove the particles. This filter was changed after each injection of waste oil.To study the recoveries on the GPC column, a standard solution containing about 400 ng g21 of each compound was used. Table 1 gives the recoveries obtained. It can be seen that only benzo[ghi]perylene has a recovery lower than 75%. This low value could be attributed to adsorption effects on the gel column. Clean-up The fraction eluted by GPC still contains a small amount of lipids and co-extracted materials which have to be eliminated.Furthermore, waste oils also contain PCBs which are eluted from the GPC column together with the PAHs and a further fractionation step is necessary, since some PCBs and PAHs are also co-eluted in the 5 PYE analytical column used.13 We studied 3% deactivated silica and alumina beds filled in glass columns of 20 3 0.6 cm id. Hexane was used as the mobile phase to elute the PCBs and hexane–DCM (95 + 5) to elute PAHs. A study of the elution profiles showed that PCBs were eluted with 10 ml of hexane from the silica column and with 15 ml from the alumina column.PAHs were eluted with hexane–DCM (95 + 5); 25 ml were necessary for the silica column and 20 ml for the alumina. In both cases, PCBs were eluted with a less polar solvent than PAHs, which were efficiently trapped by the adsorbents and were only released from the bed by increasing the polarity of the mobile phase. Table 1 Recoveries of 16 PAHs obtained after GPC separation. Standard solution in hexane of 400 ng g21 for each PAH Compound Mean concentrationa/ ng g21 (n = 4) Recovery (%) Naphthalene 350 ± 9 80 Biphenyl 325 ± 10 78 Dibenzofuran 385 ± 7 82 Acenaphthylene 373 ± 3 89 Fluorene 446 ± 6 90 Phenanthrene 386 ± 2 95 Anthracene 350 ± 3 92 Pyrene 328 ± 4 90 Fluoranthene 326 ± 2 89 Benz[a]anthracene 362 ± 4 91 Chrysene 376 ± 2 93 Benzo[k]fluoranthene 375 ± 4 92 Benzo[b]fluoranthene 376 ± 3 93 Benzo[a]pyrene 348 ± 3 91 Dibenz[a,h]anthracene 356 ± 3 94 Indene 334 ± 2 91 Benzo[ghi]perylene 279 ± 5 65 a �x ± sf, s = standard deviation, f = 1.6 (n = 4) at the 95% confidence level. 68 Analyst, 1999, 124, 67–70A recovery study with both columns (3% alumina and 3% silica) was carried out with standard solutions containing 300 ng g21 of each of the 16 PAHs. Table 2 gives the results obtained for all the PAHs under study. Comparing the chromatograms obtained with the fraction eluted from both silica and alumina columns, a better baseline was observed from the alumina column and consequently this was selected for further studies.Fig. 1 shows a general scheme of the whole analytical procedure. Determination of PAHs in waste oils Four different samples of industrial waste oils were treated and analyzed using the optimized procedure described above. Individual PAHs in the oil samples were identified through the UV spectrum obtained using a diode array detector connected to the 5 PYE column. The study carried out with diode array detection showed that other co-eluted compounds were not present with the mentioned PAHs and no other unknown PAHs appear.The identification was confirmed by GC-MS of the fraction collected from the alumina column. Table 3 gives the results obtained. As can be seen, naphthalene is the most concentrated in all the waste oils. These results agree with the Table 2 Recoveries of PAHs in silica (3%) and alumina (3%) clean-up procedures. Standard solution in hexane of 300 ng g21 for each PAH Mean recoverya (%) Compound Silica Alumina Naphthalene 89 ± 5 90 ± 6 Biphenyl 91 ± 9 92 ± 7 Dibenzofuran 90 ± 7 93 ± 9 Acenaphthylene 88 ± 8 89 ± 10 Fluorene 92 ± 9 91 ± 11 Phenanthrene 97 ± 3 98 ± 4 Anthracene 98 ± 4 99 ± 5 Pyrene 95 ± 6 97 ± 7 Fluoranthene 94 ± 7 98 ± 6 Benz[a]anthracene 87 ± 4 95 ± 5 Chrysene 97 ± 7 101 ± 8 Benzo[k]fluoranthene 85 ± 10 95 ± 11 Benzo[b]fluoranthene 80 ± 9 94 ± 12 Benzo[a]pyrene 78 ± 9 98 ± 8 Dibenz[a,h]anthracene 76 ± 7 94 ± 6 Indene 81 ± 11 91 ± 9 Benzo[ghi]perylene 74 ± 9 89 ± 11 a �x ± sf; s = standard deviation, f = 1.6 (n = 4) at the 95% confidence level.Fig. 1 Schematic diagram of the overall analytical procedure. Table 3 Slope of calibration plot for each PAH Compound Slope/mV g ng21 Naphthalene 0.603 Biphenyl 3.394 Dibenzofuran 3.432 Acenaphthylene 27.833 Fluorene 2.682 Phenanthrene 9.239 Anthracene 18.173 Pyrene 1.289 Fluoranthene 4.163 Benz[a]anthracene 3.607 Chrysene 3.446 Benzo[k]fluoranthene 4.163 Benzo[b]fluoranthene 3.130 Benzo[a]pyrene 2.736 Dibenz[a,h]anthracene 0.382 Indene 6.270 Benzo[ghi]perylene 0.756 Fig. 2 Chromatogram of a real waste oil sample (automotive oil). Analyst, 1999, 124, 67–70 69study of Ouzzani et al.14 which established that industrial virgin oils contain naphthalene. Fig. 2 shows the chromatogram of a sample of automotive waste oil. Clear differences can be pointed out within the samples. Hydraulic oil is the least contaminated by PAHs, as expected since this application does not require a heating process at high temperature and it is well known that PAHs are mainly formed in high temperature applications where organic compounds are involved.In contrast, the machinery oil is the most contaminated, with a higher number of PAHs than the other oil samples. The only sample containing benzo[a]pyrene, one of the most toxic PAHs,15 is the machinery oil. Automotive oil shows a high concentration of pyrene, as expected since it has been identified in the exhaust gases from cars.16 From an environmental point of view, the most dangerous samples are both the machinery and automotion oils.Conclusions The determination of PAHs in waste oils requires the elimination of the oily matrix and the isolation and fractionation of the different groups of organic chemicals such as PCBs and PAHs. In the case of PAHs, the elimination of the oily matrix cannot be carried out with sulfuric acid because most of PAHs are chemically degraded, and GPC with cyclohexane–DCM (70 + 30) as the mobile phase was shown to be an efficient procedure.Further fractionation of the eluted fraction to separate PCBs and PAHs is necessary and the best system is to use solid phase extraction with 3% deactivated alumina column. The elution of PAHs from this column requires a slightly polar solvent such as hexane–DCM (95 + 5) whereas PCBs are eluted with hexane in the first 15 ml fraction. Several differences in the nature and concentration of PAHs can be found when waste oils of different origin (automotive, hydraulic, machine, lamination) are analysed and these differences can be attributed to the previous use of the oil.Acknowledgements This work has been financed by project 89/96 CONSID (Diputaci�on General de Arag�on), Spain. References 1 Franklin Associates, EPA Report No. EPA/530-5W-013, Environmental Protection Agency, Washington, DC, 1985. 2 C. Nerín, C.Domeño, A. del Alamo and I. Echalyst, 1996, 121, 1731. 3 C. Nerín, C. Domeño, I. Echarri and A. R. Tornés, Toxicol. Environ. Chem., 1996, 56, 1. 4 J. Koistinen, Chemosphere, 1992, 24, 559. 5 P. de Voogt, Chemosphere, 1991, 23, 901. 6 M. K. L. Bicking and R. L. Wilson, Chemosphere, 1991, 22, 437. 7 R. A. Kern, K. O’Hara, S. Yaikow, J. L. Cercone, E. Anderson, L. Moore and M. J. Cava, J. Autom. Chem., 1991, 13, 243. 8 K. Grob and I. Kälin, J. High Resolut. Chromatogr., 1991, 14, 451. 9 H. N. Kayali, S. Rubio-Barroso and L. M. Polo Díez, J. Liq. Chromatogr., 1994, 17, 623. 10 G. Codina, M. T. Vaquero, L. Comellas and F. Broto-Piug, J. Chromatogr. A, 1994, 673, 21. 11 L. Nondek, M. Kuzilek and S. Krupicka, Chromatographia, 1993, 37, 381. 12 EPA, Code of Federal Regulations, 40 CFR, 1986, vol. 60, Ap. A, Environmental Protection Agency, Washington, DC. 13 C. Nerín, C. Domeño, P. Fernández and J. Cacho, Quím. Anal., 1998, 17, 75. 14 L. Ouzzani, M. Caude and R. Rosset, Int. J. Environ. Anal. Chem., 1992, 47, 137. 15 I. C. T. Nisbet and P. K. Layoy, Regul. Toxicol. Pharmacol., 1992, 16, 290. 16 F. Valeria and M. Pala, Fresenius’ J. Anal. Chem., 1991, 339, 777. Paper 8/07576J Table 4 Concentrations of PAHs found in real waste oils Oil PAH Concentration/ mg g21 RSD (%) (n = 4) Cutting oil Naphthalene 659 7 Dibenzofuran 51 9 Phenanthrene 13 8 Benzo[k]fluoranthene 2 9 Benzo[b]fluoranthene 2 5 Hydraulic oil Naphthalene 1930 7 Acenaphthylene 0.9 3 Anthracene 0.2 7 Automotive oil Naphthalene 1681 11 Biphenyl 154 7 Dibenzofuran 191 9 Phenanthrene 28 7 Anthracene 0.1 6 Pyrene 61 8 Benz[a]anthracene 5 7 Machine oil Naphthalene 1854 6 Dibenzofuran 316 10 Phenanthrene 27 9 Anthracene 0.5 7 Fluoranthene 0.07 11 Benz[a]anthracene 6 8 Benzo[k]fluoranthene 7 8 Benzo[a]pyrene 3 8 70 Analyst, 1999, 124, 67–70

ISSN:0003-2654    

DOI:10.1039/a807576j

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Determination of methylcyclopentadienylmanganese tricarbonyl (MMT) in aqueous samples by SPME-GC-AED Fan Yang* and Y. K. Chau National Water Research Institute, Environment Canada, Canada Centre for Inland Waters, Burlington, Ontario, Canada, L7R 4A6. E-mail: fan.yang@cciw.ca Received 14th September 1998, Accepted 20th November 1998 Solid phase microextraction (SPME) was successfuly applied to the determination of the gasoline additive methylcyclopentadienylmanganese tricarbonyl (MMT) in aqueous samples.The procedure is simple, efficient and solvent free. Ultra-trace levels of MMT (pg l21) in water can be determined with the combination of SPME and gas chromatography–plasma atomic emission detection. The linear range is between 1 and 1000 pg l21 with relative detection limits of 0.3 pg l21 (as Mn) of MMT in water. The precision for replicate analyses (n = 4) of an MMT solution (10 pg l21 as Mn) was 7.1% (RSD). Different kinds of aqueous samples were collected and extracted for the determination of MMT with the SPME technique.In spite of its short half-life in the presence of light, MMT was found in all highway runoff and sewage samples. The liquid–liquid extraction method was also used for MMT determination in highway runoff and lake water samples and the results from the two techniques were comparable. Introduction Methylcyclopentadienylmanganese tricarbonyl (MMT) is used as a substitute for alkylleads as an antiknock agent and octane improver in unleaded gasoline.MMT has been used in Canada since 1976, but only recently approved for use in the USA.1 The combustion of MMT leads to the emission of oxides of manganese to the atmosphere. MMT is highly toxic to some animals through inhalation, ingestion and skin absorption. High ambient manganese levels may cause toxic effects to the central nervous system, pulmonary toxicity and severe injury to the kidneys, liver and lungs.1–4 Although MMT is known to photolyze rapidly under sunlight, it is fairly stable in the absence of light with half-lives ranging from 0.2 to 1.5 years in aquifer materials at 25 °C.5 MMT has been found in air and roadside dirt.2,6,7 Its health concerns have been raised recently because the use of MMT in gasoline is expected to increase in the future. Solid phase microextraction (SPME) is a simple, solvent-free and efficient extraction technique.Polymer-coated fibers are used to extract analytes from gaseous or aqueous samples, and then directly inserted to the injection port of a chromatograph for quantitative and qualitative analyses.SPME has been successfully aplied to the determination of different organic compounds.8–12 With appropriate derivatization, organometallic compounds and metal ions in aqueous solution can also be extracted by SPME.13–15 SPME has exceptional advantages for volatile and semi-volatile analytes over static, purge-and-trap and liquid–liquid extraction (LLE) techniques because of its simplicity and good reproducibility. GC combined with plasma atomic emission detection (AED) is an extremely sensitive and selective technique for MMT determination, with an absolute detection limit of 0.5 fg expressed as Mn.6 In this study, trace MMT in water samples was successfully extracted with the headspace SPME technique.The combination of SPME and GC-AED provides a simple and efficient analytical method for the determination of ultra-trace levels of MMT in aqueous samples with good precision.Experimental Reagents Methylcyclopentadienylmanganese tricarbonyl (MMT) was obtained from Aldrich (Milwaukee, WI, USA). Distilled water was further purified using a Milli-Q system (Millipore, Bedford, MA, USA). Working standard solutions (1–1000 pg Mn l21) were prepared by diluting a stock standard solution in methanol with distilled water. All standard solutions were stored in amber glass bottles at 4 °C in the dark.SPME device and extraction procedure A SPME fiber holder for manual injection and a fused silica fiber coated with a 100 mm film of polydimethylsiloxane (PDMS) were obtained from Supelco (Bellafonte, PA, USA). Fibers were conditioned for 1 h at 300 °C in a helium atmosphere in the injection port of the gas chromatograph prior to use. An aqueous sample (20 ml) was placed in a 40 ml vial with a septum cap and magnetically stirred. Sampling was performed by exposing the SPME fiber to the headspace or directly dipping the fiber in the liquid into the vial for a known period of time.The fiber was then retracted into the syringe and quickly transferred to the GC inlet liner in the injection port. The fiber was manually pushed out of the syringe at 250 °C for 15 s to allow the extracted analytes to be thermally desorbed from the fiber as the GC temperature program was initiated. GC-AED parameters The GC-AED system consisted of a gas chromatograph (HP Model 5890) equipped with an HP SPME inlet liner and an HP helium microwave plasma atomic emission detector (Model 5921A) (Hewlett-Packard, Palo Alto, CA, USA).The optimum operating parameters of the GC-AED system for manganese Analyst, 1999, 124, 71–73 71have been studied previously,6 and are given in Table 1 for reference. Analysis of aqueous samples Highway run-off and lake water samples were collected in 40 ml vials and 4 l bottles, respectively, and transferred back to the laboratory immediately for MMT determination.The samples in 40 ml vials were adjusted to 20 ml and carried through the SPME headspace sampling procedure described previously. The LLE of MMT from aqueous samples was also performed for comparison with the SPME method. Sample extraction was carried out by adding 2 3 50 ml of hexane to 1 l of water sample. Similarly, three sewage samples were obtained from the Burlington Waste Water Treatment Plant and extracted by the SPME headspace technique.Results and discussion Optimization of the SPME technique Absorption–time profiles were studied by determining headspace and aqueous-phase equilibration times for MMT with 100 mm PDMS fiber. The two sampling methods have similar extraction efficiencies and the results are shown in Fig. 1. The headspace SPME technique was used throughout the study because of less matrix interference compared with liquid sampling for environmental samples. From the equilibrium curves in Fig. 1, a 15 min equilibrium time was found to be optimum for the headspace SPME extraction of MMT from water at room temperature (25 °C). The effect of temperature on the extraction efficiency of the headspace SPME technique was investigated by sampling MMT standard solutions for 15 min at different temperatures from 25 to 75 °C. Within this temperature range, the amounts of MMT extracted were constant in 15 min. The results indicated that increasing temperature did not significantly affect the sorption efficiency of MMT on the SPME fiber coating.After the fiber had been exposed to 250 °C for 15 s in the GC inlet liner, the thermal desorption of MMT was completed. Complete desorption was confirmed by re-injecting the fiber into the GCAED system for the determination of residual analytes. GC-AED determination The GC-AED experimental conditions were optimized in our previous work and the manganese 259 nm line has been shown to be the most sensitive channel for the determination of MMT.6 After extraction, the SPME fiber was manually injected into the GC inlet liner and the GC temperature program was simultaneously initiated.Fig. 2 shows the GC-AED trace for the headspace SPME extraction of MMT in highway runoff. The retention time of MMT is 4.57 min. The trace manganese compound at 4.25 min was identified as cyclopentadienylmanganese tricarbonyl (CMT), which was present in the MMT sample.6 Linearity, precision and detection limits Headspace SPME was carried out in a 40 ml vial containing 20 ml of aqueous solution.The linear range was between 1 and 1000 pg Mn l21 (correlation coefficient 0.997). The limit of detection (LOD) was determined as three times the standard deviation of the background noise measured on an aqueous sample with four replicates. The LODs for MMT are 0.3 pg Mn l21 with headspace SPME and 0.5 pg Mn l21 with LLE. Although MMT is a low volatility compound (6.3 Pa at 20 °C),5 it shows a higher concentration in the solid coating phase at equilibrium on extraction.The precision for the method was 7.1% (RSD) at the 10 pg Mn l21 level MMT. Good reproducibility of MMT analysis was achieved with the combination of the SPME technique and GC-AED determination. The method of LLE extraction of MMT from aqueous samples was evaluated. The recovery of MMT was 96 ± 3% (n = 3) at a spike level of 10 pg as Mn in a 1 l water sample. Analysis of environmental aqueous samples Environmental samples including highway run-off, sewage samples and lake water were collected, extracted by the headspace SPME method and subjected to GC-AED for the determination of MMT.The results are summarized in Table 2. MMT was present in all highway run-off and sewage samples but was not detected in lake water. The concentrations of MMT Table 1 Operating conditions for GC-AED GC parameters— Injection port Splitless Injection port temperature 250 °C Column SPB-1, 30 m 3 0.53 mm id Column head pressure Helium, 100 kPa (14.5 lb in22) Temperature program 45 °C for 0.5 min, then 30 °C min21 to 200 °C AED parameters— Transfer line SPB-1 Transfer line temperature 280 °C Cavity temperature 280 °C Spectrometer purge gas N2 at 2 l min21 Helium make-up gas 280 ml min21 Mn wavelength 259 nm Hydrogen pressure 517 kPa (75 lb in22) Oxygen pressure 172 kPa (25 lb in22) Fig. 1 Extraction time profile of 5 fg ml21 mn, for MMT solution sampled with 100 mm PDMS fiber. /, Headspace, -, aqueous-phase extraction.Fig. 2 GC-AED chromatogram of a headspace SPME extraction of highway runoff for the determination of MMT. 72 Analyst, 1999, 124, 71–73in highway run-off and lake water were determined by both SPME and LLE methods and the results were comparable (Table 2). The difference for some samples could be caused by the sample collection procedure. MMT was also found in all sewage samples and less MMT remained in effluent samples after the waste water treatment.Conclusions Little published information on the occurrence of the gasoline additive MMT in water samples is available. In order to assess the MMT risk to the environment, an efficient method is essential for monitoring the presence of MMT in aquatic systems. Headspace solid phase microextraction of MMT has substantial advantages over the traditional LLE because of its simplicity and excellent reproducibilty and it is solvent free.In this work, the combination of an SPME fiber coated with PDMS and GC-AED has been demonstrated to be an extremely efficient, sensitive and selective procedure for the determination of MMT in water. In spite of its short half-life in the presence of light, MMT was found in all highway runoff and sewage samples. Such findings warrant further investigations. References 1 H. Frumkin and G. Solomon, Am. J. Ind. Med., 1997, 31, 107. 2 V. S. Gaind, K. Vohra and F. Chai, Analyst, 1992, 117, 161. 3 P. A. McGinley, J. B. Clay and G. Gianutsos, Toxicol. Lett., 1987, 36, 137. 4 J. W. Hwang, Anal. Chem., 1972, 44, 29A. 5 A. W. Garrison, M. G. Cipollone, N. L. Wolfe and R. R. Swank, Jr., Environ. Toxicol. Chem., 1995, 14, 1859. 6 Y. K. Chau, F. Yang and M. Brown, Appl. Organomet. Chem., 1997, 11, 31. 7 M. Coe, R. Cruz and J. C. Van Loon, Anal. Chim. Acta, 1980, 120, 171. 8 C. L. Arthur and J. Pawliszyn, Anal. Chem., 1990, 62, 2145. 9 Z. Zhang, M. J. Yang and J. Pawliszyn, Anal. Chem., 1994, 66, 844A. 10 R. E. Majors, LC-GC, 1995, 13, 82. 11 W. M. Coleman, III, J. Chromatogr. Sci., 1997, 35, 245. 12 J. Y. Horng and S. D. Huang, J. Chromatogr., 1994, 678, 313. 13 T. G�orecki and J. Pawliszyn, Anal. Chem., 1996, 68, 3008. 14 Y. Cai and J. M. Bayona, J. Chromatogr. A, 1995, 696, 113. 15 L. Moens, T. D. Smaele, R. Dams, P. V. D. Broeck and P. Sandra, Anal. Chem., 1997, 69, 1604. Paper 8/07140C Table 2 Concentration of MMT in environmental samples MMT/pg Mn l21 Sample SPMEa LLE Highway runoff— Sample 1 2.0 ± 0.2 2.4 Sample 2 18.3 ± 2.6 16.3 Sample 3 1.0 ± 0.1 2.8 Sample 4 0.9 ± 0.2 1.1 Sample 5 1.7 ± 0.1 2.8 Sample 6 11.0 ± 1.1 14.5 Sewage— Raw 50.0 ± 5.1 –b Primary 36.1 ± 3.7 – Final 12.0 ± 1.3 – Lake water— Sample A nd ndc Sample B nd nd a Mean ± s (n = 3). b –, No data. c nd, Not detected. Analyst, 1999, 124, 71–73

