摘要:

Analyst, January 1995, Vol. 120 47 Determination of Creatinine in Human Serum. Statistical lntercalibration of Methods M. C. Gennaro, C. Abrigo and E. Marengo Dipartimento di Chimica Analitica dell 'Universita di Torino, Via P. Giuria 5, 10125 Turin, Italy C. Baldin and M. T. Martelletti Laboratorio Analisi USSL 76, Casale Monferrato, Alessandria, Ituly An intercalibration of methods was performed in the determination of creatinine in human plasma. Jafflt, enzymatic creatinine-deaminase (CRDI) and ion-interaction reagent high-performance liquid chromatography (HPLC) methods were employed and the results were compared using a parametric t-test for multiple samples and non-parametric Mann-Whitney, Wilcoxon and Friedman tests. The comparison did not show a statistically significant difference in the use of the three methods.The ion-interaction HPLC procedure is proposed as a reference method for creatinine assays. Keywords: Creatinine determination; plasma; ion-interaction high-performance liquid chromatography; intercalibration Introduction The concentration level of creatinine in human serum is a marker for the diagnosis of renal, muscle and thyroid disease. The physiological level of creatinine ranges between 40 and 110 pmol 1-1 and increases substantially in the presence of renal malfunctions. 1 In hospital laboratories the determina- tion is generally performed by the automated Jaffk spectro- photometric procedure,2 based on the reaction that takes place in alkaline medium between creatinine and picric acid to form an orange-red compound. A major disadvantage of the method, apart from the several pre-treatment steps required, is the low specificity, due to the side-reactions that take place between picric acid and other species generally present in serum [such as glucose, acetoacetic acid, acetone, 3,4-dihy- droxyphenylalanine (DOPA) and ascorbic, hippuric and pyruvic acid],3,4 and which can lead to an overestimation in the determination of creatinine.On the other hand, high bilirubin concentrations can mask the reaction between creatinine and picric acid and lead to negative errors.4 In order to improve the selectivity, modifications to the original Jaffk procedure have been proposeds--7 and a stopped- flow method that does not require the prior precipitation of serum proteins was developed.% More specific enzymatic methods have been applied, which make use, for example, of creatinine iminohydrolaseg or phenol-4-aminophenazone (PAP).10.11 These methods are not interference free, especially from bilirubin.4-12 The importance of creatinine determination for diagnostic purposes has recently led to the development of enzymic potentiometricl3 and FET (field-effect transistor) 14 sensors. Reversed-phase HPLC methods have also been proposed for the determination of creatinine in biological fluids,'.11-15-17 and also ion-exchangel%?lY and ion-pair?" chromatographic and high-performance capillary electrophoresis2 1 methods. In this paper, the results obtained in the determination of creatinine by a reversed-phase ion-interaction reagent HPLC method developed in this laboratory" are compared through statistical parametric and non-parametric tests, with the results obtained by two methods generally employed in hospital laboratories, namely the Jaffk procedure and the creatinine-deaminase CRDI-UV enzymic method.Experimental Apparatus Analyses were carried out using a Merck-Hi tachi Lichrograph L-6200 chromatograph, equipped with a two-channel Merck- Hitachi D-2500 chromato-integrator, interfaced with , an L-4200 UV/VIS detector and an L-3720 conductivity detector with temperature control. The precolumn and column tem- peratures were held at 25.0 & 0.2"C. A Metrohom 654 pH meter equipped with a combined glass-calomel electrode was employed for pH measurements and a Hitachi Model 150-20 spectrophotometer for absor- bance measurements. An automatic Hitachi Model 717 analyser maintained at 37 "C was used in hospital laboratories for creatinine determination.Chemicals and Reagents Ultrapure water from a Millipore Milli-Q system was used for the preparation of all the solutions. Creatinine was of analytical-reagent grade from Merck. Octylamine and picric, orthophosphoric and uric acid were of analytical-reagent grade from Fluka. Standard solutions of creatinine and uric acid (500.0 mg 1-1) were prepared in dark flasks and stored at 4 "C. The normal and pathological certified control sera were Precinorm-U and Precipath-U, respectively, from Boehr- inger-Mannheim . Sample Preparation All the samples were prepared in hospital laboratories for centrifugation at 3000s for 10 min.No deproteinization process was carried out. The serum samples were stored at 4°C in small plastic containers, filled as near to capacity as possible. No other treatment is required, apart from a 0.20 pm filtration for the HPLC analysis. All the analyses were performed within 2 d, in order to avoid any possible degradation or bacterial interference." Modified Jaff2 Method This method is based on the reaction that takes place between creatinine and picric acid in alkaline medium and the48 Analyst, January 1995, Vol. 120 spectrophotometric determination ( h = 500 nm) of the red picrate formed. As a modification to the original Jaffe method, n o protein precipitation was performed after the centrifugation step and the spectrophotometric determination was performed at a fixed time.CRDI-UV method The method was developed in the laboratories of Pasteur Diagnostici, Milan, Italy and is based o n the specific action of creatinine-deaminase (EC 3.5.4.21) enzyme (CRDI), which leads to the formation of 1-methylhydantoin and ammonia, and the spectrophotometric determination at a fixed time and at 340 nm of glutamate-NADP+ formed by the reaction of ammonia with a-ketoglutarate and NADPH+. Ion-interaction HPLC Method The ion-interaction HPLC method makes use of a C18 reversed-phase column and 5.00 mmol 1-1 octylammonium orthophosphate (pH 6.4 k 0.4) as the ion-interaction reagent. The interaction reagent represents the only component of the mobile phase which, flowing under isocratic conditions, determines the dynamic modification of the stationary phase and the formation on it of an electrical double layer.The retention process is due to an adsorption reaction on the surface of the stationary phase of ion pairs formed between the analyte and the ion-interaction reagent. The method permits the simultaneous determination of amines and anions and was employed here in the determination of creatinine and uric acid in human serum. A 5 pm Spherisorb ODs-2 (Phase Separation) column (250.0 x 4.6 mm i.d.) fully end-capped and with a carbon load of 12% (0.5 mmol g-1) was used, together with a LiChrospher RP-18 (5 ytn) guard column (150.0 x 4.0 mm i.d.). The solutions to be used in the mobile phase were prepared by adding an appropriate amount of orthophosphonic acid to the 5.00 mmol 1-1 octylamine solution to give a pH of 6.4 k 0.4.Even if the concentrations are not exactly stoichiometric, for simplicity the ion-interaction reagent is reported as octylammonium orthophosphate. No other reagent or organic modifier was added to the mobile phase. The chromatographic system was conditioned by passing the mobile phase through the column until a stable baseline signal was obtained; a minimum of 1 h was necessary. The repeatibility of measurements, with regard to both retention time and sensitivity (evaluated as peak area), was within 3% for sequential measurements (under the same conditions of eluent preparation) and the reproducibility (for different eluent preparations) was always within 5%. The mobile phase was freshly prepared every third day and after use the column was washed by passing 1 + 1 v/v water-methanol at 0.5 ml min-* for 1 h.Before injection, serum samples were diluted with ultrapure water (1 + 9 v/v) and filtered through a 0.20 ym Anotop disposable syringe filter. Results and Discussion Ion-interaction reagent chromatography, when used for com- plex matrices, offers some advantage of selectivity over reversed-phase HPLC, because only species able to form adsorbable ion pairs can be retained. The method is free from interferences from metal ions and also from neutral, colloidal or high molecular mass species (e.g., bilirubin). With regard to the other potential interfer- ents present in serum, acetoacetic acid does not absorb at 254 nm and DOPA and hippuric, uric and ascorbic acid show very different retention times to creatinine.Calibration Fig. l ( a ) shows the calibration graph of peak area (as given by the integrator) as a function of standard concentrations ranging between 0.25 and 20.00 mg 1 - 1 . The graph was fitted by the regression equation y = 53142.28~ f 460.10 (at the 95% confidence level) with a correlation coefficient r2 = 0.9988. Analysis of Certified Control Serum The ion-interaction HPLC method was then applied to the analysis of certified control sera from Boehringer-Mannheim, which contain creatinine and uric acid at normal and patholog- ical levels. The results obtained by the ion-interaction HPLC method for creatinine and uric acid are reported in Table 1 in comparison with the certified concentration range; the agree- ment is very close.Recovery Yield Study in Spiked dialysed Serum In order to check for possible matrix effects on the quantita- tive results for creatinine, a recovery yield study was perfor- med on samples of dialysed serum that do not contain creatinine. The samples were spiked with creatinine standard additions ranging between 1.0 and 12.0 mg 1-1 [Fig. 1(6)] and analysed. Comparable results were obtained by using the calibration graph [Fig. l(a)] and the standard addition method. The average results in Table 2 show an average recovery always greater than 90%. 0 10 20 0 6 12 0 4 8 Creatinine @pm) Fig. 1 (a). Calibration graph of peak areas (as given by the integrator) versus creatinine concentration. (b), Dialysed serum (standard additions method): peak areas (as given by the integrator) versus creatinine (mg I-* added).(c), Sample serum (standard additions method): peak areas (as given by the integrator) versus creatinine (mg 1-1 added). Table 1 Determination of creatinine and uric acid in certified control sera by ion-interaction reagent HPLC Creatinidmg 1-1 Uric acidlmg 1 - Samples Certified value HPLC method Certified value HPLC method Pathological control 28.9-4 1.7 32.0 k 1.6 73- I05 84.4 k 4.8 Normal control serum: 16.9-24.3 21.7 k 0.9 38.7-55.7 43.4 _+ 2.1 serum: Precipath U Precinorm-UAnalyst, January 1995, Vol. 120 49 8 r- Determination of Creatinine in Samples of Human Serum The method was then applied to the analysis of 30 samples of human serum, randomly chosen in hospital laboratories. These samples were analysed in the hospital laboratories by the modified Jaffk and the CRDI enzymic methods and by HPLC methods in the laboratory of the University of Turin.Fig. 2 shows a typical chromatogram. The reproducibilities obtained with the three methods are comparable and always within 5%. Table 3 reports the results as the average values of three experiments. Fig. 3 reports the data obtained by the HPLC method respectively versus the results obtained by (a) the CRDI enzymatic method and ( b ) the Jaffk method. The correlation coefficients (r2) are 0.9963 with the CRDI method and 0.9980 with the Jaffk method. For a statistical method intercalibration, the results obtained using the three methods were treated by parametric and non-parametric statistical tests. Creatinine I t-Test for Multiple Samples The test is particularly suitable for the statistical treatment of samples of slightly varying composition, such as those generally of interest in the analysis in biological fluids, where the analyte of interest is often present in narrow physiological I v- cq Uric acid Q) Table 2 Recovery of creatinine from spiked dialysed serum Average Addedmg 1-1 Found/mg I-' recovery (%)* 12.0 9.0 6.0 4.0 2.0 1.0 * n = 3 .11.80 10.72 6.54 3.62 2.09 0.93 98.3 119.0 109.0 90.5 104.5 93.0 t z iij 0 10 20 30 Tirnelrnin Fig. 2 Chromatogram recorded for a typical serum sample (diluted 1 : 10 v h ) , containing creatinine and uric acid: column, Spherisorb ODS-2 (250 x 4.6 mm i.d.). end-capped, 5 pm; ion-interaction reagent, 5.00 mmol 1-1 octylamine orthophosphate; flow rate, 1.0 ml min-1.Spectrophotometric detection at 254 nm. concentration range~.2~.~5 The t value is evaluated through the parameter Di calculated as the difference between the results obtained with the two methods for the same sample with Table 3 Comparison of results (average data, n = Jaffe, enzymic CRDI and HPLC methods Sample No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Jaffkl mg 1-1 16.1 14.4 10.6 8.1 14.3 13.8 9.6 24.8 87.3 100.4 139.1 47.3 67.5 47.9 12.2 9.7 8.5 5.9 97.7 12.5 124.4 83.6 18.5 14.9 84.5 39.7 10.3 8.9 84.7 18.9 CRDII mg 1-1 13.3 12.3 8.7 5.7 13.6 10.9 8.8 22.8 89.4 107.2 155.4 45.5 70.8 45.9 10.2 9.9 9.8 5.9 99.8 10.1 125.6 83.7 15.9 12.7 87.9 40.5 9.6 10.0 94.1 16.0 3) obtained by the HPLCI mg 1-1 14.7 13.5 8.9 6.2 12.4 10.5 8.5 21.9 90.8 98.5 137.7 45.9 68.8 44.4 11.1 10.6 9.2 6.1 98.6 10.9 125.2 83.6 16.5 13.2 86.5 41.2 9.8 9.2 92.8 17.4 150 I 0 25 50 75 100 125 HPLC rnethodrng t' Fig.3 (a). Comparison between CRDI-UV and ion-interaction HPLC method (r2 = 0.9963) ( h ) , Comparison between JaffC and ion-interaction HPLC method (r' = 0.9980).50 Analyst, January 199.5, Vol. I20 regard to the sign, and the mean 6 of all the individual 0, differences: D - t = - 6 S where The method was applied to the treatment of the data in Table 3. A t value of 1.1335 was calculated in the comparison between the Jaffk and the HPLC methods and t = 1.2200 in the comparison between the CRDI and the HPLC methods. When comparing the Jaffit and the HPLC methods the calculated t value was 0.9400. The tabulated value of the t-distribution at the 95% confidence level and 29 degrees of freedom is t = 2.045.I t can be concluded that at this confidence level there is no significant difference between the three methods. Mann- Whitney U- Test The Mann-Whitney U-test is a non-parametric test for the comparison of methods 1 and 2, through nl and n2 measure- ments performed with the two methods.25 All the data are ranked by assigning rank 1 to the lowest, rank 2 to the second and so on. For the two series of data the sums of the ranks R1 and R2 and the parameters U1 and U2 are calculated: The smaller of the two calculated U values is compared with the value tabulated for the U distribution for 29 degrees of freedom. The calculated values of U are are follows: Ulafti. = 468 and Ut-lPL~ = 432 in the comparison between the Jaffk and HPLC methods; UCRDI = 472 and UJaffi.= 428 in the comparison between the CRDI and Jaffe methods; and UcRDl = 453 and UHPLC = 447 in the comparison between the CRDI and HPLC methods. As the tabulated value for n l = n2 = 30 at the 95% confidence level, U = 322, is lower than all the calculated U values, it is concluded that also according to this test there is no statistically significant difference between the results of the three methods. Wilcoxon Matched-pair Signed-rank Test All the Di values calculated (with regard to the sign) as the difference between the data obtained with the two methods for the same sample are first ranked without regard to the sign, starting with the smallest value.Then the sign of Dj is considered. The null hypothesis of equivalence of the methods is taken according to which the sum (T+) of all the ranks for positive Di is close to the sum for the negative Di ( T - ) . The smaller the value of T = min (T+, T-) the larger the significance of the difference. The values of T calculated from the data in Table 3 are as follows: in the comparison (a) between the Jaffi: and HPLC methods T- = -320.5 and T+ = 144.5, (b) between the CRDI and HPLC methods T- = -264.0 and T+ = 200.0 and ( c ) between the Jaffe and CRDI-UV methods T- = -251.5 and T+ = 232.5. The tabulated value for T for n l = n? = 30 at the 95% confidence level is 137. This value is lower than all the T,,i,, values calculated in the comparisons of the three methods and it can be concludcd that also according to this test, thcre is no significant difference between the three methods.Friedman Test The non-parametric Friedman test'h-27 offers the advantagc of being able to treat and compare k blocks of data simul- taneously and is therefore very suitable in the situation considered here, in which the comparison of three methods is being performed. The data are ordered in a matrix with N rows and k columns, which contains the objects in the rows and the different treatments in the columns. In our instance the objects are the samples investigated ( n = 30) and the treatments are the data obtained with the three different methods ( k = 3). This matrix corresponds to Table 3. A rank is then assigned in every row, giving the lowest rank to the lowest value, then progressively increasing the rank as the value increases.In the present instance (k = 3), the ranks 1, 2 and 3 are adsigned to the concentration values obtained with the three methods. Then a xr? value is calculated as 1 - X 1L Xr2 = 2 (R,)' - 3n(k + 1) nk(k+ 1) ,=, where Ri is the sum of the ranks of the jth column. The calculated value of x r 2 (1.31) is compared with the values of the x? distribution for 29 degrees of freedom. As xr2 (calculated) < x~o,990 in the x' distribution, it can be concluded that the results obtained with the three methods do not differ from each other at a significance level a = 0.01. This result agrees with those obtained through the other tests applied. Conclusion The statistical tests applied to the results obtained in the determination of creatinine using the Jaffk, CRDI enzymic and HPLC methods showed no statistically significant differ- ence between the use of the three methods (at a confidence level of 95% or greater).The methods, with regard to accuracy and precision, can be used alternatively and, in particular, the ion-interaction HPLC method can be used as a reference method for the determination of creatinine. The results are accurate and reproducible and free from interfer- ences from glucose, bilirubin, acetic acid, uric acid, hippuric acid, DOPA and acetoacetic acid. No particular pre-treat- ment is required, apart from a centrifugation and 0.20 pm filtration. In addition, the method is particularly inexpensive, because no organic solvent is needed in the mobile phase.The authors thank the Minister0 dell'Universit2 e della Ricerca Scientifica e Technologica (MURST, Rome, Italy) and the Consiglio Nazionale delle Ricerche (CNR, Rome, Italy) for financial support. References Jcppcrscn, M. T.. and Hanscn, E. H . , Anal. Chim. A m , 1988. 214. 147. Rcincs. J . A.. Arin, M. J., and Dicz, M. T.. J . C'Izmnzarogr.. 1992, 607, 199. Dc Lcacy, E. A., Brown. N . N.. and Claguc. E. A., Cliti. Chem. ( Winsrori-SLilem, N . C . ) , 1989, 35. 1772. Guy, J . M., and Ixgg, E. F.. Clitz. C'lzem., 1990. 36. 1851. Kirkpatrick. M., Moody, C.. Albcrgo Batcs, D., and Shaffar. M.. Clin. C'liem. (Winston-Sakm, N . C . ) , 1987. 33. 1466.Analyst, January 1995, Vol. 120 51 6 7 8 9 10 1 1 12 13 14 15 16 Huang.S . M.. and Huang, Y. C., J . C'hromutogr., 19x8, 429. 235. Mackay, G. A., Goodall, G. I . , Woods, J . , and Young, V. H . , Clin. Chem. (Winston-Salem, N. C . ) , 1987, 33, 2124. Guticrrcz, M. C.. Gomcz-Hcns, A., and Pcrcz-Bcndito, D.. Fresenius' Z. A n d . Ciiem.. 1989, 335, 576. Collison, M. E., and Mcycrhoff. M . E., Anal. Clzim. A m , 1987. 200, 61. Magnotti, R. A., Jr., Stephcns, G. W., Rogers, R. K . , and Pescc, A. J . , Clin. Ciiem. ( Winston-Salem, N. C.), 1989, 35, 1371. Schimtt, S . , and Rick. W.. Fresenius' 2. Anal. Cfiem.. 1992, 343, 84. Werner, G., Schneider, V., and Emmcrt, J . , J . CIiromatogr., 1990, 525, 265. Campanclla, L., Sammartino, M. P., and Tomassctti, M., Analyst, 1990, 115, 827. Battilotti, M., Colpicchioni, C., Giannini, I . , Porcclli, F.. Campanella, L., Cordatorc, M., and Tomassctti. M., Anal. Chim. Acta, 1989, 221, 157. Werner, E. R., Fuchs, D., Hauscn, A., Reibncggcr, G . , and Wachter, H., Clin. Chem. (Winston-Salem, N.C.), 1987, 33. 2028. Diez, M. T., Arin, M. J.. and Resines, J . A., J . Liq. Cliromatogr. , 1992, 15, 1337. 17 18 I9 20 21 22 23 24 25 26 27 Yang. S . D.. Xu, J . M., Yang. L . , Ma. Y. N . . and Bai. F.. J . Liq. Cliromutogr.. 1989, 12, 1791. Kagcdal, B.. and Olwm. B.. J . CIzrornntogi., 1990. 527, 21. Palmisano. F . . Rotunno, T.. Gucrricri. A., and Zambonin, P. G . , J. Cliromatogr.. 1989, 493, 35. Tokuda, T., Tokicda, T., Anazawa. A., and Yoshioka, M., J . Cfzromarogr.. 1990. 530. 418. Miyakc, M., Shibukawa. A.. and Nakagawa. I' .. J . High Resolut. Chromatop. , I99 1 , 14, I8 1. Gcnnaro, M. C., and Abrigo, C., Fresenius' Z. Anal. Cizern., 1991, 340, 422. Dilcna, B. A . , Clin. Chem. ( Winston-Salem, N . C.). 1988. 34, 1 007. Christian, G . D., Analytical Cliemistry, Wilcy, Ncw York, 3rd cdn., 1980. Massart. D. L., Vandcginstc, B. G. M., Dcrning, S. N., Michottc, Y .. and Kaufman, L., Ciremornetrics: a Textbook. Elscvicr, Amsterdam, 1988. Friedman, M., J . Am. Statist. Assoc., 1937, 32, 675. Friedman, M., Am. Math. Statist.. 1940, 11, 86. Paper 4104930F Accepted August I I , 1994