ISSN:0003-2654    

DOI:10.1039/a807140c

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Use of solid phase extraction for speciation of selenium compounds in aqueous environmental samples J. L. G�omez-Ariza,*a J. A. Pozas,b I. Gir�aldeza and E. Moralesa a Departamento de Qu�ýmica y Ciencia de los Materiales, Escuela Polit�ecnica Superior, Universidad de Huelva, La R�abida, Huelva, Spain. E-mail: ariza@uhu.es b Departamento de Bioqu�ýmica, Bromatolog�ýa y Toxicolog�ýa, Facultad de Medicina, Universidad de Sevilla, Seville, Spain Received 13th July 1998, Accepted 29th October 1998 The performance and optimization of an anion exchanger phase (SAX) in line with a hydrophobic phase (C18) was studied for the preconcentration of selenite [Se(iv)], selenate [Se(vi)], dimethylselenide (DMSe), dimethyldiselenide (DMDSe), diethylselenide (DESe) and diethyldiselenide (DEDSe), using a GC-MS system to characterize the different selenium species. Optimum separation and preconcentration of the inorganic selenium species were based on their retention in an SAX cartridge, with Se(iv) eluted in 25 ml of 1 mol l21 HCOOH and Se(vi) in 25 ml of 1 mol l21 HCl.Organic selenium species were retained in a C18 cartridge and eluted in 2 ml of carbon disulfide. The detection limits of the method were 1.4, 1.6, 0.6, 6.0, 400 and 900 ng l21 for Se(vi), Se(iv), DMDSe, DEDSe, DMSe and DESe, respectively, using a 1000 ml water sample, and the repeatability was less than 10%. The method was applied to natural water samples of different origins.Introduction Selenium is present in the environment from both natural and anthropogenic inputs, at concentrations ranging between 50 and 4000 mg l21 in waters. This element has been recognized as an essential nutrient and amounts of 55–70 mg of Se have been recommended as the daily intake in humans. However, at concentrations higher than 130 mg l21, selenium becomes toxic; 10 mg l21 Se is the maximum allowed concentration in drinking water as recommended by the WHO. The toxicity, bioavailability and essential nature of selenium depend on its chemical forms,1 Se(iv) and Se(vi) being the predominant species in natural waters.Otherwise, biomethylation of inorganic selenium produces volatile dimethyl selenide (DMSe), dimethyl diselenide (DMDSe) and diethyl selenide (DESe), and this process constitutes the major mechanism for selenium detoxification. Therefore, the development of reliable techniques for selenium speciation is necessary to understand the biogeochemical cycle, mobility, uptake–depuration mechanisms and toxicity of this element.Se(iv) and Se(vi) are separated using ion pair HPLC or IC and detected by AAS, AES or ICP-MS,2,3 which are preferred to non-specific detection methods because they are less influenced by the matrix in environmental samples. Without previous separation of Se(vi), Se(iv) can be determined by GC-ECD, GC-MS, GC-IDMS or fluorimetry after the formation of volatile piazselenols or hydride generation (HG)-AAS.4,5 Speciation of volatile organic selenium compounds has been successfully achieved by GC coupled with either quartz and graphite furnace AAS, MIP-AED or non-dispersive AFD.6 Despite the very sensitive analytical methods available for selenium, the low levels of this element found in water require preconcentration to improve the detectability and remove matrix interferences. Usually, organic and inorganic interferences are eliminated from water samples using column chromatography on XAD-8 resin7 and on Dowex 50W-X8,8 respectively.Preconcentration is performed by liquid–liquid extraction of piazselenols,9 on-line coprecipitation of inorganic selenium species with La(OH)3,10 complexation and adsorption on activated carbon11 and solid phase extraction (SPE) based on anionic sorbents.3 On the other hand, the volatile selenides can be separated from water by helium stripping and swept into a cold trap with solid sorbents such as activated carbon, glasswool and GC stationary phases, and then thermally desorbed or extracted with organic solvents.6,12 However, no method for the simultaneous preconcentration of Se(iv), Se(vi), DMSe, DMDSe, DESe and diethyl diselenide (DEDSe) in water have been reported.This work was focused on the determination of inorganic and volatile organic selenium species in water samples. An SPE based on the use of two different sorbent phases, octadecyl (C18) and quaternary amoninum (SAX), in separate cartridges was optimized for the preconcentration of trace amounts of the selenium species. The SAX cartridge was stacked upon the C18 cartridge for the application of the method to different natural aqueous matrices.The final analytical determination uses a GCMS approach, which presents a sensitivity comparable to that of other methods, with an additional confirmation of the nature of each species. Experimental Reagents, standards and apparatus Analytical-reagent grade reagents and pesticide grade solvents were obtained from Merck (Darmstadt, Germany) and Sigma (St.Louis, MO, USA). SAX and C18 (600 mg of sorbent) cartridges were obtained from Alltech (Deerfield, IL, USA) and Waters (Milford, MA, USA), respectively. Stock standard solutions of Se(iv) and Se(vi) (1000 mg l21 of Se) were prepared from analytical-reagent grade selenium dioxide and sodium selenate (Merck), respectively. Organoselenium stock standard solutions were prepared at a concentration of approximately 100 mg l21 (as Se) in benzene from DMDSe (Aldrich, Gillingham, Dorset, UK), DESe, DMSe (Pfaltz and Bauer, Waterbury, CT, USA) and DEDSe (synthesized by the authors13) and were kept in a refrigerator. Aqueous working solutions were prepared daily and water used in the Analyst, 1999, 124, 75–78 75experiments was doubly distilled and de-ionized and gave blank readings in all the analyses.Plastic and glassware used for experiments were previously soaked in 0.08 mol l21 HNO3 for 24 h and rinsed carefully with water.Selenium speciation was carried out using an HP Model 5890 gas chromatograph and HP Model 5970 mass detector, with a fused silica capillary column, 25 m 3 0.20 mm id and a film thickness of 0.33 mm HP-1 cross-linked methylsilicone gum (Hewlett-Packard, Palo Alto, CA, USA). Sample aliquots of 1 ml were injected manually using the splitless injection mode. Helium was used as the carrier gas at a head pressure of 100 kPa.The interface was operated at 260 °C. Electron ionization (EI) mass spectrometry was used for detection. A scan time of 1.0 s was used over a mass range of m/z 40–500. Preconcentration of selenium species using cartridges Preconcentration of inorganic selenium was performed on a SAX cartridge conditioned with 10 ml of 3 mol l21 HCl and 10 ml of water at 5 ml min21. For organic selenium species, a C18 cartridge was conditioned with 10 ml of CS2, 10 ml of MeOH and 10 ml of water, using a flow rate of 5 ml min21.The SAX cartridge was stacked on top of a C18 cartridge and 1000 ml of water with the pH adjusted at 7–8 was passed through the cartridges at 8 ml min21. The cartridges were separated and the analytes were eluted separately, Se(iv) with 25 ml of 1 mol l21 of HCOOH and Se(vi) with 25 ml of 3 mol l21 of HCl, using a flow rate of 5 ml min21. The C18 cartridge was dried under a stream of nitrogen and the organic selenium species were eluted with 2 ml of CS2 at 1 ml min21.Determination of inorganic selenium Selenium(iv) was derivatized with 5 ml of 0.1% 4-chloro-ophenylendiamine in 0.1 mol l21 HCl at 75 °C for 7 min. The solution was allowed to cool to room temperature and the piazselenol was extracted twice with 1 ml of toluene by shaking for 1 min. The organic phase was separated and evaporated just to dryness under a stream of nitrogen. The residue was dissolved in 50 ml of hexane containing fluorodinitrobenzene (FDNB) as internal standard (230 ng l21) and analysed by GC-MS.Selenium(vi) was quantitatively reduced to Se(iv) by adding 10 ml of 5 mol l21 HCl and boiling for 30 min. After allowing the residual solution to cool, the pH was adjusted to 2.1 and selenium was derivatized. The retention times obtained were: FDNB 9.8 min (m/z 186) and piazselenol 10.5 min (m/z 218). Determination of organic sevolume of a solution of 2.0 mg l21 of 2,6-diisopropylphenol (propofol) as internal standard in CS2 was added to the extract containing the organic selenium species and the solution was analysed by GC-MS.The retention times were DMSe 2.97 min (m/z 109), DESe 5.07 min (m/z 109), DMDSe 7.11 min (m/z 188), DEDSe 8.41 min (m/z 218) and propofol 17.1 min (m/z 178). A derivatization step using FDNB was introduced to improve the sensitivity of the dialkyl diselenide determination.14 The derivatives were extracted with 4 ml of ethyl acetate (three times), the extracts were evaporated to dryness, the residues were dissolved in 50 ml of toluene containing 200 mg l21 of propofol and the solutions were analysed by GC-MS.The retention times were DMDSe derivative 16.3 min (m/z 262), DEDSe derivative 18.7 min (m/z 276) and propofol 9.1 min (m/z 178). Statistical treatment The data were analysed statistically for differences using factorial analysis of variance (ANOVA). Prior to analysis, all the data were tested for homogeneity of variance using the Barlett and Levene tests.A parametric statistical test (Student’s t-test) was applied to different hypotheses. An a-value of 0.05 was adopted as the critical level for all statistical testing giving a 95% confidence level (CSS: STATISTICA). Results and discussion Preconcentration and separation of inorganic selenium species using strong anionic exchange cartridges Two strong anion exchange phases, SAX and Acell plus QMA (Waters), were used for Se(iv) and Se(vi) preconcentration and a similar performance was found in both cases (ANOVA, p > 0.61).The parameters controlling the SPE were the pH and volume of sample, both affecting the retention of inorganic selenium species, and also the concentration and volume of the eluents, which affect the elution of these analytes. The experiments were performed (three replicates) by using 12 mg of both Se(iv) and Se(vi) in 250 ml of doubly distilled water, following the procedure described under Experimental.Loading of inorganic selenium species into the cartridge was studied using aqueous solutions of pH 3–9. Fig. 1 shows that the best recoveries were obtained for pH values typically found in natural waters (7–8) (t-test, p < 0.01). Therefore, a pH of 7 was used in further experiments and no special pre-treatment of the samples was required. Selenium preconcentration from different volumes of water ranging from 100 to 1000 ml was tested and no differences were found in the recoveries.Therefore, it is possible to use large volumes of water when selenium species levels are very low in order to load suitable amounts of this element on the cartridges. However, this point should be confirmed by the analysis of natural water, because other compounds from the matrix may overload the packing material. Elution of inorganic species of selenium was tested using 0.5–4 mol l21 aqueous HCOOH solutions. Selenium(iv) was quantitatively recovered with 1 mol l21 HCOOH and higher concentrations did not improve the results (ANOVA, p > 0.98).Elution of Se(vi) required 2 mol l21 HCl (t-test, p < 0.01). The volumes of each eluent between 10 and 40 ml were tested, and 20 ml were needed to obtain quantitative recoveries (t-test, p < 0.01). No improvement was obtained by using larger volumes of elutant (ANOVA, p > 0.50). Fig. 1 Influence of pH on the retention of Se(iv) and Se(vi) on SAX cartridges.Results represent the percentage recovery ± standard deviation for three replicates, obtained by elution with an acidic medium [HCOOH and HCl solutions for Se(iv) and Se(vi), respectively]. 76 Analyst, 1999, 124, 75–78Preconcentration and separation of organoselenium species using non-polar sorbents Cartridges with non-polar (C18 and activated carbon) and polar (CN) sorbents were tested for the preconcentration of organoselenium compounds from water samples.As the extract obtained from the cartridge has subsequently to be injected into the GC system for analysis, a minimum elution volume must be used in order to avoid losses of the volatile organoselenium species during solvent removal. Moreover, the chromatographic determination establishes restrictions in the eluent choice, which must not be eluted from the chromatographic column at the same time as the most volatile organoselenium species (DMSe), because the filament used for the EI must remain off during the solvent elution.The parameters controlling the SPE were the sample volume and the nature and volume of the solvent used as the eluent. The experiments were performed (three replicates) by using samples of 100 mg (as Se) of both DMSe and DESe and 1 mg (as Se) of DMDSe and DEDSe in 250 ml of doubly distilled water, following the analytical procedure described under Experimental. Recoveries of the four organoselenium species using both polar and non-polar sorbents and CS2 volumes of 1–6 ml are given in Table 1.Recoveries higher than 80% were obtained using both C18 and activated carbon, the non-polar character of the primary interaction between analytes and sorbents being obvious. Poorer recoveries were obtained using CN cartridges. This indicated the absence of polar interactions. Higher recoveries from C18 were obtained using at least 2 ml of solvent (t-test, p < 0.01), but 4 ml were needed for the elution of the organoselenium compounds from the activated carbon sorbent (t-test, p < 0.015), which indicated stronger retention of the organoselenium compounds in this sorbent. Higher volumes did not improve the recoveries (ANOVA, p > 0.55).Finally, lower recoveries ranging from 20 to 70% were obtained using 6 ml of more polar eluents such as MeOH and Et2O, irrespective of the sorbent. To evaluate the volume of aqueous sample that can be treated with the cartridge without overloading the packing material, recovery experiments were carried out on spiked samples ranging from 100 to 1000 ml, and no differences were found (ANOVA, p > 0.81).However, no experiments were developed to check the total adsorption capacity of the cartridge, which will depend on the presence of other substances in the sample such as hydrocarbons and fats. This point should be checked by the analysis of natural samples. Calibration and analytical quality control Calibration curves were constructed by the analysis of 1000 ml aliquots of water which were submitted to the whole preconcentration method.The calibration curves were linear up to selenium concentrations of 0.750 mg l21 for Se(iv) and Se(vi) and 130, 110, 1.20 and 2.40 mg l21 for DMSe, DESe, DMDSe and DEDSe, respectively, with correlation coefficients higher than 0.997. Detection limits (DLs), sensitivity, repeatability and reproducibility of the method for water analysis are given in Table 2.These results are comparable to those reported in the literature and obtained using HPLC-ICP-MS3 [DL 160 and 80 ng l21 for Se(iv) and Se(vi), respectively], GC-AAS15 [DL 0.8 ng l21 for methylated species and 1.6 ng l21 for Se(iv)] and GCMIP- AED12 (DL 2 ng l21 for methylated species), allowing the evaluation of these species in natural aquatic environments. However, lower detection limits, 0.0044 and 0.0069 ng l21 for DMSe and DMDSe, respectively, have been reported using 1 l samples in a purge-and-trap, low temperature GC-AFD device, but no attempt to separate ethylated selenium species was made.6 Application to environmental samples The method was validated for Se(iv) determination using a CRM (CASS-3 from Laboratory of the Government Chemist, Teddington, UK) consisting of nearshore sea-water with 0.5 mg l21 total dissolved organic matter, acidified at pH 1.6 and with an Se(iv) content of 0.020 ± 0.005 mg l21, and applied to the speciation of selenium in sea-, river and tap water samples.The results are given in Table 3. No organic selenium species were found in any samples. Consequently, samples were spiked at two levels, using high concentrations of organoselenium compounds (50, 25, 1 and 1 mg l21 for DESe, DMSe, DMDSe and DEDSe, respectively) to check the absence of sorbent overloading low concentrations (5, 2.5, 0.006 and 0.05 mg l21 Table 1 Recoveries ± standard deviations (%) of organoselenium compounds eluted from octadecylsiloxane (C18), activated carbon (AC) and cyanopropylsiloxane (CN) sorbents using CS2 as eluent Sorbent Volume/ml DMSe DESe DMDSe DEDSe C18 1 53 ± 3 47 ± 5 61 ± 5 53 ± 4 2 84 ± 5 87 ± 5 88 ± 5 89 ± 5 4 86 ± 4 88 ± 4 83 ± 4 91 ± 5 6 83 ± 4 84 ± 4 86 ± 4 86 ± 4 AC 1 39 ± 5 42 ± 8 45 ± 6 53 ± 6 2 51 ± 5 64 ± 6 65 ± 5 69 ± 5 4 79 ± 5 82 ± 4 85 ± 5 87 ± 5 6 81 ± 5 76 ± 5 74 ± 6 86 ± 5 CN 1 NDa 11 ± 4 8 ± 3 9 ± 4 2 12 ± 3 9 ± 4 14 ± 5 17 ± 6 4 33 ± 4 27 ± 6 34 ± 5 30 ± 7 6 27 ± 7 33 ± 8 24 ± 6 26 ± 5 a Not detected.Table 2 Detection limits (DL), correlation coefficient (r2), sensitivity (S), repeatability (r) and reproducibility (R) of selenium determination in water samples. The repeatability and reproducibility were assessed in solutions containing 0.150 mg l21 of Se(iv), 0.150 mg l21 of Se(vi), 16.8 mg l21 of DMSe, 33.6 mg l21 of DESe, 0.240 mg l21 of DMDSe and 0.490 mg l21 of DEDSe (as Se) Species DL/ng l21 S/l mg21 r2 r (%) R (%) Se(IV) 1.6 4.89 0.998 5.6 8.6 Se(VI) 1.4 4.35 0.997 6.6 8.5 DMSe 388 0.033 0.998 6.1 9.9 DESe 903 0.029 0.999 3.7 7.0 DMDSed a 0.6 2.30 0.998 7.1 9.2 DEDSed a 6.0 0.93 0.999 7.3 9.7 a Organselenium derivative.Table 3 Concentration of inorganic species of selenium ± standard deviation (mg l21, as Sn) (three replicate samples) in water samples collected from the southwest of Spain and in a certified reference sample. Recovery studies Se concentration ± s/mg l21 Recovery Location Type pH Species Initial Added (%) Niebla River 4.1 Se(iv) 0.033 ± 0.003 0.04 96 Se(vi) 0.025 ± 0.003 0.025 97 Portil Lake 7.8 Se(iv) 0.18 ± 0.01 0.2 94 Se(vi) 0.21 ± 0.02 0.2 94 Portil Sea 7.6 Se(iv) 0.34 ± 0.02 0.3 96 Se(vi) 0.31 ± 0.02 0.3 97 Rompido Sea 7.7 Se(iv) 10.5 ± 0.9 10 98 Se(vi) 39 ± 3 40 96 Punta Sea 7.8 Se(iv) 270 ± 20 300 97 Umbria Se(vi) 81 ± 6 100 99 Ayamonte Sea 6.8 Se(iv) 5.7 ± 0.3 5 94 Se(vi) 3.4 ± 0.3 4 93 Sevilla Tap 7.1 Se(IV) 0.98 ± 0.06 1 96 Se(vi) 0.43 ± 0.02 0.5 91 CASS-3 CRM Se(iv) 0.017 ± 0.002 (duplicate) Analyst, 1999, 124, 75–78 77for the same species) to simulate more realistic environmental levels.Recoveries of 80–100% were invariably found for the four organoselenium species in all the samples in Table 3, with no significant differences between the two concetration levels tested. High levels of both Se(iv) and Se(vi) were found in water from Punta Umbria, a fishing harbour located on the Huelva coast.Lower concentrations were found in water from Rompido and Ayamonte fishing harbours, located in estuarine areas in the Piedras and Guadiana Rivers. The lowest concentrations for both Se(iv) and Se(vi) were found in water from the Tinto River (Niebla station). These levels are in the usual range found in natural waters.16 In general, Se(iv) was present in higher concentrations than Se(vi), which is in contrast to the stability predictions, indicating slow oxidation kinetics of Se(iv) to Se(vi) in natural waters.Finally, selenium concentrations found in a tap water from Seville were also in the range usually found in drinking waters by several workers16 and below the maximum selenium concentration in drinking water recommended by the WHO. Conclusions Both inorganic and volatile organic selenium species were recovered from water samples using stacked SAX and C18 cartridges, which allow 40- and 500-fold preconcentration for inorganic and organic species, respectively.Moreover, the separation of Se(iv), Se(vi) and organic selenium compounds can be achieved by fractional elution, which allows the independent determination of these species using different analytical methods. The procedure is robust enough for the analysis of tap, river and sea-water samples at environmentally representative levels of ng l21 for Se(iv), Se(vi), DMDSe and DEDSe and low mg l21 for DMSe and DESe, using a GC-MS system. Acknowledgement The authors express their thanks to the DGICYT (Direcci�on General de Investigaci�on Cient�ýfica y T�ecnica) for Grant No. PB95-0731. References 1 R. J. Shamberger, Mutat. Res., 1985, 154, 29. 2 G. K�olbl, J. Lintschinger, K. Kalcher and K. J. Irgolic, Mikrochim. Acta, 1995, 119, 113. 3 Y. Cai, M. Caba�nas, J. L. Fern�andez-Turiel, M. Abalos and J. M. Bayona, Anal. Chim. Acta, 1995, 314, 183. 4 S. M. Gallus and K. G. Heumann, J. Anal. At. Spectrom., 1996, 11, 887. 5 F. MacLeod, B. A. McGaw and C. A. Shand, Talanta, 1996, 43, 1091. 6 C. P�echeyran, D. Amouroux and O. F. X. Donard, J. Anal. At. Spectrom., 1998, 13, 615. 7 D. R. Roden and D. E. Tallman, Anal. Chem., 1983, 54, 307. 8 L. D. Martinez, M. Baucells, E. Pelfort, M. Roura and R. Olsina, Fresenius’ J. Anal. Chem., 1997, 357, 850. 9 C. I. Measures and J. D. Burton, Anal. Chim. Acta, 1980, 120, 177. 10 S. Nielsen, J. J. Sloth and E. H. Hansen, Analyst, 1996, 212, 31. 11 T. Kubota, K. Suzuki and T. Okatani, Talanta, 1995, 42, 949. 12 M. B. de la Calle, M. Ceulemans, C. Witte, R. Lobinski and F. C. Adams, Mikrochim. Acta, 1995, 120, 73. 13 J. L. G�omez-Ariza, J. A. Pozas, I. Gir�aldez and E. Morales, Int. J. Anal. Chem, in the press. 14 J. L. G�omez-Ariza, J. A. Pozas, I. Gir�aldez and E. Morales, Anal. Chim. Acta, submitted for publication. 15 T. D. Cooke and K. W. Bruland, Environ. Sci. Technol., 1987, 21, 1214. 16 J. E. Conde and M. Sanz-Alaejos, Chem. Rev., 1997, 97, 1979. Paper 8/05447I 78 Analyst