ISSN:0003-2654    

DOI:10.1039/AN9952000047

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 53 Micellar Liquid Chromatography of Zwitterions: Retention Mechanism of Cephalosporins Carmelo Garcia Pinto, JosC Luis Pirez Pav6n and Bernard0 Moreno Cordero* Departamento de Quimica Analitica, Nutricidn y Bromatologia, Facultad de Quimica, Universidad de Salamanca, 37008 Salamanca, Spain The retention mechanism of zwitterion solutes in micellar liquid chromatography is discussed. Five cephalosporins, cephazolin, cephalexin, cephalothin, cephaloridine and cephradine, and two precursor acids of these antibiotics, 7-aminocephalosporanic acid and 7-aminodesacetoxycephalosporanic acid, were studied using aqueous sodium dodecyl sulfate as the mobile phase. The influence of variables such as the surfactant concentration, pH and the presence of several organic modifiers was also studied.The partition coefficients among the bulk solution and the micelles and the stationary phase were calculated. Keywords: Micellar liquid chromatography; cephalosporins; sodium dodecyl sulfate; z witterions Introduction Since 1977, when Armstrong and Fendlerl first intentionally used an aqueous solution of sodium dodecyl sulfate (SDS) above its critical micelle concentration (c.m.c.) as the mobile phase for the chromatographic separation of transfer RNAs, many analytical separation processes mediated by organized media have been studied.2-8 The use of micellar media has made it possible to simplify the manipulation of the samples, thus permitting, for ex- ample, the direct injection of biological fluids without any previous treatment,6-9--’2 an increase in the number of water-soluble mobile phases available, the use of new stationary phases by modification of conventional stationary phases in the presence of the surfactant,6 the design of new extraction and/or preconcentration schemes based on the cloud point phenomenon13-17 and the improvement, in several instances, of the analytical signal in chromatographic detec- tion .5318 It is well documented2-19-22 that the separation mechanism in micellar liquid chromatography (MLC) has its origin in the capacity shown by the different organized media to interact directly with structurally similar solutes; the existence of this new phase (micellar phase) permits an increase in the possibilities of separation of a solute; further, the analyte can interact with the micelles of the mobile phase through one or more of the following mechanisms: hydrophobic interactions, electrostatic interactions or interactions by means of hydrogen bonding.All of these produce a larger set of possibilities in chromatographic separations. The three-phase model proposed by Armstrong2 has been described in quantitative terms and several workers have tested its validity both for neutral and charged analytes.’,SJj However, very few studies have been carried out with zwitterion solutes in spite of the large number of molecules of biological and pharmaceutical interest that behave in this way. * To whom correspondence should be addressed. In this work we studied the retention mechanism of the following zwitterionic analytes: the cephalosporins, cephaz- o h (CFC), cephalexin (CFX), cephalothin (CFT), cephalori- dine (CFL) and cephradine (CFR) and two acid precursors of these antibiotics, 7-aminocephalosporanic acid (7-ACA) and 7-aminodesacetoxycephalosporanic acid (7-ADCA), in MLC using SDS as the mobile phase in the presence and absence of different organic modifiers.Table 1 shows the structures and acid-base characteristics of the different compounds studied. Experimental Reagents The cephalosporins were obtained as a gift from Antibi6ticos (Leon, Spain) and used without further purification. Stock solutions (0.1-0.2 mol dm-3) were prepared daily and suitably diluted in water or in the mobile phase before use. The surfactant sodium dodecyl sulfate (SDS) was obtained from Fluka (Milan, Italy) and used as received.The organic modifiers, ethanol, propan-1-01, butan-1-01, pentan-1-01 and hexan-1-01 were purchased from Fluka. Methanol and aceto- nitrile (Carlo Erba, Milan, Italy) of HPLC grade were also used. Acetate, phosphate and citrate buffers were prepared from analytical-reagent grade reagents. All solvents and analytes were filtered through 0.45 pm nylon membrane filters (Millipore, Milford, MA, USA) and ultra-high-quality water obtained from an Elgastat UHQ water-purification system was used throughout. Apparatus A modular component liquid chromatographic system was used consisting of a Spectra-Physics SP 8800 ternary pump (San Jose, CA, USA), an SP 8450 UV detector and an SP 4290 integrator. In all experiments a Rheodyne Model 7125 injection valve (Cotati, CA,.USA) with a 10 mm3 sample loop was used.The stationary phase columns were a 220 X 4.6 mm i.d. Spheri 5 ODs-224 column from Brownlee (Santa Clara, CA, USA). Procedures Mobile phases were prepared by dissolving the appropriate amount of SDS in water alone or containing the desired organic modifier and buffered at the appropriate pH. A flow rate of 1.0 cm3 min-1 was used throughout. In each series of experiments the column was conditioned by passing mobile phase for the necessary time in order to stabilize the signal before the injection of the sample. Cephalosporins were detected by measuring the UV absorption at 260 nm.54 Analyst, January 1995, Vol. 120 Methods Retention times were measured automatically. The capacity factor ( k ' ) was calculated using the following equation: k' = (VR - v())/v() where V R and V, are the volumes required to elute the peak maximum for a retained solute and for a non-retained solute, respectively. The void volume, VO, of the system was measured at all mobile phase concentrations by injecting water o r NaN03 solution and was found to be 1.57 cm3, which was used for k' calculations.The elution volume of NaN03 was in good agreement with that of pure water. The stationary phase volume, V,, was calculated as the difference between the column volume, V,, and the void volume, Vo. Even though this is a poor estimate because it includes the silica support volume, the error introduced is identical for all the compounds.23 The column volume was calculated according to the expression V, = AcL where A, and L are the cross-sectional area and the length of the chromatographic column, respectively. The value of V, was found to be 3.66 cm3 and the calculated value of the stationary phase volume was 2.09 cm3.The chromatographic phase ratio, @, was calculated using the equation: The calculated value of @ was 1.33. The relative retention values between two adjacent peaks 2 and 1 (separation factor, a) were calculated using the following equation: @ = v,/v, = k'zlk'l Table 1 Structures and pK, values of the cephalosporin studied Cephalosporin Cephalothin Cephaloridine Cephalexin Cephradine Cefazolin 7-Aminocephalosporanic acid R' COOH Basic structure R* -0 In water* In 0.10 mol dm-3 SDSt PKI, PKiz PKq PK,, 3.18 - 2.22 - 1.67 - 2.70 - 2.56 6.88 4.27 7.02 2.63 7.30 4.47 7.53 2.78 - 2.54 - 7-Aminodesacetoxycephalosporanic acid -H * Data taken from ref.56. t Values determined by potentiometric titration in this work. -H 2.02 4.42 - 4.74 -0 -H 2.95 4.87 - 4.80Analyst, January 1995, Vol. 120 55 6 4 ' 2 . The peak resolution ( R ) was calculated according to the following expression: 2(tr2 - tr 1 ) w2 + w1 R= where tr2 and trl are the retention times for two adjacent peaks and w2 and w1 are the peak widths. The chromatographic efficiency was calculated using the manual procedure of Foley and Dorsey:24 (&I2 N = 41.7 BIA + 1.25 where t, is the solute retention time, wo.l is the peak width measured at 10% of the peak height and BIA is the peak asymmetry ratio also measured at 10% of the peak height.Results and Discussion Effect of the Mobile Phase pH Cephalosporins in an aqueous solution are predominantly found in the form of zwitterions (RHk) able to react both with H+ and OH- according to the following equilibrium: OH- RH2f = RH' = RH- H+ Hence the pH of the mobile phase considerably modifies the concentration of these species, directly affecting the elution time and the elution mechanism. In general, the presence of an organized medium modifies, among others, the acid-base constants of the solubilized systems;25 this modification can be explained by the differ- ences between the properties of the bulk solution and the micellar environment and by the electrostatic attractions and repulsions between the species involved and the micelle when both are charged.Table 1 gives the potentiometiically determined pK, values of the different cephalosporins in the presence of 0.10 mol dm-3 of SDS. In all instances the difference between the pK, obtained in SDS and in water is positive, indicating that the two acid-base equilibria are less displaced in the presence of SDS than in water; this modification of the acid-base equilibrium is due to the electrostatic repulsion between the micelle, which is negatively charged, and the R- species, which leads to a decrease in the equilibrium constant for RH' $ R- + H+ and the electrostatic attraction between the RH2+ species and the micelle, which modifies the equilibrium constant for RH2+ e RH' + H+. It is known that when micellar mobile phases constituted by ionic surfactants with C18 stationary phases are used, the monomers joined to the stationary phase give it a positive or negative charge according to the type of surfactant used.This has been used by different workers for the chromatographic separation of cations or anions with surfactant-modified C18 stationary phases as an alternative to conventional ion- exchange chromatography.2632 For this reason, when the analyte to be eluted has the same charge as the modified stationary phase, the molecules are rapidly eluted owing to electrostatic repulsions, whereas the molecules with opposite charge are eluted more slowly by electrostatic attraction. The dependence of k' on pH for a constant concentration of surfactant in the mobile phase is sigmoidal, as long as no electrostatic repulsion exists between the solutes and the surfactant molecules.33 Fig.1 shows the values of k' for the species studied as a function of the pH of the mobile phase. Three different behaviours can be observed: CFT, CFZ and CFL, which show only one pK, value, which is lower than or close to 3 (see Table l ) , are found in the pH range studied predominantly in the form of RH'; because of this, as the pH of the mobile phase increases, a slight decrease is produced in the capacity factors. This decrease is greater for CFT as in the presence of SDS it has a pK, of 3.18. For the two acids 7-ACA and 7-ADCA, k' decreases to a pH near 4.8 (a value similar to that of pK, calculated in the presence of micelles) and starting from this value k' remains almost constant.Finally, for CFX and CFR the experimental results show differences from the theoretical behaviour, since as the pH increases k' should decrease to values near pK1 and remain constant for pH values between pK1 and pK2 where the net charge of the dominant RH' species is zero. However, the experimental curve shows a strong decrease to pH values near 4.5. This can be explained if it is kept in mind that these two species show a high constant of solute-micelle union (see Table 5 ) , which means that the solutes are found predominantly in the micellar phase and their elution mechanism is modified with the decrease in the 5 1 3 t 2t \ 2 3 4 5 6 7 8 (4 7-ADCA t " 2 3 4 5 6 7 :j 10 P CFR " 3 4 5 6 7 PH Fig. 1 Capacity factors versus pH of the mobile phase. Mobile phase: [SDS] = 0.10 mol dm-3. Flow rate = 1.0 cm3 min-1.UV detection at 260 nm.56 Analyst, January 1995, Vol. 120 interaction of the solute with the modified stationary phase, eluting more rapidly. As an example, Fig. 2 shows the chromatograms obtained at two pH values (3.55 and 6.72); the different behaviours of the species studied in relation to pH make the order of their 1 0 2 - 1 A 8 48 50 1 0.01 A I 0 2 4 6 8 Time/min Fig. 2 Chromatograms obtained at two pH values, (a) 3.00 and (b) 6.72, of 50 ppm of different cephalosporins (10 mm3 injected). Mobile phase: [SDS] = 0.10 mol dm-3. Flow rate = 1.0 cm3 min-1. UV detection at 260 nm. Peak assignment: 1, cephazolin; 2, 7-amino- cephalosporanic acid; 3, cephalothin; 4, 7-aminodesacetoxycephalo- sporanic acid; 5, cephaloridine; 6, cephalexin; 7, cephradine.elution vary drastically, depending on the pH of the mobile phase. This effect is especially relevant with 7-ADCA, which undergoes such a modification as to move from fifth place (pH 3.00) to first (pH 6.72). Likewise, and owing to the acid-base and structural characteristics of CFL, the order of elution is inverted with respect to CFX and CFR for pH 6.72. The pH of the mobile phase has an important effect on the separation factor and the resolution of chromatographic peaks. Tables 2 and 3 show these values for the species studied. It can be seen that owing to the change in the order of elution mentioned above, a clear conclusion cannot be drawn with respect to these parameters. However, it can be concluded that the pH with which the best resolution of peaks is obtained ( R 3 1) is 4.02.The buffer concentration in the mobile phase was also studied for values between 0.02 and 0.50 mol dm-3. The results obtained show that as the buffer concentration increases there is a decrease in both retention times and capacity factors. This confirms that there is a strong electro- static component in the retention mechanism of these com- pounds. The increase in buffer concentration produces an increase in sodium ions in the mobile phase which can bind to the negative active centres of the stationary phase modified with sodium dodecyl sulfate. This means that competition is established between the sodium ions and the cephalosporins, thus decreasing the forces of electrostatic interaction between the stationary phase and the species studied.It is also possible that increasing ionic strength decreases the adsorption of the dodecyl sulfate on the stationary phase, and this would also cause decreased retention. The results obtained show that for zwitterion compounds, the electrostatic forces of attraction or repulsion play an important role in their elution, because when the CIS stationary phase is not modified with surfactants the protona- tion of the solutes produces a decrease in their retention times. The increases observed in the retention times for all the cephalosporins studied, when the protonated forms of the different species are present, could be due to the fact that the forces of electrostatic attraction between the solute and the surface of the stationary stage modified with surfactant are stronger than those existing between the solute and the micelles.This has already been demonstrated by other workers .34 Effect of the Surfactant Concentration The variation in the concentration of SDS in the mobile phase modifies the number of micelles in it and therefore the retention times of the solutes can be modified.35-36 Armstrong and Stine37 propose three groups of solutes depending on their behaviour in chromatography with Table 2 Variation of the separation factor (a) with the pH of the mobile phase. Mobile phase, [SDS] = 0.10 mol dm-3; flow rate, 1.0 cm3 min-1: UV detection at 260 nm PH a: 3.00 a7-ACAICFZ aCFTI7-ACA aCFL/Cm a7-ADCAICFL 2.52 2.05 1.22 1.45 3.55 a7-ACAICFZ aCFTl7-ACA aCFUCFT a7-ADCNCFL aCFXI7-ADCA a:CFRICFX 2.41 2.06 1.95 1 .00 7.72 1.06 1.63 1.93 1.36 2.17 5.84 1.11 4-02 ~ ~ A C A / C F Z aCFTl7-ACA a7-ADCAICFT (kFL17-ADCA %FXICFL ~ F W C F X 4.73 a7-ACAICFZ ECFTI7-ACA OIADAICFT kFL17-ADCA a:CFXICFL ~CFRICFX 2.46 1.94 1.52 3.71 5.13 1.11 5.09 E7-ACAlCFZ a7-ADCA17-ACA kFTf7-ADCA kFLICFT ~CFXICFL WFRICFX 6.40 a7-ACAI7-ADCA aCFU7-ACA aCFT/CFZ ECFXICIT (YCFLICFX ~CFFUCFL 1.42 1.70 1.96 6.01 2.48 1.19 1.11 1.10 4.36 5.35 1.28 1.15 6.72 a7-ACN7-ADCA %FZI7-ACA a:CFT/CFZ ECFWCFT ~CFRJCFX WFLICFR 1.00 1.38 4.54 3.98 1.22 1.35Analyst, January 1995, Vol.120 57 30 20 10 micellar mobile phases: solutes that bind to micelles, solutes that do not bind and solutes that are repelled by micelles; the effect of varying the surfactant concentration in the mobile phase differs for each of these three types of solutes. For species exhibiting affinity for micelles, increases in surfactant concentration lead to a decrease in retention times, whereas for solutes that do not bind to micelles the retention times are not modified to any appreciable extent. For compounds that are repelled by micelles, increases in the surfactant concentration correspondingly increase the reten- tion times.The strength of the eluent can be said to increase with increase in micellar concentration only if the solute interacts with the micelle in the mobile phase.38 Fig. 3 shows the variations in the capacity factors as a function of the SDS concentration in the mobile phase working in a 0.10 mol dm-3 acetate buffer medium at pH 4.02. In all instances k’ decreases with increase in the concentration of SDS except for CFZ, for which it remains almost constant.This compound appears to behave as a solute that does not bind to micelles. Fig. 4 shows the chromatograms corresponding to the elution of the seven cephalosporins with (a) a conventional mobile phase [methanol-water (80 + 20 v/v)], (6) 0.05 mol dm-3 SDS and (c) 0.10 rnol dm-3 SDS. The elution behaviour of a solute when using a micellar mobile phase depends on the combined effects of three partition coefficients, as shown in the following scheme: - - - where A,, Am and A, are the analyte concentrations in the bulk of the aqueous solution, in the micelle and in the stationary phase, respectively, and Km,, K,, and Ks, are the partition coefficients of the solute between the micelle and water, between the stationary phase and water and between the stationary phase and the micelle, respectively.The relationship between the retention of the solute and the composition of the micellar mobile phase for solutes that bind to micelles, addressed theoretically by Armstrong and Nome,39 can be expressed as follows: 2.5 2.0 1.5 1.0 where k’ is the capacity factor, @ the VJV, phase ratio - (b) - - - (volume of the stationary phase and void volume, respec- tively), V is the molar volume of the surfactant in dm3 mol-1, [MI is the surfactant concentration in micellar form in mol dm-3 (that is, the total concentration of SDS minus the c.m.c.) and K,,, K,, and K,, are the above-defined partition coefficients. The third partition coefficient, K,, , which describes the direct transfer of a solute associated with the micelles to the stationary phase, is usually not taken into consideration as it is generally assumed that this direct transference is minimum.2*39-41 The plot of llk’ against [MI should be a straight line (see Table 4) and will permit the calculation of K,, and K,,.Obtaining these values makes it possible to calculate the equilibrium constant between the solute and the mobile phase by the monomer of the surfactant defined as K2 = V(K,, - l).41 To obtain the equilibrium constant by micelle, Keq = A,/(A,C), where C is the L O 0.1 0.2 0.3 7-ACA CFZ 0.5 0 0.1 0.2 0.3 [SDSymol dm3 Fig. 3 Capacity factors versus sodium dodecyl sulfate concentration. Mobile phase: acetate buffer = 0.10 mol dm-3 (pH 4.02).Flow rate = 1.0 cm3 min-l. UV detection at 260 nm. Table 3 Variation of the resolution (R) with the pH of the mobile phase. Mobile phase, [SDS] = 0.10 mol dm-3; flow rate, 1.0 cm3 min-’; UV detection at 260 nm PH R 3.00 R7-ACAICFZ RCFTl7-ACA RCFLICFT R7-ADCAICFL 2.3 2.2 0.63 1.5 3.55 R7-ACAICFZ RCFT17-ACA RCFWCFT R7-ADCAICFL RCFXI7-ADCA RCFRICFX 1.7 2.0 2.2 UP 24 0.92 4.02 R7-ACAICFZ RCFTl7-ACA R7-ADCAICFT RCFW7-ADCA RCFWCFL RCFRICFX 4.73 R~-ACAICFZ RCFTI~-ACA RADCAICFT RCFU7-ADCA RCFXICFL RCFRICFX 5.09 R7-ACAICFZ R?-ADCA17-ACA RCFT17-ADCA RCFL/CFT RCFXICFL RC F RIC FX 6.40 R7-ACAI7-ADCA RCFZI7-ACA RCFTICFZ RCFXICFr RCFYCFX RCFRICFL 1.0 1.1 1.8 2.2 19 0.95 0.73 1 .o 1.1 3.5 15 1 .o Up* 0.45 1.1 6.4 6.6 1.6 UP UP 1.7 5.5 1.2 0.72 6.72 R7-ACAIADCA RCFZn-ACA RCFTICFZ RCFWCFT RCFRICFX RCFWCFR UP UP 1.6 2.6 0.79 1.3 * UP = unresolved peak.58 Analyst, January 1995, Vol.120 concentration of micelles, K3 must be multiplied by the aggregation number of the surfactant.4' For aqueous micelles of SDS the aggregation number is 62.23 The partition coefficient K,, can be calculated as the ratio between the other two partition coefficients (Ksm = Ksw/K,,).39 At low micelle concentrations the system resembles reversed-phase chromatography and K,, controls the solutes retention. I ' 1 ' \- 0 8 16 48 56 0 16 50 58 0 8 16 24 32 Time/min Fig. 4 Chromatograms obtained for different mobile phases of 50 ppm of cephalosporin (10 mm3 injected) (u) methanol-water (20 + 80 v/v) and acetate buffer = 0.10 mol dm--7 (pH 4.02).(b) SDS = 0.05 rnol dm-3 and acetate buffer = 0.10 mol dm--7 (pH 4.02). ( c ) SDS = 0.1 mol dm-3 and acetate buffer = 0.10 rnol dm--7 (pH 4.02). Flow rate = 1.0 cm3 min-1. UV detection at 260 nm. Peak assignment as in Fig. 2. However, K,, has an increasing effect as the concentration of surfactant is increased, owing to the larger number of micelles present in the mobile phase, whereas the effect of K,, on retention is independent of micelle concentration. The values of K,,, K,,, K,,, K2 and Keq are obtained from the values of the slopes and the ordinates at the origin of the corresponding regression lines (Table 5 ) . It can be observed that the K,, values are higher than the K,, values, therefore the solute-micelle equilibrium is more important than the solute-surfactant-covered stationary phase equilibrium.The effect of the increase in the concentration of surfactant and, therefore, the number of micelles in the mobile phase, produces a greater decrease in the retention times for the compounds with higher K,, values. The values in Table 5 are in good agreement with the chromatographic behaviour of the studied compounds, as the highest calculated value of K,, corresponds to the compound that experiences the largest decrease in elution time with the increase in surfactant concentration. The negative value for the intercept on the y-axis for CFR can be attributed to its hydrophobicity. It has been shown that errors in calculation can increase with a corresponding increase in the hydrophobicity of the solutes.2,3"37 This is due to the fact that these compounds have a great affinity for the stationary phase (high K,, values) and, as a consequence, the y-axis intercepts are very close to zero.In the literature, however, there are data that show that the increase in hydrophobicity does not completely justify these negative ordinates.23 Another possible source of error lies in the need to use high concentrations of surfactant in the mobile phase in order to be able to elute the more hydrophobic solutes in a reasonably short time. At these concentrations, the c.m.c., the aggregation number and the geometry of the micelle are modified with the concentration of the surfactant, whereas in the model proposed they are assumed to remain constant.2~3~ In fact, when values of llk' corresponding to concentrations of SDS between 0.05 and 0.20 mol dm-3 are used, the ordinate at the origin is modified and ceases to be negative, although it has a very low value (5.2 x 10-6).Another parameter that can undergo modifications with variation of the surfactant concentration is the separation factor. Table 6 shows the a values as a function of the surfactant concentration, calculated as the ratio between the capacity factors. A decrease in separation factor correspond- ing to an increase in surfactant concentration can be seen; this behaviour is in agreement with the fact that a decrease in the capacity factors takes place as the surfactant concentration increases. Both the capacity factors and separation factor are paramet- ers that affect the resolution of the chromatographic peaks.Table 7 shows the R values as a function of the surfactant concentration. In all instances a decrease in R occurs with increase in SDS concentration in the mobile phase. It can be seen that up to a concentration of 0.15 rnol dm-3 all the R Table 4 Coefficients of the regression lines: Ilk' = m[SDS] + b. Mobile phase, SDS in acetate buffer (pH 4.02); flow ratc, 1.0 cm3 min-1: UV detection at 260 nm Analyte CFZ 7-ACA CFT 7-ADCA CFL CFX CFR m 2.03 -t 0.03 1.45 k 0.02 2.21 f 0.07 1.13 t 0.06 0.68 -t 0.01 0.63 k 0.01 * - a Non-bonding analyte. b ?.2 $2 - I .420 f 0.005 0.9996 0.698 f 0.004 0.9995 0.40 k 0.01 0.9984 0.024 k 0.009 0.9980 0.00 1 * 0.002 0.9997 -(0.002 f 0.002) 0.9992Analyst, January 1995, Vol. 120 59 values are near to or higher than unity.Therefore, the use of a mobile phase with a surfactant concentration between 0.05 and 0.15 mol dm-3 permits an adequate separation of the compounds studied. Effect of the Presence of Organic Modifiers in the Mobile Phase One of the greatest disadvantages of liquid chromatography using micellar mobile phases is the decrease in efficiency of the chromatographic columns. The presence of small amounts of organic components in the mobile phase can have an important effect on the interaction between the micelles and the stationary phase, altering the solubilization process of the solute and the dynamic process of the formation and nature of the micelles.",44 The effcct of organic modifiers on micellar liquid chromato- graphy has been studied by many workers.45-5() The addition of these organic compounds to aqueous SDS solutions modifies the degree of aggregation (number and distribution of the aggregates), critical micellar concentration, degree of bonding with the counter ion, density of interfacial charge, the dynamics of the aggregation process and the dynamic interac- tions of the bonding of the solutes with the modified micelles;~l-sfj the presence of such additives could therefore modify thc properties of the stationary phase modified by the surfactant and the partition coefficients and the migration of the solute.A study of the influence of organic modifiers in the mobile phase was carried out with three species of low, medium and high hydrophobicity: CFT, CFL and CFX. As organic modifiers, short-chain alcohols (methanol, propanol, butanol, pentanol, and hexanol) and acetonitrile were studied.The maximum concentration of these modifiers in the mobile phase was never higher than 10% in order to ensure the presence of micelles at all times. The presence of all the organic modifiers examined produces a decrease in the capacity factor of the solutes studied (Fig. 5 ) . The higher the percentage of alcohol added at the mobile phase and the more hydrophobic the solution, the greater is the decrease. It can also be seen that as the hydrophobicity of the modifier added increases (greater number of carbons) there is a decrease in capacity factor. Combining the concentration and hydrophobicity of the modifier and the nature of the solute, the retention time of CFX can be decreased by 75% in the presence of just 0.6% of hexanol.In Fig. 6 are shown the chromatograms correspond- ing to the elution of three species of those studied with 0.1 mol dm-3 SDS, 0.6% hexanol and 0.1 mol dm-3 SDS + 0.6% hexanol; the pH of the three mobile phases was 4.02. From these results it can be concluded that the minimum amount of hexanol employed does not contribute at all to the elution of the studied compounds (the first peak appears at about 20 min and after 3 h CFT has not eluted) and the effect on the eluting strength of the SDS mobile phase can be attributed to the Table 5 Partition cocfficients K,,, K,,,, K,,, K2 and Keq. Mobile phase, 0.10 mol dm-3 acetate buffer (pH 4.02); flow rate, 1 .0 cm3 min-I: UV detection at 260 nm Anal yte CFZ CFT CFL CFX CFR 7-ACA ' 7-ADCA Slope 2.03 I .45 2.21 1.13 0.68 0.63 -4 Intercept K,, Kmw* K,m K2 Keqt - 1.42 0.53 6.8 0.08 1.4 8.7 x 101 0.70 1.1 9.4 0.11 2.1 1.3 x 10' 0.40 1.9 2.4 x 101 0.08 5.5 3.4 x 102 0.02 3.1 x 101 2.3 x 10' 0.14 5.7 x 10' 3.5 x 103 10-3 7.5 x 10' 2.8 x 103 0.25 6.8 x 102 4.2 x 103 Negative * For the calculation of K,, a molar volume of SDS = 0.246 dm3 mol-l was used.8 + Aggrcgation number of SDS = 62." t Non-bonding analyte.Table 6 Variation of the separation factor (a) with thc SDS concentration. Mobile phase, 0.10 mol d r 3 acetate buffer (pH 4.02); flow rate, 1.0 cm3 min-1; UV dctection at 260 nm [SDS]/ mOl dm-3 ~ ~ ~ - A c . A / c F Z ffc FTI7-AC A %AD( A/( F1 kLl7-ADC'A aCFX/C FL ~~CFRICFX 0.05 1.71 2.00 1.58 7.26 2.15 1.16 0.10 1.63 1.92 I .36 4.83 2.06 1.12 0.15 1.53 1.88 1.29 3.55 2.06 1.12 0.20 1.40 1.88 1.18 3.41 1.86 1.10 0.25 I .37 1.83 1.13 3.17 1.76 1.08 Table 7 Variation of the rcsolution (R) with the Concentration of SDS.Mobile phase, 0.10 mol dm-3 acetate buffer (pH 4.02); flow rate, 1 . 0 cm3 min--': UV detection at 260 nm [SDS]/ 0.05 0.96 1.7 1.7 7.5 6.4 1.8 0.10 0.95 1.7 1 . 1 6.3 5.6 1.3 0.15 0.9 1 1 .6 0.91 5.4 4.9 1.1 0.20 0.87 1.6 0.58 4.8 3.9 0.83 0.25 0.88 1.6 0.42 3.9 3.0 0.63 mol dm-' R7-AC AIC'FZ RC I-Tl7-AC A R7-ADC'AIC'FT RC.FL17-ADC A RCFXICFL RCFRICFX60 Analyst, January 1995, Vol. 120 improvement of the mass transfer between the mobile phase and the stationary phase. The addition of organic modifiers in the micellar mobile phase therefore allows the efficiency of micellar liquid chromatography to be considerably increased.It can be seen (Table 8) that in all instances the addition of organic modifiers produces an increase in the efficiency of the column, this effect being larger than the greater number of carbons of the modifier added. Conclusions The results obtained show that in elution with micellar mobile phases of zwitterion compounds the pH of the mobile phase plays a decisive role and can be considered as a factor that increases chromatographic efficiency. Therefore, the pH modifies the electrostatic interaction between the different species and the modified stationary phase. The study of the concentration of SDS in the mobile phase allowed one to calculate the solute-micelle association con- stants for five of the compounds studied.An increase in SDS concentration produces a decrease in efficiency, which means a decrease in the separation factors, but also an important decrease in the retention times, thus saving analysis time. The SDS-modified column allows the electrostatic interaction of ionizable compounds with the stationary phase and, simultaneously, the multiple interaction predicted by the three-phase model. When at the pH of the mobile phase positively charged species exist in solution, an increase in the retention times is observed owing to the electrostatic attrac- tion between solutes and the negatively charged stationary phase, the electrostatic attraction being stronger (for some solutes) than the interaction with micelles. The addition of organic modifiers to the micellar mobile phase produces an increase in the strength of elution and also in the efficiency of the chromatographic column.This effect becomes more important with increase in the percentage of the modifier added and increase in the number of carbon atoms of the additive. We gratefully acknowledge financial support from DGICYT (Project PB91-0185) and from Junta de Castilla y Le6n (Project SA68/93). C. G. P. expresses his gratitude to the Spanish Government for the awarded grant through the PFPI. 0.8 L 0.6 0.4 0.2 I I I I I J A -. 3 k 2 1 "0.0 2.0 4.0 6.0 8.0 10.0 Organic modifier (% v/v) Fig. 5 Capacity factor versus organic modifier concentration: A, hexanol; B, pentanol; C, butanol; D, propanol; E, methanol; and F, acetonitrile. 0 4 8 24 32 0 8 16 24 32 0 4 8 12 16 Time/min Fig.6 Effect of hexanol on the elution of cephalosporins. Mobile phase: (a) 0.10 mol dm-3 SDS, ( b ) 0.6% hexanol and ( c ) 0.10 mol dm-3 SDS + 0.6% hexanol.Analyst, January 199.5, Vol. 120 61 Table 8 Variation in the efficiency of the column (number of theoretical plates, N ) with organic modifiers. Mobile phase, 0.10 mol dm-3 SDS in 0.10 mol dm-3 acetate buffer (pH 4.02); flow rate, 1.0 cm3 min-1; UV detection at 260 nm Nl103 Concentration Solute (% ) CFT 0.0 0.6 1 .o 2.0 4.0 6.0 8.0 10.0 CFL 0.0 0.6 1 .0 2.0 4.0 6.0 8.0 10.0 CFX 0.0 0.6 1 .0 2.0 4.0 6.0 8.0 10.0 1-HeOH 0.30 1 .o 1.4 2.0 0.70 1.4 1.8 2.3 2.11 3.3 3.8 4.4 I-PeOH 0.30 0.90 1.6 2.1 0.70 1.3 1.8 2.4 2.1 2.7 3.4 4.1 1-BuOH 0.30 0.90 1.7 2.0 0.70 1.1 1.4 1.9 2.1 3.0 3.7 4.2 1-PrOH 0.30 0.80 1.3 1.8 2.0 2.1 0.70 1.0 1.2 1.4 1.5 1.8 2.1 3.2 3.8 4.3 4.4 4.4 MeOH 0.30 0.70 0.80 1.1 1.1 1.2 1.4 1.4 0.70 0.90 0.90 1 .o 1.1 1.1 1.1 1.3 2.1 2.3 2.5 2.8 3.0 3.3 3.7 4.1 AN 0.30 0.90 1.0 1.1 1.2 1.5 0.70 0.90 0.90 1.0 1.1 1.6 2.1 2.4 2.7 2.8 3.1 4.2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 References Armstrong, D.W., and Fendler, J. H., Biochim. Biophys. Acta, 1977,418,75. Armstrong, D. W., Sep. Purif. Methods, 1985, 14,213. Pelizzetti, E., and Pramauro, E., Anal. Chim. Acta, 1985, 169, 1. Dorsey, J. G., Adv. Chromogr., 1987, 27, 167. Hinze, W. L., in Ordered Media in Chemical Separations, ed. Hinze, W. L., and Armstrong, D. W., American Chemical Society, Washington, DC, 1987. Hinze, W. L., Ann. Chim., 1987, 77, 167.Borgerding, M. F., Williams, R. L., Jr., Hinze, W. L., and Quina, F. H., J. Liq. Chromatogr., 1989, 12, 1367. MacIntire, G. L., CRC Crit. Rev. Anal. Chem., 1990, 21, 257. De Luccia, F. J., Arunyanart, M., and Cline Love, L. J., Anal. Chem., 1985, 57, 1564. Cline Love, L. J., Zibas, S., Norosky, J., and Arunyanart, M., J. Pharm. Biomed. Anal., 1985, 3, 511. Arunyanart, M., and Cline Love, L. J., J. Chromatogr., 1985, 342, 293. Palmisano, F., Guerrieri, A., Zambonin, P. G., and Cataldi, T. R. I., Anal. Chem., 1989, 61, 946. Saitoh, T., and Hinze, W. L., Anal. Chem., 1991, 63,2520. Garcia Pinto, C., PCrez Pavon, J. L., and Moreno Cordero, B., Anal. Chem., 1992, 64, 2334. Fernandez Laespada, Ma. E., Perez Pavon, J. L., and Moreno Cordero, B., Analyst, 1993, 118, 209.Hinze, W. L., and Pramauro, E., CRC Crit. Rev. Anal. Chem., 1993,24, 133. Moreno Cordero, B., PCrez Pavbn, J. L., Garcia Pinto, C., and Fernandez Laespada, Ma. E., Talanta, 1993,40, 1703. Xia, F., and Cassidy, R. M., Anal. Chem., 1991, 63, 2883. Armstrong, D. W., and Henry, S. J., J. Chromatogr.. 1980, 3, 657. Dorsey, J. G., Chromatographia, 1987, 2, 13. Khaledi, M. G., Trends Anal. Chem., 1988, 7. 293. Borgerding, M. F., Quina, F. H., Hinze, W. L., Bowermaster, J., and McNair, H. M., Anal. Chem., 1988, 60, 2520. Khaledi, M. G., Peuler, E., and Negh-Ngwainbi, J., Anal. Chem., 1987, 59, 2738. 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 Foley, J. P., and Dorsey, J. G., Anal. Chem., 1983,55, 730. El Seoud, 0. A., Adv. Colloid.Interface Sci., 1989, 30, 1. Knox, J. H., and Hartwick, R. A., J. Chromatogr., 1981,204,3. Cassidy, R. M., and Elchuk, S., Anal. Chem., 1982, 54, 1558. Cassidy, R., and Elchuk, S., J. Chromatogr., 1983, 21,454. Barkley, D. J., Blanchette, M., Cassidy, R. M., and Elchuk, S., Anal. Chem., 1986, 58, 2222. Barkley, D. J., Dahms, T. E., and Villeneuve, K. N., J. Chromatogr., 1987, 395, 631. Ito, K., Ariyoshi, Y., Tanabiki, F., and Sunahara, H., Anal. Chem., 1991,63,273. Janos, P., Stulik, K., and Pacakova, V., Talanta, 1992, 39, 29. Arunyanart, M., and Cline Love, L. J., Anal. Chem., 1985,57, 2837. MenCndez Fraga, P., Gonzalez, E., and Sanz-Medel, A., Anal. Chim. Acta, 1988, 212, 181. Arunyanart, M., Wienberger, R., Hirsch, R. F., and Cline Love, L. J., Anal. Chem., 1982, 54, 2223. Berthod, A., Girard, I., and Gonnet, C., Anal. Chem., 1986, 58, 1362. Armstrong, D. W., and Stine, G. Y., Anal. Chem., 1983, 55, 2317. Medina Hernandez, M. J., and Garcia Alvarez-Coque, M. C., Analyst, 1992, 117, 831. Armstrong, D. W., and Nome, F., Anal. Chem., 1981,53,1662. Yarmchuck, P., Weinberger, R., Hirsch, R. F., and Cline Love, L. J., J. Chromatogr., 1984, 283,47. Berthod, A., Girard, I., and Gonnet, C., Anal. Chem., 1986, 58, 1359. Arunyunart, M., and Cline Love, L. J., Anal. Chem., 1984,56, 1557. Lang, J., and Zana, R., J. Phys. Chem., 1986, 90, 5258. Malliaris, A., Lang, J., Sturm, L., and Zana, R., J. Phys. Chem., 1987,91, 1475. Dorsey, J. G., DeEchegaray, M. T., and Landy, J. S., Anal. Chem., 1983,55, 924. Kirkman, C. M., Zu-Ben, C., and Uden, P. C., J. Chromatogr., 1984,317, 283. Borgerding, M. F., and Hinze, W. L., Anal. Chem., 1985,57, 2183.62 Analyst, January 1995, Vol. 120 48 Berthod, A., Girard, I., and Gonnet, C., Anal. Chem., 1985, 58, 1356. 49 Mullins, F. G. P., and Kirkbright, G. F., Analyst, 1986, 111, 1273. 50 Grieser, F., J. Phys. Chem., 1981, 85, 928. 51 Lianos, P., Lang, J., Strazielle, C., and Zana, R., J. Phys. Chem., 1982, 86, 1019. 52 Buton, C. A., Gan, L. H., Hamed, F. H., and Moffatt, J. R., J. Phys. Chem., 1983, 87, 336. 53 Almgrem, M., Swarup, S., and Lofroth, J. E., J. Phys. Chem., 1985, 89, 4621. 54 Malliaris, A., J. Phys. Chem., 1987, 91, 6511. 55 Baglioni, P., and Kevan, L., J. Phys. Chem., 1987, 91, 1516. 56 Tsukinaka, Y., and Akira, T., J. Pharm. Sci., 1976, 65, 1563. Paper 4/01 1860 Received February 28, 1994 Accepted July 5, 1994