ISSN:0003-2654    

DOI:10.1039/a805447i

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Bioassays for the detection of growth-promoting agents, veterinary drugs and environmental contaminants in food† Laurentius A. P. Hoogenboom,* Astrid R. M. Hamers and Toine F. H. Bovee State Institute for Quality Control of Agricultural Products (RIKILT-DLO), PO Box 230, 6700 AE Wageningen, The Netherlands Received 29th June 1998, Accepted 19th November 1998 Residues of growth-promoting agents, veterinary drugs and environmental contaminants in food products are routinely analyzed with chemical–analytical methods, using physical and spectrometric properties of a compound.Since residue limits are in general based on biological properties of compounds, bioassays offer in theory a good alternative. As a consequence, these assays are much more suitable for the detection of mixtures of compounds with common biological properties, including possibly unknown agonists. Using modern molecular biological techniques, a new generation of bioassays has been developed, showing in general a higher sensitivity and specificity for the target compounds.The CALUX (chemical activated luciferase expression) assay was developed for the detection of polyhalogenated compounds, based on their affinity for the aryl hydrocarbon (Ah) receptor. This paper focuses on the specificity of the assay. The benzimidazole compounds oxfendazole, fenbendazole, febantel, thiabendazole, mebendazole, omeprazole, lanzoprazole and benomyl were shown to give a positive response in the assay.Similar results were obtained with dexamethasone, corticosterone and cortisol, which in addition were able to enhance the response obtained with TCDD. Similarly to the flavonoids a- and b-naphtoflavone, the reported Ah receptor antagonist 4-amino-3-methoxyflavone showed a strong positive response at a concentration of 400 mm, but failed to inhibit the response obtained with TCDD. It is concluded that the chances of false-negative results appear to be minimal and can be recognized. False-positive or, better, unwanted results are in theory more likely to occur.Possible solutions to avoid or detect these type of results are discussed. In general, these kinds of assays offer great possibilities for screening of food samples. In addition to the further optimization of these assays, future work should be focused on the development of rapid, simple and selective extraction procedures. Introduction Food for animal or human consumption may contain residues of many different xenobiotics, such as environmental contaminants, mycotoxins, pesticides, veterinary drugs and growthpromoting agents.For many of these compounds, maximum residue limits (MRLs) have been established, in most cases based on the known effects of these compounds in laboratory animals. Monitoring programs are required to check food items for the presence of residues, normally based on the physicochemical properties of these compounds. In particular in the case of potent bioactive compounds with low MRLs, or mixtures of compounds, such as polychlorinated aromatic hydrocarbons, such methods are often laborious and very expensive.As a result, only limited monitoring can be performed. The use of bioassays may overcome most of these problems and offer a number of additional advantages. In general, such assays are based on the biological effects of a class of compounds, possibly following their binding to specific receptors.End-points can be based on the effects underlying the MRLs, but also on the pharmacological effects, on which the possible misuse is based. In addition, the assays are suitable for detecting mixtures of compounds or possibly unknown agonists. Bacteria are widely used to screen for the presence of antibiotics and antibacterial drugs in food.1–3 Rats and mice are used routinely to screen for the presence of respectively diarrheic (DSP) and paralytic (PSP) shellfish poisons, owing to the absence of suitable analytical methods for the different compounds.Another very interesting bioassay was developed for pesticides with acetylcholine esterase inhibiting properties, using crude extracts from flies or pig liver. The simplicity of this assay actually allows its use as a field test.4 At present there is increased concern about the large number of compounds with hormonal activity in the environment. In particular, compounds with estrogenic activity are thought to be responsible for reported decreased fertilities both in animals and men.This includes both natural and synthetic compounds.5 Traditionally, compounds with estrogenic activity have been detected with the rodent uterotrophic assay. Before suitable analytical methods became available, this test was even used for the detection of illegally used estrogens in injection sites of treated animals.6 More recent assays are based on the binding of compounds to the estrogen receptor7 or the proliferative effects on cells that contain the estrogen receptor.8 The newest generation of tests use yeast or mammalian cells containing the estrogen receptor and a reporter gene coupled to an estrogen responsive element (ERE).9–12 In response to the binding of estrogens to the receptor and the subsequent binding to the ERE, the reporter gene is transcribed resulting in increased cellular levels of the reporter gene product (e.g., luciferase).These examples demonstrate that the natural response of a certain microorganism, cell line or tissue may be selected as a suitable end-point.However, increased knowledge on mechanisms behind biological effects and new molecular-biological techniques allow modifications of cells, resulting in more sensitive, more specific and quicker assays. A clear example is the CALUX (chemical activated luciferase gene expression) bioassay, which was initially developed for the detection of dioxins and dioxin-like PCBs.Previously, it was shown that † Presented at the Third International Symposium on Hormone and Veterinary Drug Residue Analysis, Bruges, Belgium, June 2–5, 1998. Analyst, 1999, 124, 79–85 79these compounds can be detected based on their binding to the cytosolic aryl hydrocarbon (Ah) receptor,13,14 or the subsequent induction of the cytochrome P450 1A related de-ethylation of 7-ethoxyresorufin (EROD activity).15,16 The CALUX assay is based on the use of rat or mouse hepatoma cells that were stably transfected with the so-called p-GudLuc 1.1 construct containing the firefly luciferase gene under transcriptional control of the mouse dioxin responsive element (DRE).17–19 As a result, the cells show an increased production of luciferase in response to binding of compounds to the Ah receptor and the subsequent binding of the complex to the DRE.Results obtained in different laboratories show that the cells respond to very low concentrations of the most toxic dioxin, 2,3,7,8-tetrachlorodibenzo- p-dioxin (TCDD).Other dioxins and PCBs with known affinity for the Ah receptor also showed an effect, more or less reflecting their appointed international toxic equivalency factors (I-TEFs).19,20 These I-TEFs in combination with the ITEQ principle were developed in order to deal with the mixtures of dioxins and PCBs, of which only a limited number are capable of binding to the Ah receptor, each with their own affinity and resulting biological potential.21–24 The I-TEF value of a compound expresses its potency in comparison with the most toxic congener TCDD (I-TEF of 1).Levels of the different congeners are multiplied by this TEF value and subsequently summarized to obtain the total content, expressed in I-TEQs. In the case of the CALUX assay the less potent PCB congeners showed reduced activities, which may result in an underestimation of their contribution. Using mixtures of different 2,3,7,8-substituted dioxin congeners, it was also confirmed that the test results were in accordance with the I-TEQ principle, showing additivity of the individual responses of the congeners. 20 The suitability of the bioassay for measuring dioxins in environmental, plasma and food samples has been described previously,25–27 including bovine milk and coconut oil. Furthermore, the method for bovine milk, based on the CALUX assay and a clean-up procedure with acid silica columns, has been validated for determining the relatively low levels of dioxins around the current residue limit of 6 pg of I-TEQ per gram of fat.20 This paper focuses on the specificity of this assay, and discusses ways to avoid or recognize false-negative or false-positive results.Experimental Materials pGudLuc 1.1-transfected H4IIE cells were obtained from Dr. A. Brouwer of the Division of Toxicology, Department of Food Technology and Nutritional Sciences, Agricultural University, Wageningen, The Netherlands, who was involved in the development of the cells.Fetal calf serum (FCS) was purchased from Gibco (Breda, The Netherlands), TCDD from Schmidt (Amsterdam, The Netherlands), a-minimal essential medium (a-MEM), Eagle’s minimum essential medium without Phenol Red (EMEM-PR2), penicillin, streptomycin, bovine serum albumin (BSA), benomyl, benzo[a]pyrene, a-naphthoflavone from Sigma (St. Louis, MO, USA), thiabendazole from Merck, Sharpe & Dohme (Haarlem, The Netherlands), fenbendazole from Hoechst (Frankfurt, Germany), mebendazole from Janssen Pharmaceutica (Beerse, Belgium) and dimethyl sulfoxide (DMSO) (Uvasol grade) from Merck (Darmstadt, Germany). 4-Amino-3-methoxyflavone (AMF) was a kind gift from Dr.S. Safe (Texas A&M University, College Station, TX, USA), oxfendazole and febantel were a kind gift from Dr. C. Montesissa (University of Bologna, Bologna, Italy) and lanzoprazole and omeprazole were a kind gift from Dr.P. Maurel (INSERM, Montpellier, France). EROD bioassay The EROD assay was performed as described previously.28 Briefly, wild type H4IIE cells were exposed to the test compounds for 24 h. Cells were washed twice with EMEMPR2 and incubated for 15 min in EMEM-PR2, which was subsequently replaced by a solution of 5 mm ethoxyresorufin in EMEM-PR2 (prepared from a stock standard solution of 2.5 mm in DMSO). After incubation for 15 min, the medium was aspirated and stored at 220 °C until analysis. The cells were stored at 220 °C for protein determination Media samples were thawed and centrifuged and an aliquot of 100 ml was mixed with 100 ml of a solution of sulfatase (2 mg ml21) in sodium acetate buffer (0.2 m, pH 5), followed by incubation for 1 h at 37 °C.This step was included for the deconjugation of sulfate and glucuronide conjugates of hydroxyresorufin. Following the addition of 350 ml of Tris-HCl (0.15 m, pH 7.8), the fluorescence was determined in an LS 50 spectrofluorimeter (Perkin-Elmer, Norwalk, CT, USA), at lex 510 nm and lem 586 nm.The concentration of resorufin was calculated by comparison with a calibration curve. The EROD activity is expressed as pmol of resorufin formed per incubation well. For determining the protein content, cells were dissolved in 0.5 ml of sodium dodecyl sulfate (SDS)–NaOH (5%, 0.4 m) and diluted with 0.5 ml of water. The protein content was determined with the method of Lowry, as modified by Peterson,29 using the Bio-Rad (Richmond, CA, USA) DC (detergent compatible) protein assay.An aliquot of 10 ml of sample was mixed with 25 ml of reagent A (copper tartrate– sodium hydroxide–SDS) and 200 ml of reagent B (Folin reagent) on a microtiter plate. The absorption at 750 nm was determined in a microplate reader and compared with a calibration curve for bovine serum albumin. CALUX bioassay Rat H4IIE hepatoma cells, stably transfected with an AhRcontrolled luciferase reporter gene construct (pGudLuc 1.1), were cultured in a-MEM culture medium supplemented with 10% v/v FCS, 50 IU ml21 penicillin and 50 mg ml21 streptomycin.Cells were grown confluent in 24-well plates (Costar, Badhoevedorp, The Netherlands) and exposed in triplicate to standards or mixtures of standards for 20–24 h. Normally, each well contained 0.5 ml of medium including 0.5% v/v DMSO, used as a vehicle. Standards were diluted with DMSO and subsequently dissolved in the medium.In some studies conditioned medium was used, which was harvested from 3 d old monolayer cultures. Following exposure, cells were washed three times with 0.5 ml of phosphate-buffered saline and lysed in 75 ml of lysis reagent (Promega, Leiden, The Netherlands). After 15 min, the cell lysates were transferred into tubes and frozen at 280 °C. For the determination of the luciferase activity the samples were thawed on ice and centrifuged for 3 min at 13 000g.An aliquot of 20 ml of supernatant was transferred into a 96-well microtiter plate. The luciferase activity was determined using a Luminoskan RS luminometer (Labsystems, Breda, The Netherlands), which automatically injected 100 ml of a luciferin assay mixture (Promega, Leiden, The Netherlands) just prior to the measurement. The protein content was determined in a microtiter plate according to the method of Bradford,30 using the Bio-Rad assay dye reagent. Cell lysates were diluted 100-fold and 40 ml of this homogenate were mixed with 200 ml of the reagent (Coomassie Brilliant Blue G-250 in phosphoric acid–methanol).The absorption at 650 nm was determined in a microplate reader and compared with a calibration curve for bovine serum albumin. Results are expressed as relative light units (RLU) per incubation well. In some cases, the data were fitted using a one- 80 Analyst, 1999, 124, 79–85site ligand curve fit (SlideWrite Plus Version 6.00) according to the equation response = (max.response) 3 [agonist]/(EC50 + [agonist]) where [agonist] is the concentration of test compound and EC50 represents the concentration of agonist showing a half maximum response. In contrast to previous reports, the protein content of the incubation wells is no longer used to correct the data, but only to monitor for possible cytotoxic effects of compounds. A clear dose-related decrease in the protein content will be reported separately. In general, the average protein content in the different experiments varied between 170 and 260 mg per incubation well, with an intra-assay variation of around 8% (RSD).Results and discussion Dose- and time-related response of TCDD Fig. 1 compares the EROD and CALUX activities of the transfected cells after exposure to different concentrations of TCDD for 24 h. It is evident that the cells respond to low concentrations of TCDD, resulting in a very similar doserelated increase in both the EROD and luciferase activity.These data show that the CALUX bioassay has at least as much potential as the more classical EROD bioassay for determining dioxins. Fig. 2(A) shows the effect of incubation time on the luciferase activity after exposure to different concentrations of TCDD. In addition to the normal practice, cells were also incubated in conditioned medium, recovered from 3 d old monolayer cultures. In particular in the latter case, there is a clear timerelated increase in the CALUX activity at both 2 and 50 pm TCDD, which is not observed in the control cells.When exposed in fresh medium, the control cells also show a response, which disappears after prolonged exposure. This effect is also observed at a low concentration of TCDD. These data indicate the presence of an unknown agonist in fresh medium which is degraded by the cells. As a result, it does not interfere with the assay when using longer incubation intervals, but it may elevate the control levels at shorter periods.Fig. 2(B) shows the effect of the amount of medium per incubation well on the response of the cells to TCDD. At each concentration, there is a clear increase in the luciferase activity with increase in volume. Using a one-ligand curve fit, EC50 values were calculated to be 28, 15 and 10 pm for the volumes of 0.5, 1 and 2 ml per well. This shows that a two- to fourfold increase of medium results in an elevated response, normally observed at a two- to threefold higher concentration.This indicates that the cells respond to the absolute amount of TCDD per well, rather than the concentration. This observation may help to decrease further the limit of quantification for dioxins in samples, where one of the major problems is the resolubilization of the highly lipophilic dioxins in the incubation medium, following evaporation of the extraction solvent. In the case of milk samples, for example, the 20 ml of DMSO used for resolubilization had to be diluted in 4 ml of medium to keep the DMSO concentration at a non-toxic level (0.5%).20 In the case of milk samples, incubation with a larger volume of medium, possibly in combination with smaller incubation wells (48- and 96-well plates),31 may help to decrease further the current limit of quantification (1 pg of I-TEQ per gram of fat) to levels well below those observed in practice (1–2 pg of I-TEQ per gram of fat).False-negative results A difficult issue with bioassays is the specificity of the assay. False-negative results will make the test unreliable, whereas false-positive results might be acceptable to a certain extent, in particular when the assay is used for screening.In the case of the EROD and CALUX assays, a complicated, only partly resolved, Fig. 1 Dose-response curves for TCDD in the CALUX (5) and EROD (2) assays. Results are expressed as the mean and s for n = 3. GudLuc 1.1 transfected H4IIE-cells were incubated with TCDD for 24 h, washed and incubated with 7-ethoxyresorufin.Subsequently, lysates were prepared of the cells and measured for luciferase and protein content. The protein content per well was unaffected by the TCDD concentration. Fig. 2 Effect of incubation time and medium (A), and medium volume per well (B) on the response of transfected cells to TCDD. Cells were incubated with 0 (5, 2), 2 (<, ¶) or 50 (-, 8) pm TCDD in either fresh (closed symbols) or conditioned (open symbols) medium.In the experiment described in (B), cells were exposed for 16 h to different concentrations of TCDD in 0.5 (5), (normally used), 1 (<) or 2 (-) ml of fresh medium. The latter results were fitted using a one-site ligand curve fit. Results are expressed as the mean ± s for n = 3. The protein content of the wells was unaffected by either time, type of medium or well volume. Analyst, 1999, 124, 79–85 81pathway underlies the biological effect.Following the initial binding of an agonist to the Ah receptor, the complex must be transported to the nucleus and bind to DNA at the DRE, resulting in an increased transcription of the reporter gene and translation of the mRNA to the active protein. It is obvious that general cytotoxicity will result in a decreased response of the cells, but this can be recognized by a visual check of the cells after exposure or the subsequent determination of cellular protein or DNA in the cell lysates. For this reason, it is preferable to use the protein data for assessing cytotoxic effects, rather than for correcting the luciferase content in the wells, especially since it is unclear how cytotoxic effects will affect the response of the cells.The validity of protein