ISSN:0003-2654    

DOI:10.1039/AN9952000053

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 63 1 -Met hoxycarbon y I i ndol izine-3,5- dicarbaldehyde as a Derivatization Reagent for Amino Compounds in High-performance Capillary Electrophoresis Shigeyuki Oguri, Chikako Uchida and Yasuko Miyake Department of Home Economics, Aichi-Gakusen University, 28 Kamikawanari, Hegoshi-cho, Okazaki City, Aichi, 444, Japan Yasuyoshi Miki and Kazuaki Kakehi Faculty of Pharmaceutical Sciences, Kinki University, 3-4-1 Kowakae. Higashi-Osaka, 577, Japan The indolizine derivative 1 -methoxycarbonylindolizine-3,5- dicarbaldehyde (IDA) was synthesized from 24 1,3-dioxolan-2- y1)pyridine in four steps. The reactivity of the reagent towards primary amines was investigated by using alanine (Ala) as a model compound. The reagent easily reacted with Ala in 20 mmol l-1 phosphate-borate buffer at pH 10 containing 2540% v/v of ethanol in the dark at room temperature, and the reaction was completed within 15 min. The IDA derivative of alanine (IDA-Ala) showed a strong absorption at 280 nm (E = 3.31 x 104 I mol-1 cm-1) and 409 nm (E = 2.18 x 104 I mol-1 cm-1).The derivative also showed fluorescence at 482 nm when irradiated at 282 or 414 nm. These wavelengths did not overlap with those of IDA. IDA-Ala showed a 1.5 times stronger absorption than that produced by the ninhydrin method, and the fluorescence intensity of IDA-Ala was about three times that produced by o-phthalaldehyde-Ala. On analysis of IDA-Ala by high-performance capillary electrophoresis (HPCE) in the capillary zone electrophoresis (CZE) mode with detection at 280 nm, the calibration graph showed good linearity over the range 0.0172-21.5 pmol ml-1 of alanine.The detection limit for alanine was about 5 nmol ml-1. The relative standard deviations in the determination of IDA-Ala at 10.8 and 21.5 pmol ml-1 were less than 2% (n = 5). Application to the analysis of a mixture of amino acids by HPCE using the micellar electrokinetic chromatography mode is also described. Keywords: I -Methoxycarbonylindolizine-3,5-dicarbaldehyde, amino acids; high-performance capillary electrophoresis; micellar electrokinetic chromatography capillary zone electrophoresis Introduction Amino compounds having primary amino groups such as amino acids or aminocyclitol antibiotics lack chromopores or fluorophores in their molecules. Derivatization is required for their detection with high sensitivity using high-performance liquid chromatography (HPLC) or high-performance capillary electrophoresis (HPCE) with photometric and fluorimetric detections.1-6 Recently, HPCE has become an attractive separation method owing to its high resolution. In spite of a wide variety of detection techniques developed for HPCE ,7-10 UV detec- tion is still one of the most popular methods for commercially available HPCE systems. We have studied the synthesis of indolizine analogues, which have shown UV absorption and fluorescence. 11 Some of the intermediate compounds in their syntheses possibly react with various biological compounds having amino or aldehyde groups. In addition, the indolizine derivatives contain hetero- aromatic N-containing rings in their molecules, and seem to be advantageous for separation by HPCE owing to their electric charge.In this paper, we report the synthesis of the indolizine derivative 1-methoxycarbonylindolizine-3,5-dicarbaldehyde (IDA), from 2-( 1,3-dioxolan-2-yl)pyridine in four steps, and an investigation of its reactivity towards primary amines using amino acids as model compounds. Experimental Chemicals Standard samples of amino acids were purchased from Takara Kohsan (Tokyo, Japan). A standard solution of an amino acid mixture (Type H) for amino acid analysers was obtained from Wako (Dosho-machi, Osaka, Japan). Other reagents were of the highest or HPLC grade commercially available. All aqueous solutions were prepared by using water purified with a Milli-Q purification system (Millipore, Milford, MA, USA).Reagents for derivatization of amino acids For ease of handling, the dried reagent (IDA) in a tube was prepared and used in this study. The derivatization procedure was easily started by placing the reaction buffer and the sample solution in a tube. IDA was then dissolved in ethyl acetate by sonication to give a 1.00 mg ml-1 solution. A portion (100 pg per 100 PI, 0.43 pmol) was transferred into a 0.6 ml polypropylene microcentrifuge tube. The solvent was carefully evaporated to dryness by placing the tubes in a desiccator under reduced pressure. The reagent tubes were then tightly capped and stored in a refrigerator in the dark until use. Reaction buffer. The reaction buffer was prepared by mixing an equal volume of ethanol and 20 mmol 1-1 phosphate-borate buffer (Equal volumes of 20 mmol 1 - 1 sodium dihydrogenphosphate and 20 mmol 1- sodium tetra- borate were mixed and the pH was adjusted to 10 by adding 1 mol 1-1 sodium hydroxide).IDA solution. A 40 pl volume of the reaction buffer was added to the reagent tube. The solution was prepared just before use. Aqueous solution of alanine as standard sample. An aqueous solution of alanine (21.5 pmol ml-1) was prepared by dissolution of alanine (19.2 mg) in 10.0 ml of water. A standard solution of alanine containing 3-nitrophenol64 Analyst, January 1995, Vol. 120 (internal standard; MNP) was also prepared in the same manner except for the use of a 2.5 mg ml-1 aqueous solution of MNP as the solvent. These solutions were stored in a refrigerator. Apparatus Proton nuclear magnetic resonance (NMR) spectra were recorded on a JEOL FX-200 spectrometer at 200 MHz using tetramethylsilane as the internal standard.Infrared (IR) spectra were recorded with a Hitachi EPI-G2 spectropho- tometer. UV/VIS absorption spectra were obtained with a Hitachi 220A spectrophotometer using a 1 cm quartz cell. For measurement of the excitation and emission spectra, a Hitachi F-3010 spectrofluorimeter with a 1 cm quartz cell was employed. HPCE was performed on a JASCO CE-800 system with a JASCO 807-IT data processor. A capillary tube of fused silica (50 pm i.d., 50 cm effective length) was used throughout. The window (0.7 mm) for detection was made by removing the polyimide coating at the 10 cm position from the cathodic end.The applied voltage with 20 kV throughout. Sample solutions were introduced into the capillary tube from the anodic side by hydrostatic injection by raising the tube 15 cm higher than the level of the cathodic electrode for 5 s. The electropherograms were recorded by monitoring the UV absorption at 280 nm. The carrier electrolyte was 20 mmol 1-1 phosphate-borate buffer (pH 10) for the capillary zone electrophoresis (CZE) mode, and 20 mmol 1-1 sodium dodecyl sulfate (SDS)4O mmol 1-1 phosphate-borate buffer (pH 7) containing methanol at a concentration of 3% v/v for the micellar electrokinetic chromatography (MEKC) mode. Synthesis of I -methoxycarbonylindolizine-3,5-dicarbaldehyde (IDA) 1 - (tert- Butoxycarbonylmethyl) -2- (I, 3-dioxolan-2-y1)pyrid- inium bromide (2) A solution of 2-(1,3-dioxolan-2-yl)pyridinel2 (1; 15.1 g, 0.1 mol) and tert-butyl bromoacetate (19.5 g, 0.1 mol) in dry acetonitrile (100 ml) was refluxed for 7 h and the reaction mixture was kept overnight at room temperature.The precipitated crystalline material was collected and washed with diethyl ether to give 26.8 g (78%) of 2: m.p. 110-112 "C (acetonitrile); IR (Nujol, cm-l), 1740; NMR (CDCI3), 6 1.51 (m, 9H), 4.0-4.2 (m, 4H), 5.97 (s, 2H), 6.43 (s, lH), 8.16 (ddd, J 8,6,2 Hz, 1H, H-5), 8.24 (dd, J 8 , 2 Hz, H-3), 8.63 (br t, J 8 Hz, lH, H-4), 9.77 (dd, J 6, 1 Hz, 1H, H-6). Analysis: calculated for C14H20N04Br: C, 48.57; H, 5.82; N, 4.05; found: C, 48.58; H, 6.15; N, 4.85%. Methyl 3- tert- butoxycarbonyl-5- (I, 3-dioxolan-2- yl) indolizine- 1 -carboxylate (3) To a suspension of 2 (51.9 g, 0.15 mol) in tetrahydrofuran (1500 ml) were added potassium carbonate (62.1 g, 0.45 mol) and methyl propiolate (15.12 g, 0.18 mol) and the mixture was stirred for 9 d at room temperature.The insoluble material was removed by filtration and the filtrate solution was concentrated in vacuo. The residue was applied on a column of silica gel and eluted with hexane-ethyl acetate (20 + 1-10 + 1). The fractions containing 3 were collected and evaporated to dryness to afford 35.0 g (67%) of 3: m.p. 138-139 "C (hexane); IR (Nujol, cm-I), 1690; NMR (CDCI3), b 1.63 (s, 9H), 3.7-4.05 (m, 4H), 3.91 (s, 3H), 6.85 (s, lH), 7.2-7.3 (m, 2H, H-6 and H-7), 7.78 (s, lH, H-2), 8.3-8.4 (m, lH, H-8). Analysis: calculated for C18H21NOh: C, 62.24; H, 6.10; N , 4.03; found: C, 62.19; H, 6.11; N, 4.21%.I-Mt.thoxycarbonylindolizine-5-curbaldehyde (4) A solution of 3 (20.82 g, 0.06 mol) in a mixture of 10% hydrochloric acid (60 ml) and tetrahydrofuran (600 ml) was refluxed for 6 h and the mixture was concentratcd to about one quarter of the original volume. Water was added to the mixture, which was then extracted with chloroform. The chloroform layer was washed with water and dried over sodium sulfate. The solution was evaporated and purified on a column of silica gel with hexane-ethyl acetate (10 + 1) as the eluent to afford 11.58 g (95%) of 4: m.p. 135-137 "C (methanol); IR (Nujol, cm-I), 1675, 1695; NMR (CDCI3), b H-2), 7.41 (dd, J 7.2 Hz, lH, H-6), 8.57 (br d, J 9 Hz, 1H, H-8), 8.84 (d, J 3 Hz, lH, H-3), 9.86 (s, 1H, CH).Analysis: calculated for CI1H9NO3: C, 65.02; H, 4.46; N, 6.89; found: C, 65.07; H, 4.55; N, 7.03%. 3.92(~,3H),7.18(dd,J9.7H~,lH,H-7),7.38(d,J3H~,lH, l-Methoxycarbonylindolizine-3,5-dicarbaldehyde (IDA) To a solution of 4 (12.18 g, 0.06 mol) in dry dimethylform- amide (1 16 ml, 1.5 mol) was added phosphorus oxytrichloride (17 ml, 0.18 mol) at 0 "C under an argon atmosphere, and the mixture was stirred for 1 h at room temperature. The reaction mixture was poured into water and made alkaline (pH 9.0) with 5% potassium carbonate solution. The solution was extracted with chloroform and the chloroform layer was washed with water and dried over sodium sulfate. The solution was concentrated to form a precipitate, which was collected by filtration to yield 7.41 g (53%) of IDA.The mother liquid was further concentrated to afford a residue, which was purified on a column of silica gel with hexane-ethyl acetate (5 + 1) as the eluent to recover 1.06 g (8%) of IDA: m.p. 164-165 "C (methyl acetate); IR (Nujol, cm-I), 1640, 1700; NMR (CDCLJ, 6 3.97 (s, lH), 7.5-7.65 (m, 2H, H-6 and H-7), 8.09 (s, 1H, H-2), 8.55-8.65 (m, 1H, H-8), 9.67 (s, 1H), 10.11 (s, IH). Analysis: calculated for C12H9N04: C, 62.34; H, 3.92; N, 6.06; found: C, 62.40; H, 4.04; N, 6.06%. Measurement of Absorption and Fluorescence Spectra of the Derivatized Product of Alanine with Ninhydrin, o-Phthalaldehyde (OAP) and IDA The derivatization reactions of alanine with ninhydrin and o-phthaldehyde were performed according to the litera- ture.13,14 Ninhydrin method Ninhydrin (200 mg) was dissolved in a mixture of 4 mol 1-1 sodium acetate buffer (pH 5.5,2.5 ml) and 2-methoxyethanol (7.5 ml). The solution (40 p1) was mixcd with a standard solution of alanine (40 pl) in a 0.6 ml polypropylene microcentrifuge tube. The mixture was kept for 15 min at 90 "C in a water-bath and cooled to room temperature. OPA method A solution of OPA was prepared by dissolution of OPA (28.8 mg) in methanol (1 ml) and 0.1 mol 1-1 sodium hydroxide (9 ml). A solution of 2-mercaptoethanol(2-ME) was prepared by dilution of 2-Me (18.7 PI) with 0.1 mol 1 - 1 sodium hydroxide solution (10 ml). Volumes of 20 pl of each of OPA, 2-ME and the standard solution of alanine was mixed and kept for 10 min at room temperature.IDA method A mixture of IDA solution (40 PI) and the standard solution of alanine (20 pl) was kept for 15 min at room temperature in the dark.Analyst, January 1995, Vol. 120 65 For measurement of absorption spectra, 10 pl of the reaction solution obtained by the ninhydrin method and the IDA method were diluted with water (3 mi) and used for the observation of absorption spectra. A portion of the mixture (10 pl) obtained from the OPA and the IDA method was also diluted with water (50 ml) and used for measurement of fluorescence spectra. Optimization of Derivatization For studies of the effect of pH on the course of the reaction of IDA and alanine, buffer solutions of pH 6-12 were used. Buffers of pH 6 9 were prepared by mixing 20 mmol I - ' sodium dihydrogenphosphate and 20 mmol 1-1 sodium tetra- borate.Buffers of pH 10-12 were prepared by mixing 500 ml of 20 mmol l-1 sodium dihydrogenphosphate and 500 ml of 20 mmol 1-1 sodium tetraborate and adjusting the pH with 1 mol 1-1 sodium hydroxide solution. The buffer solutions were mixed with an equal volume of ethanol and used as the reaction buffer. IDA (100 pg, 0.43 pmol) in a tube was dissolved in the reaction buffer (40 pl). After addition of the standard solution of alanine containing MNP (20 pl), the mixture was kept at room temperature in the dark. An aliquot of the mixture was subjected to the HPCE in the CZE mode. Comparison of the Stabilities of the Derivatized Products Between the Present Method and the OPA method A mixture of IDA solution (40 pl) and a standard solution of alanine containing MNP (20 PI) was kept at room temperature and an aliquot was introduced into the HPCE capillary at the intervals specified in Table 1.The reagent solution in water (20 pl) was also kept in the dark, and an aliquot was introduced into the capillary to confirm the stability. The reaction between OPA and alanine was performed in the same manner as described above. An aliquot was introduced into the capillary without dilution at the same intervals as described for the present method. With OPA-Ala, a detection wavelength of 340 nm was employed for the HPCE systems. Results and Discussion Synthesis of IDA l-Methoxycarbonylindone-3,5-dicarbaldehyde (IDA) was synthesized from 2-( 1,3-dioxoIan-2-yl)pyridine (1) as the starting material as shown in Scheme 1.Pyridine derivative 1 was treated with tert-butyl bromoacet- ate to give a pyridinium salt (2). This pyridinium salt was then condensed with methyl propiolate in the presence of potas- sium carbonate to afford methyl 3-terr-butoxycarbony1-5-( 1,3- dioxolan-2-y1)indolizine-1-carboxylate (3). Deprotonation and decarboxylation of the indolizine derivative (4) followed by formylation with dimethylformamide and phosphorus oxychloride gave IDA in the total yield of 30% from 1. From the proton NMR investigation of IDA, disappearance of the proton signals at 7.38 and 8.84 ppm in 4 showed subsitution of the 3-position, which is more reactive than the 4-position. The two signals observed at 9.67 and 10.1 1 ppm showed clearly the presence of two aldehyde groups.Thus, IDA was easily confirmed to have the structure shown in Scheme 1. Table 1 Stabilities of IDA, IDA-Ala and OPA-Ala. Each value represents the peak-area ratio of IDA, IDA-Ala or OPA-Ala to MNP obtained by the HPCE method in the CZE mode. The detection wavelength for IDA and IDA-Ala was 280 nm and that for OPA-Ala was 340 nm Time/h ~~ Sample 0.5 1 .0 4.0 6.0 24 IDA 0.51 0.51 0.52 0.51 0.49 IDA-Ala 1.71 1.73 1.70 1.65 1.68 OPA-Ala 4.89 4.84 4.28 3.99 2.33 F E I n CI 0 10 20 30 40 50 60 Time/min Fig. 1 Optimization of the reaction conditions. The reaction course was traced using HPCE. Analytical conditions for HPCE: carrier electrolyte. 20 mmol 1-1 phosphate-borate buffer (pH 10); capillary tube, 50 cm x 50 pm (i.d.1 fused silica; applied voltage, 20 kV; detection wavelength, 290 nm; hydrostatic injection (15 cm, 10s).pH: A, 6; B, 7 ; C, 8; D, 9; E, 10; F. 11; G, 12. 1 2 3 7 4 Scheme 1 IDA66 Analyst, January 1995, Vol. 120 a, 2 2 GI Optimization of Reaction Conditions Effect of p H and reaction time Alanine was used as the standard primary amino compound for the optimization study of the derivatization reaction. The effect of pH on the derivatization efficiency is shown in Fig. 1. The time course of the derivatization is also shown in Fig. 1 . At around neutral pH (6-8), the reaction proceeded slowly. On the other hand, the reaction was completed within 10 min under weakly basic conditions (pH 9 and 10). However, a decrease in the yield was observed at higher pH, probably owing to degradation of the product.At pH 10, the yield became the highest and constant after 10 min of reaction. 4 A (4 \ - Effect of ethanol as an organic additive For complete dissolution of the reagent and the products, an organic solvent should be added to the reaction buffer. Ethanol was chosen for this. The effect of the ethanol concentration on the reaction efficiency is shown in Fig. 2. By using buffer solutions containing ethanol at a concentration of 25-50%, the yields were almost constant. Spectral Characteristics of IDA-Ala The absorption spectra of the reagent (IDA) and the reaction solution of the IDA derivative of Ala (IDA-Ala) are shown in a F5 t \I 1.4 t 4.1 3.1 2.1 1.1 1.2 1.3 Buffer (pH 10) to ethyl alcohol ratio (v/v) Fig. 2 Other conditions as in Fig. 1 . Effect of the concentration of ethanol in the reaction buffer.A 200 300 400 500 Wavelengthhm Fig. 3 Absorption spectra of A, IDA-Ala and B, IDA in watcr. Fig. 3. IDA-Ala showed strong absorption at 280 nm (E = 3.31 x 10~1mol-~cm-~)and409nm(~ =2.18 X 10~lmol-lcm-l). On the other hand, the reagent did not show absorption maxima at these wavelengths. The reagent showed little absorption at 409 nm where IDA-Ala showed the strongest absorption. The fluorescence spectra of IDA and IDA-Ala are shown in Fig. 4.The excitation maxima (292 and 414 nm) observed from IDA-Ala were different from those observed for IDA (312 and 344 nm), as shown in Fig. 4(a). The emission spectra of IDA-Ala showed maxima at 482 nm [irradiated at 292 and 414 nm, Fig. 4(a) and ( b ) , respectively]. When IDA-Ala was irradiated at 414 nm, the fluorescence observed at 482 nm had little interference from the reagent.Stabilities of the Reagent and the Derivatized Product The stabilities of the reagent and the product (IDA-Ala) were investigated by HPCE. As a reference method OPA-Ala was also investigated in the same manner. Both the reagent (IDA) 150 > u) a, c .- c .- ,.Analyst, January 199.5, Vol. 120 67 and the product (IDA-Ala) were stable for at least 24 h , as shown in Table 1. On the other hand, the peak intensity of OPA-Ala rapidly decreased and became one third of the original value after 24 h. Calibration and Reproducibility The calibration graph for IDA-Ala showed good linearity over the range 0.0172-21.5 ymol ml-1 with a satisfactory correla- tion ( r = 0.999), as shown in Table 2.The detection limit of alanine was 5 nmol ml-1 at a signal-to-noise ratio of 3. Repeated analyses ( n = 5 ) of IDA-Ala at 10.8 and 21.5 pmol ml-1 gave excellent reproducibility with relative stan- dard deviations of 1.75% and 1.33%, respectively, as shown in Table 3. Comparison of the Present Method with Ninhydrin and o-Phthalaldehyde (OPA) Method The absorption spectrum of IDA-Ala is shown in Fig. 5 together with that of Ruhemann’s Purple chromphore (abbre- viated as ‘ninhydrin-Ala’ in this paper) obtained from the ninhydrin method. The absorbance of IDA-Ala at 280 nm showed almost a 2.5 times stronger absorption than that of ninhydrin-Ala at 570 nm. The absorbance at 414 nm was 1.5 times higher. The fluorescence spectra of IDA-Ala are shown in Fig.6 together with that of OPA-Ala. The fluorescence intensity of IDA-Ala at 482 nm (irradiated at 414 nm) was about three times stronger than that of OPA-Ala at 461 nm (irradiated at 342 nm). Table 3 Prccision of dcrivatization procedurc using HPCE in thc CZE mode. Each valuc rcprescnts the pcak-area ratio of IDA-Ala to MNP Alaninc conccntration/yrnol ml- 1 Sample 10.8 21.5 1 0.848 1.70 2 0.868 1.71 3 0.888 1.75 4 0.878 1.72 5 0.879 1.75 Mean’ 0.872(1.75%) 1.70(1.33%) ’ The valucs in parcnthcses arc rclativc standard deviations ( n = 5 ) . a, C 0.5 9 n 200 400 600 800 Wavelengthhm Fig. 5 and thc prcscnt method. A. IDA-Ah; and B, ninhydrin-Ala. Comparison of the scnsitivity bctwccn thc ninhydrin method Separation of the Labelled Amino Acids by Micellar Electrokinetic Chromatography After derivatization of a mixture of standard amino acids by the present method, the reaction mixture was analysed by HPCE. Proline and hydroxyproline were not derivatized by the present method; separation of the other amino acids is shown in Fig.7. Tryptophan and threonine did not give clear peaks. This was probably due to oxidative degradation under the alkaline conditions, as occasionally observed for the alkaline degrada- tion of glycoproteins. Separation by simple zone electrophor- esis did not afford a satisfactory separation especially for the group Phe, Lys and Arg (data not shown). Addition of SDS and a small volume of methanol enhanced the resolution of these amino acids, although the separation among Glu, Asp and Gly became slightly worse.The two aldehyde groups in the IDA molecule are not equivalent. Only one of the aldehyde groups (probably that at the 5-position owing to its higher reactivity) seems to react with an amine to form a Schiff base. The Schiff base produced becomes stable by formation of a conjugated diene with the other aldehyde group at the 3-position in solution. Further, the conjugated dienes produced may show strong batho- chromic shifts in their spectra. Although the Schiff base seems stable in solution, the attempt to isolate the reaction product was not successful owing to its instability in the isolated state. 200 A 350 500 650 Wavelengthhm Fig. 6 Comparison of thc sensitivity in spectral characteristics bctwccn OPA method and the present method.A, IDA-Ala excited at 414 nrn; B, IDA-Ala cxcitcd at 292 nrn; and C, OPA-Ala excited at 342 nm. 0 10 Time/min 20 Fig. 7 Scparation o f a mixture of 15 amino acids by MEKC. Conditions: carricr clcctolytc, 20 rnrnol I-‘ SDS-40 mrnol I-‘ borate-phosphatc buffer (pH 7.0) containing rncthanol at a conccn- tration of 3% v/v; applied voltagc, 20 kV; other conditions as in Fig. 1. 1. Scr; 2, Cys; 3 , Glu; 4. Asp; 5. Gly; 6. A h ; 7, Val; 9, His; 10, Mct; 1 1 , Ilc; 12, LCU; 13. NH,; 14. Phc; 15, Lys; 16. Arg, 17, Tyr. Pcaks 8 and 18 arc due to the reagent. Each amino acid concentration uscd was 1.43 vimol rnl-1.68 Analyst, January 1995, Vol. 120 Conclusion The indolizine dialdehyde derivative IDA was prepared and used for labelling of amino acids. The derivative has strong UV absorption at 280 and 409 nm and also fluorescence at 482 nm (with irradiation at 282 or 414 nm). The reagent did not show interferences in the determination of the amines. Using the present derivatization method, a mixture of amino acids was derivatized and separated by HPCE in the MEKC mode. The analysis is performed within 40 min including derivatiza- tion and separation. References 1 Barth, H. G., Barber, W. E.. Lochmuller, C. H., Majors, R. E., and Regnier, F. E., Anal. Chem., 1988, 60, 387R. 2 Roth, M., Anal. Chem., 1971, 43, 880. 3 Benson, J. R., and Hare, P. E., Proc. Natl. Acad. Sci. USA, 1975, 72, 619. 4 Lindroth, P., and Mopper, K., Anal. Chem., 1979, 51, 1667. 5 6 7 8 9 10 11 12 13 14 Wilkinson, J . M., J. Chromatogr. Sci., 1978, 16, 547. Carpino, L. A., and Han, G. Y.. J. Org. Chem.. 1972.37.3404. Chervet, J. P., Ursem. M., Salzmann, J. P . . and Vannoort, R. W., J . High. Resolut. Chromatogr., 1989. 12, 278. Heiger. D. N.. in High Performance Capillary Electrophoresis, report, Hewlett-Packard, Waldbronn. 1992, p. 100. Wang, T.. Aiken, J . H., Huie, C. W., and Hartwick, R. A., Anal. Chem., 1991, 63, 1372. Tsuda. T., Sweedler, J . V.. and Zare, R. N., Anal. Chem.. 1990, 62. 2149. Miki, Y., Hachiken, H., Yoshikawa, M., Takemura, S., and Ikeda, M., Heterocycles, 1991, 32, 655. Bradsher, C. K., and Parham, J. P., J. Org. Chem., 1963,28,83. Moore, S., and Stein, W. H., J. Biol. Chem., 1954, 211, 907. Morineau, G., Azoulay, M., and Frappier, F., J. Chromatogr., 467, 209. Paper 4102577F Received May 3, 1994 Accepted September 9, 1994