content as a measure of crude cytotoxicity has been demonstrated in a multicenter study by comparison with more specific assays such as the MTT reduction, Neutral Red uptake of LDH release assays.32 In practice, however, effects may be more subtle.A disadvantage of the EROD assay, for example, is the fact that cytochrome P450 1A can be irreversibly inhibited by a number of compounds. Exposure of cells to an agonist in the presence of such an inhibitor would result in a false-negative test result. A known class of cytochrome P450 1A inhibitors is benzimidazole drugs,33–35 some of which are used as anthelmintics in food-producing animals.Fig. 3 shows the effect of the anthelmintic benzimidazole drug oxfendazole on the TCDD induced EROD (A) and luciferase (B) activity. At relatively low doses, oxfendazole completely inhibits the EROD activity (A), whereas in the case of the CALUX assay there appears to be an additive rather than an inhibitory effect (B). This demonstrates the advantage of selecting a reporter gene encoding for an enzyme, such as luciferase, for which no inhibitors are known thus far. Another possible interference with a positive response is the competition between certain compounds for binding to the Ah receptor, as shown for certain PCBs.17 In this particular case, the interaction occurred at relatively high concentrations and it remains to be shown whether absolute and relative concentrations of these compounds in samples could lead to such falsepositive results.In general, false-negative results can be detected by testing extracts in both the absence and presence of a positive control.A decreased response obtained with cells exposed to the sample extract in the presence of, e.g., TCDD would indicate the presence of interfering compounds. The potential of this approach is demonstrated in Fig. 3(A) by the decreased EROD activity in cells exposed to TCDD in the presence of oxfendazole. False-positive results In theory, the CALUX assay responds to any compound capable of binding to the Ah receptor. In addition to certain dioxins and PCBs, these assays have, for example, been shown to be sensitive to a number of mutagenic polycyclic aromatic hydrocarbons (PAHs).As a result, the CALUX assay was shown to be suitable for the rapid screening of coconut oil for the presence of this type of PAHs.25 However, as shown in Fig. 4, low concentrations of benzo[a]pyrene (BaP) are only detected after the relatively short incubation time of 4 h. Following exposure to high concentrations of BaP, a clear response is also observed after 20 h.This effect is thought to be due both to the metabolism of BaP by the cells and to the instability of the luciferase formed during the first few hours of exposure.36 This is supported by the results obtained with the newly developed CAFLUX assay, expressing the more stable enhanced green fluorescent protein instead of luciferase.37 Based on actual concentrations in samples, this feature allows a discrimination between the effects of PAHs and dioxins, by varying the incubation interval. Fig. 4 again demonstrates that the short-term incubation of cells in conditioned medium results in much lower luciferase concentrations in the control cells (400 versus 6575 RLU per well). After 20 h the difference was much less (125 versus 625 RLUs per well), again indicating the presence of an unstable Ah receptor agonist in the medium. This is supported by the fact that the combined effect of BaP and this agonist appears to be additive, as shown by comparison of the response in fresh and conditioned medium after 4 h of Fig. 3 Effect of oxfendazole alone (2) or in combination with 100 pm TCDD (5) in the EROD (A) and CALUX (B) assays. pGudLuc transfected cells were incubated for 24 h with various concentrations of oxfendazole in the absence or presence of 100 pm TCDD. Subsequently, the EROD and luciferase activity were determined. Results are expressed as the mean ± s for n = 3. The protein content per well was unaffected by TCDD or oxfendazole.Fig. 4 Effect of benzo[a]pyrene in the CALUX assay, following exposure of p-GudLuc transfected H4IIE cells for 4 (5, 2) and 20 h (<, ¶) in either fresh (closed symbols) or conditioned (open symbols) medium. Results are expressed as the mean ± s for n = 3. The protein content per well was unaffected by the concentration of benzo[a]pyrene, incubation time or the type of medium. 82 Analyst, 1999, 124, 79–85incubation (Fig. 4). As observed with 50 pm TCDD [Fig. 2(A)], the response at high concentrations of BaP was higher in the presence of conditioned medium. At present it is unclear whether this is due to the degradation of a specific compound by the cells or rather the excretion of a specific factor into the medium. As shown in Fig. 3, a slight positive effect was also observed with oxfendazole, belonging to the group of the benzimidazoles, which are used as anthelmintics, fungicides and antigastrics. Similar effects were observed with other members of this group such as febantel, fenbendazole, thiabendazole, mebendazole, lanzoprazole, omeprazole and benomyl (Table 1).Incubation with 1 mm thiabendazole resulted in a response (18 406 ± 2218 RLU per well) similar to the plateau value obtained with 500 pM TCDD (17 714 ± 836 RLU per well). These type of compounds have been shown to induce cytochrome P450 1A related enzyme activities in vivo and in vitro,38–40 although there are strong indications that these compounds are not real Ah receptor agonists.41–43 The positive response in the CALUX assay excludes the possibility that the increased EROD activity in exposed animals is due to stabilization of cytochrome P450 1A.However, based on current regulations, these compounds should not be present in food derived from food-producing animals at levels that could induce a positive effect. A third group of compounds, showing a slight positive effect in the assay, are the corticosteroids, such as hydrocortisone, corticosterone and dexamethasone.As shown in Fig. 5, these compounds alone show a small and dose-related positive effect (A), but can also dramatically enhance the effect of TCDD (B). In the latter case the EC50 value shifted from 16 pm for the cells exposed to TCDD only to 8, 6 or 4 pm for cells exposed to TCDD in the presence of 100 nm of corticosterone, 100 nm of hydrocortisone or 10 nm of dexamethasone, respectively. There are strong indications that the effect is due to an indirect effect of these compounds on the Ah receptor pathway, probably an increase in the concentration of the receptor.44,45 It is very likely that these compounds are present in certain types of samples, in particular blood plasma, resulting in a false-positive result or overestimation of levels of Ah receptor agonists.As indicated by the appearance of a plateau level at higher concentrations of the hormones [Fig. 5(A)], this effect can be prevented by adding, e.g., 100 nm of dexamethasone to the incubation medium.The reason for the increased response observed when cells were exposed to only corticosteroids is unclear. However, the response was not influenced by using conditioned medium, excluding the augmentation of the cellular response to the ‘medium factor’ (see above) by corticosteroids. One possible way to investigate whether a positive response in the assay is caused by a real Ah receptor agonist would be the use of specific blockers of the Ah receptor. The flavonoids anaphthoflavone (aNF), reported to be a partial antagonist,46,47 and 4-amino-3-methoxyflavone (AMF), shown to be a pure antagonist in human breast tumor MCF-7 cells,48 were tested for this purpose.However, both aNF (data not shown) and AMF (Table 2) showed a strong luciferase induction themselves at concentrations of 50 and 400 mm, respectively, comparable to the effect observed at 3–10 pm TCDD. In comparison with TCDD, these two compounds had a relative potency of about 1026 and 1027, respectively (CALUX-TEF).As in the case of TCDD, the effect of 400 mm AMF was strongly induced by 10 nm dexamethasone (Table 2). At a lower concentration of AMF that did not induce the luciferase activity (4 mm), there was no clear inhibition of the signals induced by 3 pm TCDD or 10 mm oxfendazole. Similar was the case at higher concentrations of TCDD and appears to be in contrast with the strongly decreased CYP1A1 mRNA levels observed in MCF-7 cells.48 More specific antagonists of the Ah receptor are necessary to demonstrate the value of this approach, but such compounds remain to be discovered.At this stage, an alternative and more promising approach is the in vitro gel retardation assay, based on the binding of a compound to the Ah receptor and subsequently to DNA.49 Table 1 Effect of the benzimidazole compounds oxfendazole (OXF), fenbendazole (FBZ), febantel (FBT), thiabendazole (TBZ), mebendazole (MBZ), lanzoprazole (LPZ), omeprazole (OMP) and benomyl (BEN) in the CALUX assay.Cells were incubated with 1, 10, 100 or 1000 (only TBZ) mm of the compounds for 16 ha Concentration/mm Compound 1 10 100 1000 OXF FBZ FBT TBZ MBZ LZP OMP BEN 1308 ± 188 1429 ± 275 978 ± 232 459 ± 35 449 ± 36 991 ± 15 346 ± 4 309 ± 11 2131 ± 347 3324 ± 597 1378 ± 279 696 ± 66 1536 ± 93 3080 ± 135 388 ± 19 403 ± 16 3638 ± 37 2506 ± 413 1446 ± 165 4291 ± 162 2306 ± 20 12020 ± 45 2010 ± 110 1206 ± 37 18406 ± 2218 a Results are expressed as RLU per well (mean ± s for n = 3).Nonexposed cells showed a response of 395 ± 13 RLU per well. Cells exposed to 5, 50 or 500 pm TCDD showed a response of 4378 ± 179, 12 143 ± 594 and 17 714 ± 836 RLU per well, respectively. The protein content per well was slightly decreased at 100 mm OXF, FBZ or FBT and 1000 mm TBZ. Fig. 5 Effects of hydrocortisone (2), corticosterone (<) and dexamethasone (5) alone (A) or in combination (B) with various concentrations of TCDD in the CALUX assay.In the latter case cells were incubated for 24 h with various concentrations of TCDD in the absence (8) or presence of 100 nm hydrocortisone, 100 nm corticosterone or 10 nm dexamethasone. The latter results were fitted using a one-site ligand curve fit. Results are expressed as the mean ± s for n = 3. Protein content per well was unaffected by the concentration of the corticosteroid alone or in combination with TCDD.Analyst, 1999, 124, 79–85 83In general, it will be questionable whether a positive result is really false or simply due to a possibly unknown agonist, which has been overlooked by more specific chemical–analytical methods. As such, these assays are extremely suitable for the detection of unknown or new agonists with potential health risks for the consumer. On the other hand, current regulations are directed towards specific compounds or classes of compounds and not towards compounds with a certain biological effect.The use of simple but selective clean-up procedures, such as the acid silica procedure used for bovine milk,20,26 may be suitable to make the assay more specific for a certain class of agonists. Conclusions The new generation of bioassays has great potential, in particular as rapid screening assays. In the case of the CALUX assay for polyhalogenated aromatic hydrocarbons, the sensitivity of the assay allows its use for detecting residues at MRL levels.False-negative results are limited and can easily be recognized, whereas the occurrence of false-positive or, maybe better, undesired positive results can be reduced by the use of specific assay conditions or the application of more selective clean-up procedures. Other suitable areas for these kind of assays may include a number of different hormonal activities (estrogens, androgens, glucocorticoids, b-agonists), but also certain classes of myco-, phyto- or phycotoxins.In the ideal case this may result in assays that can be used under field conditions. In addition to the development and improvement of specific assays, it is important to develop relatively simple clean-up procedures with high and reproducible recoveries of analytes. As demonstrated above, special attention should be paid to the purity or simple purification of chemicals used for extraction or exposure of cells. Current developments in the area of high throughput screening (HTS) will further increase the sensitivity and speed, and thereby reduce the costs of these assays.To support these developments, it is essential that this new generation of bioassays is validated and introduced into the regulatory field as screening assays, supported by well recognized analytical methods. Acknowledgements The development and validation of the CALUX assay for Ah receptor agonists were financially supported by grants from the Dutch Ministry of Agriculture, Nature Management and Fisheries and the Dutch Technology Foundation (STW).References 1 R. Bogaerts and F. Wollf, Fleischwirtschaft., 1980, 60, 672. 2 C. Lund, in Proceedings of the 2nd World Congress on Foodborne Infections and Intoxications, Berlin, 1986, p. 819. 3 J. F. M. Nouws, N. J. G. Broex, J. P. M. Den Hartog and F. Driessens, Arch. Lebensmittelhyg., 1988, 39, 135. 4 F. R. Tejada, L. S. P. Madamba and A. W. Tejada, Phil. Agric., 1996, 79, 217. 5 S. Safe, Environ. Health Perspect., 1995, 103, 346. 6 R. Kroes, L. G. Huis in’t Veld, P. L. Schulter and R. W. Stephany, in Anabolic Agents in Animal Production, FAO/WHO Symposium Rome, March 1975, ed. Lu and Rendel, Georg Thieme, Stuttgart, 1976, p. 192. 7 C. J. M. Arts, M. J. van Baak, G. R. M. H. Haenen and J. van der Greef, in Euro Residue II, ed. N. Haagsma, A. Ruiter and P. B. Czedik-Eysenberg, University of Utrecht, 1993, p. 143. 8 A. M. Soto, Environ. Health Persp., 1995, 103 (Suppl. 7), 113. 9 S. F. A. Arnold, Science, 1996, 272, 1489. 10 P. Balaguer, M. S. Denison and T. R. Zacharewski, Organohalogen Compd., 1995, 23, 215. 11 E. J. Routledge and J. P. Sumpter, Environ. Toxicol. Chem., 1996, 15, 241. 12 J. Legler, L. Bouwman, A. J. Murk, A. Brouwer, J. Stronkhorst and D. Vethaak, Organohalogen Comp., 1996, 29, 347. 13 S. Safe, Chemosphere, 1987, 16, 791. 14 N. J. Bunce, Organohalogen Compd., 1995, 23, 209. 15 S. W. Kennedy, A. Lorenzen, C.A. James and B. T. Collins, Anal. Biochem., 1993, 211, 102. 16 T. W. Sawyer, A. D. Vatcher and S. Safe, Chemosphere, 1984, 13, 695. 17 J. M. M. J. G. Aarts, M. S. Denison, M. A. Cox, A. C. Schalk, P. A. Garrison, K. Tullis, L. H. J. de Haan and A. Brouwer, Eur. J. Pharm. Environ. Toxicol., 1995, 293, 463. 18 P. A. Garrison, K. Tullis, J. M. M. J. G. Aarts, A. Brouwer, J. P. Giesy and M. S. Denison, Fundam. Appl. Toxicol., 1996, 30, 194. 19 J. T. Sanderson, J. M. M. J. G. Aarts, A.Brouwer, K. L. Froese, M. S. Denison and J. P. Giesy, Toxicol. Appl. Pharmacol., 1996, 137, 316. 20 T. F. H. Bovee, L. A. P. Hoogenboom, A. R. M. Hamers, W. A. Traag, T. Zuidema, J. M. M. J. G. Aarts, A. Brouwer and H. A. Kuiper, Food Add. Contam., in the press. 21 S. Safe, Crit. Rev. Toxicol., 1990, 21, 51. 22 S. Safe, Crit. Rev. Toxicol., 1994, 24, 87. 23 International Toxicity Equivalency Factor (I-TEF) Method of Risk Assessment for Complex Mixtures of Dioxins and Related Compounds, NATO/CCMS (North Atlantic Treaty Organization, Committee on the Challenges of Modern Society), Report No. 176, North Atlantic Treaty Organization, Brussels, 1988. 24 U. G. Ahlborg, G. C. Becking, L. S. Birnbaum, A. Brouwer, H. J. G. M. Derks, M. Feeley, G. Golor, A. Hanberg, J. C. Larsen, A. K. M. Liem, S. Safe, C. Schlatter, F. Wærn, M. Younes and E. Yrjänheikki, Chemosphere, 1994, 28, 1049. 25 T. F. H. Bovee, L. A. P. Hoogenboom, W. A. Traag, T. Zuidema, J. H.J. Horstman, J. M. M. J. G. Aarts, T. J. Murk, A. Brouwer, M. S. Denison and H. A. Kuiper, Organohalogen Compd., 1996, 27, 303. 26 A. J. Murk, J. Legler, M. S. Denison, J. P. Giesy, C. van de Guchte and A. Brouwer, Fundam. Appl. Toxicol., 1996, 33, 149. 27 A. J. Murk, P. E. G. Leonards, A. S. Bulder, A. S. Jonas, M. J. C. Rozemeijer, M. S. Denison, J. H. Koeman and A. Brouwer, Environ. Toxicol. Chem., 1997, 16, 1583. 28 L. A. P. Hoogenboom and A. R. M. Hamers, Organohalogen Compd., 25, 53. 29 G. L. Peterson, Anal. Biochem., 1977, 83, 346. 30 M. Bradford, Anal. Biochem., 1976, 72, 248. 31 M. S. Denison, W. J. Rogers, M. Fair, M. Ziccardi, G. Clark, A. J. Murk and A. Brouwer, Organohalogen Compd., 1996, 27, 280. 32 A. Fautrel, C. Chesn�e, A. Guillouzo, G. De Sousa, M. Placidi, R. Rahmani, F. Braut, J. Pichon, H. Hoellinger, P. Vint�ezou, I. Diarte, C. Melcion, A. Cordier, G. Lorenzon, M. Benicourt, B. Vannier, R. Fournex, A. F. Peloux, N. Bichet, D. Gouy and J.P. Cano, Toxicol. In Vitro, 1991, 5, 543. Table 2 Luciferase activities in cells exposed for 24 h to 10 mm oxfendazole, 10 nm dexamethasone or 3 pm TCDD in the presence of 0, 4, 40 or 400 mm of 4-amino-3-methoxyflavone. Results are expressed as RLU per well (mean ± s for n = 3)a Concentration of AMF/mm None TCDD (3 pm) OXF (10 mm) DEX (10 nm) 0 4 40 400 364 ± 4 443 ± 11 631 ± 16 4453 ± 34 2231 ± 94 2148 ± 66 1449 ± 24 5381 ± 173 2136 ± 86 3415 ± 83 2484 ± 99 6581 ± 153 2153 ± 141 2676 ± 142 4571 ± 122 16235 ± 522 a There was no dose- or compound-related decrease in the protein content per well. 84 Analyst, 1999, 124, 79–8533 R. Gugler and J. C. Jensen, Gastroenterology, 1985, 89, 1235. 34 R. J. Chenery, A. Ayrton, H. G. Oldham, S. J. Norman and P. Standring, Biochem. Pharmacol., 1988, 37, 1407. 35 M. Murray, A. M. Hudson and V. Yassa, Chem. Res. Toxicol., 1992, 5, 60. 36 J. F. Thompson, K. F. Geoghegan, D. B. Lloyd, A. J. Lanzetti, R. A. Magyar, S. M. Anderson and B. R. Branchini, J. Biol. Chem., 1997, 272, 18 766. 37 J. M. M. J. G. Aarts, A. Jonas, L. C. van den Dikkenberg and A. Brouwer, Organohalogen Compd., 1998, 37, 85. 38 D. Diaz, I. Fabre, M. Daujat, B. Saint Aubert, P. Bories, H. Michel and P. Maurel, Gastroenterology, 1990, 99, 737. 39 X. Rey-Grobellet, N. Ferre, C. Eeckhoutte, G. Larrieu, T. Pineau and P. Galtier, Biochem. Biophys. Res. Commun., 1996, 220, 789. 40 S. Krusekopf, U. Kleeberg, A. G. Hildebrandt and K. Ruckpaul, Xenobiotica, 1997, 27, 1. 41 M. Daujat, B. Peryt, P. Lesca, G. Fourtanier, J. Domergue and P. Maurel, Biochem. Biophys. Res. Commun., 1992, 188, 820. 42 P. Lesca, B. Peryt, G. Larrieu, M. Alvinerie, P. Galtier, M. Daujat, P. Maurel and L. A. P. Hoogenboom, Biochem. Biophys. Res. Commun., 1995, 209, 474. 43 M. Daujat, S. Charrasse, I. Fabre, P. Lesca, Y. Jounaidi, C. Larroque, L. Poellinger and P. Maurel, Eur. J. Biochem., 1996, 237, 642. 44 F. J. Wiebel and P. Cikryt, Chem.–Biol. Interact., 1990, 76, 307. 45 B. D. Abbott, G. H. Perdew, A. R. Buckalew and L. S. Birnbaum, Toxicol. Appl. Pharmacol., 1994, 128, 138. 46 J. A. Blank, A. N. Tucker, J. Sweatlock, T. A. Gasiewicz and M. I. Luster, Mol. Pharmacol., 1991, 40, 607. 47 M. Merchant, V. Morrison, M. Santostefano and S. Safe, Arch. Biochem. Biophys., 1992, 298, 389. 48 Y. F. Lu, M. Santostefano, B. D. Cunningham, M. D. Threadgill and S. Safe, Arch. Biochem. Biophys., 1995, 316, 470. 49 P. Balaguer, A. Joyeux, M. S. Denison, R. Vincent, B. E. Gillesby and T. Zacharewski, Can. J. Physiol. Pharmacol., 1996, 74, 216. Paper 8/04950E Analyst, 1999, 124, 79–8

ISSN:0003-2654    

DOI:10.1039/a804950e

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Immobilisation of thiabendazole specific antibodies on an agarose matrix for application in immunoaffinity chromatography† Elizabeth Horne,a Tiernan Coyle,a Michael O’Keeffe*a and David L. Brandonb a Teagasc, The National Food Centre, Dunsinea, Castleknock, Dublin 15, Ireland b Food Safety and Health Research Unit, Western Regional Research Centre, Agricultural Research Service, US Department of Agriculture, 800 Buchanan Street, Albany, California 94710, USA Received 6th July 1998, Accepted 20th November 1998 The covalent immobilisation of two monoclonal antibodies (each of which binds both thiabendazole and its metabolite, 5-hydroxythiabendazole) on an agarose matrix is described.Immunoaffinity columns, made using these immunosorbents, bound radiolabelled thiabendazole and a number of eluents, were evaluated, with ethanol–water (70%) being selected as the most suitable. Re-equilibration of the immunosorbent was possible by purging with water, resulting in a reusable column.The columns were found to be stable under the conditions employed. The ability of an antibody to combine with an antigen with high selectivity and affinity is a characteristic feature which has raised interest in their use in immunoaffinity chromatography (IAC). IAC columns may be applied to sample pre-treatment through antigen isolation and concentration from a biological extract. Sample purification may be achieved by a washing step with suitable solvent(s) prior to elution.Historically, the use of immunoaffinity chromatography has been limited by the low availability of antibodies against target analytes coupled with the cost of antibody and immunoaffinity column production (milligram amounts of antibody per column may be required). However, the cost of IAC is offset somewhat if columns are reusable and may be reactivated by equilibration following use. A number of applications have been reported in the area of residue analysis for anabolic agents and veterinary drugs.For example, an immunosorbent prepared by the immobilisation of a monoclonal anti-salbutamol antibody (which exhibits a 75% cross-reactivity with clenbuterol) on a divinyl sulfone activated Sepharose 4B was used in the trace determination of salbutamol and clenbuterol in tissue samples.1 Other IAC sorbents reported include an anti-oestrogen immunosorbent prepared by immobilisation of anti-oestrogen antiserum on cyanogen bromide activated Sepharose 4B,2 and immunosorbents for 19b-nortestosterone and its metabolite 19a-nortestosterone,3,4 and for trenbolone and its metabolite 17a-trenbolone,5 prepared by coupling the appropriate rabbit antiserum to tresyl cyanogen bromide activated Sepharose and tresyl chloride activated Sepharose, respectively. A multiimmunoaffinity chromatography column for multi-anabolic (nortestosterone and methyl testosterone) residue analysis of meat was prepared by combining the isolated immunoglobulin G fractions from antisera raised against both anabolic steroids and coupling them to tresyl activated Sepharose.6 Applications of IAC in veterinary drug analysis have been reported such as those for determination of chloramphenicol in tissues7 and in milk and eggs,8 of dexamethasone in animal feed9 and in urine,10 and of avermectin in biological samples.11 This paper reports the immobilisation of two monoclonal antibodies,12 each of which binds both thiabendazole and 5-hydroxythiabendazole, on agarose gel.Radiolabelled thiabendazole was used to evaluate the binding efficiency of the columns and elution modes. Experimental Materials Two monoclonal antibodies, each of which binds thiabendazole and its metabolite 5-hydroxythiabendazole, were developed using the haptens 2-(2-succinamido-4-thiazolyl)benzimidazole (antibody no. 300) and 5-succinamido-2-(4-thiazolyl)benzimidazole (antibody no. 430) (Fig. 1). Hapten synthesis, formation of protein conjugates, antibody production and characterisation (Table 1) were performed as previously described.12 An AminoLink Plus Immobilisation kit obtained from Pierce and Warriner Ltd.(Chester, UK) was used for antibody immobilisation. AminoLink Plus coupling gel is a cross-linked beaded agarose support that has been activated to form aldehyde functional groups. The kit included plastic minicolumns (each containing 2 ml gel supplied as a 50% slurry in water with 0.02% sodium azide as preservative), polyethylene porous disks, neutral pH coupling buffer (0.1 mol l21 sodium phosphate, 0.15 mol l21 sodium chloride, pH 7.2, PBS), quenching buffer (1 mol l21 TRIS–HCl, pH 7.4), wash solution (1 mol l21 sodium chloride) and reducing agent (sodium cyanoborohydride solution, 5 mol l21).Radiolabelled thiabendazole (labelled with 14C in the benzene ring) was provided by Dr. Pierre Galtier (Institut National de la Recherche Agronomique (INRA), Toulouse, France).The methanolic stock solution received (100 ml) contained 500 mg thiabendazole with activity of 44.8 mCi (1.66 MBq). The stock solution was diluted to give a working solution at concentration 50 ng ml21 (4.48 3 1023 mCi ml21, 166 Bq ml21) in water. Methanol and water (HiperSolv grade), cocktail T (scintillation cocktail), ethylene glycol, sodium chloride and sodium dihydrogen orthophosphate (AnalaR grade) from BDH (Poole, Dorset, UK) and ethanol (pro analysi grade) from Merck (Darmstadt, Germany) were used.Sodium azide (SigmaUltra grade) was obtained from Sigma (Poole, Dorset, UK). † Presented at the Third International Symposium on Hormone and Veterinary Drug Residue Analysis, Bruges, Belgium, June 2–5, 1998. Analyst, 1999, 124, 87–90 87Apparatus A liquid scintillation counter, 1219 Rackbeta (LKB Wallac, Turku, Finland), and a GENESYS 5 UV/VIS spectrophotometer (Milton Roy, Rochester, NY, USA) were used. Antibody immobilisation The immunoaffinity chromatography columns were prepared according to the kit standard protocol.All solutions which were used for antibody immobilisation, column washing or elution were allowed to flow through the gel bed under gravity. The columns and reagents were equilibrated at room temperature. Antibodies 430 and 300 were supplied in phosphate buffered saline (pH 7.0) and diluted to a concentration of 1 mg ml21 with coupling buffer. The storage solution was drained from the column (without letting the gel bed become dry) and the gel was equilibrated by passing coupling buffer (5 ml) through the column and draining it.Antibody solution (2 ml) and sodium cyanoborohydride solution (40 ml) were added and the reaction slurry was mixed by gentle end-over-end rocking for 6 h at room temperature. The coupling solution was drained from the column and the gel was washed with coupling buffer (5 ml) and the effluents were combined and retained.Remaining active sites on the gel were blocked to prevent interaction with analytes during immunoaffinity chromatography. This was achieved by washing the gel with quenching buffer (4 ml) and then mixing it with quenching buffer (2 ml) and sodium cyanoborohydride (40 ml) by gentle end-over-end rocking for 30 min. The column was drained and washed with wash solution (at least 4 3 5 ml quantities) and the effluent was checked for the presence of uncoupled antibody; further washing was carried out until the absorbance of the effluent at 280 nm was zero.The gel was washed with degassed storage solution (3 3 5 ml, 0.05% aqueous solution of sodium azide) and a polyethylene porous disc was placed above the column bed (within 1 mm of the gel surface). When storage solution reached the top of the porous disc, the flow automatically stopped. To store the column, storage solution was added (2 ml) and the column was sealed and stored in an upright position protected from light at 4 °C.Determination of immobilisation yield The absorbance of the antibody coupling solution was measured (280 nm) prior to the immobilisation step. The absorbance of the combined effluents [coupling solution (2 ml) and coupling buffer wash (5 ml)] was measured (280 nm) following the immobilisation procedure in addition to the absorbance of the wash solution effluents. The immobilisation yield was calculated using these absorbance values. Evaluation of eluents All solvent mixtures, solutions and buffers were filtered and degassed prior to use. The column was equilibrated at room temperature.The storage solution was drained from the column which was then washed by passing water (5 ml) through it, leaving it ready for use. The elution of radiolabelled thiabendazole from the immunosorbent columns was examined by application of an aqueous thiabendazole solution (2 ml, 50 ng ml21, 4.48 3 1023 mCi ml21, 166 Bq ml21). The column was washed with water (5 ml) or phosphate buffer (5 ml, 0.1 mol l21, pH 7.2; a pre-elution buffer used only in the evaluation of sodium chloride solutions (1 mol l21 and 3 mol l21) as eluents).The eluent (35 ml) was allowed to pass through the column. The eluates from the application, washing and elution steps were collected as 1 ml fractions into scintillation vials. Cocktail T (10 ml) was added and the radioactivity in each vial was counted (5 min) on a liquid scintillation counter.The following solvent mixtures or buffers were evaluated: ethanol–water (10 + 90, 20 + 80, 30 + 70, 40 + 60, 50 + 50, 60 + 40, 70 + 30 v/v), methanol–water (60 + 40 v/v), ethylene glycol–water (50 + 50 v/v; pH 8.0) sodium chloride [1 mol l21 and 3 mol l21 in phosphate buffer (0.01 mol l21, pH 7.2) and phosphate buffer (0.1 mol l21 at pH 3.5, 6.0 and 10.0)]. To prepare the column for storage following use, it was washed with water (2 ml) and storage solution (5 ml).Once the column had drained, storage solution (2 ml) was added and it was stored sealed and upright in darkness at 4 °C. Results and discussion Three immunoaffinity chromatography columns were prepared by immobilisation of antibody 430 on agarose and two columns prepared by immobilisation of antibody 300. In all cases, 1 mg Fig. 1 (i) Thiabendazole, (ii) 5-hydroxythiabendazole, (iii) 2-(2-succinamido- 4-thiazolyl)benzimidazole hapten and (iv) 5-succinamido-2-(4-thiazolyl) benzimidazole hapten.Table 1 Specificities of monoclonal antibodies to thiabendazole and related compounds, determined by competitive ELISA IC50, a mmol l21 (Cross-reactivity,%) Compound Antibody 430b Antibody 300c Thiabendazole 0.018 (100) 0.069 (100) 5-Hydroxythiabendazole 0.1 (18) 0.81 (8.5) 5-Aminothiabendazole 0.032 (56) 1.1 (6.3) Cambendazole 0.0057 (316) > 100 ( < 0.07) Methylbenzimidazole carbamate > 100 ( < 0.02) 10 (0.7) Albendazole > 100 ( < 0.02) > 100 ( < 0.07) Mebendazole > 100 ( < 0.02) > 100 ( < 0.07) Fenbendazole > 100 ( < 0.02) > 100 ( < 0.07) a Concentration of compound which inhibits binding of the horseradish peroxidase-conjugated hapten to solid-phase antibody by 50%. b Elicited with 5-succinamidothiabendazole.c Elicited with 2-succinamidothiabendazole (adapted from ref. 12). 88 Analyst, 1999, 124, 87–90of antibody was applied per millilitre of agarose gel. The immobilisation efficiency at pH 7.2 was determined by measurement of antibody solution absorbance prior to and after the immobilisation procedure.In the case of the three antibody 430 columns the efficiency was determined to be in the range 76 to 86%, while the efficiency for the two antibody 300 columns was determined to be 84 and 85%. A higher coupling efficiency (approximately 20% higher) for bovine serum albumin on AminoLink plus coupling gel has been reported at pH 10 compared to pH 7.2.13 Antibody immobilisation may be performed at pH 10 only if the antibody is known to be stable at this pH.In the absence of such information, pH 7.2 was selected for this study. Antibodies 300 and 430 were covalently immobilised on agarose using stabilised Schiff’s base bonding. The commercially available gel had been activated to form aldehyde groups which would react with primary amine groups on the antibody surface under the conditions described. The bond was stabilised using the reducing agent sodium cyanoborohydride.The orientation of the antibody could not be controlled during immobilisation as primary amine groups are randomly distributed over an antibody surface. However, at least a part of the antibodies (300 and 430) remained active and retained their ability to bind thiabendazole on immobilisation. This was demonstrated by application of an aqueous radiolabelled thiabendazole solution (100 ng) to the columns and counting of the fractionated eluate (1 ml fractions).The possibility of non-specific interaction between the agarose gel and analytes (e.g., thiabendazole and its metabolites) was reduced by blocking active sites which remained after antibody immobilisation. TRIS–HCl buffer was used for this purpose. Due to the time and cost required in the production of immunoaffinity chromatography columns, it is highly desirable that reusable columns would be available. In order to achieve this, the antibody–antigen interaction must be reversible using mild reagents in order to avoid permanent denaturation of the antibody. In addition, the chemical stability of the gel must be taken into account when selecting analyte elution conditions as dissolution of or damage to the gel would result in antibody loss.The pH range of elution reagents for AminoLink Plus coupling gel is pH 2–11. The maximum organic modifier content of an elution mixture which will not adversely affect the gel is 70%.A number of approaches were taken to the elution of thiabendazole from a column containing immobilised antibody 430 (Table 2): (i) Water–organic modifier mixtures. Organic modifiers disrupt the hydrophobic interactions within the antibody, and those between the antibody and antigen, thereby changing the antibody structure and allowing release of an antigen. Organic modifier–water mixtures of 10 to 60% ethanol–water, 60% methanol–water and 50% ethylene glycol–water were unsatisfactory in that analyte elution was not observed at 10 to 20% ethanol–water mixtures or using 50% ethylene glycol–water and large elution volumes were required for ethanol–water mixtures with higher ethanol content.A small elution volume is required to make the immunoaffinity column a useful tool in sample concentration and clean-up. Ethanol–water (70 + 30 v/v) eluted thiabendazole in a volume of 8 ml with a recovery of 88%. The eluate may be concentrated easily by evaporation due to its high ethanol content.Although the immobilised antibodies were denatured on contact with organic modifiers, they were reactivated on flushing with water, resulting in a reusable immunoaffinity column. (ii) Buffer solutions. Buffers of high or low pH, that is greater than 3 pH units from the antibody’s pI, are used commonly to elute analytes from immobilised antibody columns. These buffers change the ionic interactions within the antibody thereby changing its structure and allowing antigen release.Due to considerations regarding the stability of the agarose gel the buffer pH values examined were 3.5, 6.0 and 10.0. No elution of radiolabelled thiabendazole was observed when 30 ml of the eluents were passed through the column. (iii) Chaotropic ions. Solutions of chaotropic ions in buffer are used for analyte elution from immunoaffinity columns as they disrupt the water structure around an antibody. This results in a disruption of the hydrophobic interactions in the antibody structure, and between the antibody and antigen, giving rise to antigen release.Common chaotropic ions used are chloride, iodide, perchlorate and thiocyanate ions with chloride being one of the weaker types. The chloride ion was examined as a chaotropic eluting reagent in this study as the stability of antibody 430 was unknown and strong chaotropic ions may render an immunoaffinity column unusable. Solutions of sodium chloride (1 and 3 mol l21) in phosphate buffer (0.01 mol l21, pH 7.2) were evaluated.It is common to flush the column with pre-elution buffer prior to eluting the column with chaotropic ions. The pre-elution buffer is usually the same as the buffer used to prepare the chaotropic ion solution. As was the case with buffer solutions, no elution of thiabendazole was observed when 30 ml of the eluent was passed through the column. The requirement for relatively harsh elution conditions, i.e., use of organic modifier–water mixtures, may indicate that the immobilised antibody binds thiabendazole very strongly.The use of ethanol–water (70 + 30 v/v) was also found to be suitable for the elution of radiolabelled thiabendazole from immobilised antibody 300; while most of the thiabendazole eluted in a volume of 8 ml (Fig. 2), elution commenced only after application of 10 ml ethanol–water to the column. Application of immunoaffinity columns The columns were used in studies on protein-bound residues of thiabendazole.A number of methods were evaluated for the release of bound thiabendazole or its metabolites from rabbit hepatocytes, pig and mouse livers and the extracts were applied to the immunoaffinity columns. Immobilised antibody 300 would be expected to bind released residues which have intact bendzimidazole moieties but not necessarily intact thiazole moieties, while immobilised antibody 430 would be expected to bind only released residues with an intact thiazole structure.Application of extracts was followed by column washing with water (12 ml) during which non-immunoactive residues and matrix co-extractives were eluted. Column washing with ethanol–water (70 + 30 v/v) (12 ml for antibody 430 and 20 ml for antibody 300) resulted in elution of immunoactive substances. The immunoaffinity columns were reusable, under the Table 2 The assessment of eluents for elution of thiabendazole from an immunoaffinity column containing immobilised antibody 430 Eluent Elution volume/ml Recovery (%) Ethanol–water: 10 + 90 NEa 20 + 80 NE 30 + 70 > > 30 NMb 40 + 60 > 30 NM 50 + 50 28 72 60 + 40 26 103 70 + 30 8 88 Methanol–water, 60 + 40 26 70 Ethylene glycol–water, 50 + 50, pH 8.0 NE Sodium chloride (1 mol l21 and 3 mol l21) in phosphate buffer, pH 7.2 NE Phosphate buffer (0.1 mol l21), pH: 3.5, 6.0, 10.0 NE a No elution: elution did not occur when 30 ml of the eluent was passed through the column.b Not measured. Analyst, 1999, 124, 87–90 89conditions of use described, for at least 6 months (antibody 430). Conclusion Monoclonal antibodies capable of binding thiabendazole and its metabolite, 5-hydroxythiabendazole, were successfully immobilised on agarose gels to form immunoaffinity columns. The antibodies’ binding capacity for thiabendazole was maintained on immobilisation and ethanol–water (70 + 30 v/v) was identified as a suitable eluent. Reactivation of the columns was possible by flushing with water following antigen elution, rendering the columns reusable.This work was carried out within the framework of the EC-AIR programme (Project no. AIR2-CT93-0860, ‘Development of metabolically competent in vitro models for food safety evaluation’); the authors (EH, TC and MOK) acknowledge financial support from this programme. References 1 K. Pou, H. Ong, A. Adam, P. Lamothe and P. Delahaut, Analyst, 1994, 119, 2659. 2 A. Farjam, A.Brugman, H. Lingeman and U. A. Th. Brinkman, Analyst, 1991, 116, 891. 3 L. A. van Ginkel, R. W. Stephany, H. J. van Rossum, H. van Blitterswijk, P. W. Zoontjes, R. C. M. Hooijschuur and J. Zuydendorp, J. Chromatogr., 1989, 489, 95. 4 W. Haasnoot, R. Schilt, A. R. M. Hamers, F. A. Huf, A. Farjam, R. W. Frei and U. A. Th. Brinkman, J. Chromatogr., 1989, 489, 157. 5 L. A. van Ginkel, H. van Blitterswijk, P. W. Zoontjes, D. van den Bosch and R. W. Stephany, J. Chromatogr., 1988, 445, 385. 6 L. A. van Ginkel, R. W. Stephany, H. J. van Rossum, H. M. Steinbuch, G. Zomer, E. van de Heeft and A. P. J. M. de Jong, J. Chromatogr., 1989, 489, 111. 7 C. van de Water and N. Haagsma, J. Chromatogr., 1987, 411, 415. 8 C. van de Water, D. Tebbal and N. Haagsma, J. Chromatogr., 1989, 478, 205. 9 D. Courtheyn, N. Verheye, V. Bakeroot, V. Dal, R. Schilt, H. Hooijerink, E. O. van Bennekom, W. Haasnoot, P. Stouten and F. A. Huf, in Proc. EuroResidue II Conference, ed. N. Haagsma, A. Ruiter and P. B. Czedik-Eysenberg, University of Utrecht, Faculty of Veterinary Medicine, Utrecht, The Netherlands, 1993, p. 251. 10 S. M. R. Stanley, B. S. Wilhelmi and J. P. Rodgers, J. Chromatogr. B, Biomed. Appl., 1993, 620, 250. 11 J. Li and C. Qian, J. AOAC Int., 1996, 79, 1062. 12 D. L. Brandon, R. G. Binder, A. H. Bates and W. C. Montague, Jr., J. Agric. Food Chem., 1992, 40, 1722. 13 A Technical Guide to Protein Immobilisation, Pierce and Warriner (UK) Ltd., Chester, UK. Paper 8/05220D Fig. 2 Elution profile of [14C] thiabendazole (circa 2000 counts min21 applied) from immunoaffinity columns containing immobilised antibody 430 ( ______ ) and immobilised antibody 300 ( - - - - ). 90 Analyst, 1999, 124, 87–90