ISSN:0003-2654    

DOI:10.1039/AN9952000063

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 69 Evaluation, Mechanism and Application of Solid-phase Extraction Using a Dithiocarbamate Resin for the Sampling and Determination of Mercury Species in Humic-rich Natural Waters HAkan Emteborg, Douglas C. Baxter, Michael Sharp and Wolfgang Frech Department of Analytical Chemistry, UnieH University, S-901 87 UmeH, Sweden A method for the simultaneous determination of mercury species at sub-ng 1-1 levels in humic-rich natural waters was evaluated. Methyl- and inorganic mercury were preconcentrated from a freshly collected humic-rich water sample following batchwise addition of pre-cleaned dithiocarbamate resin. The sample was shaken for 8-22 h, allowing mercury complexed by the humic substances to be transferred to and enriched on the resin. A mathematical model describing the recovery as a function of shaking time was developed and the model derived indicated that the rate-determining step for collection of mercury species is the release of these species from two major groups of sites present on the humic substances.The dithiocarbamate resin was collected by filtration and transferred to a column which was then installed in a closed flow injection system where the mercury species were eluted with an acidic thiourea solution. Next, the mercury species were extracted into 500 p1 of toluene as diethyldithiocarbamate complexes and butylated with a Grignard reagent. The butylated forms were injected (G13 pl) on to a gas chromatograph equipped with a non-polar capillary analytical column, separated and detected at 253.7 nm by atomic emission spectrometry after excitation in a microwave-induced helium plasma.The recovery of methylmercury was 82.6 k 5.6% at a dissolved organic carbon (DOC) concentration of 44 mg 1-1, and increased to over 92.5% at lower DOC levels. For inorganic mercury, the recovery was poorer, 45.8 k 3.3% at a DOC concentration of 12 mg 1-1. The detection limits for the methyl- and inorganic mercury were 0.04 and 0.28 ng 1-1 (3s criterion), respectively, for a 1 1 sample, the latter being limited by the poorer recovery and contamination problems. The accuracy of this mercury speciation method was assessed, in part, by means of a laboratory intercomparison and use of an independent analytical technique. Keywords: Methylmercury and inorganic mercury; humic substances; natural waters; capillary gas chromatography with micro wave-induced helium plasma atomic emission detection; dithiocarbamate resin; kinetic study Introduction It has been recognized for decades that humic and fulvic acids (HA/FA) play an important role in the complexation and transportation of metal ions such as Fe3+, Hg'+, PW+, Cu2+ and AP+ in freshwater systems.1-3 Mercury and iron are most strongly complexed by HA/FA, and consequently the avail- ability of these metal ions for isolation and determination is strongly affected by even low concentrations of HA/FA in a water sample.' The complexing capability of HA/FA origi- nates from at least two major groups of sites which can be discerned using various techniques.4 In natural waters the loading of mercury species is very low, typically at the ng 1-1 level.In this context, it is therefore important to note that the fraction of free mercury is very low in comparison with the total concentration. Most sampling procedures for mercury determinations in natural waters employ additions of preser- vatives to ensure the integrity of the samples. Examples of preservatives that have been used are 1% hydrochloric acid, 0.025% K2Cr207, 2% nitric acid, freezing, trivalent gold together with dilute nitric acid and even humic substances.611 It is conceivable that preservation might lead to extraneous additions of mercury in addition to alteration of the speciation in the sample,7 e.g., methylmercury may be decomposed under oxidizing conditions as reported by Leermakers et al.,7 although the literature is contradictory on this point.5--7 Inorganic mercury can be reduced to metallic mercury and evaporate or form highly stable amalgams unless oxidizing conditions prevail. Clearly, optimum preservative conditions for both inorganic and methylmercury seem to be difficult to obtain. As far as sample containers are concerned, those made from amber-coloured borosilicate glass or poly- tetrafluoroethylene (PTFE) have generally shown the best properties for storage of ultra-trace mercury solutions."-7 For the determination of total mercury in natural waters, numerous methods have been described. 12.13 However, to date only two methods suitable for the isolation and determi- nation of methylmercury in humic-rich waters have been reported,8.14 the major features of these being outlined below.Speciation of methylmercury in humic-rich waters can either be based on distillation and derivatization prior to gas chromatography with atomic fluorescence spectrometric detection (GC-AFS) as described by Bloom and co- workers,8.15 or, following extraction, by GC with electron- capture detection (ECD) as described by Lee and Mowrer.14 In the method using derivatization,Is a distillation step is employed as the ethylating reagent used (sodium tetraethyl- borate) is consumed by matrix components when added directly to humic-rich waters. Although the distillation step is time consuming, requiring 5-7 h for 50 ml samples, higher and more reproducible recoveries for methylmercury are obtained compared with liquid-liquid extraction.Methylmercury in the distillate is then ethylated and purged from the solution on to a carbotrap. Note that inorganic mercury can neither be distilled nor determined simultaneously using this method. After the purging is completed, the carbotrap is installed at the injector end of the packed analytical column. Methylethyl- mercury is desorbed thermally, separated, pyrolysed and detected using AFS. The major advantage of this method is the extremely good detection limit for methylmercury of 0.006 ng 1 - 1 using a 50 ml sample volume.15 Lee and MowrerlJ70 Analyst, January 199.5, Vol. 120 preconcentrated methylmercury from humic-rich waters on to a sulfhydryl cotton fibre (SCF)-gauze mounted on a glass frame which was immersed in a 4 1 reservoir.The methylmer- cury was eluted from the gauze with hydrochloric acid and enriched in a column containing SCF wool after adjustment of the eluate to pH 4. Methylmercury was then eluted again with hydrochloric acid, extracted into benzene and determined, as the chloride, by capillary GC-ECD. By using a 4 1 sample and 400 pl of benzene in the final extract, a detection limit of 0.05 ng 1-1 was achieved. This batch-column procedure provided a variable, non-quantitative recovery of methylmercury, and as a result of the use of a non-specific (with respect to mercury) detector the chromatograms exhibited numerous interfering peaks. Here, we propose a sampling, enrichment and work-up procedure which is free from additions of preservatives and therefore does not alter the chemical composition of the water sample dramatically.This is achieved by addition of a highly mercury-selective dithiocarbamate resin directly to a freshly collected humic-rich water sample. In this way an in situ enrichment of mercury species from the water sample is accomplished. Following preconcentration, the resin is trans- ferred to a micro-column and the mercury species are eluted with an acidic thiourea solution. After neutralization, the eluted species are extracted into toluene as diethyldithiocar- bamate complexes and butylated with a Grignard reagent. These derivatives are separated and determined by capillary GC with microwave-induced plasma atomic emission spec- trometric detection (GC-MIP-AES). Detection limits of 0.04 ng 1-1 for methylmercury (10 p1 injections) and 0.28 ng 1 - 1 for inorganic mercury can be obtained using a 1 1 water sample and 500 pl of toluene in the final extract.An interesting review by Langford and Gutzmanl6 describ- ing the use of kinetic data for metal speciation has been published concerning N?+, Cu2+, Fe3+ and Al3+ in systems containing natural ligands such as fulvic acids and hydrous iron(m) oxides. Studies are performed by adding an excess of colour-forming reagent for complexation of the analyte ions and spectrophotometric detection is used to follow the time course of the complexation reaction. Added metal concentra- tions are typically in the pmol 1 - 1 range, which are far above environmental levels for most metals. Applying pseudo-first- order kinetics, non-linear regression and/or Laplace trans- forms, conclusions on the number of groups of sites present in the natural ligands can be made, as well as values of the rate constants for dissociation of the metal monitored from these sites.Kinetically crucial parameters such as pH, temperature and ionic strength must be controlled. To date only a few papers concerning kinetically controlled speciation have been Table 1 Operating conditions for the gas chromatograph. Carrier gas, helium; flow rate. 18 ml min-I; injection volumes, S13 p1 Column oven- Initial column temperaturePC 50 Initial hold time/min 1 Ramp ratePC min-1 40 Isothermal hold timdmin 1 Final column temperature/"C 180 Injector temperaturePC 180 Injector Reluyi- Initial relay -1 Switch delay timdmin 2.3 Final relay + I Method complete end time/min 5.25 * Initial relay ( - I ) , column effluent vented; final relay ( + I ) .column effluent directed to microwave-induced plasma. published for metal concentrations of environmental rel- evance. 17 No statement was made concerning the way in which additions of buffer and ionic medium may affect the specia- tion. 16.17 I t was mentioned, however, that the values obtained for the rate constants are only true relatively speaking in comparison with the natural system. It was also stated that methods for the examination of kinetic speciation with high resolution and high sensitivity are extremely hard to design. It is clear that most of the literature to date on metal speciation in natural systems concerns systems in equilibrium.A kinetic study of a natural system may give additional, important information on the intermittent presence and availability of certain species of interest during changing conditions. This paper demonstrates the possibility of performing kinetically controlled speciation studies of an analyte species of great environmental concern at the ng-* level, namely the methylmercury ion. It should also be possible to extend these studies to the inorganic mercury ion in the same matrix. Experimental Instrumentation The instrumentation has been described in detail pre- viousIy,18-2* only slight alterations having been made, and thus a brief description will be given here. A Varian (Palo Alto, CA, USA) Model 3300 gas chromato- graph with a 15 m x 0.53 mm i.d.wide-bore fused-silica column coated with a 1.5 vm non-polar dimethyl polysiloxane stationary phase (DB-1, J & W Scientific, Rancho Cordova, CA, USA) was used throughout. The analytical column was coupled to an atmospheric pressure microwave-induced helium plasma sustained in a Beenakker TMolo cavity (AHF Ingenieurburo, H. Feuerbacher, Tiibingen, Germany) via a pneumatic four-way valve (Valco, Houston, TX, USA) automatically controlled by the gas chromatograph. After the solvent peak had passed, the valve switched, thereby re- routing the column effluent to the plasma through a 0.25 mm i.d. deactivated fused-silica capillary transfer line (J & W Scientific). The transfer line from the four-way valve passed through the right wall of the column oven, via a heated interface maintained at about 150 "C, into the MIP torch assembly as detailed elsewhere.22 Emission from mercury and carbon atoms excited in the MIP was detected at the 253.652 and 247.857 nm lines, respectively, using an MPD 850 multichannel spectrometer (Applied Chromatography Systems, Luton, Bedfordshire, UK).The signals were moni- tored with a Sekonic (Tokyo, Japan) SS-250-F dual-pen recorder and Varian Star PC software for integration and evaluation of chromatographic data (mercury channel only). Operating conditions for the gas chromatograph are given in Table 1. For preconcentration of mercury species from water, two different enrichment techniques involving the same type of dithiocarbamate resin were used, viz., a column method and a batch method.The column method has been described previously. 18 For both methods, elution of mercury species from the resin was achieved with an acidic thiourea solution using a flow injection system (FIAS-200, Perkin Elmer, Uberlingen, Germany). When using the batch method, the resin was collected in the column, which was then simply installed in the flow injection system. Two different columns were used for elution in the batch-method experiments. For analytical applications where approximately 230 mg of resin (wet mass) were required for 1 1 samples, a glass column, 50 X 3 mm i.d., volume 350 p1 (Omnifit, London, UK), was used. For sample volumes ~ 2 5 0 ml where only approximately 45 mg of resin were required, and also in column-method applica- tions, a smaller 60 p1 volume Kel-F column (poly-Analyst, January 1995, Vol.120 71 chlorotrifluoroethylene; Malmo Fluorocarbon, Malmo, Sweden) was used. Additionally, the laboratory-constructed Kel-F column (commercially available at the first author's address; 9.7 x 2.8 mm i.d.) is equieped with frits made from Vyon-P (polypropylene; PIAB, Akersberga, Sweden) in order to retain the dithiocarbamate resin in the column. In initial studies on the efficiency and elution profiles for methylmercury and inorganic mercury using the larger glass column, total mercury measurements were performed by electrothermal atomic absorption spectromety (ETAAS). For these experiments, a Perkin-Elmer 4100 ZL instrument was used. Instrumental and operating conditions were as previ- ously de~cribed.~3 For total mercury measurements, a Pt-lined graphite atomizer cold vapour (CV) ETAAS system was used, details on the instrumentation and sample preparation tech- nique having been reported previously. 1224 Reagents and Materials All the chemicals used were of analytical-reagent grade and in some instances further purified to lower the reagent blank (see below).For the experiments described here, sample and reagent containers were 250-1000 ml PTFE bottles (Nalgene; Nalge, Rochester, NY, USA) or 200 ml polysulfone vessels (Nalgene) for total mercury determinations using CV- ETAAS. For collection of the resin added to water samples, a 1 1 receiver with a 0.5 I polysulfone filter holder (300-4100 Nalgene) was used. Durapore SV-type 5 ym pore size filters were obtained from Millipore (Milford, MA, USA) and were acid washed prior to use as described by Hiraide et al.25 All bottles and filter holders were acid washed using distilled sub-boiling nitric acid and deionized water obtained from a Milli-Q apparatus (Millipore) ('Milli-Q water') and afterwards rinsed with copious amounts of Milli-Q water.All standard solutions were stored in glass vessels at 4 "C and were prepared gravimetrically . Stock standard solutions of methylmercury chloride (MeHgCI, 303.6 mg 1 - 1 Hg) (Merck, Darmstadt, Germany) and ethylmercury chloride (EtHgCI, 500 pg 1 - 1 Hg) (K & K Laboratories, Plainview, NY, USA) were prepared by dissol- ving the salts, for MeHgCl in Milli-Q water, and for EtHgCl in ethanol to 227.3 mg 1-1 which was further diluted to 500 pg 1-1 with Milli-Q water.Inorganic mercury solutions were pre- pared by diluting a 1000 mg 1-1 certified standard (HgCI,; Analytical Standards, Kungsbacka, Sweden), which was also used to verify the stability of the MeHgCl and EtHgCl stock standard solutions at regular intervals. Dilute mercury solu- tions were prepared from the stock standard solutions immediately before use. Sodium hydroxide (semiconductor grade; Aldrich, Milwaukee, WI, USA) was placed in 25 ml platinum crucibles (Adelmetall, Stockholm, Sweden) and heated for 12 h at 400 "C. A solution of 1 mol 1-1 NaOH was prepared by dissolving the salt in an appropriate volume of Milli-Q water. Thiourea (ACS grade, Merck) was dissolved in Milli-Q water (5% m/m) and stored in a 200 ml polysulfone bottle containing Chelite S resin (Serva, Heidelberg, Germany) to purify the solution with respect to mercury.18-74 Aliquots of this thiourea were acidified with an appropriate volume of concentrated nitric acid (sub-boiling distilled in an all-quartz apparatus; Hereaus Quarzschmelze, Hanau, Germany) to pH 1 .O immediately before use. The complexing agent, 0.5 mol 1-1 diethyldithiocarbamate (DDTC), was prepared by dissolving the sodium salt (ACS 99+ grade, Aldrich) in Milli-Q water. This solution was purified from mercury contamination by pumping two cycles through an ethanol-washed large-capacity column containing a CI8 bonded-phase silica packing material.26 The resin containing the dithiocarbamate groups was prepared from macroporous glycidyl methacrylate spheres (Spheron E 300, 0.04-0.063 mm; Lachema, Brno, and Chemapol, Prague, Czech Republic) and synthesized ammonium dithiocarba- mate.27 Approximately 11 g of Spheron E 300 containing 0.3 k 0.1 mmol epoxide groups per gram of resin (dry mass) were activated with 8 g of ammonium dithiocarbamate dissolved in 60 ml of Milli-Q water in a round-bottomed flask revolving for 20 h on a rotary evaporator, immersed in an oil-bath kept at 50 "C.The degree of funtionalization of the epoxide groups was assessed by determining the sulfur content in the resin in a commercial laboratory. Elemental analysis revealed a degree of functionalization of 80% with a 25% loss of dithiocarba- mate groups over 26 months, indicating a good long-term stability of the resin. The activated resin was rinsed with dilute nitric acid and Milli-Q water and then packed in the Kel-F column or the larger glass column.Resin for batch-method experiments was purified by pumping approximately 15 ml of acidic thiourea solution through the glass column. The column was then emptied on an acid-washed Millipore filter-paper. When handled outside the column, all manipulations with the purified resin were performed in a clean bench (class 100; Ceag Schirp, Reinraumtechnik, Selm-Bork, Germany). The resin was divided using acid-washed PTFE tubing and placed in acid-washed 2 ml PTFE sample cups (Perkin Elmer). The procedure of weighing the resin was necessary to ensure that all material added could be entirely collected in the columns used for batch-method applications. To the sample cups containing the purified and weighed resin, a few drops of Milli-Q water were added to prevent the resin from drying.Finally, the sample cups were sealed with Parafilm (American National Can, Greenwich, CT, USA). Procedures Preconcentration and elution All water samples were handled in a clean bench and sealed with Parafilm prior to further manipulation in the laboratory. Before preconcentration with the column method, 3 ml of acidic thiourea solution were pumped through the column followed by a rinsing step (Milli-Q water) to remove all mercury species from the column and tubing. Details on operating conditions for the flow injection system and preconcentration procedures for the column method can be found in work reported earlier. 18 The collected mercury species were then eluted with typically 0.9 ml of acidic thiourea solution into 10 ml screw-capped glass centrifuge tubes.This volume of acidic thiourea was found to be sufficient for elution when using the smaller Kel-F column as determined in previous work. However, when using the glass column, larger volumes of the acidic thiourea solution were required to achieve 100% desorption of the enriched mercury species from the dithiocarbamate resin. Typically >4 ml of acidic thiourea solution were required, as can be seen in Fig. 1. Enrichment of mercury species using the batch method was performed by adding the purified resin directly to the water sample. The sample cups were rinsed into the collection vessels with a few drops of Milli-Q water to avoid losses of the resin in this step.The sample flasks were then shaken slowly on an automatic shaker (IKA-VIBRAX-VXR, Labassco, Partille, Sweden), set such that the individual spheres of the resin were suspended in the water column and not lying on the bottom of the flask. For collection of the resin an acid-washed 0.5 1 polysulfone filter holder with a 1 1 receiver was used (Nalgene). With the concave side up, the 5 ym pore size filters (Millipore) were mounted in the filter holder. After filtration of the 200-1000 ml water samples containing the resin, the72 Analyst, January 1995, Vol. 120 filter-paper covered with resin was transferred to a small plastic funnel with the convex side of the filter towards the funnel wall using plastic tweezers. The columns were connec- ted to the funnels via silicone-rubber tubing of 10 mm length.The columns themselves were attached to connectors on the lid of the filter holder via another piece of silicone-rubber tubing. Suction was applied to the receiver, and the resin was easily transferred from the filter-paper to the column with a few drops of Milli-Q water. Note that all manipulations were made in the clean bench. After collection, a frit was placed in the open end of the column and the lid was screwed in position. Next, the column was installed in the flow injection system and the mercury species were eluted as described above. Extraction and derivatization In the centrifuge tube, containing the eluted mercury species in acidic thiourea solution, 110 1-11 of 1 moll-’ NaOH solution per ml of acidic thiourea were added, followed by 1 ml of pH 9 borate buffer (Merck) to obtain a suitable pH value for further extraction.As complexing agent, 1 ml of 0.5 mol 1-1 DDTC was added. Next, 0.5 ml of toluene (distilled-in-glass quality; Burdick and Jackson, Muskegon, MI, USA) was added and the tube was shaken for 10 min. Toluene was found to provide significantly better recoveries than less polar solvents such as hexane. It is also difficult to withdraw a hexane phase with a micropipette owing to the low viscosity and high volatility of hexane. Pasteur pipettes must therefore be used and the volume withdrawn must be determined gravimetrically . A 0.4 ml portion of the toluene phase containing the DDTC-mercury complexes was withdrawn with a micro- pipette (Gilson, Villiers Le Bel, France), transferred into another 10 ml centrifuge tube standing in an ice-water bath and 200 1-11 of 2.24 mol 1-1 butylmagnesium chloride in tetrahydrofuran (Aldrich) were added.The mixture was shaken gently and allowed to stand for 5-10 min, after which 300 1-11 of 0.6 mol 1-1 hydrochloric acid were added to quench the excess of Grignard reagent. After centrifugation at 5000 rpm (3200g) for approximately 5 min (centrifuge from Wifug, Bradford, Yorkshire, UK), the organic phase was withdrawn using a glass Pasteur pipette and placed in a 2 ml screw-capped septum-equipped vial. The samples were stable for several weeks when stored below -18 “C. Collection of water samples To test the recovery of mercury species added to natural water samples [total organic carbon (TOC) concentrations ranging from 0 to 44 mg 1-1 as determined by the Swedish standard 100 80 h $? v 60 c - Q) c 5 40 E a 20 I I I I I Volume of acidic thiourea/ml 0 0.5 1.5 2.5 3.5 4.5 Fig.1 Elution profiles for, A. methylmercury and, ., inorganic mercury from 230 mg of dithiocarbamate resin packed in Omnifit glass column. method SS 02 81 991 after enrichment on the dithiocarbamate resin using the batch method, a variety of water samples were collected. Fresh water (TOC = 0 mg 1-1) was taken directly from a tap in the laboratory (total mercury <0.6 ng 1 - 1 as determined by CV-ETAAS”). Sub-surface rivulet water was collected in 1 1 acid washed PTFE bottles from Savariin, 15 km north of Umei (pH 7, TOC 14 mg I-]), in September and November 1993.Samples were also collected from a stream (pH 4, TOC 35-44 mg 1-1) draining a marsh surrounded by low sedge peat near Svartberget Forest Research Area, 70 km west of Umei close to the River Vindel on two occasions in September and October 1993. The bottles were immersed below the surface (approximately 15 cm) and filled to the mark. An acid-washed open-topped ‘lid’ covered with a plastic gauze (1 mm2 holes) was used to prevent larger particles from being collected. At this stage, the resin was added directly to the samples destined to be analysed with respect to methyl- and inorganic mercury. Samples to be used in adsorption profile experiments (recovery as a function of shaking time, see Fig. 3) were spiked with 5 ng 1-1 of methylmercury at the sampling site, allowing the added mercury species to equilib- rate with the sample for 1-8 h, after which the resin was added. Samples were also taken without any additions and used for the determination of total and dissolved organic carbon (TOC/DOC) and diluted with tap water to give ‘semi-natural’ water samples having various natural organic ligand concen- trations.For the determination of total mercury, samples were preserved with 5 ml of concentrated nitric acid and 5 ml of 1% K2Cr207 solution. 12 Measurements of blanks For column-method applications, blank measurements were performed as described in a previous paper.’* For the batch-method application, cleaning, handling and manipula- tion of the resin in contact with the sampler cup, PTFE sample bottle, PSF filter holder, filter, funnel and finally the column were investigated with respect to blanks.Blanks for methyl- mercury were below the detection limit of 0.04 ng 1 - 1 expressed as Hg using a 13 1-11 injection. Inorganic mercury blanks were higher owing to larger risks of contamination. When mercury species were eluted with 4 ml of acidic thiourea solution from the larger glass column, a blank, originating from the reagents, corresponding to 1.5 ng 1-1 in a 1000 ml sample was obtained. Results and Discussion When spiking HNFA-containing samples for performing recovery tests, there is an obvious risk that the added species will saturate the available binding sites present on the humic substances. Further, the added species may attach to more labile functional groups than those occupied by the incipient species.Therefore, it is desirable that the amount of added species is low and sufficient time is allowed for equilibration with the sample to avoid overestimation of the analyte recovery. In the experiments described here, the water samples were spiked with very small amounts of mercury species to avoid such problems. An additional effect has been observed by Powell and Florence,’8 that of displacement of metals from an Fe-colloid-humic substance when adding mercury in the 40-200 pg 1-1 concentration range, for the in situ plating of thin mercury film electrodes for anodic stripping voltammetry. Even though metal ions such as sodium are not particularly strongly bound to humic substances, medium addition may nevertheless alter the intrinsic distribution (i.e., the labile fraction will increase). Further, increasing salinity (adding ionic medium) causes humic substances to aggregate and precipitate in estuaries.2 If this happens in the modelAnalyst, January 1995, Vol. 120 73 100 80 h v 8 2 60 > &40-' [r 2o rn solution, the results obtained will clearly not reflect the situation in the humic-rich water. Previous work using the dithiocarbamate resin column method has shown that the recoveries of mercury species from spiked marsh water samples were extremely low and irrepro- ducible although the method was found to be very efficient for the enrichment from sea- and fresh waters.18 The reason for the low recovery from marsh waters was believed to be slow mass transfer of humic-bound mercury compounds to the dithiocarbamate groups of the resin. In Fig.2 it can be seen that the content of DOC, which largely consists of humic (HA) and fulvic acids (FA) in these waters,2,29 has a very strong influence on the recovery of added methyl- and inorganic mercury. This makes the column method unsuitable for samples other than sea-, rain, spring and fresh waters with a very low content of DOC. It can also be seen that the recovery for inorganic mercury is generally lower than for methylmercury with increasing concentrations of DOC. More- over, Fig. l demonstrates that inorganic mercury is more strongly retained than methylmercury by the dithiocarbamate resin, showing that inorganic mercury is more tenaciously bound to the resin.The conclusion must be that methylmer- cury-FA/HA complexes are weaker than the corresponding inorganic mercury complexes, conferring on methylmercury a higher degree of availability in humic-rich waters. This difference may probably be attributed to differences in the charge and size of the Hg2+ and MeHg+ ions. Preliminary investigations showed that 40% of added methyl- and ethylmercury could be recovered after 4 h with the resin in passive contact with a tap water sample. Obviously shaking of the sample flasks was essential, especially for marsh water where HA/FA competes with the resin for mercury species. The recoveries of methyl- and inorganic mercury were investigated as a function of shaking time using the batch- method sampling procedure and GC-MIP-AES separation and detection of mercury species.As can be seen from Table 2, recoveries are acceptable and reproducible for methylmer- cury. For inorganic mercury the recoveries are poorer although the reproducibility is acceptable. The reason for A A - - - A I I I I I I t I I I I I q l m I t 0 1 2 3 4 5 6 7 8 9 1 0 DOC/rng I-' Fig. 2 Recovery of 10 ng 1-1 of, A, methylmercury and, ., inorganic mercury as a function of dissolved organic carbon (DOC) concentra- tion using the column method. these lower recoveries can be found in the higher stability of the inorganic mercury-HA/FA complexes as noted above. Two replicate recovery tests performed for inorganic mercury in a water with a DOC concentration of 44 mg 1-1 gave a value similar to that reported in Table 2 for inorganic mercury recovered from a water with 12 mg 1-1 DOC.Additionally, recovery tests for methylmercury after different shaking times and for various concentrations of DOC were performed to yield adsorption profiles (see Fig. 3). The samples were spiked at the collection site, allowing the added methylmercury to equilibrate and form HNFA complexes for 1-8 h prior to addition of dithiocarbamate resin. Recoveries were high after equilibration for 8 h, demonstrating that methylmercury is not lost from the sample at concentrations of a few ng 1-1 in waters with a high content of HA/FA. Indeed, the literature reports using humic acid as a preservative for trace mercury(i1) solutions, supporting our observations.' 1 These results show that addition of any preservative other than the resin itself is unnecessary, hence the methodology suggested is advan- tageous for sampling of mercury species from waters rich in HA/FA.Mathematical Description of Adsorption Profiles Obtained for Methylmercury from Waters Rich in Humic Substances During the work obtaining the time-concentration data (adsorption profiles) the temperature was 20 k 2 "C and the pH was 4.0, controlled by the intrinsic buffering capacity of the HA/FA present in the samples. No medium was added as the intention was to perform the measurements as close to natural conditions as possible, and further, additions of ionic medium may lead to mercury contamination. Finally, altera- tion of speciation cannot be overlooked even though it is unlikely to take place in this case.It should also be mentioned that the number of experimental data points used was more than indicated in Fig. 3; typically 6-9 points were measured. This may be a limitation but further work is under way to increase the sample throughput rate. The adsorption profiles for methylmercury from waters with two concentrations of DOC (see Fig. 3), show that larger amounts of DOC suppressed the rate of uptake on the dithiocarbamate resin (DTC-). With the objective of describing the curve form mathematically, a model was derived based on kinetic Table 2 Recovery of 5 ng 1-1 of methylmercury and 25 ng 1-1 of inorganic mercury added to 100-200 ml water samples using the batch method. Injection volumes, 6 p1. Data expressed as mean values 2 one standard deviation for the number of replicates given in parentheses Shaking Species Sample Sampling site time/h DOC/mglkl pH Recovery (%) recovered Tap water Umed University 22 -0 7 92.5 k 4.5 (4) MeHg+ Rivulet water Savarin 22 12 6.8-7 96.0 k 8.8 (4) MeHg+ Marsh water Svartberget 22 44 4-4.5 82.6 k 5.6 (4) MeHg+ Rivulet water Savarh 20 12 7 45.8 k 3.3 (3) Hg2+74 Analyst, January 199.5, V d .120 considerations (see Fig. 4). It was also expected that the model might provide an explanation for the slower uptake of mercury from waters rich in humic substances. The model is based on the following assumptions: (i) that the rate-determining step lies in the release of methylmercury from two major groups of sites on the humic substances; (ii) that any methylmercury which is liberated from the humic substances immediately reacts with the DTC groups on the resin through a very fast reaction (k3); and (iii) that the concentrations of complexing groups on the resin and on the free ligands are in large excess and fairly constant throughout the experiment so that pseudo- first-order kinetics can be applied.One group of sites exhibits slow release of methylmercury (strong sites) from HNFA with a low rate constant ( k 2 ) resulting from the presence of phenolic,4 sulfur and nitrogen sites. The other group exhibit- ing a faster release of methylmercury-ligand complexes (weak sites) with a higher rate constant ( k l ) , consists mainly of carboxylic sites.4 This assumption is supported by the fact that the elemental composition of aquatic HA/FA comprises as much as 10% of sulfur and nitrogen relative to the oxygen content.* It is also recognized that the elemental composition of aquatic HAEA with respect to sulfur and nitrogen is characteristic of different environments, varying between about 7% in terrestrial and 0.5% in marine fulvic acids.3.30 Recovery (%) = 100[aAt + (1 - a)&] (1) where At = 1 + [k3exp(-k,t) - klexp(-k3t)/(k,-k3)] 1x denotes the fraction of the total ligand amount (= 1) exhibiting a high rate constant for release (weak sites) of methylmercury from the humic substances.Assuming that k3 >> kl = k2 makes it possible to simplify eqn. (1) as follows: Recovery (%) = 100[ 1-aexp(-klt) - (1-a)exp(-k2t)] (2) This,approximation of eqn. (1) is valid even for k3 values only five times greater than k l and k 2 , as shown by numerical simulations.Different values of a ranging from 0.60 to 0.85 were evaluated using eqn. (2) and an a of 0.70 gave the best overall fit to experimental data obtained at various DOC concentrations, indicating that 70% of all sites exhibit fast release of methylmercury. Results obtained using a gel filtration method for the study of metal binding by lacustrine fulvic acids and related compounds have shown the presence d + M e H g D T C 1 3fMeHg+ + DTC- MeHgL" I Analytical window Fig. 4 Model for the releasc of rncthylmercury (MeHg+) from two different groups of sites present in humic substances ( L ' , L ) and subsequent binding to the dithiocarbamate resin (DTC-). Table 3 Values obtained for k l , k3 and r2 using eqn. (2). Error tcrms for k l and k2 are given as 95% confidcncc intervals DOC/mgI-l k,/10-4s-1 k?/lO-'s-l r2 8.8 3.0 k 0.6 2.3 ? 0.7 0.9923 17.5 1.3 f 0.2 0.9 k 0.3 0.0932 35.0 1.1 k0.2 2.4 k 0.9 0.9946 44.0 0.9 k 0.1 1.1 k 0.5 0.993 1 All data 1.3 k 0.3 1.6 k 0.6 0.0494 of two distinct groups of binding sites with different affinities for copper.-' It must be emphasized that the concentration- time data should be plotted as differences in log (concentra- tion) for a fixed time interval.The plot is then searched for the number of linear regions which roughly corresponds to groups of sites exhibiting similar complexation formation character- istics.'" Ratios of strong sites related to weak sites as reported by Mantoura and Riley4 were 30 : 70 (strong : weak) for copper in a lake water fulvic acid, in agreement with the results reported here for methylmercury. The nature of methylmer- cury binding to humic substances has been discussed by Lee and Iverfeldt,31 who indicated that methylmercury prefers the lower molecular mass humic fractions largely corresponding to the FA, and when forming complexes the methylmercury cation favours linear binding.All data in Table 3 were obtained using eqn. (2). As can be seen from the r2 values, eqn. (2) fits the experimental data very well. For comparison, Sojo and De Haan" reported rate constants for the release of iron from two distinct groups of binding sites in a humic-rich lake water, 'kl' = 1.6 X 10-2 s- 1 and 'k2' = 5 x 10-4 s-1, where 62% of the sites were found to be 'fast' (k,) and 38% exhibited lower rate constants ( k 2 ) .It is indicated that iron is bound to polymeric hydroxides and possibly fulvic acids making up the two major ligands for iron in this natural water. The values reported here for methylmer- cury are lower, indicating that MeHg is more slowly released than iron, although it should be noted that there may be differences in the nature of the sites present in these cases, and differences in metal to ligand ratios may also affect the values obtained for the rate constants. A single exponential term in the description of the curve form could not explain the experimental points satisfactorily (i.e., one major groups of sites). As can be seen in Table 3, kl decreases as the concentration of DOC increases. The observed rate of the reaction is that of methylmercury binding to the rcsin, which is limited by the rate-determining steps in the release of methylmercury from the humic and fulvic acids.As the reverse reactions in the release of methylmercury from the humic substances have not been accounted for in the model and k-?), an increase in the ligand concentration will affect the observed rate for the release. This is clearly reflected for k l as given in Table 3. As can be seen from the data obtained for k2, there is no clear trend in the results as observed for k l . To reach some kind of understanding of the apparently large spread in k2 it is essential to keep in mind that k2 is smaller than k l and determined with a poorer precision, as can be seen from the confidence intervals in Table 3.As the model of eqn. (2) includes exponential terms, slight differ- ences in the curve form due to experimental errors (5-7%) give fairly large differences for various DOC concentrations. The mean ratio between k , and k2 of 8.3 k 3.6 indicates an eightfold difference in the rate of release of methylmercury from the two different groups of sites. I t should also be observed that rate constants are normally determined with poorer precision as errors in the individual analytical measure- ments are propagated and magnified in the final result. Performance of the Proposed Method and of Instrumentation Spectral interference Even though the enrichment of mercury species on the dithiocarbamate resin is selective, a small fraction of the humic substances will be found in the final toluene extract owing to non-specific adsorption of humic substances on the resin.Indeed, brownish humic particles could be seen in the interface between the aqueous and organic phases after the extraction was completed. I t is not known, however, if the humic substances present in the final extract cause theAnulyst, January 1995, Vol. 120 75 interference discussed below. As stated previously, when large masses of organic compounds enter the MIP a conti- nuum band emission from carbon results. 18 As the selectivity is often reported as the ratio of the responses to the analyte of interest and to carbon at the wavelength monitored, the carbon response at the mercury channel, termed a 'ghost' signal, can be corrected for a feedback circuit linked to the output from the carbon channel. By careful adjustment of the 'ghost' response, the number and intensity of the interfering peaks could be compensated for in all but a few intense carbon signals.The dip in the baseline at 3 min seen in Fig. 5 is attributable to an over-compensation of the carbon signal affecting the mercury channel. This interference is separated in time from the specific signal and therefore does not affect the determination of mercury species. I t should also be mentioned that with the ageing of the walls of the ceramic torch tubes the intensity of the carbon peaks decreases considerably during the lifetime of the torch. Detection limits The absolute detection limit for the GC-MIP-AES system in its present configuration is 0.4 pg (3s criterion) as either methylmercury or inorganic mercury.Concentration detec- tion limits depend on the volume of the water sample, recoveries of mercury species, volume of toluene used for extraction, injection volumes and procedural and reagent blanks. By using a 1 1 sample and a 13 p1 injection, the detection limit for methylmercury is estimated to be 0.04 ng I-', on the basis of 3s, as no reagent blank was observed for methylmercury. The detection limit for inorganic mercury was calculated by performing four replicate measurements o n a blank following the batch method protocol and found to be 0.28 ng I-' employing a 6 pl injection. The detection limit obtained for inorganic mercury using the column method's was 0.15 ng I-'. In this instance the recovery for inorganic mercury was quantitative for sea- and fresh waters low in humic substances, whereas the recovery for inorganic mercury from a humic-rich water sample using the batch method was only approximately 50% after 20 h of shaking, resulting in a poorer detection limit.The blank signal from inorganic mercury corresponds to 1 .S ng 1-1 and results predominantly from reagent contamination. Improvement of the detection limit for inorganic mercury could be accommodated by injection of larger sample volumes, 2 6 pl. However, the improvement is not directly proportional to the volume injected, as the noise level o n the mercury channel increases slightly owing to correction for the large carbon emission signals, as discussed in the spectral interference section. The detection limits reported here are sufficient for speciation of methylmercury and inorganic mercury in most humic-rich natural waters.Calibration Calibration in all experiments described was effected using external standards, which were found to have a linear response with correlation coefficients Y > 0.995. Use of ethylmercury as an internal standard is inappropriate, however, as only 50% of this species added to humic-rich water samples could be recovered. This is in contrast to results 2.7 min I0.* mV I I ' 4 0 1 2 3 4 Ti me/m i n Fig. 5 Chromatograms displaying a blank (A) and three replicates (B, C, D) of methylmercury in rivulet water collected from Savarh in Novcmbcr 1993. Peaks correspond to a detcrmined concentration of 0.22 k 0.02 ng 1 - 1 using 9 PI aliquots injected on the GC-MIP-AES system.SCC text for explanation of thc dip at 3 min. Table 4 Analytical application and laboratory intcrcomparison with indcpcndcnt analytical techniques for thc dctcrmination of methylmercury and total mcrcury in humic-rich natural waters. Data cxprcsscd as mcan valucs k one standard dcviation for the number of replicates given in parcnt hcscs Concentration of mercuryhg I-' Sample type Analytical tcchniquc Mcthylrncrcury Inorganic mcrcury Total mcrcury Rivulet water" GC-MIP-AES 0.22 k 0.02 (3) 2.05 k 0.13 (3) 2.27 k 0.25 (3)i GC-AFSi- 0 . 19, 0.2 1:;: - - CV-ETAAS - - 2.7 k 1.4 (4) CV- ETA AS - - 4.0 k 1.6 (4) Marsh wa tcrs GC-M I P- A ES 0.30. 0.3 I 4.2.4.9 3.5.5.2' ' Collcctcd from Siivarh, pH 7. DOC 12 mg I-'. i Sum of mcthylmcrcury and inorganic mcrcury.8 Collcctcd at Svartbergct Forcst rcscarch arca, pH 3 , DOC 44 rng I-'. Analyses pcrforrned by Swcdish Environmcntal Rcscarch Institute, Gothcnburg, using the mcthod of Horvat el a/. ' 576 Analyst, January 1995, Vol. 120 reported for recoveries of ethylmercury from waters with a very low content of humic substances using the column method.18 The reason for this is assumed to be decomposition in the presence of organic substances, as has been observed previously in biological matrices.1"20 It is also worth mention- ing that tri- and diethyllead compounds exhibit much poorer stability in various types of water than the analogous methylated forms,32 further demonstrating the lower inherent stability of ethylated organometallic species.33 Peak-area signal evaluation from the atomic emission detector was used throughout the experiments.All results reported here were corrected for recovery as reported in Table 2. Accuracy The assessment of the accuracy for speciation of mercury species at ng I-' levels must be considered to be very difficult as there are no certified reference materials available. Demonstrated accuracy for certified reference materials of biological origin at pg g-1 levels is of little relevance for ultra-trace speciation although, as in this instance, it might provide evidence for the validity of the extraction and derivatization steps.2" One way to assess the accuracy of the results is to make small-scale laboratory intercomparisons and use a reference method if possible.To this end we compared results obtained using the proposed method with those for two other procedures. Total mercury in water samples was determined by CV-ETAAS.12 The other method used for comparison was GC-AFS following distillation of methylmer- cury from humic-rich water samples15 and implemented at the Swedish Environmental Research Institute, Gothenburg. As can be seen from Table 4, the results for methylmercury in Savarh determined by GC-MIP-AES and GC-AFS are in excellent agreement. Results generated using CV-ETAAS for total mercury correlate reasonably with the sum of methyl- mercury and inorganic mercury concentrations determined using the present method. It is thus indicated that the proposed batchwise enrichment GC-MIP-AES procedure gives accurate and reliable results, although a more extensive intercomparison would be required for a thorough evaluation.As several independent methods for mercury speciation in natural waters are now available,6-8,15 such an intercompari- son is feasible and is currently being planned. Analytical Application Table 4 gives results obtained for methylmercury and inor- ganic mercury in marsh and rivulet water with concentrations of dissolved organic carbon DOC ranging from 12 to 44 mg I-'. The concentration of mercury species was found to be positively correlated with the DOC concentration, which is in agreement with the observations of Lee and Hultberg.34 The chromatograms in Fig. 5 show signals for methylmercury from three samples of a humic-rich rivulet water and a blank.The signal for inorganic mercury is not displayed as it was off-scale. Measurements of methylmercury in samples collected from the same rivulet in September 1993 revealed a concentration of 0.20 & 0.03 ng 1-1 for four replicates. Unfortunately, no long-term measurements have been made to establish whether there are any seasonal variations in the concentration of methylmercury as described by other workers .34 Conclusions Using a highly selective and efficient sample enrichment, separation and detection system, the simultaneous determina- tion of methylmercury and inorganic mercury is now possible at natural levels in waters with high concentrations of humic substances. However, improvements in the performance for inorganic mercury, in terms of recovery and contamination reduction, are clearly required.The batch method protocol also provides an 'in situ' sampling and enrichment step which precludes the use of preservatives for storage of water samples. In addition, together with the model developed describing adsorption profiles, it is now possible to obtain information on the affinity of methylmercury and inorganic mercury for humic substances, in addition to information on rate constants for the release of methylmercury from the humic substances. A certified reference material for methyl- mercury and total mercury at the ng 1-1 level would be very useful for the increasing number of laboratories performing mercury speciation analyses in natural waters. Long-term stability studies of mercury species in sterilized humic-rich waters, which are naturally preserved, might be valuable for evaluating such waters as conceivable candidates for reference materials.This work was financially supported by the Swedish Environ- mental Protection Board and by the Centre for Environmental Research in Umei. We are indebted to B. Glad and K. Irgum (UmeA University) for very fruitful discussions concerning the preparation of the dithiocarbamate resin, and E. Lundberg (Norrby Marine Research Laboratory) for the loan of the FIAS-200 system. We are also very grateful for the measure- ments performed by E. Lord and J. Munthe at the Swedish Environmental Research Institute, Gothenburg. Finally, we express our gratitude to Bodenseewerk Perkin Elmer, Uber- lingen, Germany, for the loan of the 4100 ZL instrument.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 References Stumm, W., and Morgan, J. J., Aquatic Chemistry, Wiley, New York, 2nd edn., 1981. Boggs, S., Jr., Livermore, D. C., and Seitz, M. G., Rev. Macromol. Chem. Phys., 1985, C25, 599. Humic Substances and Their Role in the Environment, ed. Frimmel, F. H., and Christman, R. F., Wiley, New York, 1988. Mantoura. R. F. C., and Riley, J. P., Anal. Chim. Acta, 1975, 78, 193. Baeyens, W., Trends Anal. Chem., 1992, 11,245. Ahmed, R., and Stoeppler, M., Anal. Chim. Acta, 1987, 192, 109. Leermakers, M., Lansens, P., and Baeyens, W., Fresenius'J. Anal. Chem., 1990,336, 655. Bloom, N., Can. J . Fish. Aquat. Sci., 1989, 46, 1131. Certificate, ORMS-I, National Research Council of Canada, Ottawa, Ontario, 1992. Analytical Quality Control Committee, Analyst, 1985, 110, 103. Heiden, R. W., and Aikens, D. A., Anal. Chem., 1983, 55, 2327. Baxter, D. C., and Frech, W., Anal. Chim. Acta, 1989,225,175. Bloom, N., and Fitzgerald, W. F., Anal. Chim. Acta, 1988,208, 151. Lee, Y. H., and Mowrer, J., Anal. Chim. Acta, 1989,221,259. Horvat, M., Liang, L., and Bloom, N. S . , Anal. Chim. Acta, 1993, 282, 153. Langford, C. H., and Gutzman, D. W., Anal. Chim. Acta, 1992, 256, 183. Sojo, L. E., and De Haan. H., Environ. Sci. Technol., 1991,25, 935. Emteborg, H., Baxter, D. C., and Frech, W., Analyst, 1993, 118, 1007. Bulska, E., Emteborg, H . , Baxter. D. C., Frech, W.. Ellingsen, D., and Thomassen. Y., Analyst, 1992, 117.657. Emteborg, H., Hadgu, N., and Baxter, D. C.. J. Anal. At, Spectrom., 1994, 9, 297. Bulska, E., Baxter, D. C., and Frech, W., Anal. Chim. Acta, 1991,249, 545. Cammann, K., Lendero, L., Feuerbacher, H.. and Ballsch- miter, K . . Fresenius' 2. Anal. Chem., 1983, 316, 194.Analyst, January 1995, Vol. 120 77 23 Emteborg, H., Bulska. E.. Frech, W., Baxter, D. C.. J. Anal. 30 At. Spectrom.. 1992. 7. 405. 24 Baxter, D. C.. and Frech, W.. Anal. Chim. Acra, 1990,236,377. 31 25 Hiraide. M., Tillekratne, S. P., Otsuka. K.. and Mizuieke. A., Anal. Chim. Acta, 1985. 172, 215. 32 26 Sperling. M.. Yin, X.. and Wclz, B., J. And. At. Spectrom., 1991, 6, 295. 33 27 Redeman. E., Icke. R. N., and Alles. G. A . , in Orgunic Syntheses, ed. Horning, E. C., Wiley, New York, 1955, vol. 3. pp. 763-765. 34 Powell, H. K. J . , and Florence, T. M., Anal. Chim. Actu, 1990, 228, 327. Reuter, J. H . , and Perdue, E. M.. Geochim. Cosmochim. Actu, 1977, 41. 325. 28 29 Stuermer. D. H., and Payne. J . K., Geochim. Cosmochim. Acta, 1976, 40, 1109. Lcc, Y. H., and Iverfeldt. A., Water Air Soil Pollut., 1991, 56. 309. Van Cleuvenbergen, R . . Dirkx, W., Quevaviller, Ph., and Adams. F., Int. J. Environ. Anul. Chem.. 1992, 47, 21. Craig. P. J.. and Brinkman, F. E., in Orgunometullic Com- pounds in the Environment, ed. Craig, P. J., Longman, Harlow, 1986. ch. 1. pp. 1-64. Lee. Y. H., and Hultberg. H . , Environ. Toxicol. Chem., 1990. Paper 4102381 A 9, 883. Received April 22, 1994 Accepted July 6, 1994