ISSN:0003-2654    

DOI:10.1039/a805220d

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Preparation and application of a trimethoprim ion-selective piezoelectric sensor Weifeng Li, Xiaoli Su, Huijuan Guo, Wangzhi Wei and Shouzhuo Yao* College of Chemistry and Chemical Engineering, Hunan University, Changsha 410082, China. E-mail: xieqj@public.cs.hn.cn Received 29th September 1998, Accepted 18th November 1998 PVC film containing various activants was coated on one electrode of a thickness-shear mode piezoelectric quartz crystal (PQC) to fabricate a trimethoprim (TMP) ion-selective piezoelectric sensor for the determination of TMP.The method is based on the selective adsorption of TMP ion on the modified film and the sensitive mass response of the PQC. The method was found to be sensitive, rapid and easy to handle with no need for previous sample treatment. The logarithm of frequency shift of PQC was found to bear a linear relationship to the logarithm of the concentration of TMP over the range 2 3 1028–2 3 1023 m with a detection limit of 2 3 1028 m at pH 4.0.Recoveries were from 99–102.9%. Two activants, TMP phosphotungstate and TMP silicotungstate, were synthesized and investigated. Influencing factors were examined in detail and optimized. The results for real samples obtained by the proposed method agreed with those obtained by a conventional method. 1. Introduction Trimethoprim, 2,4-diamino-5-(3,4,5-trimethoxybenzyl)pyrimidine (TMP), widely used in conjunction with sulfamethoxazole (SMX), is an effective antimicrobial agent used, for example, in the treatment of urinary tract infections and as a powerful bacteriostatic agent, because the antimicrobial activity is greater than that when the sulfa-drug is used alone.There are various analytical procedures for the assay of TMP alone or mixed with sulfamethoxazole, the most important being non-aqueous titrimetry,1 fluorimetry,2 HPLC,3–5 UV spectrophotometry,6 HPTLC,7 ion-selective electrode (ISE) methods8,9 and differential pulse polarography and cyclic voltammetry.10 The titrimetric method suffers from lack of selectivity and other basic substances may cause interferences.Tedious sample clean-up procedures including extraction and purification are always required for HPLC and TLC methods. UV spectrophotometry requires expensive instrumentation. A drawback of the ISE method is that its response behavior is affected by the high background conductivity of the solution and the high impedance and the low capacity of the electrode, and its applications have been limited to a certain extent. It is necessary, therefore, to develop efficient, sensitive and reliable analytical methods for the rapid determination of trimethoprim.Selective adsorption–desorption processes of the component ions of insoluble salts at their solid–aqueous interface have been studied by several groups and some fundamental characteristics of the adsorption mechanism of solid films were offered. These theoretical results were proved to agree well with experiments and applied to real sample detection.11–15 In our work, two insoluble salts, TMP phosphotungstate and TMP silicotungstate, were synthesized and applied for the assay of trimethoprim.A piezoelectric quartz crystal (PQC), on the basis of the Sauerbrey equation,16 is used in a microgravimetric technique capable of measuring as small as nanogram mass changes easily on its surface.16,17 Recently, PQC has been employed as a useful tool for the in situ, quantitative measurement of adsorption processes at the nanogram level at PQC electrode–solution interfaces, such as specific adsorption processes of various anions on metal electrode surfaces,18 underpotential deposition processes,19 self-assembly adsorption of alkanethiols on Au20 and selective complexion processes between a PQC surface modified using a Langmuir–Blodgett film and analytes in solution.21 Although the PQC itself is not a selective detector and responds to any mass change, surface modification of the PQC using a chemically selective reagent leads to a useful chemical sensor which works in aqueous systems.22 Compared with the conventional detection methods, as a mass response microgravimetric method PQC possesses greater sensitivity to mass changes and has better frequency stability, rapid response and a lower frequency–temperature coefficient.Further, it is simple to fabricate and convenient to operate.It is not surprising that the PQC has recently been used so widely, e.g., for the determination of pharmaceutical compounds23 and in the life sciences.24,25 On the basis of the above-mentioned principle, a novel allsolid- state trimethoprim ion-selective piezoelectric sensor was fabricated and applied to real sample assays. 2. Experimental 2.1. Reagents All chemicals used were of analytical-reagent grade. Trimethoprim was obtained from Peking Medicine Factory (Peking, China).Doubly distilled water was used throughout. 2.2. Preparation of solutions 2.2.1. Buffer solution. Sodium hydroxide solution (1 m, 50 mL) was mixed with 428 mL of 1 m acetic acid and the mixture diluted to 1 L. This solution, which had a pH of 4.0 and an ionic strength of 0.05 m, was used as the background solution throughout. 2.2.2. Trimethoprim standard solutions. A suitable amount of trimethoprim powder was dissolved 36% acetic acid solution and the solution was adjusted to pH 4.0 with sodium hydroxide solution.By successive dilution, a series of working standard solutions from 1021 to 1026 m were prepared. Analyst, 1999, 124, 91–95 912.2.3. Preparation of body fluids. According to ref. 26 artificial gastric juice was prepared by mixing 5 g of pepsin with 8.2 mL of dilute hydrochloride solution and diluting to 500 mL with water. A similar operation was performed for artificial intestine juice preparation. 2.3. Preparation of ion-pair complexes8 Trimethoprim phosphotungstate (TMP-PPT) was prepared by mixing 20 mL of 0.02 m phosphotungstic acid solution and 40 mL of 0.01 m trimethoprim solution.The precipitate was filtered off on a Porosite-4 sintered-glass crucible, washed several times with water and dried under vacuum. Trimethoprim silicotungstate (TMP-SCT) ion-pair complex was prepared in a similar manner. 2.4. Fabrication of the ion-selective piezoelectric (ISP) sensor The procedure for the fabrication of a TMP-PPT–PVC ISP sensor for the surface modification on the Ag electrode of the PQC was as follows: 10 mg of TMP-PPT ion-pair complex prepared by the aforementioned method, about 0.1 mL of dibutyl phthalate and 200 mg PVC powder were mixed thoroughly and dissolved in 5 mL of tetrahydrofuran. A small volume of this solution was spread on one Ag electrode surface of a PQC while it rotated rapidly.The tetrahydrofuran slowly evaporated at room temperature from a mixture placed on the surface of the quartz crystal and a transparent uniform film was formed on the crystal surface.A similar procedure was carried out for a TMP-SCT–PVC ISP sensor. All ISP sensors thus prepared can be stored in air when not in use. 2.5. Apparatus The ISP sensor used was a 9 MHz quartz crystal (12.5 mm diameter) with silver electrodes (6 mm diameter) on each side. An Iwatsu SC-72001 universal frequency counter (Japan) was employed to record frequency. The quartz crystal was fixed to a detection cell made of PTFE using silicone rubber, in which only one side of the quartz crystal was allowed to contact solution.The crystal holder was directly connected to an ICTTL oscillating circuit21 supplied by a JWY-30B dc voltage regulator, and the dc working voltage was set at 5 V. Temperature control (37 ± 0.1 °C) was achieved using a constant-temperature water-bath and a computer (Model 4192A, Hewlett-Packard, Palo Alto, CA, USA) was used for data analysis. 2.6. Procedure The modified sensor should be immersed in 1 3 1024 m trimethoprim solution for 3 h for preconditioning before detection is carried out, and then washed with water until the frequency returns to its initial value (frequency for the modified quartz crystal oscillator in air, fa). To eliminate the effect of the film thickness and concentration of activant, the same modified sensor was used repeatedly to complete a series of measurements and the detection conditions were strictly maintained constant when experiments were carried out in stirred buffer solution (pH 4.0) and the temperature was kept at 37 ± 0.1 °C when room temperature was about 20 °C.A preconditioned ISP sensor was immersed in background electrolyte solution and a steady resonant frequency (f1, frequency shift < 1 Hz in 3 min) was obtained. Then a series of standard solutions or sample solutions from low to high concentration were injected into the background solution in steps to raise the concentration of trimethoprim ion.At constant time intervals, both time and the corresponding frequency of the ISP sensor were recorded until the frequency became stable (fi). The frequency shift for the concentration of each solution was calculated as Df = fi 2 f1 (1) 3. Results and discussion 3.1. Theoretical considerations The proposed method is based on the selective adsorption– desorption of trimethoprim ion across a PVC active film. When any ion is adsorbed or desorbed into or from the modified film, or several ions with different molecular mass exchange with one another across the film, the surface mass changes of the modified film can be measured by an ISP sensor, even though these changes may be very small.For the AT-cut quartz crystal, considering the general case of overtones, the relationship27 between the mass change and frequency shift can be expressed as Df = 22.26 3 106f0 2hDm/A (2) where Df is the change in the oscillation frequency, h is the overtone order, Dm/A is the change in the mass of the film per unit area and f0 is the oscillation frequency of the fundamental mode of the quartz crystal.Hence the mass change can be expressed by the frequency shift. On the other hand, when the sensor is immersed in trimethoprim solution, an equilibrium is established between the two phases. At a low concentration ( � 0.01 m), the relationship between the concentration of the tested ion and the amount adsorbed, namely the mass change of the modified film, can be expressed by the Freundlich isotherm28 equation: Dm = KfC1 n (3) where n and Kf are constants and n > 1.Combining eqns. (2) and (3), we obtain Df = 22.26 3 106f0 2KfhC1 n (4) For a given AT-cut quartz crystal, the area A is a constant, with the following mathematical consequence: Df = KAC1 n (5) where KA = 2.26 3 106f0Kfh/A is a constant. Its logarithmic form is log (2Df) = 1 n log C + log KA (6) 3.2.Sensor performance ISP sensors, made with TMP salts of different anions, were prepared and their performances were compared. The same ISP sensor was used for a series of experiments. Fig 1 shows a typical temporal course of the response of the ISP sensor modified with TMP-SCT–PVC film. After conditioning in the background solution without any TMP ion (platform A, frequency f0), with increase in TMP concentration the frequency of the ISP sensor decreased rapidly. For each concentration of TMP ion, there is a corresponding platform.After completing a series of detections at various TMP ion concentrations, the solution was removed from the detection cell and water was injected repeatedly until the frequency of the ISP sensor gradually increased to reach a value (fr) close to the steady oscillating frequency (f0) in the background solution 92 Analyst, 1999, 124, 91–95obtained after the first conditioning. In other words, the solution was continuously diluted.Generally, there was finally always a small difference between fr and f0, mainly due to the adsorption of TMP ions at the sensor surface or the probability of the dissolution of activants in the renewed TMP ion-free background solution. Fig. 2 shows a comparison of two kinds of ISP sensors, modified using TMP-PPT–PVC film and TMP-SCT–PVC film. The linear calibration graphs of log (2Df) versus log C, where C is the concentration of trimethoprim ion, can be described by the respective equations log (2Df) = 3.36 + 0.25 log C (r = 0.987) log (2Df) = 3.49 + 0.31 log C (r = 0.99) Obviously, the ISP sensor modified using TMP-SCT–VC film exhibits better performances than the other.Therefore, most of our experiments were carried out using the TMP-SCT–PVC ISP sensor. The adsorbed trimethoprim species was found to be easily washed off the modified film and the ISP sensor can be used for 6 months without physical damage. pH effect The effect of pH on the response of an ISP sensor modified using TMP–SCT-PVC film was studied in solutions of different TMP concentration (in the background solution), in which the pH was varied by adding hydrochloric acid and/or sodium hydroxide solution from a 50 ml microsyringe.At pH values between 3 and 7, no significant effect on frequency response was observed (Fig. 3). At pH > 7, trimethoprim tends to precipitate, causing the frequency to increase. A similar phenomenon was observed at pH < 3; this is due to a shift of the equilibrium between trimethoprim ion and free trimethoprim in the solution. 3.4. Thickness of the film After conditioning, the modified ISP sensor was immersed in background solution until a stable baseline was finally obtained, then a series of sample solutions where injected to raise the concentration of TMP. During the experiments, frequency responses were measured using the same ISP sensor and the experimental conditions were kept the same. These treatments allow effects due to the viscoelasticity of the coating to be neglected; subsequent test results proved that this is valid.The thickness of the film has a significant effect on the response. An over-thin film causes lower sensitivity and a narrower response range, while an over-thick film leads to a long-term shift in the frequency and instability, and even non-oscillation of the ISP sensor. In this work, about a 104 Hz shift resulting from the film has usually been used. 3.5. Selectivity When trimethoprim is determined by other methods, many foreign basic substances interfere with the determination, and these must be removed beforehand. The effects of some of these substances and other common compounds on the response of the TMP ISP sensor were examined. Fig. 4 shows the response of the PQC modified using only PVC film. It is obvious that this PQC did not exhibit any clear selectivity for TMP ion. Background solution containing 0.05 m Na+ was used throughout, which means that the sodium cation did not seriously interfere with the response of the ISP sensors. First, we define kit = Dfi/Dft as the response-selectivity coefficient, where Dfi is the frequency-shift response of the ISP sensor to 1 3 1024 M of the interfering ion and Dft is the frequency-shift response to 1 3 1024 m of trimethoprim ion.The results (Table 1) show that the ISP sensor shows rather good selectivity for trimethoprim, and even most sulfa-drugs do not interfere significantly. It can be seen that no significant interferences are caused by various inorganic compounds and pharmaceutical Fig. 1 Course of the observed frequency response of the trimethoprim ISP sensor with sample-solution injections. Final concentration of trimethoprim (A) 0; (B) 2 3 1028; (C) 2 3 1027; (D) 2 3 1026; (E) 2 3 1025; (F) 2 3 1024; (G) 4 3 1023 m. Detection was carried out using the TMP-SCT–PVC ISP sensor in the background solution.Fig. 2 Concentration dependence of the frequency response determined using the ISP sensor modified with (2) TMP-SCT–PVC film and (5) TMPPPT –PVC film. C is the concentration of trimethoprim. Fig. 3 Influence of the solution pH on the frequency response of the TMPSCT –PVC ISP sensor: (2) 2 3 1025 and (~) 2 3 1024 m trimethoprim solution. Analyst, 1999, 124, 91–95 93ingredients. When the molecular mass is considered, kit becomes K = Dfi/Mi Dft/Mt = Kit Mt Mi (7) where Mi and Mt are the molecular mass of the interfering cations and trimethoprim, respectively. Here, K can express the selectivity more precisely. 3.6. Applications of the ISP sensor The applicability of the method to the assay of pharmaceutical preparations was examined. The response curves in Fig. 5 show good agreement with the Freundlich isother Tables 2 and 3 give the recoveries for aqueous solution, artificial gastric juice and artificial intestine juice and a comparison between the proposed method and the non-aqueous titration method.The results were satisfactory. 3.7. Comparisons of several methods in the determination of trimethoprim The comparisons in Table 4 show that the proposed method has some advantages over other detection techniques, e.g., the reagents and instruments required are simpler and cheaper than those required in HPLC, TLC and UV methods. Furthermore, as a mass sensor, the influence of the electric double-layer capacitance, the Faradaic impedance and the conductive properties of the modified membrane is negligible.No significant effect was caused by the background; good responses can easily be obtained and the stability of the frequency signal is better than that obtained in the potentiometric method. 4. Conclusions The development and investigation of new pharmaceutical sensors are of increasing interest in the chemical and pharma- Fig. 4 Response of the PQC sensor modified using only PVC film.All cations were measured in the background solution. (2) Trimethoprim; (½) benzydamine; (5) isoniazid; (») urea; (3) quinine. Table 1 Ion selectivity of the TMP-SCT ISP sensor Interfering ion kij = Dfi/Dfb a K = kij(Mt/Mi)a Potassium chloride No interference Magnesium chloride No interference Barium chloride 0.01 0.03 Copper chloride 0.02 0.07 Urea 0.03 0.14 Glucose 0.02 0.04 Atropine sulfas 0.12 0.06 Benzydamine 0.06 0.06 Lactate 0.03 0.09 Isoniazid 0.04 0.09 Tetramethylammonium bromide 0.03 0.13 Ammonium chloride No interference Quinine hydrochloride 0.07 0.06 Theophylline 0.02 0.02 Nicotinamide 0.05 0.12 Morphine hydrochloride 0.07 0.07 Sulfadimethoxine 0.08 0.09 Sulfaguanidine 0.10 0.15 Sulfacetamide 0.10 0.12 Sulfadiazine 0.13 0.14 a fi = frequency response of the sensor to 2.0 3 1024 m interfering ion; fj = frequency response of the sensor to 2.0 3 1024 m TMP ion; Mi = molecular mass of the interfering ion; Mt = molecular mass of the trimethoprim ion.Table 2 Determination of trimethoprim using the TMP ISP sensor in different media Artificial gastric juice Artificial intestine juice Added/mg Founda/mg Recovery (%) Added/mg Founda/mg Recovery (%) 2.7 2.6 ± 0.05 96.3 1.9 2.0 ± 0.03 105.3 9.4 9.3 ± 0.09 98.9 4.6 4.5 ± 0.14 97.8 17.6 17.2 ± 0.24 97.7 11.4 11.7 ± 0.51 102.6 23.8 24.3 ± 0.83 102.1 17.2 17.4 ± 0.95 101.2 28.1 28.4 ± 0.93 101.1 23.8 23.5 ± 1.4 98.7 35.2 35.7 ± 1.31 101.4 30.1 30.2 ± 1.75 100.3 Mean ± s 99.6 ± 2.2 101 ± 2.7 a Mean ± standard deviation (n = 3).Fig. 5 Graph of log (Df) versus log C for the TMP-SCT–PVC ISP sensor: (2) in aqueous solution (in the background solution); (½) in artificial gastric juice solution; (3): in artificial intestine juice solution. C = concentration of trimethoprim. 94 Analyst, 1999, 124, 91–95ceutical fields. As a selective, economic, convenient and sensitive analytical tool, the ISP sensor is expected to find wider applications for the determination of various pharmaceutical substances.Acknowledgement This work was supported by the China Natural Science Foundation. References 1 US Pharmacopeia, XX Revision, American Pharmaceutical Association, Washington, DC, 1980, pp. 744–751. 2 Z. G. Pang, B. Q. Wang, J. J. Zhou and N. Wang, Fenxi Huaxue, 1994, 22, 363. 3 P. Nachilobe, J. O. Boison, R. M. Cassidy and A. C. E. Fesse, J. Chromatogr. B, Biomed. Appl., 1993, 127, 243. 4 S. C. Li, Y. H. Cao and Y. Li, Yaowu Fenxi Zazhi, 1995, 15, 36. 5 V. Hormazabal, I. Steffenak and M. Yudestad, J. Chromatogr. A, 1993, 648, 183. 6 N. Zhang, X. G. Zhou, R. C. Guan and Y. Q. Zhang, Yaowu Fenxi Zazhi, 1995, 15, 38. 7 J. K. Lalla, S. U. Bhat, N. R. Sanda and M. U. Shah, Indian Drugs, 1997, 34, 275. 8 S. Z. Yao, J. Shiao and L. H. Nie, Talanta, 1987, 34, 983. 9 M. A. Ahmed and M. M. Elbeshlawy, Anal. Lett., 1995, 28, 2123. 10 N. X. Wan and Z. B. Xie, Yaoxue Xuebao, 1987, 22, 848. 11 E. G. Harsanyi, K.Tóth, L. Pólos and E. Pungor, Anal. Chim. Acta, 1983, 152, 163. 12 T. R. Berube, R. P. Buck, E. Lindner, K. Tóth and E. Pungor, Anal. Chem., 1991, 63, 946. 13 K. Uosaki, Y. Shigematsu, H. Kita, Y. Umezawa and R. Souda, Anal. Chem.,1989, 61, 1980. 14 Y. Tani, Y. Unezawa, K. Chikama, A. Hemmi and M. Soma, J. Electroanal. Chem., 1994, 378, 205. 15 K. Iitaka, Y. Tani and Y. Umezawa, Anal. Chim. Acta, 1997, 338, 77. 16 G. Z. Sauerbrey, Z. Phys., 1959, 155, 206. 17 D. A. Buttry and M.D. Ward, Chem. Rev., 1992, 92, 1355. 18 M. Deakin, T. Li and O. Melroy, J. Electroanal. Chem., 1988, 243, 343. 19 M. Deakin and O. Melroy, J. Electroanal. Chem., 1988, 239, 321. 20 K. Shimadzu, I. Yagi, Y. Sato and K. Uosaki, Langmuir, 1992, 8, 1385. 21 Y. Ebara, H. Ebato, K. Ariga and Y. Okahata, Langmuir, 1994, 10, 2268. 22 R. Cox, D. Gomez, D. A. Buttry, P. Bonnesen and K. N. Raymond, Anal. Chim. Acta, 1994, 561, 71. 23 L. H. Nie, S. L. Tan and S. Z. Yao, J. Pharm. Sci., 1991, 80, 17. 24 S.H. Si, Y. J. Xu, L. H. Nie and S. Z. Yao, J. Biochem. Biophys. Methods, 1995, 31, 135. 25 B. Konig and M. Gratzel, Anal. Lett., 1993, 26, 1567. 26 Chinese Pharmacopoeia, Chemical Engineering Press, Beijing, China, 3rd edn., 1985, vol. 2, pp. 82. 27 S. Yao, S. Tan and L. Nie, Sci. China, Ser. B, 1993, 36, 796. 28 Adsorption from Solution at Solid/Liquid Interfaces, ed. G. D. Parfitt and C. H. Rochester, Academic Press, London, 1983. Paper 8/07581F Table 3 Determination of trimethoprim using the TMP-SCT ISP sensor and the pharmacopoeia method ISP sensor Non-aqueous titrimetry26 Added/mg Founda/mg Recovery (%) Added/mg Found/mg Recovery (%) 6.8 6.9 ± 0.04 101.5 3.5 3.4 97.1 12.5 12.3 ± 0.17 98.4 7.2 7.2 100.0 19.7 20.0 ± 0.29 98.4 16.7 16.9 101.2 21.4 21.6 ± 0.42 101.5 24.1 24.3 100.8 26.2 26.5 ± 0.93 101.1 27.0 27.4 101.5 29.5 29.7 ± 0.87 100.7 32.9 32.6 99.1 Mean ± s 100.7 ± 1.2 99.9 ± 1.7 a Mean ± standard deviation (n = 3). Table 4 Comparisons of the proposed method with other methods for the determination of trimethoprim Method Application Calibration range Detection limit Recovery (%) RSD (%) Ref. HPLC Feeding stuffs 0.25–4 mg g21 100–105 2.2–5.8 5 HPLC-MS Serum 10–200 ng mL21 1.7 3 10211 m 82–89 3 HPTLC Pharmaceuticals 50–250 ng 25 ng 7 Fluorimetry Pharmaceutical 0.12–3 mg mL21 95.5–105.7 2 Spectrophotometry Solution 0.019–0.21 mg mL21 98.1–101.1 5 ISE Aqueous solution 0.01–10 mm 2.4 3 1026 m 9 Cyclic voltammetry Aqueous solution 0.55–5.5 mm 2.0 3 1027 m 10 ISP sensor Aqueous solution 2 3 1028–2 3 1023 m 2 3 1028 m 98.4–101.5 1.16 (n = 6) This work Analyst, 1999, 124, 91–95 95

ISSN:0003-2654    

DOI:10.1039/a807581f

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Construction and evaluation of ion selective electrodes for perchlorate with a summing operational amplifier: application to pyrotechnics mixtures analysis Ricardo Pérez-Olmos,*a Ainoa Rios,a María P. Martín,a Rui A. S. Lapab and José L. F. C. Limab a Departamento de Química Analítica, Escuela Universitaria de Ingeniería Técnica Industrial, Plaza de la Casilla nº3, 48012 Bilbao, Spain b CEQUP/Departamento de Química-Fisica, Faculdade de Farmacia, Rua Anibal Cunha 164, 4050 Porto, Portugal Received 4th September 1998, Accepted 17th November 1998 The construction and evaluation of an electrode selective to perchlorate with improved sensitivity, constructed like a conventional electrode (ISE), but using an operational amplifier to sum the potentials supplied by four membranes (ESOA) is described.The two types of electrodes, without an inner reference solution, were constructed using octylammonium chloride as sensor, a mixture of dibutylphthalate and o-nitrophenyloctyl ether as solvent mediator and PVC as plastic matrix, the membranes being obtained by direct application onto a conductive epoxy resin support.After the comparative evaluation of their working characteristics they were used in the determination of perchlorate in propellants, fulminating powders and fireworks. On all occasions, the results obtained by the ESOA were found to be similar to those obtained by the conventional ISE, but with higher precision. The limit of detection of the direct potentiometric method developed was found to be 0.1 g kg21, and the precision and accuracy of the method, when applied to eight different samples of pyrotechnic mixtures, expressed in terms of mean relative standard deviation and average percentage of spike recovery were 0.4% and 100.5%, respectively.Adequate agreement was found between the results obtained by the potentiometric method and the reference methods, since the calculated relative errors ranged from +0.4% to 20.5%.Introduction Although ammonium and potassium perchlorates are more expensive than potassium chlorate, they have replaced it, as oxidizer, in the composition of many pyrotechnic mixtures used as explosives and propellants for military and civilian purposes. This is because perchlorates contain more available oxygen per unit weight and are more stable and less sensitive to mechanical action than chlorates.1 They are widely used today throughout the world for mining, excavation and demolition.In the fireworks industry, combined with other compounds, they are used to produce coloured flames, noise and light.2 The recommended analytical methods for the determination of perchlorate in explosives and propellants usually consist in titrimetric procedures. In some of them the ammonium ion is titrated directly as an acid with methanolic potassium hydroxide in the presence of acetone employing potentiometric detection of the end-point,3 or with aqueous sodium hydroxide in the presence of formaldehyde using phenolphthalein as visual indicator.4 In other methods, the perchlorate ion suffers a reduction process by means of the Parr bomb technique and the resultant chloride is determined by the Volhard method.5 A modification of this method has been developed6 in which mercuric nitrate is used as titrant and a chloride selective electrode as end-point indicator.The use of other instrumental techniques such as ion-exchange chromatography and capillary electrophoresis has been proposed.The matrix of these types of samples contains many organic and inorganic compounds, so that the recommended analytical methods are tedious, suffer from interferences and present lack of selectivity if other ammonium salts or chlorates and chlorides are present. For that reason, some authors have proposed the potentiometric titration of perchlorate using organic reagents as titrants (hexadecyltrimethylammonium and hexadecylpyridinium chloride,7 tetraphenylarsonium chloride8 and zephiramine9) and different types of perchlorate selective electrodes as end-point detectors.One of the motivations of this work was to develop a simple, quick, reliable and economic potentiometric method that uses a perchlorate selective electrode and direct potentiometry as analytical technique of measurement, allowing the determination of this anion without taking into account the type of pyrotechnic mixture analyzed (propellants, explosives, fireworks, etc.) or the cation contained in the salt (ammonium, potassium, sodium, etc.).This new analytical procedure could be an advantageous alternative to all the different reference methods. Since the introduction of the first perchlorate selective electrode,10 ion-association complexes of a metal chelate, longchain quaternary ammonium salts, organic dyes or neutral carriers as sensors have been developed.7,9,11–17 The majority of these electrodes were constructed by the conventional procedure of Moody and Thomas,18 with internal reference solution, or as solid state electrodes by the techniques of Freiser19 and Lima and Machado.20 The elimination of the inner reference solution by direct application of the selective membrane on a conductive surface,20 associated with the use of long-chain quaternary ammonium compounds and suitable solvents plasticizers, results in perchlorate selective electrodes with good reproducibility and stability.21 The construction of a perchlorate selective chemical field-effect transistor (CHEMFET) has also been reported.22 The fact that ISE present a constant sensitivity over a wide range of concentrations, albeit a great advantage in many circumstances, also has a disadvantage. For example, an error of ± 0.5 mV in measuring a potential corresponds to a relative error Analyst, 1999, 124, 97–100 97in concentration of approximately 2% in the case of electrodes sensitive to monovalent species, and of approximately 4% for those sensitive to divalent ones.23–25 Several solutions to this kind of problem have been suggested by some authors on the basis of cells connected in series,26–28 or alternatively, by totalling the potentials of two conventional electrodes immersed in the same vessel.29–31 This work refers to the construction of ISE, sensitive to perchlorate, which comprise four sensor membranes placed in the same electrode body with summing operational amplifiers (ESOA), which must be more sensitive (quadruple slope) than conventional electrodes.The construction process used in this work was the same as that previously reported,32 and for comparative purposes, conventional ISE were also constructed. The evaluation of the behavior of all the units constructed and their applicability to the determination of perchlorate, by direct potentiometry, in different types of pyrotechnic explosives such as propellants, fulminating powders and fireworks is reported.The samples were also analyzed by titrimetric procedures according to some military standard methods adopted as reference techniques.4,5 Experimental Apparatus The potentials were measured with a Crison, 2002 (Barcelona, Spain) digital potentiometer ( ±0.1 mV sensitivity). Electrode switchers of the same brand were coupled to the potentiometer. The Ag/AgCl double-junction reference electrode was Orion, 90-02-00 (Cambridge, MA, USA).A 0.033 mol L21 ammonium sulfate solution was placed in the outer compartment and the solution used in the inner compartment was the same as provided with the reference electrode. For pH determinations, namely those used for tracing the Reilley diagrams, an Orion, 91-02-00 combined glass electrode was used. Measurements were carried out in double-walled cells at 25.0 ± 0.2 °C by means of a thermostatized water bath Selecta, Tectron 3473100 (Barcelona, Spain). Reagents and solutions Analytical grade reagents were used without further purification throughout.The solutions were prepared with distilled and deionized water, with a conductivity of less than 0.1 mS cm21. A 0.33 mol L21 ammonium sulfate solution as ionic strength adjuster solution (ISA) was used. The 0.1 mol L21 perchlorate stock solution was prepared by dissolving sodium perchlorate which had been previously ovendried at 110 °C for 24 h and standardized with a standard solution of tetraphenylarsonium chloride.33 When necessary, the working solutions were prepared by rigorous dilution of the corresponding stock solution.Construction of the ISE and ESOA The construction process was the same as that previously reported,32 in which the conductor was a mixture (1 : 1.2 m/m) of graphite powder (Merck, Darmstadt, Germany) and a nonconductive epoxy (1 g of Araldite M with 0.4 g of hardener HR, both of them supplied by Ciba-Geigy, Barcelona, Spain).At the end of the electrode body, consisting of a perspex tube (od 12 mm, id 10 mm, length 15 cm), a septum of the same material that divided it into four equal parts was applied. The conductor support, consisting of a mixture of epoxy resin and graphite, and four shielded cables were placed in these compartments. The support was left to harden overnight in an oven at 50–60 °C. Then, a 2 mm-deep cavity was cut out in which the membrane was placed in a similar manner, as that previously described for conventional electrodes.20 The membrane used in the construction of the ISE and the ESOA was prepared from 0.36 g of tetraoctylammonium chloride, Fluka (Buchs, Switzerland), a mixture of 3.75 g of dibutylphthalate (Fluka) and 4.75 g of o-nitrophenyloctyl ether (Fluka) and 0.18 g of PVC (Fluka) dissolved in 6 mL of tetrahydrofuran (Merck).After successive applications of the membranes onto the solid conductive support, the different units were left to dry for 2 d at room temperature, to guarantee a complete tetrahydrofuran evaporation.After preparation, they were conditioned for 3 d in a 0.1 mol L21 solution of the primary ion. The summing device was composed of four voltage following stages implemented and calibrated so that at the exit the potential would be equal to the sum of potentials supplied by each of the voltage followers. High input impedance ( > 1014 ½) operational amplifiers were used in the voltage follower circuits. The whole setup was installed in a metallic box together with a symmetric power supply source (±12 V).Proposed procedure An accurately weighed sample (about 5 g) of previously homogenized and dried pyrotechnic mixture was extracted in 200 mL of boiling deionized water for 3 h. After filtering, washing with hot water and cooling, the extract was made up to 250 mL with deionized water in a calibrated flask. An aliquot of 5.00 mL of the extracted sample was pipetted into a 100 mL calibrated flask and made up to volume with deionized water. After that, 50.0 mL of this solution was transferred to a plastic beaker, 5 mL of a 0.33 mol L21 ammonium sulfate solution was added, as ISA solution, and the perchlorate selective and reference electrodes were immersed.The perchlorate concentration in the sample solution was determined by direct potentiometry. Results and discussion The response characteristics were evaluated by repeatedly preparing calibration curves for solutions between 1026 and 1021 mol L21, covering the linear and non-linear response zones.The ionic strength was adjusted to 0.1 mol L21 with ammonium sulfate, and the lower limits of linear response, the practical limits of detection, the slopes and the reproducibility of the potential values were established.34 The response time was determined by spiking a dilute solution (1025, 1024 and 1023 mol L21) with a more concentrated one so as to obtain a 10 times more concentrated solution, and recording the time required for a stable potential (±0.1 mV).The extent of the proton and hydroxide interferences was evaluated by measuring, at different pH values, the potentials of the electrodes in solutions of constant concentration of the primary ion (1024, 1023, 1022 or 1021 mol L21) and tracing the Reilley diagrams. The potentiometric selectivity coefficients were determined for three different concentration values of the primary ion and various anion interferences (1024, 1023 and 1022 mol L21), by using the separated solutions method.The evaluation of the working characteristics of the ESOA for perchlorate was made simultaneously with ISE constructed, with conventional shape, using the same type of membranes and some of the results