ISSN:0003-2654    

DOI:10.1039/AN9952000069

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 79 Optimization of Atomization Parameters in the Speciation of Organotin Compounds by Hydride Generation-Gas Chromatography-Electrothermal Atomic Absorption Spectrometry P. M. Sarradin, F. Leguille, A. Astruc,* R. Pinel and M. Astruc Laboratoire de Chimie Analytique, CURS, Universite' de Pau et des Pays de l'tldour, 64000 Pau, France The detection of organotin hydrides by electrothermal (quartz furnace) atomic absorption spectrometry coupled on-line with hydride generation is closely dependent on several parameters such as the geometry of the quartz cell and the flow rates of additive gases. The influence of the design of the atomizer and the effects of the gas flow rates were evaluated in this study. The design of the quartz cell atomizer is not critical provided that a few rules are followed: the helium flow carrying organotin hydrides should be mixed with hydrogen in as small as possible dead volume (no mixing chamber) and oxygen introduction should be made as close as possible to the light path.The ratios of gas flow rates must be optimized for each cell design, the most important being the H2 : 0 2 ratio chosen to produce highly reducing working conditions. Detection limits obtained after optimization ranged from 30 to 130 pg depending on the conditions and species. Keywords: Optimization; atomization; organotin speciation; hydride generation-gas chromatography-electrothermal atomic absorption spectrometry; quartz furnace Introduction Detection by atomic absorption spectrometry coupled with hydride generation (HG) was introduced around 1970 for the determination of arsenic and selenium.1 The hydrides volatilized in the reactor are transferred directly to the detector or accumulated in some trap, usually a cold trap.' In speciation studies of organometallic compounds, several substituted hydrides are simultaneously produced, transferred and trapped. After differential thermal volatilization these hydrides are further separated by on-line gas chromatography (GC) and detected with electrothermal (quartz furnace) atomic absorption spectrometry (QFAAS). This is a simple method using inexpensive equipment, especially convenient when dealing with freshwater or sea-water samples for which no pre-treatment, except acidification, is necessary. This method has been applied to speciation studies of various elements such as antimony and bismuth,3 arsenic,- sele- nium7.8 and as early as 1979 for tin." However there is no commercial equipment available and workers have to con- struct their own.Most papers published in this area have presented analytical data obtained with different environmental samples and each group has used different experimental conditionsl(kl8 (Table 1). Commonly used gases are helium to purge and vehiculate the hydrides and hydrogen and oxygen or air as atomization * To whom correspondence should be addressed. additives in the quartz cell. The hydrogen and oxygen or air flow rates used have varied widely.10 Organotin Speciation Procedure In acidic solutions, inorganic tin cation, and also many organotin cations, are reduced by NaBH4 to stannanes of general formula R,,SnH(4-l,).A flux of carrier gas is used to flush out these analytes to a cold trap, i.e., a short GC packed column (60 x 0.5 cm i.d.) immersed in liquid nitrogen. Helium or hydrogen may be used for this purpose; helium is generally preferred (to reduce explosion risks). Evolved hydrides are cold-trapped at liquid nitrogen temperature. After the end of this collection step, gentle heating of the GC column produces differential volatilization of the retained compounds followed by GC separation. The retention times of the various organotin compounds studied are linearly related to their relative molecular masses and neither the composition nor granulometry of the station- ary phase has a marked effect.lSq1" This led some workers to consider that this step is more a fractional distillation than a real chromatographic separation.1".2° More detailed descriptions of these experimental conditions have already been published. 19-22 The sensitivity of the on-line QFAAS detection of individual organotin hydrides and that of other hydrides10.23 is known to be dependent on the presence Table 1 Derivatization methods and atomization parameters used for the determination of organotin compounds by GC-QFAAS Furnace Derivati- Carrier gas/ Additive gas/ tempera- zation ml min-I ml min-I ture/oC Ref. Methylation Hydridiz- ation Ethylation Hydridiz- ation Hydridiz- ation Pentylat ion Pcntylation Hydridiz- ation He 20 N2 60 He 60 He 60 He 40 Ar 6 Ar 32 He400 HZ 60 Air 40 H2 300 H2 300 Air 15 Air 16 Air 140 Air 140 Air 45 Air 9 Hz 220 Hz 220 H2 350 H2 470 H2 1200 0 2 90 800 11 9 50 12 950 13 500 14 Nofurnacc 15 900 16 17 950 1880 Analyst, January 1995, Vol.120 of hydrogen and oxygen in the quartz furnace,7.18724.25 on the furnace temperature and on the quality of the quartz surface.7.~.1() I n this work, two fully automated assemblies for the speciation of organotin compounds by HG-GC-QFAAS were constructed from different components and the effects of various parameters affecting atomization conditions such as cell geometry and gas flow rates were studied. The relative importance of these parameters was evaluated and the optimization of gas flow rates was studied by both an experimental design method and a simplex procedure.Experimental Equipment Two different, fully automated, assemblies were constructed and compared. In the older one (A-I) a TRS 80 microcom- puter drove the whole analytical process through a laboratory- made interface, absorbance signals being measured with an IL 15 1 atomic absorption spectrometer (hollow-cathode lamp, h = 286.3 nm) and treated by a Varian 4270 integrator.lg.20 The quartz cell was heated with a commercial ceramic furnace (Perkin-Elmer MHS1). The second assembly (A-11) was driven by a Shimadzu CR4A integrator with a PC16N I/O card and a laboratory- made interface, absorbance being measured by a Varian 10 atomic absorption spectrometer.26 A Nichrome wire wrapped around the quartz tube isolated in refractory material was used to heat the quartz cell, the heating power being adjusted with a variable transformer.The reactor and GC column were identical in both assemblies. Quartz Furnace Atomizer The quartz furnace atomizer has almost always been an open-ended T-tube of varying length, diameter and design (Stallard et al.14 used a close-ended cell). Large tubes (100-150 mm long, 3-15 mm diameter) are most cornmon,18,2s small ones (10 X 3 mm) rarely being used.'() Atomization gases, H2 and 0 2 or air, are injected into the main flow of carrier gas at various positions close to the atomization area. Four different quartz atomizers (1-4) were tested in this study (Table 2). Cells 1 and 2 differed in the position of the O2 inlet. The other two (3 and 4) were similar to cells 1 and 2 as regards 0 2 injection design but included mixing chambers (approximately 2.5 cm3) (Fig.1) where the H2 and carrier gas flows were mixed before 0 2 addition. In all experiments the cells were heated to 950 "C, which was the optimum temperature chosen from previous work and known to be a non-critical factor. Optimization of Gas Flow Rates The He flow rate varies during an experiment owing to the variable head loss produced by thermal expansion of the GC column material linked to the wide temperature variations Table 2 Parameters of the quartz cells shown in Fig. 1 (all dimensions in mm) Cell No. Parameter 1 2 3 4 I 140 140 140 140 4 12 12 12 12 do* 3 3 3 3 d' 6.5 1 6.5 1 (- 196 to +220 "C), so only the initial flow rate values may be considered. To optimize the sensitivity of the detection of organotin compounds it is necessary with each kind of cell to study the effects of the flow rates of the various gases involved, and this implies numerous experiments excluding the possibility of univariate searches. Therefore , two complementary approaches were used: the experimental design method and the simplex method.Experimental design method Table 3 describes the three factors retained for each cell studied and their ranges of variation selected from preliminary experiments. The 23 experimental design matrix indicates eight sets of experimental conditions (Table 4). First- and second-order interactions are considered following ref 27. The individual atomic absorption signals produced by the hydride derivatives of six organotin cations [mono-, di- and tri-methyltin cations (MMT, DMT, TMT, respectively) and mono, di- and tri-butyltin cations (MBT, DBT, TBT, respec- tively] were simultaneously studied in each of the eight experiments.A 10 ng amount (as tin) of each compound was introduced as 100 yl of a 100 ng ml-1 standard solution in I t I Umm 70 60 50 40 30 20 10 0 l " ' I I I : , Helium Cells 1 and 2 70 60 50 40 Vmm 30 20 10 0 Helium Cells 3 and 4 Fig. 1 temperature measurements. Designs of the different quartz cells tested. L and Y , axes of Table 3 Factors and rangcs of variations retained Factors/ml min-1 Level He 0 2 H2 (1) (2) (3) + 300 100 200 150 50 100 -Analyst, January 1995, Vol. 120 81 100 ml of ultrapure water (Milli-Q) acidified with 0.5 ml of Suprapur (Merck) acetic acid. From the set of experimental data, the effect Ex of each factor and of their interactions may be evaluated from the equa tion27 Ex = ( l / n ) 2 k Ri where Ex is the effect of the xth factor, n is the total number of experiments ( n = 8), Ri is the ith measured value for the variable, and k is assigned to Ri in the summation as given in the matrix in Table 4.The effect of each factor was then compared with the uncertainty AE, calculated using n i = 1 AE = Ay (t/&) where Ay is the experimental uncertainty and t is a coefficient related to the retained confidence level. Here, AE = 0.6 or 0.16, depending on the assembly used, with t = 0.05 at the 2s confidence level. Simplex method Optimization of the O2 and H2 flow rates was further improved by using a simplex procedure at a constant high value of input He flow rate (300 ml min-1).The chromato- graphic peak areas were registered for butyltin compounds only, the main object of our studies, during duplicate analyses of mixtures of MBT, DBT and TBT using speciation assembly A-I1 with cells 1 and 3. Independent simplex procedures were performed with the two cells, using H2/02 initial flow rates of 280/30, 320/30 and 300/80 ml min-1. Development of an Improved Atomizer In order to study more precisely the effect of the position of the O2 inlet, a fifth quartz cell with an adjustable position of the O2 inlet was designed (Fig. 3). Table 4 Experimental design matrix Experi- Factors Interactions ment NO. (1) (2) (3) (1-2) (1-3) (2-3) (1-2-3) 1 + + + + + + + 2 3 + 4 5 + + + 6 7 + 8 He O2 H2 - + - - + + + + + + - - - + - - + + + + - - - - - - - - - + + + + - - - - - - - - - - - 300 MMT DMT TMT MBT DBT TBT MMT DMT TMT MBT DBT TBT Fig.2 Effects of A, hydrogen, B, helium and C, oxygen flow rates on the atomic absorption signal of methyl- and butyltins, with cells 1 and 2 and assembly A-I (a), and cells 1, 2 and 4 and assembly A-I1 (b). The chromatographic peak areas of butyltin compounds were registered during duplicate analyses of a mixture of MBT, DBT and TBT (5 ng of each in 100 ml of ultrapure water acidified with acetic acid) (H2 = 340 ml min-1, 0 2 = 20 ml min-1, He = 300 ml min-1, assembly A-11). Results and Discussion Experimental Design Method Influence of gas flow rates The results are shown in Fig. 2, illustrating the effects of the gas flow rates on the atomic absorption response for each compound studied, with three cells and two speciation assemblies.The effects of the various gas flow rates can be classified in the order H2 > O2 >> He, with both speciation assemblies A-I and A-11. The H2 flow rate has a marked positive effect whereas O2 has a negative effect. Moreover, the second- and third-order interactions (not presented here) are also signifi- cant. The He flow rate shows only a very slight positive effect. This means that during the gas flow rate adjustment for optimum sensitivity of the assembly, the H2 : 0 2 : He ratios (especially H2 : 02) have to be considered in addition to the individual flow rates. Another important conclusion is that similar effects are observed for all the analytes, but with different amplitude, TBT being the less affected species.The most important analyte in these studies is tributyltin (TBT) owing to its high environmental significance. The choice of the He flow rate has little impact on TBT determinations and the best detection limits will be obtained under reducing conditions (i.e., high H2 : 0 2 ratios). The best experimental conditions defined by this test were obtained with both assemblies for He, O2 and H2 flow rates of 150, 50 and 200 ml min-1, respectively, but very similar sensitivities were obtained with an He flow rate of 300 ml min-1. Influence of the design of the quartz cell atomizer If the same general behaviour prevailed in all the configura- tions tested (three cells, two assemblies), noticeable variations of the importance of the effects of H2 and 0 2 flow rates were observed.The sensitivity of tin detection with cell 2 was highly dependent on gas flow rates with both assemblies. This cell has no mixing chamber and the O2 inlet is situated directly in the light path . Hydrogen Teflon ferrules 518 Chromatographic Reducing union 518.38 Swagelock 318 Oxygen sliding Swagelock 318 Oxygen Fig. 3 Modified cell design.82 Analyst, January 1995, Vol. 120 Detection with cell 1 is also highly dependent on the gas flow rates with assembly A-I but much less with assembly A-11. This cell has the same design as cell 2 except for the position of the O2 inlet, fairly remote from the light path. Therefore, it appears that the Perkin-Elmer MHS-1 furnace used in assembly A-I is more efficient than the laboratory- made furnace used in assembly A-11.It may also be noted that the H2-02 flame lacks stability when cell 1 is used with assembly A-I. These effects may be linked to a lack of temperature uniformity using assembly A-11, influencing the atomization kinetics. Cells that included mixing chambers, such as cell 4, have a much lower sensitivity to gas flow rates. Our conclusion is that the best conditions are obtained in reducing atmospheres with flow rates H2 >> O2 using a carefully designed heating device, a cell with a very low dead volume (no mixing chamber) and an O2 inlet very close to the optical path. Simplex Optimization Table 5 presents the optimum conditions obtained by simplex optimization of H2 and O2 flow rates, together with the various sensitivities of detection of butyltin species.Slightly different optimum H2 : O2 ratios were evidenced with cells 1 and 3, which differ only in the presence of a mixing chamber in cell 3. Optimum sensitivities for the detection of butyltin com- pounds are improved 5-6-fold when the dead volumes are minimized, i.e. , when there is no mixing chamber. Development of an Improved Atomizer Using the fifth cell with an adjustable O2 inlet position (Fig. 4), maximum sensitivities were obtained when the O2 inlet was close to the light path. However, the ‘reactive zone’ seems to be larger than expected as a nearly constant sensitivity could be observed for 0 < d’ d 13 mm. When d’ is larger than 13 mm, the sensitivity decreases for the three species by up to 30% for MBT (least affected species) and more than 50% for Table 5 Gas flow rates optimized by the simplex method and corresponding sensitivities using speciation assembly A-I1 Optimized flow rate/ml min-I Sensitivity/area units ng- Cell H2 0 2 MBT DBT TBT 1 340 20 56.6 30.1 21.9 3 300 30 11.5 6.1 3.3 I T I 0 5 10 15 20 25 dlmm Fig. 4 Peak areas obtained with different positions of the 0 2 inlet, assembly A-11. A.MBT; B, DBT; and C, TBT. Gas flow rates: H2. 340; O?, 20; and He, 300 ml min-l. DBT and TBT. The production of tin atoms takes place out of the T-tube and recombination phenomena reduce drastically the free tin atom concentration in the light path. Effect of the quartz wall surface After several experiments, slight decomposition of the quartz surface was observed (body of the cell and O2 inlet tube), a white deposit appeared on the quartz surface.This was followed by a serious decrease in sensitivity (up to 50% decrease in the signal on some occasions). This could not be due to chlorides because of the cold trapping step. This effect may be produced by the combination of a large excess of hydrogen and a high temperature leading to an alteration (reduction?) of the quartz surface. Similar observations have been made by Mayer et al.8 and by Agterdenbos et al.7 in As and Se determinations. Periodic cleaning of the cell with 40% HF for 15 min28 restores the quartz surface and the sensitivity. Discussion and Conclusion The observed effects of gas flow rates on sensitivity are similar to those summarized by Dedina’o in investigations on the atomization of hydrides in H2-02 flames (without He) and which he studied in detail for SeH2.The decay of the concentration of free tin atoms in the light path is not primarily due to flushing effects as the sensitivity increases at high total gas flow rates. However, this decay is fast as optimum sensitivity is always attained when the H2-02 flame (and, hence, the reaction zone) is very close to (or even in) the light path. This can be compared with the observations of Stallard et a1.14 The dependence of sensitivity on gas flow rates is difficult to correlate precisely with data reported by Dedinalo owing to the different conditions used (He = 0, H2 = 0-12 000 and O2 = 3-210 ml min-1 were the flow rates used by Dedina, compared with He = 275-825, H2 = 100-500 and O2 = 0-100 ml min-1 in this study).The main difference is that the optimum H2 : 0 2 ratios in this study, although far above stoichiometry, are much lower than those found by Dedina for selenium detection. The total gas input flow rate (optimized to approximately 900-1400 ml min-1 in this study) is comparable to the H2 flow rate (500-1400 ml min-1) optimized for selenium determination with a geometrically similar flame in a tube quartz atomizer (Fig. 6.3 in ref. 10). This implies that the reaction kinetics of the atomization-recombination processes of selenium (from SeH2) or tin (from organotin hydrides) are of the same order of magnitude. In the mixing chambers used in the design of cells 3 and 4, very low O2 concentrations prevail.The 5-6-fold lower sensitivity obtained with quartz cells involving these mixing chambers indicates that partial hydride decomposition occurs there, which may be due to interactions with the quartz surface7-8928 followed by a partial decay of the population of free tin atoms before they enter the light path. To obtain the best sensitivities, the cell geometry should be similar to those of cells 1 and 2 or the improved design of cell 5 with an adjustable O2 inlet, i.e., absence of a mixing chamber, minimized dead volumes and an O2 inlet close to the light path. The improved atomizer is to be preferred as it gives the possibility of adjusting periodically the position of the O2 inlet; this is important as the very narrow quartz tube introducing 0 2 is altered during extended use of the cell and may be noticeably shortened. The atomization efficiency is highly dependent on the state of the quartz surface as an important decrease in sensitivity may be observed when the surface is altered (appearance of a white deposit).This alteration may accelerate the recombina-Analyst, January 1995, Vol. 120 83 tion of radicals and, thus, decrease the H radical concentration necessary for the atomization of organotin hydrides, leading to a depression of the signa1.3,7,8 This could also explain the sensitivity variations observed during weeks or even days of work with the same quartz cell that make cell cleaning necessary. The sensitivities observed for the various organotin com- pounds studied, expressed as a function of tin mass, vary from species to species (see Table 5 ) .In addition to possible intrinsic differences in the rates of production of tin atoms from these different molecules, one may note that during a single speciation experiment the He flow rate (and, therefore, the total gas flow rate) changes continuously, in addition to the furnace temperature owing to gas cooling effects. As the retention times of organotin hydrides are different they are transferred into the atomization cell under different condi- tions of gas flow rate and temperature. Both parameters have been demonstrated") to play a role in atomization efficiency. Gas flow rates must be optimized for each combination of cell design and electrical heating device; a furnace carefully designed to ensure a homogeneous temperature in the quartz cell is desirable.Simultaneous optimization of the flow rates of the three gases, He, H2, 0 2 , is recommended, but the flow rate of helium is much less critical than that of either H2 and 02. As the HG-GC part of the speciation equipment enforces some constraints on He flow rate, such as the necessity to carry hydrides from the reactor to the cold trap, and a variable head-loss through the GC column during the thermal cycle, the adjustment of the helium flow rate may be considered as an independent problem and the detailed optimization proce- dure restricted to the adjustment of H2 and O2 flow rates, a simplex procedure then being useful. References Holak, W., Anal. Chem., 1969. 41, 1712. Godden, R. G., and Thomerson, D.R., Analyst, 1980 105, 1137. Asami, T., Kubota, M., and Saito, S., Water Air Soil Pollut., 1992, 62, 349. Bax, D., Van Elteren, J. T., and Agterdenbos, J., Spectrochirn. Acta, Part B, 1986, 41, 1007. Wagenen, S. V., and Carter, D. E., Anal. Chem., 1987,59,891. 6 7 8 9 10 11 12 13 14 1.5 16 17 18 19 20 21 22 23 24 25 26 27 28 Chen, H.. Brindle, I. D., and Lc, X. C.. Anal. Chem.. 1992.64, 667. Agterdenbos, J . , Van Noort, J. P. M., Peters, F. F., and Bax, D., Spectrochirn. Acta, Part B , 1986,41, 283. Mayer, D., Haubenwallner, S., and Kosmus, W., Anal. Chim. Acta. 1992, 268, 315. Braman, R. S., and Tompkins, M. A., Anal. Chern., 1979, 51. 12. Dedina, J., Prog. Anal. Spectrosc., 1988, 11, 2.51. Forsyth. D. S., Sci. Total Environ., 1989, 89, 299. Clark, S., and Craig, P. J., Appl. Organornet. Chern., 1988, 2, 33. Ashby, J. R., and Craig, P. J . , Appl. Organornet. Chern., 1991, 5 , 173. Stallard, M. O., Cola, S. Y., and Dooley, C. A., Appl. Organornet. Chern.. 1989, 3, 105. Valkirs, A. O., Seligman, P. F., Olson, G. J.. Brinckman, F. E., Matthias, C. L., and Bellama, J. M., Analyst, 1987, 112, 17. Lobinsky, R., Dirkx, W. M. R., Ceulemans, M., and Adams, F. C., Anal. Chem., 1992, 64, 159. Dirkx, W. M. R., Van Mol, W. E.. Van Cleuvenbergen, R. J. A., and Adams, F. C., Fresenius' 2. Anal. Chern., 1989, 335, 769. Donard, 0. F. X., Rapsomanikis, S., and Weber, J. H., Anal. Chem., 1986, 58, 772. Desauziers, V., Thesis, Pau University, 1991. Lavigne, R., Thesis, Pau University, 1989. Desauziers, V., Leguille, F., Lavignc, R., Astruc, M., and Pinel, R., Appl. Organornet. Chern., 1989,3, 469. Donard, 0. F. X., Randall, L., Rapsomanikis. S., Weber, J. H., Int. J. Environ. Anal. Chern., 1986, 27, 55. Dedina, J., and Rubeska, I., Spectrochirn. Acta, Part B , 1980, 35, 119. Welz, B., and Melcher, M., Analyst, 1983, 108, 213. Michel, P., in Oceans 87 Proceedings, vol. 4, International Organotin Symposium, IEEE, New York, 1987, pp. 1340-1343. Leguille, F., Thesis, Pau University, France, 1992. Goupy, J . , La Mdhode des Plans d'Experience, Dunod, Paris, Dedina, J., in Fortschritte in der Atomspektrometrischen Spureanalytik, ed. Welz, B., Verlag Chemie, Weinheim, 1984, pp. 29-47. 1988, pp. 147-157. Paper 4103186E Received May 31, 1994 Accepted August 26, I994