obtained are shown in Table 1. This parallel study proved that the behavior of the ESOA was, as a whole, 98 Analyst, 1999, 124, 97–100similar to the corresponding ISE with the same type of membrane, except for the slope which was quadruple that of the conventional ISE. The electrodes showed an initial drift of potential, and stability only occurred if the perchlorate ISE and ESOA had been conditioned for three days in a 1023 mol L21 solution of the primary ion.Theoretically, in the all-solid-state PVC membrane electrodes without an internal reference solution there is no well-defined internal reference potential.However, in practice, these electrodes provide reproducible potentials. The establishment of a constant electrode potential requires a period to stabilize the internal reference potential in the graphite conductive epoxy–PVC boundary, by means of the O2–H2O couple as has been previously suggested.35,36 All the units had a long lifetime (generally greater than 10 months) which is quite good for electrodes whose membranes are based on mobile carrier sensors. This is probably due to the absence of an internal reference solution and because the membrane is directly applied on the conductive support.20 From the results obtained during the evaluation of the units constructed it is possible to affirm that the perchlorate ESOA presents good working characteristics, similar to those of conventionally shaped electrodes. Its increased sensitivity is due to the sum of potentials supplied by the four membranes they incorporate, without however a decrease in the potential stability.It was decided to start the analytical applications of the electrodes constructed by developing a simple and quick procedure for the determination of perchlorate in different types of pyrotechnic mixtures such as propellants, fulminating powders and fireworks. Then eight samples, whose perchlorate concentration had been previously determined by the reference methods,4,5 were analyzed by the proposed potentiometric procedure. In each case, the precision was evaluated by its application to eleven samples of the same product and was expressed in terms of its standard deviation.Standard perchlorate additions were used to evaluate the accuracy of the method and the average percentage of spike recovery, for three different additions, was calculated. The data obtained (Table 2) confirm that the potentiometric methods had good precision and accuracy. The ESOA had better precision and recoveries compared to conventional ISE and the reference techniques, when simultaneously applied to the determination of the perchlorate contained in the samples.To test whether the potentiometric and the reference methods differ in their precision, a significance F test (two-tailed test) was carried out. The calculated F-values for all the samples were less than the critical F-value, with the exception of the Table 1 General working characteristics of the perchlorate ISE and ESOA Characteristics ISE ESOA Lower limit of linear response/mol L21 8.2 3 1026 5.1 3 1026 Practical limit of detection/mol L21 1.3 3 1026 1.2 3 1026 Slope/mV log21C) 57.3 ± 1.0 255.8 ± 3.8 Response stability/mV d21 ±0.5 ±0.7 Response time/s 1023 mol L21 13 13 1024 mol L21 15 15 Working pH range 1023 mol L21 3.5–12 3.5–12 1024 mol L21 3.5–12 3.5–12 Lifetime (months) > 10 > 10 Potentiometric selectivity coefficients (Kpot X.I) X 1024 mol L21 1023 mol L21 1022 mol L21 1024 mol L21 1023 mol L21 1022 mol L21 SO4 22 2.0 3 1024 7.9 3 1025 4.0 3 1025 2.0 3 1024 1.0 3 1024 5.0 3 1025 I2 1.0 3 1021 6.3 3 1022 4.0 3 1022 1.3 3 1021 7.9 3 1022 5.0 3 1022 Br2 2.5 3 1021 4.0 3 1022 6.3 3 1023 2.5 3 1021 5.0 3 1022 7.9 3 1023 Cl2 7.9 3 1022 1.3 3 1022 3.2 3 1023 1.0 3 1021 1.6 3 1022 4.0 3 1023 F2 2.5 3 1022 3.2 3 1023 6.3 3 1024 2.0 3 1022 2.5 3 1023 5.0 3 1024 NO32 7.9 3 1022 2.5 3 1022 7.9 3 1023 1.3 3 1021 3.2 3 1022 1.3 3 1022 NO22 5.0 3 1022 7.9 3 1023 2.5 3 1023 5.0 3 1022 1.3 3 1022 4.0 3 1023 ClO32 6.3 3 1022 3.2 3 1022 1.3 3 1022 7.9 3 1022 3.2 3 1022 1.3 3 1022 IO42 6.3 3 1021 8.3 3 1021 1.1 6.3 3 1021 8.1 3 1021 1.1 Table 2 Comparative study of the precisions and accuracies attained in the determination of perchlorate in pyrotechnic mixtures by simultaneous application of the potentiometric method (ISE and ESOA) and the reference methods Sample Reference methoda ISE ESOA Xb Rc Xb Rc REd F Xb Rc REd F Propellants 1P 620.3 ± 4.7 100.9 623.6 ± 3.7 100.5 +0.5 1.61 622.6 ± 2.5 100.1 +0.4 3.53 2P 782.3 ± 4.5 101.7 783.6 ± 3.9 99.4 +0.2 1.33 780.7 ± 2.5 100.7 20.2 3.24 3P 777.4 ± 3.7 101.3 776.9 ± 3.2 100.8 20.1 1.34 775.8 ± 2.1 100.2 20.2 3.10 Fulminating powders 1FP 621.2 ± 6.0 101.2 624.0 ± 3.6 101.1 +0.5 2.78 622.1 ± 3.3 101.0 +0.2 3.31 2FP 768.6 ± 4.3 101.5 770.3 ± 2.1 100.7 +0.2 4.19 770.9 ± 2.0 100.5 +0.2 4.84 Fireworks 1F (red) 584.3 ± 3.0 101.9 584.0 ± 4.2 102.3 20.1 1.96 583.4 ± 1.6 101.8 20.2 3.52 2F (green) 432.0 ± 2.6 101.3 430.5 ± 2.6 99.1 20.3 1.00 432.8 ± 1.9 99.2 +0.2 1.87 3F (white) 519.5 ± 3.3 100.9 517.2 ± 2.7 100.8 20.4 1.49 520.0 ± 1.8 100.3 +0.1 3.36 a MIL-A-23442(OS)4 for propellants and MIL-A-23946(AS)5 for fulminating powders and fireworks. b Mean perchlorate concentration and standard deviation for eleven determinations (g kg21).c Mean spike recovery. d Relative error of the proposed method versus the reference methods. The critical Fvalue, considering a 95% confidence level and 10 degrees of freedom, for a two-tailed test, is 3.72.Analyst, 1999, 124, 97–100 99sample 2FP which shows a slight discrepancy, so there is no significant difference between the two standard deviations at the 95% confidence level. The limit of detection of the proposed method was obtained as recommended by the Analytical Methods Committee,37 and was found to be 2.3 g kg21 and 0.1 g kg21 for the ISE and ESOA, respectively. After that, 18 samples of propellants and 16 samples of fireworks were simultaneously analyzed by the proposed potentiometric method and the reference methods.The results obtained and the data calculated by regression analysis are shown in Table 3. The confidence limits of the slope, intercepts of the regression lines and the t-values for the correlation coefficients at a 95% confidence level and n 2 2 degrees of freedom were also determined. From these data, it is possible to state that percentages of difference obtained between the proposed method and reference methods, when applied to fireworks, ranged from 20.06 to +1.01, with an average value of +0.46, when ISE is used, and from –0.16 to +0.56, with an average value of +0.26, when ESOA is applied.When the proposed and the reference methods were applied on propellants the results obtained ranged from –1.45 to +1.70, with an average value of +0.80, for ISE determinations, and from –0.36 to +0.66 with an average value of +0.29, for ESOA determinations.On the other hand, the calculated slopes and intercepts do not significantly differ from the ideal values of 1 and 0, respectively. Thus there is no evidence of systematic difference between the proposed method and the reference methods. From the correlation coefficients, the t-values calculated were always greater than the tabulated tvalues, so significant correlations exist between the proposed method and the reference methods.38 The authors gratefully acknowledge the financial support of: Departamento de Educación, Universidades e Investigación del Gobierno Vasco, Spain (Project: PI 96/11).We also thank J. Martínez and M. A. Madariaga from Unión Española de Explosivos S.A. and J. L. Valdecantos from Valecea Pirotecnia for the samples provided and to Prof. Dr. F. M. Goñi from the Department of Biochemistry of the University of the Basque Country for the literature supplied and his helpful discussions. References 1 D. Price, A. R.Clairmont and I. Jaffee, Combust. Flame, 1967, 11, 415. 2 J. A. Conkling, Chemistry of Pyrotechnics, Marcel Dekker, New York, 1985, p. 60. 3 US Department of Defense, Method 216.1, Military Standards 286 B, US Department of Defense, Washington DC, 1969. 4 US Department of the Navy, MIL-A-23442(OS), US Government Printing Office, Washington DC, 1966. 5 US Department of the Navy, MIL-A-23946(AS), US Government Printing Office, Washington DC, 1966. 6 W. Selig and G. L. Crossman, Informal Report UCID-15623, Lawrence Radiation Laboratory, Livermore, CA, 1970. 7 J. Gu, L. Luo and J. Yang, Fenxi Huaxue, 1989, 17, 1159. 8 R. J. Baczuk and R. J. Dubois, Anal. Chem., 1968, 40, 685. 9 T. Tamura and M. Kataoka, Bunseki Kagaku, 1984, 33, 591. 10 J. W. Ross, in Ion Selective Electrodes, ed. R. A. Durst, US Government Printing Office, Washington DC, 1969, p. 57. 11 S. Alegret, A. Florido, J. L. F. C. Lima and A. A. S. C. Machado, Quim. Anal., 1986, 5, 36, and references cited therein. 12 J. Liu, Y. Masuda and E. Sekido, Analyst, 1990, 115, and references cited therein. 13 J. R. Fernandes and L. T. Kubota, Anal. Lett., 1993, 26, 2555. 14 Z. S. Alagova, N. V. Rozhdestvenskaya, G. A. Khripun and O. K. Estefanova, Zh. Prikl. Khim. (St. Petersburg), 1990, 63, 2643. 15 A. K. Jain, M. Jahan and V. Tyagi, Analyst, 1987, 112, 1355. 16 J. Pan and Y. Liu, Fenxi Huaxue, 1981, 9, 593. 17 A. K. Jain and V. Tyagi, Anal. Chim. Acta, 1990, 231, 69. 18 G. J. Moody and J.D. R. Thomas, in Ion Selective Electrodes in Analytical Chemistry, ed. H. Freiser, Plenum, New York, 1978, vol. 1, p. 287. 19 H. Freiser, in Ion Selective Electrodes in Analytical Chemistry, ed. H. Freiser, Plenum, New York, 1980, vol. 2, p. 85. 20 J. L. F. C. Lima and A. A. S. C. Machado, Analyst, 1986, 111, 799. 21 S. Alegret and A. Florido, Analyst, 1991, 116, 473. 22 J. Casabó, L. Escriche, C. Pérez-Jiménez, J. A. Muñoz, F. Teixidor, J. Bausells and A. Errachid, Anal.Chim. Acta, 1996, 320, 63. 23 P. L. Bailey, Analysis with Ion-Selective Electrodes, Heyden, London, 2nd edn., 1980, p. 3. 24 V. V. Cosofret, Membrane Electrodes in Drug-Substances Analysis, Pergamon Press, Oxford, 1982, p. 7. 25 R. A. Durst, in Ion Selective Electrodes in Analytical Chemistry, ed. H. Freiser, Plenum, New York, 1978, vol. 1, p. 311. 26 R. Stepak, Fresenius’ J. Anal. Chem., 1983, 315, 269. 27 A. Parczewski and R. Stepak, Fresenius’ J. Anal. Chem., 1983, 316, 29. 28 K.Suzuki, K. Tohda and T. Shikai, Anal. Lett., 1987, 20, 1773. 29 A. Parczewski, Talanta, 1987, 34, 586. 30 A. Parczewski, Talanta, 1988, 35, 473. 31 A. Karocki, K. Madej and A. Parczewski, Chem. Anal. (Warsaw), 1989, 34, 383. 32 R. Pérez-Olmos, A. Rios, R. A. S. Lapa and J. L. F. C. Lima, Fresenius’ J. Anal. Chem., 1998, 360, 659. 33 D. Midgley and K. Torrance, Potentiometric Water Analysis, Wiley, Chichester, 1978, p. 391. 34 IUPAC, Pure Appl. Chem., 1981, 53, 1913. 35 A. Hulanicki and M. Trojanowicz, Anal. Chim. Acta, 1976, 87, 411. 36 R. W. Cattral, D. M. Drew and I. C. Hamilton, Anal. Chim. Acta, 1975, 70, 269. 37 Analytical Methods Committee, Analyst, 1987, 112, 199. 38 J. C. Miller and J. N. Miller, Statistics for Analytical Chemistry, Wiley, Chichester, 1984, p. 108. Paper 8/06919K Table 3 Results obtained (g kg21) for the determination of perchlorate in pyrotechnic mixtures by simultaneous application of the references and proposed potentiometric methods including their regression equations Propellants Fireworks and fulminating powders Cref CISE CESOA Cref CISE CESOA 933.6 941.2 933.5 614.3 612.1 613.9 905.2 912.1 908.1 703.5 706.8 707.3 782.9 788.7 785.4 758.6 760.6 757.4 872.1 878.1 876.1 440.1 440.3 440.6 845.2 851.0 846.8 424.3 427.9 425.9 793.1 798.9 796.4 337.4 337.8 338.0 763.5 768.9 766.3 503.9 503.6 505.9 739.7 745.9 742.7 372.3 374.6 374.1 707.3 714.4 709.9 561.4 565.4 563.9 649.7 654.2 649.5 591.6 591.0 592.2 612.2 619.6 616.5 390.0 392.9 392.2 594.3 604.4 597.5 633.3 639.0 635.6 531.7 539.1 535.2 472.5 474.3 473.6 474.8 480.7 475.7 351.7 354.9 351.3 579.3 570.9 577.2 493.5 498.0 495.0 685.8 697.4 687.9 536.2 541.6 538.0 560.3 565.4 562.3 627.5 633.0 629.6 CISE = (1.005 ± 0.007) Cref + (2.21 ± 5.19) CISE = (1.001 ± 0.005) Cref + (1.94 ± 2.52) r = 0.9996 t = 141.38 r = 0.9998 t = 187.05 CESOA = (1.002 ± 0.003) Cref + (1.0 1± 2.14) CESOA = (1.000 ± 0.003) Cref + (1.12 ± 1.42) r = 0.99993 t = 388.04 r = 0.99994 t = 374.15 The tabulated t-values, considering a 95% confidence level and 16 or 14 degrees of freedom, are 2.12 and 2.14, respectively. 100 Analyst, 1999, 124, 97–100

ISSN:0003-2654    

DOI:10.1039/a806919k

出版商:RSC    

年代:1999

数据来源: RSC