ISSN:0003-2654    

DOI:10.1039/AN9952000079

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 85 Determination of Trace Amounts of Phosphorus in Silicate Materials by Simultaneous Inductively Coupled Plasma Atomic Emission Spectrometry F. J. Valle Fuentes Instituto de Ceramica y Vidrio, CSIC, Arganda del Rey, 28500 Madrid, Spain S. del Barrio Martin Instituto Tecnoldgico Geominero de Espafia, Rios Rosas 23, Madrid 28003, Spain Inductively coupled plasma atomic emission spectrometry (ICP-AES) implemented on a simultaneous spectrometer was used for the determination of trace amounts of phosphorus in rocks, soils and sediments. Samples were decomposed by using a mixture of concentrated hydrofluoric, perchloric and nitric acid in Teflon beakers. Overlap of emissions from other elements present in the samples with the polychromator spectral line at 214.914 nm for phosphorus was studied.The corresponding inter-element coefficients for overlap correction were determined and the optimum position for background readings was established. The limit of determination achieved was 24 ppm referred to the solid sample. The proposed method was tested by applying it to the analysis of rock, soil and sediment certified reference materials, and determining its short-time precision and method precision, which turned out to be O.24-O.68YO and 1.70-4.16% (relative standard deviation), respectively. Keywords: Inductively coupled plasma atomic emission spectrometry; phosphorus determination; rock; soil; sediment Introduction Phosphorus occurs widely in the Earth's shell which contains approximately 0.23% of P2O5.Most of the phosphorus is present in minerals of the apatite [3Ca3P205Ca(CI,F)2] group. Small amounts of V04, As04 and Si04 can substitute for Po43-, and Na, Sr, U, Th and lanthanoid metals may replace Ca in the parent structure. In relation to the natural geological sedimentation cycle,' apatite is relatively insoluble in alkaline or neutral waters. Its solubility increases with increasing acidity and decreasing temperature. Thus, although apatite may be residually concentrated in the first stages of weathering, it eventually breaks down. The soluble phosphorus from apatite binds to organic matter during sedimentation to produce primarily ampellites and black shales. The phosphorus content in igneous rocks ranges from 600 ppm in granites to 1200 ppm in the most alkaline rocks. Exceptionally, some alkaline igneous rocks and highly developed granites may contain even higher proportions.Phosphorus is of great geochemical interest for several reasons. Thus, (i) it is a good lithological indicator associated with magmatic rocks, (ii) it separates certain facies between alkaline rocks or granites and (iii) it differentiates clay lithologies (black lutites or ampellites) from other types of lithology in sedimentation media. For the above reasons, the inclusion of phosphorus determi- nations in geochemical programmes is of great significance. The general literature on silicate materials abounds with references to the determination of phosphorus. The instrumental techniques most frequently used for this purpose include spectrophotometry,2-12 atomic absorption spec- trometryl3-15 and X-ray fluorescence spectrometry.1620 Direct current plasma (DCP) and inductively coupled plasma atomic emission spectrometry (ICP-AES) have also been used for determining phosphorus in silicate materials despite their poor sensitivity to this element and the weak emission lines21 typically obtained (phosphorus is one of the elements with the highest ICP detection limits.22 Bankston et al.23 used the 214.914 nm line for the DCP determination of phosphorus following sample dissolution by LiB02 fusion in a graphite crucible. Cook and Miles,24 Burman25 and McLaren et a1.26 used the lines at 178.287, 213.618 and 214.914 nm, respec- tively, to determine phosphorus by ICP after dissolving samples with HF in a platinum or (PTFE) vessel.Xu et al.27 investigated spectral interferences in the determination of phosphorus in steel by using an dchelle spectrometer. They found the P I 213.618 nm line to be free from any interference from the iron line, but partly overlapped with the wing of the Cu I1 213.598 nm line. Kanda and Taka28 suggested that the 214.914 nm line is suitable for soil and sediment analysis. However, the line is subject to a severe background shift interference from the Cu I1 214.897 nm line. The excellent multi-element excitation capabilities of the ICP technique have been shown also to result occasionally in serious spectral interferences arising from line overlap and background shifts in geochemical determinations of phospho- rus. Background shifts can often be corrected by making off-line background measurements using a quartz refractor technique.In those instances where some spectral interfer- ence is unavoidable, correction coefficients are determined by nebulizing ultrapure matrix solutions following calibration under the same working conditions as those to be used in the subsequent analyses. If the interference originates in a polychromator channel, then the correction factor can readily be applied to the interfered element. The most sensitive lines for phosphorus determination by air-path ICP-AES are the UV lines at 213.618 and 214.914 nm. The purpose of this work was to investigate spectral interferences with the emission line of phosphorus at 214.914 nm (viz., the phosphorus line set in our spectrometer polychromator) and obtain a coincidence scan for those of the interferents.Interferent scans were required for appropriate selection of a corrective procedure.86 Analyst, January 1995, Vol. 120 Experimental Apparatus and Operating Conditions A Jarrell Ash ICAP-62 simultaneous spectrometer equipped with an ICP source was used. The spectrometer was furnished with a polychromator of 1500 grooves mm-1 accommodating 35 photomultiplier tubes. The specifications of the ICP source were as follows: frequency, 27.12 MHz; induced power, 1000 W; and reflected power, <5 W. The operational conditions used were as follows: plasma gas (argon) flow rate, 17 1 min-1; aerosol carrier gas (argon) flow rate, 0.75 I min-1; coil cooling water flow rate, 1.5 1 min-I; sample introduction, pneumatic cross-flow nebulizer; sample delivery, Gilson Minipuls-2 peristaltic pump (2 ml min-I); measurement time, 10 integra- tions at 5 s per step; and observation height, 16 mm.Reagents Stock standard solutions of high purity (Riedel-de Haen), HF, HN03 and HCI (Suprapur grade, Merck) and distilled water of resistivity less than 18 Mi2 cm were utilized to make all standard and sample solutions. Samples The samples tested were eight different certified silicate reference materials supplied by the National Bureau of Standards (NBS), the Canadian Certified Reference Materials Project (CCRMP), the Rat Fur Gegenseitige Wirschaftsshhi- life (RGW), the Bureau of Standards of the Republic of South Africa (SAR) and the Bureau of Standards of the Republic of China (GBW).Decomposition of Samples Samples were decomposed by using the following procedure. In a 100 ml Teflon beaker were placed 1.0000 g of finely ground, previously dried (110°C) sample, 10 ml of concen- trated HF, 5 ml of concentrated HCIO4 and 5 ml of concentrated HN03. The mixture was heated to dryness at 190°C on a hot-plate to remove tetrafluorosilicate and residual HF. Then, 5 ml of concentrated HCI and 5 ml of distilled water were added and the residue was heated at 100 "C until complete dissolution. The resulting solution was finally diluted to 100 ml with distilled water and stored for subsequent analysis. Preparation of Standard solutions for Spectral Interference Study Thirteen single-element working standard solutions were prepared by diluting to 100 ml previously prepared stock standard solutions containing 2 ml of concentrated HN03 and 10 ml of concentrated HCI.The element concentrations were as follows: aluminium 1000, iron 1000, calcium 1000, magne- sium 500, titanium 250, tungsten 100, cadmium 100, silver 100, zirconium 100, copper 12, vanadium 10, niobium 10, tin 10 and phosphate 3.26-10 pg ml-1 (as P). These concentrations are similar to or slightly higher than those typically encoun- tered in silicate rocks, soils and sediments. Results and Discussion Spectral Interferences Spectral interferences were identified and appropriate wavelengths for background correction were chosen by performing short-wavelength scans near the analyte wavelength. Figs. 1 4 show typical overlapped scans obtained for phosphorus at 214.914 nm.Fig. 1 includes the scans for pure solutions of aluminim, iron, phosphorus and a blank reagent. As can be seen, aluminium caused a strong interference. Pritchard and Lee29 ascribed this effect of aluminium to two complementary effects, namely (i) the presence of a doublet at 214.874 and 214.964 nm that is only observed above 3500 pg ml-1 A1 and (ii) the presence of a broad, unbroken band that extends to higher wavelengths. The effect of iron was less marked: one of its lines (214.917 nm) overlapped that of phosphorus, and another three lines (214.850, 214.962 and 215.018 nm) only affected its spectral background structure. Calcium, magne- sium and titanium gave a flat background that raised that of the phosphorus line slightly but did not alter its measurement position (the scans have been omitted from the figure).Fig. 2 shows the scans obtained for the copper, vanadium and niobium solutions. Even though these three elements gave peaks near the phosphorus emission line (viz., 214.842 nm for V, 214.872, 214.903 and 214.950 nm for Nb and 214.897 for Cu), only the Cu line and the Nb line at 214.903 nm overlapped that of phosphorus (the others only affected thc background measurement position). Fig. 3 shows the scans for tungsten, zirconium and tin. As can be seen, the emission of phosphorus overlapped two W lines (214.914 and 214.885 nm) and one, uncatalogued Zr line (approximately 214.89 nm). However, the emission of Sn (214.874 nm) and an unknown W line at approximately 214.96 nm should only affect background measurements.While the W, Zr and Sn concentrations in the standards used allowed their interfering effects to be revealed, these levels are hardly reached in rocks, soils and sediments (with the exception of specific mineralization zones). Fig. 4 shows the spectral scans obtained for a certified soil sample, a blank reagent and two multi-elemental standards of the following compositions: standard A, 1000 pg ml-1 Al, Fe and Cu, 500 pg ml-1 Mg, 250 pg ml-1 Ti and Na, 12 pg ml-1 Cu and 10 pg ml-1 Nb and V; and standard B, the same as standard A plus 3.26 pg ml-1 P. The similar characteristics of the spectra for both standards and the soil sample and the difference between the two 11785 t , .- g W c - 775 I Wavelength + Fig. 1 Spectral scans in the vicinity of the P 1214.914 nm line for A, AI (1000 pg ml-1); B, blank; C, Fe (1000 pg ml-I); and D, P (3.26 pg ml-1).6734 2. fn W 4- .- c - 775 t 214.897 n ~ , ,A E Wavelength + Fig. 2 Spectral scans in the vicinity of the P 1214.914 nm line for A, Cu (12 pg ml-1); B, V (10 pg ml-I); C, Nb (10 pg ml-I); and D, P (3.26 yg ml-I).Analyst, January 1995, Vol. 120 87 standards in the P emission region, confirm that the model used to study potential interferences was correct. The determination of phosphorus entailed a prior selection of the background measurement position and the establish- ment of correction coefficients for inter-elemental effects. Measurement of Background Irrespective of the contribution from the attacking reagents used, the spectral background in the vicinity of the 214.914 nm line for phosphorus was essentially provided by A1 and Fe and, to a lesser extent, the Cu, V, Nb, W and Zr lines mentioned in the previous section.Despite the complexity of ICP emission, accurate back- ground measurements could be made by using a quartz refractor plate behind the spectrometer entrance slit. The plate shifted the beam falling on the grating, thereby effectively changing the angle of incidence. The magnitude of the shift was proportional to the refractor plate angle relative to the incident beam. This allowed spectral scans of kO.1 nm to be achieved in the vicinity of the line to be measured. As can readily be seen from the spectral region provided by this procedure (Fig. 4), the optimum position for measuring the spectral background was 214.876 nm.The background intensity was subtracted from that of the analytical line by using a multiplexed analogue-to-digital converter and com- puterized data processing. Correction of Spectral Overlap In order to overcome spectral overlap of the phosphorus line with those of some interferents that resulted in spurious analyte concentrations, the following equation was used: cP = A p - 2 K p j ~ j J where cp is the corrected concentration of phosphorus, AP its apparent concentration, cj the concentration of interferent j ; and Kpj the inter-element correction coefficient of element j for phosphorus. Kpj values (Table 1) were calculated by 2357 > u) a Y .- Y - 705 2: 21 4.96 nm Wavelength + Fig. 3 Spectral scans in the vicinity of the P I 214.914 nm line for A, W (100 pg ml-1); B, Zr (100 yg ml-1); C, Sn (100 pg ml-l); and D, P (3.26 pg ml-l).Wavelength + Spectral scans in the vicinity of the P I 214.914 nm line for A, Fig. 4 standard A; B, standard B; C, sample GSD-12; and D, blank. measuring the emission intensities at 214.914 nm for the single-element solutions. KPCu takes into account the interfer- ence of Cu 214.897 nm in the locations of peak and background measurements. The negative KPNb value is due to the interference of Nb 214.872 nm, which affects the back- ground more than the interference of Nb 214.903 nm, which, in turn, affects the peak. Limit of Determination for Phosphorus The limit of determination for phosphorus was considered, with a probability greater than 99.87%, as the concentration providing ten readings whose average intensity, decreased by three times their standard deviation, was not overlapped with the average blank intensity (standard A) increased by three times its standard deviation.In order to calculate both the determination limit and the phosphorus concentration in rocks, soils and sediments that the proposed method could discriminate, nine standards containing 0, 0.25, 0.50, 0.75, 1.00, 1.25, 1.50, 1.75 and 2.00 pg ml-1 of phosphate were analysed under the optimum experimental conditions. Table 2 shows the average intensity, standard deviation and variation range for the readings obtained according to the k 3s criterion. As can be seen, there was an essentially linear relationship between the average intensity and its standard deviation (Fig.5). Such a linear relationship allowed a determination limit of 0.75 pg ml-1 (equivalent to 24 ppm solid sample) for phosphorus to be calculated and confirmed that attacking solutions applied to the materials studied differing by as little as 0.25 pg ml-1 (equivalent to 8 ppm solid sample) in their phosphorus contents could be discriminated with a probability greater than 99.87% over the concentration range 24-100 ppm. The inflection point in the curve of concentration versus intensity (0.75 pg ml-1 of P) is explained because the lower concentra- tions are below the limit of determination and, in practice, they do not produce variations in the intensity reading. Table 1 Inter-element correction coefficients for some of the interferents studied relative to the 214.914 nm line for phosphorus Interferent Unm K p j (pg ml-1 P/mg ml- interferent) A1 Fe c u Nb Zr W Background continuum 1.6 X 214.897 0.005 2 14.872 -0.05 214.903 214.889 0.003 214.914 0.016 214.917 1.5 x 10-4 Table 2 Data used to establish the limit of determination for phosphorus and the concentration that the proposed method can discriminate in silicate materials Standard concentration/ yg ml-l P o p P 0 0 0.25 0.082 0.50 0.163 0.75 0.245 1 .oo 0.326 1.25 0.408 1.50 0.484 1.75 0.571 2.00 0.652 Average reading 23 300 23 500 23 700 24 100 25 100 26 100 27 200 28 200 29 300 Standard deviation 120 122 125 130 141 150 159 170 181 Reading variation range 22 940-23 660 23 134-23 660 23 7 10-24 490 25 650-26 550 27 790-28 57 1 23 325-24 075 24 523-25 523 26 723-27 677 28 757-29 84388 Analyst, January 1995, Vol.120 Precision and Accuracy The precision and accuracy of the proposed method were determined by analysing the above-mentioned eight certified reference materials. Table 3 summarizes the results obtained in the short-term precision testing (ten successive determina- tions on one sample specimen) and method precision testing (ten determinations on ten different sample specimens). The mean value ( x ) , standard deviation (s) and relative standard deviation (s,) for each assay are given. The precision and accuracy obtained with the two test series were acceptable. Conclusion The proposed ICP-AES method allows the determination of phosphorus in silicate materials by use of a direct-reading spectrometer.By correcting spectral line overlap (using empirical inter-elemental coefficients) and choosing an appropriate spectral background measurement position, the 2.00 - 1.50 E I - g 1.00 0.50 160 v) 120 0 22 24 26 28 30 Intensity readings x lo3 Fig. 5 intensity readings for the standards. Variation of the standard deviation (s) with the average Table 3 Results obtained in the determination of phosphorus (as % P205) in eight rock, soil and sediment certified reference samples Standard so-1 so-2 GBWO 7401 GBWO 7307 GBWO 7312 NBS 2704 BM SARM 3 (GSD-1) (GSP-7) (GSD- 12) Concentration (%) ~ Certified 0.062 0.300 0.0735 0.0820 0.0235 0.10 0.046 0.026 Found 0.061 0.305 0.0695 0.0755 0.0212 0.097 0.044 0.023 Short-term precision, 0.62 0.54 S (%) 0.24 0.57 0.62 0.68 0.49 0.60 Method precision, 1.70 2.30 s r (Yo) 1.95 2.81 4.16 3.32 3.10 3.55 element can be determined with adequate accuracy and precision from the 214.914 nm line in the spectrometer polychromator, which is subject to severe interference from other elements.References 1 Cathcart, J . , and Gulbrandsen, R. A., US Geol. Surv. Prof. Pap., No, 820. 1971, p. 515. 2 Baadsgaard, H., and Sandell, E. B., Anal. Chim. Acta, 1954, 11, 183. 3 Riley, J. P., Anal. Chim. Acta, 1958, 19, 413. 4 Langer, K., and Baumann, P., Fresenius’ 2. Anal. Chem., 1975, 277,359. 5 Bodkin, J. B., Analyst, 1976, 101, 44. 6 Whitehead, D., and Malik, S . A., Analyst, 1976, 101,485. 7 Watkins, P. J., Analyst, 1979, 104, 1124. 8 Iosof, V., and Neacsu, V., Rev. Roum. Chim., 1980, 25,589. 9 Kuroda, R., and Ida, I . , Fresenius’ 2. Anal. Chem., 1983, 316, 53. 10 Chalmers, R. A., Analyst, 1953, 78, 32. 11 Sala, J. V., Hernands, V., and Canals, A., Analyst, 1986, 111, 965. 12 Ingamells, C. 0.. Anal. Chem., 1966, 38, 1228. 13 Shapiro, L., and Brannock, W., US Geol. Surv. Bull., 1962, No. 14 Kirkbright, G. F., and Marshall, M., Anal. Chem., 1973, 45, 1610. 15 Kirkbright, G. F., Smith, A. M., and West, T. S., Analyst, 1967, 92, 411. 16 Gladney, E. S., Burns, C. E., and Roelants, J., Geostand. Newsl., 1982, 7, 3. 17 Lee, R. F., and McConchie, D. M., X-Ray Spectrom., 1982,11, 55. 18 Flanagan, F. J., Geochim. Cosmochim. Acta, 1973,37, 1189. 19 Abbey, S., Geol. Surv. Can. Pap., 1983, No. 83. 20 Schroeder, B., Thompson, G., Sulanowska, M., and Ludden, J . N . , X-Ray Spectrom., 1980, 9, 198. 21 Wohlers, C. C., ICP Inf. Newsl., 1985, 10, 593. 22 Winge, R. K., Peterson, V. J., and Fassel, V. A., Appl. Spectrosc., 1979, 33, 206. 23 Bankston, D. C., Humphris, S. E., and Thompson, G., Anal. Chem., 1979, 51, 1218. 24 Cook, J. M., and Miles, D. L., Analyst, 1985, 110, 547. 25 Burman, J. O., ICP Inf. Newsl., 1977,3,33. 26 McLaren, J. W., Berman, S. S., Boyko, V. J., and Russell, D. S., Anal. Chem., 1981, 53, 1802. 27 Xu, J., Kawaguchi, H., and Mizuike, A., Appl. Spectrosc., 1983,37, 123. 28 Kanda, Y., and Taira, M., Anal. Chim. Acta, 1988,207, 269. 29 Pritchard, M. W., and Lee, J . , Anal. Chim. Acta, 1984, 157, 313. 1144-A. Paper 4102589J Received May 3, 1994 Accepted August 2, I994

ISSN:0003-2654    

DOI:10.1039/AN9952000085

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 89 Atomic Absorption Spectrometric Determination of Copper and Lead in Silicon Nitride and Silicon Carbide by Direct Atomization Toshihiro Nakamura, Yuji Noike, Yoshiyuki Koizumi and Jun Sat0 Department of Industrial Chemistry, Meiji University, Higashimita, Tama-ku, Kawasaki, 214 Japan A method involving atomic absorption spectrometry (AAS) with direct atomization has been developed for the determination of trace amounts of Cu and Pb in Si3N4 and Sic using a graphite furnace. The samples were ground to a particle size of less than 20 pm and mixed with 2.5 times the amount of graphite powder in a corundum mortar. A 0.5-3 mg amount of the mixed sample was weighed in a tared graphite cup and atomized in a cup-type graphite furnace. The absorbance was determined by integration of the spectral lines in the absorbance versus time spectrum.Calibration was effected using aqueous standard solutions. The absorbance versus time spectrum of Cu shows double peaks. On the basis of the X-ray diffraction patterns and scanning electron micrographs of the Si3N4-graphite powder mixture heated at the temperature of each atomization step, the first peak could be assigned to the Cu vaporizing from Si, and the second peak to the Cu vaporizing from Sic. The results for 12 a-Si3N4, two p-Si3N4 and two SIC samples were in good agreement with the values obtained by electrothermal AAS for liquid samples. The relative standard deviations for Cu (0.11-53.7 ppm d m ) and Pb (0.361.55 ppm m/m) are 1.639% (n = 5) and 4.3-17% (n = 5), respectively.The determination limits, corresponding to twice the standard deviation of the blank measurements, are 92 pg for Cu and 53 pg for Pb. Keywords: Silicon nitride; silicon carbide; copper and lead determination; direct atomization; electrothermal atomic absorption spectrometry Introduction Silicon nitride ceramics have high chemical resistance and excellent mechanical properties at high temperature, and they are used as heat-resisting high-strength materials for turbine blades and ceramic engines. As the existence of the b-phase in raw a-Si3N4 powder and the presence of metallic impurities such as Ca, Ti and Fe at the grain boundary of the crystallites of Si3N4 ceramics adversely affects the mechanical properties of these ceramics, such as the three-point bending strength at high temperature,' trace amounts of metallic impurities have routinely been determined by inductively coupled plasma atomic emission spectrometry (ICP-AES) ,2 particle-induced X-ray emission spectrometry (PIXE)3 and atomic absorption spectrometry (AAS).4-6 Usually, ICP-AES and AAS require complicated and tedious procedures for the dissolution of Si3N4 samples.Also, these processes are sometimes accom- panied by accidential contamination. Many attempts have been made to carry out analyses by direct atomization of solid samples, and the sampling method has been reviewed by Langmyhr7 and Ng.8 Katsukov et al.9 determined Fe in Si3N4 and Sic by direct atomization using electrothermal atomic absorption spectrometry (ETAAS). Graule et a1.10 determined 13 trace elements in slurry samples by direct atomization ICP-AES.Because the direct atomiza- tion method is free from dissolution and dilution procedures, the development of a highly sensitive and rapid determina- tion, with few contamination problems, is feasible. However, as a solid or liquid matrix usually generates considerable amounts of background absorption, the effective reduction of the background requires the use of further techniques, including the Zeeman effect methodllJ2 and the Smith-Hieftje method13 in correcting the interference. The difficulties in the introduction of samples into the furnace have been reduced by the insertion of (1) a platform into the graphite tube,l4 (2) a cup into the graphite tube furnace15 and (3) a graphite cup into the cup-type furnace.16 The selection of calibrating materials is important when a solid sample is atomized directly.Few investigations have been carried out into the use of aqueous standards for calibration, with the exception of Cu in carbonate rocks,17 minor elements in silicate rocks,18 Mn and Cu in quartz and rock crystals19 and metals in Zr02.20 Headridge and Nicholson,21 and Atsuya and Itoh,22 however, have used synthetic solid standards having similar compositions to actual samples, which gave more accurate results than those obtained with aqueous standards. The chemical and physical state of the determinants at the atomizing temperature, on which the absorbance and absor- bance versus time profiles of the determinant depend signifi- cantly, have not yet been investigated, with the exception of Zn, Ag, Pb and Bi in steel23 and Cu in Al meta1.24 Although the absorbance versus time profile for liquid samples shows a single peak, determinants in nitrides, carbides and oxides having high melting-points give complicated profiles.The present investigation is concerned with a further application of direct atomization AAS to determine Cu and Pb in Si3N4 and Sic ceramics by using a graphite cup and a graphite furnace. Discussion will be extended to: (1) particle size of sample, (2) amount of graphite powder to be added, (3) chemical state of samples at pyrolysis and atomization steps, (4) optimization of atomization programme, (5) effectiveness of chemical modifier and (6) optimum amounts of samples to be introduced.The effectiveness of aqueous standard solu- tions for the calibration of solid samples will also be dis- cussed. The proposed method was applied successfully to the determination of Cu and Pb in several Si3N4 and Sic ceramic powders. Experimental Apparatus A Hitachi 2-8000 polarized Zeeman atomic absorption spectrometer was used in conjunction with monoatomic90 Analyst, January 1995, Vol. 120 hollow cathode lamps, a photothermal controller and a Hitachi AA data-processor which compensates for back- ground absorbance. Pure Ar (99.995%) was used as the inert gas. The atomization was conducted in a Hitachi No. 180-7402 cup-type graphite cuvette with a graphite cup (3.0 mm diameter x 3.8 mm high externally, 2.5 mm diameter x 3.0 mm high internally) constructed according to Atsuya and Itoh16722 from a spectroscopic graphite rod (Nippon Carbon).The graphite crucible and the graphite cup were heated to 2600 "C prior to use to remove any contaminating material. Pulverized samples were weighed on a Mettler Model M3 microbalance and were placed in the graphite cup using a spoon made from a sheet of parchment paper. The miniature cup was handled with titanium tweezers. Aqueous samples were injected into the graphite cup within the furnace using an Eppendorf Model 4700 micropipette. The powder mill was an Ishikawa Model ALM grinder with a corundum mortar and pestle. The powder mixer was an Iwaki Model MA-1 mill with a polyethylene cylinder (24 x 12 mm i.d.) and a single ball (9 mm diameter). X-ray diffraction patterns were recorded with a Rigaku Rint 1200 X-ray diffractometer with a Cu anode X-ray tube and a graphite monochromator.A secondary electron image was obtained with a JEOL JSM-840 scanning electron microscope. Particle size distribution was measured with a Shimadzu Model SA-CP3L centrifugal particle size analyser . Reagents and Samples Silicon nitride samples were prepared from 12 a-phase-rich powdered materials, one fi-phase-rich powdered material (FP-1) and one p-phase-rich sintered material (Jb-1). Two Sic powders were also used as samples. Graphite powder of about 20 pm particle size was prepared from spectroscopic carbon (Nippon Carbon) by grinding in an agate mortar. Prior to use, it was checked for Cu and Pb contamination. Copper and Pb standard solutions of 1000 ppm m/m were prepared according to the Japanese Industrial Standard (JIS) KO102 procedure.25 Chemical modifier solutions were prepared from nitrates of Mg, Ni, Rh and Pd dissolved in de-ionized water.All reagents used were of extrapure grade. De-ionized water was purified with a Kurita Demi-Ace Model DX-16 de-mineralizer, and filtered through a membrane filter (Millipore AAWPO 700, pore size, 0.8 pm diameter). Procedure The samples, ground to a particle size of less than 20 pm, were mixed with 2.5 times the amount of graphite powder in a corundum mortar. Prior to use, the cuvette and the miniature cup were baked at 2600 "C, and checked for the absorbance of Cu and Pb. A 0.5-3 mg amount of the mixed sample was inserted into the cuvette. The sample was atomized under the instrumental conditions described in Table 1, according to the heating programme described later (Table 2).The absorbance was determined by integration of the spectral lines in the absorbance versus time spectrum. Analyte concentrations were determined against calibration graphs constructed from a suite of standard solutions. Results and Discussion Particle Size As solid samples are essentially heterogeneous, they usually give poor reproducibility in direct atomization analysis. In particular, solid substances are generally polycrystals, and trace elements are distributed in a wide variety of spatial conditions: some are diluted homogeneously within whole crystals, whereas others are concentrated at grain boundaries. Hence, when solid samples are employed in analyses, estab- lishment of the optimum experimental conditions concerning particle size is required in order to provide better repro- ducibility. It has been reported that samples of smaller particle size tend to provide a better reproducibility and an intense absorbance, and that they also exhibit a sharp profile of the absorbance versus time spectrum.Silicate rocks18 and quartz19 with a particle size of less than 5 pm were reported to have provided a good reproducibility. According to Wilson,26 this particle size is suitable for obtaining accurate results. Powders of raw a- and (3-Si3N4 have particle sizes of 0.4-1 pm. They are homogeneous and independent of the procedure used in their production. The direct atomization of the powders provided a good reproducibility [relative stan- dard deviations (s,) ranged from 2 to 9%].However, when being calcined, crystallite agglomerates and impurities which exist homogeneously in Si3N4 crystallites may be forced to diffuse to the grain boundaries. The reproducibility of the absorbance, which is, therefore, assumed to be dependent on the particle size, was examined with calcined materials. The calcined sample composed of the fi-phase (Jp-1) was ground in a corundum mortar and the particle sizes were classified. Mixtures with graphite powder, at a mixing ratio of 1:2.5, were atomized directly and the absorbance was recorded. Samples (modal diameter 1.9 pm) without grinding had an s, of 23% for 35 pprn m/m Cu, and 42% for 0.85 ppm m/m Pb. A longer grinding time decreased the s, and, when the particle size was less than 7 ym, the s, levelled off (5-6%), although the absorbances of Cu and Pb increased slightly (about 5%).These observations indicate that Si3N4 powder can be used without grinding, but when calcined, it should be ground thoroughly to a particle size of less than 10 pm. Table 1 Instrumental conditions for the direct atomization of Cu and Pb in Si3N4 and Sic Element c u Pb Cuvette Analytical line/nm 324.8 283.3 Slit-widthhm 1.3 1.3 Carrier gas flow rate/cm3 min-1 Ar, 200 Ar, 200 Interrupted gas flow ratekm3 min-1 Cup type + graphite cup Lamp current/mA 5.0 7.5 Ar, 30 Ar, 30 Table 2 Experimental conditions for the determination of Cu and Pb in Si3N4 and Sic Element c u Pb Particle size*/ym 5 (0.3-15) Mixing ratio of sample to graphite powder 1 : 2.5 Samples amounthg 1-2 0.5-3 1 : 1-3 Calibrating standard Calibrating rangehg Absorbance Aqueous standard 0-30 0-3 Peak area Atomization conditionst- DryingPC, s 80-120.30 80-120.30 Pyrolysing/"C, s 1500.30 1200,30 Atomizing/"C, s 2500,15 2000,15 * Modal diameter.t The value in each pair is temperature in "C and time in s.Analyst, January 1995, Vol. 120 91 Addition of Graphite Powder In direct atomization of solid samples within a graphite furnace, the mixing of graphite powder with the samples enhances the absorbance, and prolongs the lifetime of the I I I I 0 0.25 0.5 0.75 1 .o AmounVmg Fig. 1 Variation in the atomic absorbance of A, Cu and B, Pb in 0.25 mg of Si3N4 powder as a function of amount of graphite powder added.-- I C T I o"o F' B 0 10 0 10 0 10 0 10 Time/s Fig. 2 Variations in absorbance versus time spcctrum of Cu and Pb for 0.5 mg of Si3N4 powder as a function of amount of graphite powder added. A, Si3N4; B, Si3N4-graphite (1 + 1 ) ; C, Si3N4-graphite (1 + 2); D, Si3N4-graphite (1 + 2.5); and E, Sic-graphite (1 + 2.5). furnace." The determination of nine elements in silicate rocks has demonstrated the effectiveness of this approach.18 For Si3N4 samples, the addition of graphite powder can also be expected to be effective. The variation in the absorbances of Cu and Pb as a function of the amount of graphite powder added to a-Si3N4 (Aa-1) is shown in Fig. 1. The absorbances increased with an increase in the amount of graphite powder added and levelled off above 0.5 mg (mixing ratio of sample : graphite powder, 1 : 2).The absorbance versus time profile of Cu under these conditions is shown in Fig. 2. With increases in the amount of graphite powder, the absorbance versus time profiles become sharper and the spectral lines begin to exhibit double peaks. This tendency was also observed in the analysis of @Si3N4. In the direct atomization of Cu in Sic, the absorbance peak appears at the same elapsed time as the latter peak of the double peak from Si3N4. It is considered that Si3N4 may be converted into Sic in the furnace, from which Cu may be vaporized. Fig. 3 shows the secondary electron images of the Si3N4 powder and the mixture with graphite powder observed with a scanning electron microscope (SEM) before and after atomi- zation, together with the wall of the graphite cup after the atomization.The SEM image obtained before heating [Fig. 3(a)] shows that the particle size of Si3N4 powder is less than 1 pm and distributed uniformly. Silicon nitride appears to adhere to the surface of coarse graphite particles [Fig. 3(b)]. After heating at the atomization temperature (2500 "C) of Cu, the inner wall of the graphite cup is covered with pentagonal dodecahedron-shaped crystals of 2-10 pm particle size, which are assumed to be Sic. The mixture of Si3N4 and graphite powder is converted into a similar aggregate of crystals. Fig. 4 shows the X-ray diffraction patterns of the residual material in the graphite furnace after heating the mixture. In the X-ray diffraction pattern before heating, the (002) line from graphite, and lines from a-Si3N4 and the impurity (3-Si3N4 are visible.Heating for 15 s at 1500"C, the pyrolysing temperat- ure of Cu, produces no change in the crystal components. At a temperature of 2000 "C, the lines from the a-phase disappear and the (111) line from Si and (110) and (200) lines from 8F Fig. 3 Secondary electron images of Si3N4 powder (a) and Si3N4 mixture [(b). (c) and (d)l before [(a and (b)] and after [(c) and ( d ) ] heating at 2500°C for 15 s in a graphite cuvette.92 Analyst, January 1995, Vol. 120 polytype S i c begin to appear. Although the intensities of the lines from the P-phase are reduced at this temperature, the effect is not as marked as for the lines from the wphase. When heated at the atomizing temperature of Cu (2500"C), the diffraction lines from Si3N4 and Si disappear, and only those from 8F Sic and graphite are visible.These results indicate that the mixture of Si3N4 and graphite powder may be converted into Sic in the graphite furnace through the following processes: 1500 "C 1800 "C 2000 "C 2500 "C a-Si3N4 + C(G) + Si + C + Si + SiC(8F) + C + SiC(8F) + C p-Si3N4 + C(G) + Si3N4 + SiC(8F) + Si + C + SiC(8F) + C Considering the chemical composition of the samples at the temperature of each atomization step, the first peak of the double peaks observed in the absorbance versus time spec- trum of Cu in the Si3N4-graphite mixture may be due to the Cu vaporizing from Si, and the second peak to the Cu vaporizing from Sic. It is assumed that Pb can also exhibit double peaks in a similar manner to Cu.However, as Pb has a low atomizing temperature, the second peak may be small and close to the first peak. Although the addition of graphite powder could not eliminate the split in the absorbance versus time spectral lines in the present experiments, the samples were mixed with graphite powder at a mixing ratio of 1:2.5, because this enhanced the absorbance markedly. Heating Programme Fig. 5 shows the variations in the absorbances of Cu and Pb in Si3N4 as a function of the temperature of each heating step. Drying was carried out at 80-120°C for 30 s. The pyrolysis time was 30 s and the atomization time was 15 s. Highly pure argon was employed as the inert gas under the conditions described in Table 1. Silicon nitride powder was mixed with graphite powder at a mixing ratio of 1 : 2.5, and 1.0 mg of the mixture was used in this investigation.The optimum atomization temperatures were established empirically as follows. For Cu: pyrolysis at 1500°C for 30 s, atomization at 2500 "C for 15 s; for Pb: pyrolysis at 1200 "C for 30 s, atomization at 2000°C for 15 s. Modifier In order to prevent the loss of analytes in the pyrolysis step, Mg, Ni, Rh and Pd were employed as chemical modifiers.28 This procedure serves to promote the formation of free atoms in the atomization stage. Palladium, in particular, has often been used as a modifier for the elements of Groups 13-15.29 In order to promote the sensitivity, the effectiveness of Ni and Pd as sensitizers was examined. The addition of an aqueous solution of Ni(N03)2 or Pd(N03)2 to 1 mg of the Si3N4- graphite powder mixture reduced the linewidth of the absorbance versus time spectrum by about 10%. The absor- bances were not enhanced.Sample Amounts In the atomization of solid samples, the use of too large an amount of sample results in a lowering of the absorbance. This is partly because the sample itself reduces the conductivity of heat, and partly because the sample material is not atomized completely. On the other hand, too small an amount of sample leads to large errors in weighing. The variation in absorbance 20 30 40 28 (degrees) Fig. 4 X-ray diffraction patterns of Si3N4-graphite powder (1 + 2.5) calcined in a graphite cuvette. a, a-Si3N4; p, P-Si3N4; G, graphite; Si, metallic Si; Sic 8F, 8F polytype of Sic.(a) Not calcined; (b) calcined at 1500 "C; (c) calcined at 2000 "C; and (d) calcined at 2500 "C. - 0 ' I I I I 1 1000 1500 2000 2500 Temperature/% Fig. 5 Variation in the atomic absorbance of Cu (0,O) and Pb (A, A) in Si3N4-graphite powder (1 + 2.5) as a function of pyrolysing (0, A) and atomizing (0, A) temperature. Amount of Si,N,/mg 0 0.5 2 2 4.0 ?i g 2.0 8 2 3.0 T) c a, c - 1 .o 0 50 100 Amount of Cdng Fig. 6 amount of Cu. 0, in Si3N4; and 0, in aqueous standard. Variation in the atomic absorbance of Cu as a function ofAnalyst, January 1995, Vol. 120 93 with sample amount was examined with amounts of Si3N4- graphite powder (1 : 2.5) from 0.1 to 4.0 mg, which is the upper limit of the capacity of the cup. The results are shown in Figs. 6 and 7, where calibration graphs are constructed from a suite of standard solutions.With increasing amounts of Si3N4 powder, the absorbances initially increase proportionally. However, the linearity in the absorbances of Cu and Pb breaks down beyond 0.5 and 2.0 mg, respectively: the upper limits of calibration are 40 ng of Cu and 6 ng of Pb. The calibration graphs constructed from suites of standard solutions show the same slopes as those of a series of solid samples, and the calibration graphs are linear up to 50 ng of Cu and up to 6 ng of Pb, which is almost identical with the results obtained for the direct atomization of Si3N4. Therefore, the lowering of the absorbance is considered to be due to a pressure effect. The above observations indicated that Amount of Si,NJmg 0 0.5 e 0.4 3 8 9 0.3 0 E 0.2 c - 0.1 0 5 10 Amount of Pblng Variation in the atomic absorbance of Pb as a function of Fig.7 amount of Pb. A, in Si3N4; and A , in aqueous standard. 1-2 and 0.5-3 mg of the sample-graphite powder mixture (1 : 2.5) provides suitable experimental conditions for the determination of Cu and Pb, respectively. Application Concentrations of Cu and Pb in 14 Si3N4 and two Sic samples, obtained under the proposed experimental conditions, are given in Table 3. Copper and Pb were quantified against calibration graphs constructed from a suite of standard solutions. Each value in Table 3 is the average of five measurements; the s, are 39% at 0.11 pprn m/m and 1.6% at 53 ppm m/m of Cu, and 17% at 0.36 pprn d m and 4.3% at 1.55 ppm m/m of Pb, respectively.Values obtained by ETAAS for liquid samples are also given in Table 3 for comparison. The values agree well with each other. The determination limits, corresponding to twice the standard deviation of the blank measurements, are 92 pg for Cu and 52 pg for Pb. The authors thank the Machine Shop, Meiji University, for making the graphite cups. This work was supported financially by a 1993 grant-in-aid from Meiji University. References 1 Itoh, N., Sasamoto, T . , Sata, T., Komeya, K., and Tsuge, A., Yogyo-kyokai-shi, 1982,90, 209. 2 Ishizuka, T . , Uwamino, Y., and Tsuge, A., Bunseki Kagaku, 1984, 33,486. 3 Miyagawa, Y . , Saito, K., Niwa, H., Ishizuka, T . , and Miya- gawa, S., Bunseki kagaku, 1985,34, 766. 4 Tsuge, A., Uwamino, Y., and Ishizuka, T., Yogyo-kyokai-shi, 1985, 93, 182. 5 Belaya, K.P., Kustova, L. V., and Kichina, T. M., Zavod. Lab., 1987, 53, 33. 6 Matsunaga, H., and Hirate, N., Bunseki Kagaku, 1988, 37, T215. 7 Langmyhr, F. J . , Analyst, 1979, 104,993. Table 3 Cu and Pb concentrations in Si3N4 and S i c samples Element c u Pb S-ETA AS* L-ETA AS? S-ETA AS* L-ETA AS? Concentration Concentration Concentration Concentration Sample (PPm) s r (YO)' ( P P ~ ) ( P P 4 sr (%)* ( P P 4 Aa-1 24.1 3.2 25.7 0.55 14 0.57 Aa-2 20.7 3.9 21.6 1.55 4.3 1.51 Aa-3 21.8 6.6 20.3 0.97 9.6 0.87 Aa-4 22.2 4.0 21.1 0.92 9.0 0.88 B a-2 0.47 18 0.29 0.49 17 0.42 Ca-1 5.1 6.3 5.3 0.57 14 0.60 Ca-2 0.54 22 0.70 0.46 13 0.53 Da-1 0.71 4.0 0.75 0.36 11 0.35 EE- 1 0.13 39 NDS 0.37 7.6 0.41 Fa-2 12 22 10 0.39 16 0.42 FP-1 0.93 4.5 1.1 26 3.1 28 Ga-1 11.4 1.6 10.3 0.40 14 0.40 Ha-1 8.2 2.6 8.4 0.86 6.8 0.88 Jp-1 35.1 7.7 54.2 0.84 7.9 0.80 Sc-17 28.8 4.9 25.0 0.62 9.8 0.59 Sc-27 0.11 15 NDS NDS NDB * Solid ETAAS, proposed method.t Liquid ETAAS. One gram of sample was dissolved with HF-HN03 in a Teflon pressure vessel and the absorbance of a 10 mm3 portion of the * n = 5 . 7 Sic. sample solution was recorded. ND = Not detected.94 Analyst, January 1995, Vol. 120 8 9 10 11 12 13 14 15 16 17 18 19 Ng, K. C., Sample Introduction in Atomic Spectroscopy, ed. Sneddon, J . , Elsevier, Amsterdam, 1990, pp. 147-163. Katsukov, D. A., Kruglikova, L. P., and L’Vov, B. V., Zh. Anal. Khim., 1975, 30, 238. Graule, T., von Bohlen, A., and Broekaert. J. A. C., Fresenius’ Z. Anal. Chem., 1989,335, 637. Koizumi, H., and Yasuda, K., Anal. Chem., 1976, 48, 179. Koizumi, H., Yasuda, K., and Katayama, M., Anal. Chem., 1977, 49, 1106. Smith, S. B., and Hieftje, G. M., Appl. Spectrosc., 1983, 37, 419. Chakrabarti, C. L., Wan, C. C., and Li, W. C., Spectrochim. Acta, Part B, 1980,35, 93. Price, W. J., Dymott, T. C., and Whiteside, P. J., Spectrochim. Acta, Part B, 1980, 35, 3. Atsuya, I., and Itoh, K., Bunseki Kagaku, 1982,31, 708. Nakamura, T., Okubo, K., and Sato, J . , Anal. Chim. Acta, 1988, 209, 287. Nakamura, T., Oka, H., Morikawa, H., and Sato, J., Analyst, 1992, 117, 131. Nakamura, T., Sasagawa, R., and Sato, J., Bunseki Kagaku, 1992, 41, 89. 20 21 22 23 24 25 26 27 28 29 Nakamura, T., Oka, H., Ishii, M., and Sato, J., Analyst, 1994, 119, 1397. Headridge, J. B., and Nicholson, R. A., Analyst, 1982, 107, 1200. Atsuya, I., and Itoh, K., Bunseki Kagaku, 1982, 31, 713. Takada, K., and Hirokawa, K., Talanta, 1982, 29, 849. Takada. K., Anal. Sci., 1987, 3, 221. Japanese Industrial Standard K0102, Japanese Standards Asso- ciation, Japan, 1971. Wilson, A. D., Analyst, 1964, 89, 416. Langmyhr, F. J., Stubergh, J. R., Tomassen, Y., Hassen, J. E., and Dolezal, J., Anal. Chim. Acta, 1974, 71, 35. Ni, Z.-M., and Shan, X.-Q., Spectrochim. Acta, Part B , 1987, 42, 937. Matsumoto, K., Anal. Sci., 1993, 9, 447. Paper 4102927E Received May 17, 1994 Accepted July 26, 1994

ISSN:0003-2654    

DOI:10.1039/AN9952000089

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 95 Comparison of Reflux and Microwave Oven Digestion for the Determination of Arsenic and Selenium in Sludge Reference Material Using Flow Injection Hydride Generation and Atomic Absorption Spectrometry Rajananda Saraswati,* Thomas W. Vetter and Robert L. Watters, Jr.? Inorganic Analytical Research Division, Chemical Science and Technology Laboratory, National Institute of Standards and Technology, Gaithersburg, M D 20899, USA A microwave oven digestion procedure was developed for rapid dissolution of sludge samples (NIST SRM 2781) for the determination of As and Se. Microwave oven digestion with HN03 and H2S04 provides results comparable to those obtained by reflux column digestion using HN03, H2S04 and HC104. The experimental details for sample preparation and the flow injection hydride generation atomic absorption spectrometric method are discussed.The effects of matrix and various acid concentrations on the extraction and absorbance were also studied. The proposed method has detection limits of 0.15 and 0.17 ng ml-1 for As and Se, respectively. Keywords: Microwave oven digestion; flow injection; arsenic and selenium determination; sludge; standard reference material Introduction The determination of arsenic and selenium in various complex matrices such as soils, plants and biological samples by the heated quartz cell atomization with hydride generation technique has been described by several workers. 1-4 Domestic sewage sludge is a challenging matrix for the determination of As and Se because of its complex mixture of organics and inorganics.Sludge or related material is usually decomposed by wet digestion with concentrated nitric, sulfuric and perchloric acids for the determination of total arsenic and selenium. 195-7 Although complete recovery is possible by wet digestion, there are dry mineralization procedures that are also widely used for some of the biological samples because of their simplicity and safety.*-10 However, these traditional sludge decomposition techniques for the determination of arsenic and selenium are tedious and time consuming. The microwave decomposition technique seems to be the most preferable for preparing samples for the determination of As and Se. Although there are a number of fast and efficient acid decomposition procedures using microwave ovens for the determination of As and Se in a variety of samples,3.11,12 no reports have been published on the decomposition of sludge material by a microwave technique.Flow injection analysis (FIA) methodologies can be com- bined with atomic absorption spectrometry (AAS) to improve the speed, precision and sensitivity of the hydride generation * On leave from the Defence Metallurgical Research Laboratory, Hyderabad, India. To whom correspondence should be addressed. AAS method even further. Through FIA it is possible both to miniaturize and to automate large-scale and manual analytical procedures. FIA has found wide application in many fields13-17 and its use is rapidly increasing because of the simplicity of the method. The aim of this work was to develop both a rapid digestion procedure using a microwave oven and an accurate method for the determination of arsenic and selenium in the proposed NIST Standard Reference Material (SRM) Domestic Sewage Sludge (SRM 2781) by FIA hydride generation and AAS.The method involved microwave oven digestion and the reduction of As and Se ions by sodium tetrahydroborate followed by sweeping the hydrides into an absorption cell and detection. Experimental Instrumentation * A commercial microwave oven (CEM Model MDS-2100 with 950 W power) was used for decomposing the samples. Digestion vessels (100 ml) consisting of a Teflon-PFA inner liner with a poly(ether imide) outer casing that can withstand a pressure of 1380 kPa (200 psi) and a temperature of 200°C were used.Absorption measurements were made using a commercial FIA system (Perkin-Elmer FIAS-200) interfaced to an atomic absorption spectrometer (Perkin-Elmer Model 5000). Specific instrument parameters for FIA are detailed in Table 1. The FIA was computer controlled (386SX running MS-DOS 5.0) with two peristaltic pumps, four switching valves, a sample injection loop, a mixing coil and a liquid-gas separator. Dilute (10% v/v) hydrochloric acid (HCI) was used in the carrier stream to sweep the sample from the injection loop to the mixing coil where it reacted with a solution of 0.25% (w/v) sodium tetrahydroborate (NaBH4) stabilized in 0.05% sodium hydroxide (NaOH). The gaseous hydrides of As or Se producd were separated from the solution phase (in the liquid-gas separator) and swept by a stream of argon to a resistively heated (9OOOC) quartz absorption cell mounted in the light path of a electrodeless discharge lamp (As or Se) in the atomic absorption spectrometer. Peak height measurement was used for accurate quantification.' To describe experimental procedures adequately it was occasionally necessary to identify commercial products by manufacturer's name or label. In no instance does such identification imply endorsement by NIST, nor does it imply that the particular product or equipment is necessarily the best available for that purpose.96 Analyst, January 1995, Vol. 120 Reagents High-purity acids obtained by a sub-boiling distillation pro- cess18 at the National Institute of Standards and Technology (NIST) and high-purity de-ionized water (18 Mi2 cm-1) were used throughout for preparing solutions.Microwave oven decomposition vessels and all glassware used for the prepara- tion and storage of solutions were pre-cleaned by soaking in nitric acid (HN03) and rinsed with high-purity de-ionized water. Standard solutions of As and Se were prepared by serial dilution of SRM 3103 and SRM 3149, spectrometric arsenic and selenium standard solutio-ns, to the required levels. The reductant solution was prepared by dissolving 2.5 g 1-1 analytical-reagent grade NaBH4 in 0.05% NaOH solution. The reductant solution was filtered using vacuum filtration through a 0.45 p.m Millipore filter. Reagent solutions were prepared fresh daily. Sample Drying Separate samples of SRM 2781 were dried in a vacuum oven at ambient temperature for 24 h to obtain mass loss data.SRM 2709 (San Joaquin Soil) was dried in oven at 110°C for 2 h. Undried samples were analysed to avoid the effects of volatile element losses. The results for the undried samples were multiplied by the mean undried to dried sample mass ratio obtained from the separate drying experiment. All certified values are referenced to the most stable basis mass obtainable for the material, viz., the dried mass. Reflux Digestion Procedure This procedure was similar to that of Rains and Menis.19 One sample from each of six bottles of SRM 2781 was prepared, together with two samples of SRM 2709 as a control and two blanks. Undried samples, weighing approximately 1 g each, were transferred into 250 ml distillation flasks and 10 ml of concentrated HN03 and 10 ml of sulfuric acid (H2S04) were added.The flasks were then placed on a reflux condenser and the samples were refluxed (approximately 100 "C) for about 3 h. Then 10 ml of perchloric acid (HC104) were added and the samples were heated for about 5 h until fumes of H2SO4 appeared (approximately 300 "C). The samples were cooled and 5 ml of HCl were added. The samples were heated for 30 Table 1 FIA instrumental parameters for the determination of As and Se in SRM 2781 Parameter As Se Wavelength Source Expansion Measurement time FIAS fill time FIAS inject time Reductant Carrier Cell temperature Sample loop Carrier gas flow rate Reductant flow rate Waste flow rate Reaction coil length Stripping coil length 193.7 nm Perkin-Elmer electrodeless discharge lamp at8W l x Peak height 15 s 10 s 15 s 0.25% NaBHj in 0.05% NaOH 10% HCI 900 "C 10 ml min-1 6 ml min-1 24 ml min-1 110 mm 300 mm window 40 PI 196.0 nm Perkin-Elmer electrodeless discharge lamp at6W l x Peak height 15 s 10 s 15 s 0.25% NaBHj in 0.05% NaOH window 10% HCI 900 "C 40 yl 10 ml min-1 6 ml min- * 24 ml min-1 110mm 300 mm min at near boiling (approximately 110 "C), transferred into 100 ml calibrated flasks and diluted to volume with water.The resultant solution contained a small amount of white precipi- tate, which was determined to be silica by electrothermal AAS (ETAAS). The precipitate was allowed to settle to the bottom of the flask before aliquots were taken for analysis. Microwave Oven Digestion Procedure Four procedures using different combinations of acids were attempted with the microwave oven digestion.Each pro- cedure incorporated a preliminary stage, at relatively lower temperatures and pressures, as a precaution against rapid generation of decomposition gases or exothermic reactions that can sometimes occur with samples containing organic matter. Approximately 0.25 g samples and blanks were digested with acids in microwave oven decomposition vessels using the experimental conditions detailed in Table 2 for Methods A-C and Table 3 for Method D. After the digestion programme was completed, the vessels were cooled in ice-water for 15 min. Then, for each vessel, the cap was carefully removed and the contents were transferred into a 100 ml calibrated flask and diluted to volume with 2% v/v HCI.Method A Samples were digested with 6 ml of concentrated H2S04 and 8 ml of concentrated HCl and were subjected to a maximum temperature of 190°C and a maximum pressure of 410 kPa (60 psi). Method B Samples were digested with 6 ml of 30% hydrogen peroxide (H202), 2 ml of concentrated HCl and 6 ml of concentrated Table 2 Experimental conditions for microwave oven digestion of sludge (SRM 2781) by Methods A, B and C Stage Parameter 1 2 Power (%) 70 100 Pressure/kPa 69 1170 Run time/min 20 35 Time at Flmin 10 20 TemperaturePC 120 190 Fan speed (%) 100 100 * Time at P indicates the amount of time the digestion was held at the controlling parameter ( P ) of pressure or temperature. The microwave oven would heat the sample at the power level indicated until the pre-set parameter was reached and hold the sample at that parameter for the indicated amount of time.For these samples the temperature was the controlling parameter as the pre-set pressures were never reached. Table 3 Experimental conditions for microwave oven digestion of sludge (SRM 2781) for Method D Stage Parameter 1 2 3 4 5 Power (YO) 100 80 80 80 80 Pressure/kPa 280 690 1030 1310 1310 Run time/min 15 10 10 20 15 Time at Flmin 5 3 3 10 10 TemperaturePC - 190 190 190 190 Fan speed (%) 100 100 100 100 100 * See Table 2.Analyst, January 1995, Vol. 120 97 H2S04, and were subjected to a maximum temperature of 190°C and maximum pressure of 360 kPa (52 psi). Method C Samples were digested with 6 ml of 30% H202 and 6 ml of concentrated H2SO4, and were subjected to a maximum temperature of 190 "C and a maximum pressure of 220 kPa (32 psi).Method D Samples were digested with 5 ml of concentrated HN03 and 5 ml of concentrated H2SO4, and were subjected to a maximum temperature of 175°C and a maximum pressure of 1310 kPa (190 psi). In addition, prior to diluting the samples to volume, the contents of each vessel were transferred into a 125 ml Pyrex Erlenmeyer flask. All the HN03 was removed (as indicated by the cessation of fumes) by about 20 min of gentle heating of the solution (approximately 25 ml) on a hot-plate that had a surface temperature of about 175 "C. Reduction of As and Se As"' and SeIV show maximum affinity to form their hydrides, and it was therefore essential to pre-reduce AsV and Sevl to As"' and SerV before performing analysis by hydride genera- tion.Arsenic A 10 ml volume of the above sample solution was transferred into a 50 ml calibrated flask. Then 5 ml of reducing agent, consisting of 10% m/v potassium iodide (KI) and 5% m/v ascorbic acid solution, were added to the samples and calibration standards, which were allowed to react for about 1 h in order to reduce As prior to analysis. Selenium Samples (and standards) to be analysed for Se were pre- reduced with 1 + 1 HCl. A 10 ml volume of the original solution was taken in a 100 ml calibrated flask and diluted to volume with 1 + 1 HCI. The contents were heated in a water-bath (9OOC) for about 25 min and then cooled to room temperature before analysis. Analysis Procedure Sample solution concentrations were determined by establish- ing a linear response with a calibration graph prepared using four standard solutions ranging from 5 to 20 ng ml-L of As and from 2 to 10 ng ml-1 of Se that were matched in acid and reducing agent concentrations with the samples.The recovery of single spikes within the linear range for As (93%) or Se (90%) in the sample matrix was determined and corrections were applied by the single standard addition method, in which a sample was diluted 1 + 1 with either a blank or a standard of known As or Se concentration. Results and Discussion The conventional reflux wet digestion procedure for the complete recovery of As and Se from sludge samples takes a very long time and involves a dangerous combination of acids such as HN03, H2S04 and HC104, accompanied by heating.As a better alternative, a microwave oven digestion procedure was developed and tested for a sludge reference material (SRM 2781). Different combinations and volumes of HN03, H2SO4, HCl and H202 and different time and power settings of the microwave oven using single- and multi-stage power and time settings were tested to ensure the total recovery of As and Se. Of Methods A to D discussed above, only Method D, using HN03 and H2SO4, yielded a complete recovery of As and Se. The recoveries of As and Se obtained by the different digestion procedures are given in Table 4. The combination of HCl and H2SO4 used in Method A is not workable because of the lack of an oxidizing agent that could effectively oxidize organoarsenic or organoselenium compounds in the sludge. However, some of the inorganic As and Se were leached by the combination of these acids, which yielded the lower values obtained by this method.The low recovery obtained by Methods B and C, using HCI, H202 and H2S04, might be a result of incomplete digestion of the sludge, which includes incomplete decomposition of organoarsenic or organosele- nium compounds. Although H202 is an oxidizing agent, it seems that it was not fully effective at the temperature and pressure that were used. It was not possible to reach higher temperatures as the microwave oven digestion vessels used were not capable of withstanding temperatures higher than 200 "C. The lack of oxidation using HCI and/or H202, relative to that obtained with HN03, is evident from the pressures obtained in the microwave-oven decomposition (Table 4).Only decomposition with HN03 yielded high pressures, which are indicative of the decomposition of organic compounds to gases. The condition of the solutions after microwave oven digestion gives another indication of the lack of oxidation in the absence of HN03. The digested solutions from Methods A-C were light yellow and had black particles in them, whereas the solutions from Method D were of similar colour Table 4 Results obtained for the analysis of sludge (SRM 2781) using different microwave oven digestion methods Maximum Maximum No. of Mean As Mean Se temperature pressure determi- concentration found/ concentration found/ Digestion method attained/"C attained/kPa nations pg g- I* I.18 g-'* A 190 410 3 0.26 f 0.05 0.16 k 0.12 B 190 360 4 0.36 f 0.03 0.17 k 0.01 C 190 220 3 0.38 k 0.19 0.41 f 0.10 D 175t 1310 6 7.94 f 0.57 17.53 k 1.07 (HN03 + H~SOJ) (HCI + H2SO4) (HCI + H202 + H2S04) 0 4 2 0 2 + H2S04) * All uncertainties are expressed as the expanded uncertainty, U , calculated according to the CIPM approach.25 t The maximum temperature of 175 "C obtained for Method D occurred in the first minute of stage 4 and because the microwave power was limited by the pressure parameter the temperature levelled off within 4 min to between 155 and 160°C and stayed in that range for the remainder of the digestion.98 Analyst, January 1995, Vol. 120 but contained white particles.The solutions from the reflux procedure also contained white particles.The black precipi- tate (Methods A-C) and white precipitate (Method D and reflux procedure) were allowed to settle to the bottom of the flask before aliquots were taken for analysis. In a separate experiment, the black precipitate, which was probably car- bon, dissolved when the solution was heated (approximately 50°C) and 2-3 ml of concentrated HN03 were added. In another experiment the white precipitate was determined to be silica by ETAAS. Acid combinations that excluded HN03 were attempted initially because it is well known that the presence of HN03, even at low concentrations (below 0.01 mol ]-I), will suppress arsine generation.’ Nitric acid also seriously interferes with the preliminary reduction of As to lower oxidation states with KI-ascorbic acid by oxidizing iodide ion to an iodine precipi- tate.Although excess of HN03 creates many problems in analysis by FIA, the complete extraction of As or Se with HN03 was critical to the success of the experiment, as incomplete recovery of these elements could not be tolerated. The microwave oven digestion procedure (Method D) with the HN03-H2S04 mixture was further studied to optimize the effect of acid concentration on the FIA determination of As and Se and the recovery of these elements from the sludge. The amount of HN03 required for complete recovery of these elements from the samples was determined by digesting identical samples with various concentrations of a mixture of HN03 and H2S04. The optimum amount of HN03 required was 5-6 ml when combined with 5-6 ml of H2SO4, with a sample mass of 0.25 g in the digestion vessel.Fig. 1 shows the effect of the amount of HN03 used for the microwave oven digestion on the absorbance value obtained for As and Se. Arsenic and selenium are known to be present in more than one oxidation state in a sample solution.20 Success in determining these elements by the flow injection hydride generation technique depends on the elements having been brought into one definite oxidation state before the reduction by NaBH4. This is essential because the rate of reaction of reduction to the hydride differs from one oxidation state to another, resulting in different sensitivities in the final excita- tiodatomization step. Usually, higher oxidation states possess lower sensitivity in the hydride generation processes because of different rates of reaction.21.22 Arsenic was reduced by 0.08 I I I g g T I v) 0.03 3 0.02 0.01 0 1 2 3 4 5 6 Nitric acidlml Fig.1 Effect of nitric acid on the extraction of A , arsenic and B, selenium. KI-ascorbic acid solution and Se was reduced by 1 + 1 HCI as discussed earlier. The optimum amounts of KI and ascorbic acid to be added for the reduction of As into its lower oxidation state were investigated by adding different amounts of reagent to a series of sample solutions having the same As concentration. It was found that 0.01 g ml-1 of KI and 0.005 g ml-1 of ascorbic acid were sufficient to reduce 0.1 pg ml-1 of As to its lower oxidation state. However, when As and Se in the same solution are to be determined simultaneously as the hydrides, by FIA-AAS, a problem arises.KI-ascorbic acid solution, which is an effective reductant for AsV, will also reduce Sevl and SeIV to elemental selenium, whereby the hydride formation of the ions suffers. On the other hand, in the presence of dilute HCI (1 + l), which is used for reducing Sevl, the reduction potential of AsV to Asi1’ is lowered, thus changing the kinetics of arsine generation, which leads to lower sensitivities.23 Therefore, separate aliquots were taken from the original solution and reduced separately for As and Se . The effects of carrier solution and NaBH4 concentration on the generation and detection of hydrides were determined by varying the concentration of these components. The best absorbance peak was obtained when the concentration of carrier solution was in the range 10-15% and the reductant concentration in the range 0.2-0.3%.The effect of carrier gas flow rate was studied by changing the argon flow rate. The signal increased with increase in the flow rate up to 70 ml min-1, after which it reached a plateau. At argon flow rates below 30 ml min-* serious memory effects were created because of inefficient sweeping of the hydrides to the quartz cell. With the arrangement of 10% HCI as carrier solution, 0.25% reductant solution and a 70 ml min-1 argon flow rate, the hydride generation worked smoothly and the sensitivity and reproducibility of hydride generation were maximized. Measurement Uncertainty Good linearity was observed in the ranges 5-20 ng ml-1 of As and 2-10 ng ml-1 of Se.The detection limits, defined as the As or Se concentration corresponding to three times the standard deviation of the blank, were 0.15 ng ml-1 of As and 0.17 ng ml-1 of Se. The characteristic concentrations (0.1 absor- bance) were 0.01 pg ml-1 of As and 0.03 pg ml-1 of Se. The accuracy of this method was tested by determining As and Se in NIST SRM 2709 San Joaquin Soil.24 Uncertainties were calculated using the CIPM approach25 and are reported here as a relative percent of the sample value. Type A uncertainties are calculated from the sample measurement imprecision based on the results for six samples of sludge and Type B uncertainties are calculated from estimates of the measurement systematic uncertainty. The Type A uncertainty was 1.6% by reflux and 2.4% by microwave oven digestion for arsenic.For Se the Type A uncertainty was 1.6% for both the reflux and microwave oven methods. The Type B uncertainty was 2.4% for As and 2.6% for Se. The total expanded Table 5 Comparison of results obtained by reflux digestion and microwave oven digestion methods SRM 2781* SRM 2709 (contro1)t Method Aslpg g-lg Selyg g-l$ Aslpg g-l$ Selpg g-l+ Reflux digestion 7.88 k 0.46 17.34 f 1.07 16.78 k 0.98 1.56 f 0.10 Microwave digestion 7.94 k 0.57 17.53 f 1.07 16.89 f 0.99 1.68 f 0.10 Certified value23 - - 17.7 k 0.8 1.57 ? 0.08 (Method D) * n = 6 . t n = 2. * All uncertainties are expressed as the expanded uncertainty, U, calculated according to the CIPM approach.”Analyst, January 1995, Vol. 120 99 uncertainty for As was calculated to be 5.8% using reflux and 7.1% using the microwave oven method.For Se the total expanded uncertainty was calculated to be 6.2% for both the reflux and microwave oven methods. The total expanded uncertainty of the samples was converted into a relative uncertainty to estimate the uncertainty of the controls as only two samples of SRM 2709 were analysed. The analysis of these materials yielded values and uncer- tainties (Table 5 ) that show no evidence of bias compared with the certified values. The results obtained by reflux decompo- sition and microwave oven decomposition are comparable (Table 5 ) . Conclusion Microwave oven digestion of sludge with HN03-H2S04 in a pressurized PFA vessel provides a reasonable alternative to the conventional wet digestion procedure for the determina- tion of As and Se.As evidenced by the results of flow injection hydride generation atomic absorption spectrometry, the detection limits, precision and accuracy are similar for the two methodologies. Decomposition with HN03 is necessary as microwave oven digestion with HCI and/or H202, in combina- tion with HZS04, results in low yields of As and Se. Although HN03 can suppress arsine generation, its effect can be minimized by evaporation with gentle heating. Sample prep- aration times are reduced from 10-12 h using a wet digestion reflux to about 1-2 h using the microwave oven. One author (R.S.) acknowledges the support of the Depart- ment of Science and Technology, New Delhi, India, and NIST. References 1 Maher, W. A., Talanta, 1983, 30, 534.2 Van Der Veem, N. G., Keukens, H. Chim. Acta, 1985, 171, 285. J., and Vos, G., Anal. 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Ybanez, N . , Cervera, M. L., and Montoro, R., J. Anal. Ar. Spectrom., 1991, 6, 379. Webb, D. R., and Carter, D. E . , J . Anal. Toxicol., 1984,8,118. Morita, M., Uehiro. T., and Fuwa, K . . Anal. Chem., 1980, 53. 1806. Agemian, H., and Thomson, R., Analyst, 1980, 105, 902. Jin, K . , Ogawa. H., and Taga, M., Bunseki Kagaku. 1983,32, 171. Sturgeon, R. E . . Willie. S. N., and Berman, S. S., J. Anal. At. Spectrom., 1986, 1, 115. Krynitsky, A. J., Anal. Chem., 1987, 59, 1884. Brumbaugh. W. G., and Walther, M. J . , J. Assoc. Off. Anal. Chem., 1989. 72, 484. Nakashima, S., Sturgeon, R. E., Willie, S. N., and Berman, S. S., Analyst. 1988, 113, 159. Matusiewicz, H . , Sturgeon, R. E., and Berman, S. S., J . Anal. At. Spectrom., 1989,4, 323. Tyson, J . F., Spectrochim. Acta. Rev., 1991, 14, 169. Kuroda, R . , Oguma, K., Kitada, K., and Kozuka, S., Talanta, 1991, 38, 1119. Maria, C. G., and Townshend, A., Anal. Chim. Acta, 1992, 261, 137. Rgiicka, J . , and Hansen, E. H., Flow-Injection Analysis, Wiley-Interscience, New York, 2nd ed., 1990. Valcarcel, M.. and Luque de Castro, M. D., Flow-Injection Analysis: Principles and Applications, Ellis Horwood, Chiches- ter, 1987. Kuehner, E. C., Alvarez, R., Paulsen, P. J., and Murphy, T. J., Anal. Chem., 1972,44, 2050. Rains. T. C., and Menis, 0.. J. Assoc. Off. Anal. Chem.. 1972, 55, 1399. Bye, R., Talanta, 1990, 37, 1029. Castillo, J . R., Mir, J. M., Martinez, C., and Gomez, M. T., Fresenius' 2. Anal. Chem., 1986, 325, 171. Brooke, P. J . , and Evans, W. H., Analyst, 1981, 106, 514. Agget, J., and Aspell, A. C., Analyst, 1976, 101, 341. Certificate of Analysis, Standard Reference Material 2709 (San Joaquin Soil), National Institute of Standards and Technology, Gaithersburg, MD. 1992. Guide to the Expressions of Uncertainty in Measurement, ISO, Geneva, 1993. Paper 31068680 Accepted April 22, 1994

ISSN:0003-2654    

DOI:10.1039/AN9952000095

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 101 Determination of Cadmium and Lead in Vegetables After Activated-carbon Enrichment by Atomic Absorption Spectrometry Mehmet Yaman Firat University, Science and Arts Faculty, Department of Chemistry, Elazig, Turkey Seref Giiqer Inonii University, Science and Arts Faculty, Department of Chemistry, Malatya, Turkey An enrichment method has been developed for the determination of cadmium and lead in vegetable matter by flame atomic absorption spectrophotometry after preconcentration with 8-hydroxyquinoline or cupferron on activated-carbon. An enrichment factor of up to 100 was achieved by using both of the mentioned complexing reagents. The optimum lowest pH values were found as 4.8 and 4.4 with cupferron, and 5.3 and 5.8 with oxine, for cadmium and lead, respectively.The effect of decomposition procedures, such as dry and wet ashing, on the recovery were also studied and compared. The relative standard deviations were found to be 2% for cadmium (7 pg 1-1 Cd) and 3% for lead (70 pg 1-1 Pb). Vegetable samples from fertilized farmland and roadside areas of Elazig, Turkey were analysed by using the proposed method. Keywords: Cadmium; lead; activated-carbon; oxine; cupferron; atomic absorption spectrometry; vegetable Introduction Determination of trace elements below pg g-1 levels in vegetables has an increasing importance in food and agricul- tural chemistry. Cadmium and lead are among the toxic elements that can be harmful to plants and, through their introduction into the food chain, to human health.Toxic metals may enter the food chain from a number of different sources, such as soil, sewage-sludge and fertilizers.' The levels of cadmium and lead in soils from Malatya, Turkey2 and the relationship between their concentration in soil and apricots has been investigated.3 Electrothermal atomic absorption spectrometry (ETAAS) has mostly been used for the determination of cadmium and lead in biological matrices.4-9 However, improving the sensi- tivity, by using an enrichment step in the analytical meihod, flame atomic absorption spectrometry (FAAS) can also be used for their determination.10 Table 1 Operating parameters for the atomic absorption spectropho- tometer Parameters Cadmium Lead Wavelengthhm 228.8 283.3 HCl Current/mA 8.0 10.0 Air flow rate/] min-1 10.0 11.0 Acetylene flow rate/l min-1 2.7 2.8 Slit width/nm 0.7 0.7 Solution flow rate/ml s-l* 0.15 0.15 * In TC2 signal position, the meter reaches its final steady value at 6 s.22 Activated-carbon has been used for the enrichment of elements in different matrices2,'71*-16 but not for the determi- nation of lead and cadmium in vegetable matter.In our previous work, vanadium levels in vegetable matter were determined by using an 8-hydroxyquinoline-activated carbon enrichment method with FAAS.17 In this study, a similar enrichment method was modified by using 8-hydroxyquinoline, or cupferron, for the determination of cadmium and lead in vegetables. Experimental Apparatus A Perkin-Elmer (Norwalk, CT, USA) Model 370 atomic absorption spectrophotometer equipped with Perkin-Elmer single- and multi-element hollow cathode lamps were used for the determinations.The optimum conditions for atomic absorption spectrometry are shown in Table 1. The pH was measured with an Electro-Mag M-822 pH meter (Instanbul, Turkey). For the enrichment procedure, a Labor Teknik MK-300 magnetic stirrer (Ankara, Turkey), Hettich EBA 111 centrifuge (Tiitlingen, Germany) and an Electro-Mag Model 1810 ashing furnace were used. Reagents All chemicals were of analytical-reagent grade, unless other- wise stated. Doubly distilled water was used for the prepara- tion of solutions. In the digestion procedures, concentrated nitric acid (65%, 14.4 rnol 1-I), sulfuric acid (98%, 18.0 rnol I-I), hydrogen peroxide (35%, 11.4 rnol 1-1) and perchloric acid (65%, 9.9 rnol I-*) were used.Stock solutions of cadmium and lead (1000 pg 1-1) were prepared by dissolving CdS04-8H20 (Merck, Darmstadt, Germany) and Pb(NO& (Merck) in 0.1 rnol 1-1 nitric acid. The other stock solutions were prepared from reagent-grade metals, or their salts, and acidified with high-purity nitric or hydrochloric acid (both 0.1 rnol 1-1). Buffer solutions of pH 2.2-6.0 were prepared by adding 0.1 rnol 1-1 HCI or 0.1 rnol 1 - 1 NaOH solution to 0.1 rnol 1 - 1 sodium citrate solution. For other buffer solutions (pH 6.0-7.0 and pH 8.0-10.0) 0.066 rnol 1-1 Na2HPO4.2H20 and 0.066 rnol 1-1 KH2P04 and 13.2 rnol 1-1 ammonia and ammonium chloride solutions were used, respectively. The solutions of 0.2% (0.014 mol 1-1) 8-nydroxyquinoline (oxine) and N-nitroso phenyl hydroxilamine (cupferron, 0.013 rnol 1-1) were prepared by dissolving 0.2 g of reagents in 100 ml of ethyl alcohol.The activated-carbon (Merck) was purified by treating with 12.1 rnol 1-1 hydrochloric acid (Merck) for 3 h, washing with102 Analyst, January 1995, Vol. 120 water, drying at 110°C and treating with aqua regia [hydro- chloric acid-nitric acid (3 + 1)J for 24 h. The mixture was filtered through filter paper (Schleicher and Schuell No. 589 blue ribbon, Dassell, Germany), washed with water and dried at 110°C. The dried activated-carbon was prepared as a suspension of 25 mg ml-1 in water. Digestion of Samples Five types of vegetable were chosen for analysis. These were spinach, cabbage, lettuce, carrot and tomato, which were taken from fertilized farmland and roadside areas in Elazig, Turkey.Dry ashing The vegetable matter was dried at 105 "C, ground in agate and homogenized. Samples (2-10 g) were weighed into evaporat- ing dishes and ashed at 470-500°C in a furnace for 3 h. The ashed samples were dissolved in nitric acid-perchloric acid (2 + 1) (for 10 g of dried matter, 4 ml of acid mixture was used) and diluted to 200 ml with distilled water. Wet ashing Samples were dried at 105 "C, weighed (2-10 g) and digested in nitric acid-hydrogen peroxide-perchloric acid (for 10 g of dried matter, 20 ml of 14.4 mol 1-1 nitric acid + 10 ml of 11.4 mol 1-1 hydrogen peroxide + 10 ml of 9.9 mol 1-1 perchloric acid) at 170°C until a clear digest was obtained (approxi- mately 3 h). In addition, 18.0 moll-' sulfuric acid (for 10 g of dried matter, 2.0 ml of acid) was added to form cadmium sulfate which prevents volatilization.18-19 Then, the clear digest was diluted to 200 ml with water.Determination of Matrix Components Direct atomic absorption measurements were also made to characterize matrix components of plant tissues. It is import- ant to study the effect of matrix components as some of these components form chelates with complexing agents and, therefore, can interfere with the enrichment pH in adsorption steps. Matrix components were found (mg kg-1: mg of element per kg of dried food) in the ranges given: Ni: 2.0, Mn: 9.0-30.0, Cu: 2.0-9.0, Al: 15-800, Zn: 1 5 4 0 , Fe: 40-700, Mg: 1000-5000 and Ca: 2000-10000. Enrichment Procedure The pH of cadmium and lead solutions (200 ml) of desired concentration, each containing matrix components at the upper concentration levels, were adjusted to optimum value and buffer solutions (25 ml) were added.After adding 35 ml of complexing reagents, such as oxine or cupferron, and 5 ml of activated-carbon suspension (25 mg ml-I), the pH of the solutions were adjusted again, as necessary. The mixture was stirred mechanically for 1 h and filtered through a filter paper (Schleicher and Schuell No. 589 blue ribbon). The residue was dried at 105°C for 1 h. Nitric acid (14.4 mol 1-1, 4 ml) was added to the residue in a glass beaker and evaporated to dryness. Then 2 mol 1-1 HN03 (2 ml) was added and, after centrifuging twice at 2000 rpm, the clear solution was separated for measurements. The Cd and Pb contents of the solutions (for each element, a different solution was obtained) were determined by means of a flame atomic absorption spectrometer.The steps to the enrichment procedure are outlined in Fig. 1. Results and Discussion The parameters that are thought to affect the enrichment and measurement steps in the procedure are listed below. In developing the analytical scheme, each parameter was investi- gated for 200 ml of 7 yg 1-1 Cd and 70 pg-* Pb solutions with the matrix components at their maximum concentrations. The effect of each parameter listed was tested 5 times. The relative standard deviations, s,, of the data points ranged from 1.5 to 3.5%. Effect of pH on the Recovery The effect of pH on the recovery of each element is shown in Figs. 2 and 3.The optimum pH for recoveries of up to 95% were found in the pH range of 5.3-6.8 for Cd and 5.8-9.0 for Pb, by using oxine. The corresponding values for cupferron were 4.8-7.8 for Cd, and 4.4-8.0 for Pb. Effect of Other Parameters on the Recovery The recoveries of Cd and Pb at the optimum lowest pH values have been examined by using different amounts of oxine or cupferron, by adding a constant amount of activated-carbon. It was found that the recoveries of Cd and Pb increased by up to 95% by using 60 mg of oxine for both Cd and Pb I Adjust pH of solution (200 ml) to desired value I 1 I Add the buffer solution (25 ml) I [Add the complexing reagent (35 ml) and the activated-carbon suspension (5 ml)l I 1 .c Stir for 1 h and filter through filter paper + I Dry at 105°C for 1 h and add 14.4 rnol I-' nitric acid (4 ml) 1 1 I Evaporate to dryness and add 2 rnol I-' HNO, (2 ml)] 1 I Centrifuge (two steps) and separate off the clear solution] 1 [Determine the Cd and Pb contents of the clear solution by flame atomic absorption spectrophotometry I Fig.1 Analytical scheme of the enrichment procedure.Analyst, January 1995, Vol. 120 103 3 90 2 80 v k 70 60 determination, 40 mg of cupferron for Cd and 50 mg of cupferron for Pb determination. The results are shown in Figs. 4 and 5 . The effect of the amount of activated-carbon on the recovery was also studied. It was found that 125 mg of activated-carbon by adding the optimum amount of oxine, and 100 mg of activated-carbon by adding the optimum amount of cupferron, were sufficient for both Cd and Pb.The results are , 100 I E60 840 a 30 20 rjl 10 2 3 4 5 6 7 8 9 10 PH Fig. 2 The effect of pH on the recovery of cadmium with A, oxine and B, cupferron. The concentration of cadmium was 7 pg 1-l(200 ml) and the solution contained matrix components at upper levels (see text). 90 a0 70 h Y $ 6 0 $ 5 0 a 40 2 8 30 20 lo r 0 ' I I I I I I 3 4 5 6 7 8 9 PH Fig. 3 The effect of pH on the recovery of lead with A, oxine and B, cupferron. The concentration of lead was 70 pg 1-1 (200 ml). (Solution contained matrix components at upper levels). shown in Figs. 6 and 7. The results in Figs. 8 and 9 show that the optimum contact periods, when cupferron was used, were 30 and 40 min stirrings for Cd and Pb, respectively. However, corresponding stirring periods were slightly longer when oxine was used: 45 and 50 min for Cd and Pb, respectively. 100 90 h $ 80 Y $ 70 8 60 50 30 40 ! 0 10 20 30 40 50 60 70 80 90 100 Ligand addedmg Fig.5 Determination of the optimum amount of ligand, using 200 ml of 70 pg 1-1 lead solution containing matrix components at upper levels. A, Oxine and B, cupferron. 100 I I U 50 I I 1 1 I I I 0 25 50 75 100 125 150 175 200 Activated-carbon addedmg Fig. 6 Determination of the optimum amount of activated-carbon, using 200 ml of 7 pg 1-1 cadmium solution containing matrix components at upper levels. A, Oxine and B, cupferron. 50 ' " I I I I I I Activated-carbon addedmg 0 25 50 75 100 125 150 175 200 Fig. 7 Determination of the optimum amount of activated-carbon, using 200 ml of 70 pg 1-1 lead solution containing matrix components at upper levels.A, Oxine and B, cupferron. $51 0 8 70 II: 60 0 10 20 30 40 50 60 70 80 90 100 Contact time/min Fig. 8 Determination of the optimum contact time, using 200 ml of 7 pg 1-1 cadmium solution containing matrix components at upper levels. A, Oxine and B, cupferron. 100 I 1 100 I 1 20 ' I I I I I I I I 0 10 20 30 40 50 60 70 80 Ligand added/mg Fig. 4 Determination of the optimum amount of ligand, using 200 ml of 7 pg 1-1 cadmium solution containing matrix components at upper levels. A, Oxine and B, cupferron. 60 1 1 1 1 1 l 1 I l l 0 10 20 30 40 50 60 70 80 90 100 Contact time/min Fig. 9 Determination of the optimum contact time, using 200 ml of 70 pg 1-1 lead solution containing matrix components at upper levels.A, Oxine and B, cupferron.104 Analyst, January 1995, Vol. 120 Selection of Complexing Reagents Recoveries of up to 95% were achieved when both oxine and cupferron were used for Cd and Pb enrichment. However, cupferron was preferred for enrichment of both Cd and Pb in vegetable samples as the optimum pH was lower when this complexing agent was used. Working at low pH values is advantageous because the major elements such as Ca, Mg and Fe in vegetable matrices can precipitate as hydroxides at high pH values. In addition, the major elements such as Ca and Mg may form chelate compounds with complexing reagents at high pH values.2O Furthermore, at these optimum pH ranges, less activated-carbon and cupferron were needed for recoveries of up to 95%. Finally, optimum contact time was very short when cupferron was used as the complexing agent.Calibration Graph and Precision Calibration curves were obtained by using cupferron for 200 ml of 1.0-7.0 pg 1-1 Cd (pH 5.0 k 0.2) and 200 ml of 10-70 pg 1-1 Pb (pH 4.7 k 0.2) solutions containing matrix components at upper concentration levels (at the optimum conditions determined previously). The enrichment proce- dure in Fig. 1 was applied to these solutions. The clear solutions were analysed by means of flame atomic absorption spectrophotometry . Calibration curves were linear in the concentration range of 1.0-7.0 pg 1-1 Cd and 10-70 pg 1-1 Pb. The equations of the curves were as follows: y = 7 . 6 0 ~ - 0.6 Y = 1.00 for Cd and y = 0.5457~ + 0.8478 Y = 1.00 for Pb Relative standard deviations were found to be 2% for Cd at a concentration of 7.0 pg 1-1, and 3% for Pb at a concentration of 70 pg 1-1, for 12 replicate enrichment procedures.Accuracy and Applications To ensure that this method was valid, recovery of Cd and Pb from vegetable samples fortified with these elements were found and are shown in Tables 2 and 3. It was found that the Cd and Pb added to vegetable samples were recovered by at least 95% by using the wet ashing procedure. Recovery levels for Cd and Pb by using wet ashing were always higher than those by using the dry ashing procedure for all vegetable samples. However, recoveries for Cd in dry ashing were much lower than those of Pb. These results may be attributed to the loss of Pb and especially Cd in the dry ashing step.Although it is almost impossible to discriminate amongst the loss arising from volatilization, adsorption or contamination from reagents, it is believed that volatilization of the elements as volatile salts is the prime source for the observed low recoveries. Adsorption can be excluded as the procedure was followed in exactly the same way, using the same glassware, and the same reagents were used throughout. The effect of contamination on the higher recovery levels observed with the wet ashing procedure was eliminated by subtracting values obtained for blanks. Furthermore, it can be inferred that standard deviation, s, values for wet ashing must be higher, if contamination had occurred to any appreciable degree, as acid was used many times compared to dry ashing.However, as seen from Tables 2 and 3, s for these procedures are almost of the same magnitude. Therefore, the effect of contamination may be reliably overlooked. Although the exact mechanism for the losses needs radioisotope techniques, the above discussion on contamination and numerous reports on the loss of volatile elements in graphite furnace atomic absorbtion spectrometry determinations led us to believe that loss through volatilization is responsible for the lower recovery levels in the dry ashing procedure. As a matter of fact, the loss of lead in biological materials following different drying and ashing procedures was shown to occur using 203Pb as a tracer.21 Table 2 Results and recovery of Cd from vegetable matter (dried mass basis). The results are mean values f s, (n = 9) Cd addedlmg kg-1 Cd foundlmg kg- Recovery (%) Sample Dry ashing Wet ashing Dry ashing Wet ashing Dry ashing Wet ashing Spinach 0 leaves 0.15 Cabbage 0 leaves 0.20 Lettuce 0 leaves 0.20 Carrot 0 roots 0.15 Tomato 0 fruits 0.15 0 0.15 0 0.20 0 0.20 0 0.15 0 0.15 0.30 k 0.05 0.40 -t- 0.03 0.61 f 0.04 67 107 0.35 k 0.06 0.48 k 0.05 0.76 k 0.05 65 105 0.45 k 0.08 0.56 + 0.06 1.02 f 0.07 55 110 0.09 4 0.02 0.19 k 0.03 0.30 k 0.03 66 100 0.15 k 0.02 0.26 k 0.03 0.36 k 0.03 73 107 0.45 t 0.06 0.55 k 0.07 0.80 t 0.10 0.15 _C 0.03 0.20 t 0.02 Table 3 Results and recovery of Pb from vegetable matter (dried mass basis).The results are mean values k s, (n = 9) Pb addedlmg kg-1 Pb foundmg kg- Recovery (% ) Sample Dry ashing Wet ashing Dry ashing Wet ashing Dry ashing Wet ashing Spinach 0 leaves 0.50 Cabbage 0 leaves 0.50 Lettuce 0 leaves 0.50 Carrot 0 roots 0.50 Tomato 0 fruits 0.50 0 0.50 0 0.50 0 0.50 0 0.50 0 0.50 0.73 k 0.13 1.16 k 0.10 1.15 t 0.15 1.58 k 0.12 0.65 -t 0.10 1.03 k 0.12 0.60 k 0.10 1.00 -t- 0.11 0.53 k 0.07 0.81 k 0.09 0.82 k 0.10 1.34k0.11 86 104 1.30 f 0.14 1.83f0.11 86 106 0.85 k 0.12 1.36k0.13 76 102 0.70 t 0.10 1.15k0.12 80 96 0.73 k 0.08 1.22 kO.10 76 98Analyst, January 1995, Vol.120 105 The optimized enrichment method using cupferron was applied to the determination of Cd and Pb in various foodstuffs, such as spinach, cabbage and lettuce leaves, carrot roots and tomato fruits, which were taken from fertilized farmland and roadside areas in Elazig, Turkey. The results are summarized in Tables 2 and 3.The given values are the means of three completely independent digestions of three separate portions of the same sample obtained from different locations, or at different times. Hence, nine measurements were performed for each sample type. It was found that the lettuce and cabbage leaves had the highest Cd and Pb contents among the examined foodstuff. ConcIusions A method was developed for the determination of trace amounts of Cd and Pb in vegetable matter by FAAS. Matrix components were characterized and the upper concentration levels of matrix components were added to all standard Cd and Pb solutions. Thus, interferences from matrix com- ponents were minimized. The levels of Cd and Pb in the reagent blanks in the wet ashing procedure were 0.6 and 8 pg 1-1 with s values of 0.1 and 0.8 pg 1-1, respectively.The detection limits, defined as three times the s values of the blanks were, therefore, 0.3 pg 1-1 and 2.4 pg 1-1 for Cd and Pb, respectively. The procedure described above may allow the detection of 0.06 and 0.48 ng of Cd and Pb in biological materials, respectively, per gram of dried vegetable. Although slightly different pH values for Cd and Pb were used, simultaneous analysis of both metals may be achieved using the same pH, e.g., pH 5.0, in the enrichment procedure. However, the outlined enrichment procedure yields only 2 ml of solution, which may allow, at most, two measurements of absorption with the classical aspiration technique, although one measurement is effected in this work.If, on the other hand, the signal sampling of the instrument is adequate, injection technique may allow up to 20 measurements from just one digestion procedure. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 References Coutate, T. P., Food, The Chemistry of Its Components, 2nd edn. Royal Society of Chemistry, Cambridge, 1992, pp. Giiqer, S., and Demir, M., Anal. Chim. Acta, 1987, 196, 277. Demir, M., GiiGer, S., and Esen, T., J. Agric. Food Chem., 1990, 38, 726. Yin, X., Schlemmer, G., and Welz, B., Anal. Chem., 1987,59, 1462. Dolinsek, F., Stupar, J., and Vrscaj, V., J. Anal. At. Spectrom., 1991, 6,653. Beek, H. V., Greefkes, H. C. A., and Baars, A. J., Talanta, 1987, 34, 580. Frank, R., Suda, P., and Luyken, H., Bull. Environ. Contam. Toxicol., 1989, 43, 737. Ohta, K., Aoki, W., and Mizuno, T., Microchim. Acta, 1990,1, 81. Puchades, R., Maquieira, A., and Planta, M., Analyst, 1989, 114, 1397. Malamas, F., Bengston, M., and Johansson, G., Anal. Chim. Acta, 1984, 160, 1. Berndt, H., Jackwert, E., and Kimura, U., Anal. Chim. Acta, 1977, 93,45. Wanderboght, B. M., and Van Grieken, R. E.. Anal. Chem., 1977, 49, 311. Berndt, H., and Messerschmidt, J. F., 2. Anal. Chem., 1981, 308, 104. Ramadevi, P., and Naidu, G. R. K., Analyst, 1990, 115, 1469. Jackwert, E., 2. Anal. Chem., 1974,271, 120. Jackwert, E., Lohmar, J., and Wittler, G., 2. Anal. Chem., 1973, 266, 1. Giiqer, $., and Yaman, M., J. Anal. At. Spectrom., 1992,7,179. Feinberg, M., and Ducauze, C., Anal. Chem., 1980, 52,207. Erwin, J. M., and Ivo, N., Analyst, 1992, 117, 23. Mincewski, J., Chwustowska, J., and Dybczynski. R., Separa- tion and Preconcentration Methods in Inorganic Trace Analysis, ed. Masson, M. R., Chichester, Wiley, 1982, p. 200. Xue, Z. L., and Wang, Y. X., J. Radioanal. Nucl. Chem., Lett., 1987, 119,425. Analytical Methods for AAS (catalogue), Perkin-Elmer Cor- poration, Norwalk, 1976. Paper 4100487F Received January 26, I994 Accepted June 28, 1994 265-273.

ISSN:0003-2654    

DOI:10.1039/AN9952000101

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Analyst, January 1995, Vol. 120 107 Rapid Determination of Calcium, Magnesium, Sodium and Potassium in Milk by Flame Atomic Spectrometry After Microwave Oven Digestion Miguel Angel de la Fuente and Manuela Juarez* Instituto del Frio (CSIC), Ciudad Universitaria sln, 28040 Madrid, Spain A rapid method for determining Ca, Mg, Na and K in skim, semi-skim and whole milk was studied. These elements can be accurately determined in milk by atomic absorption (Ca and Mg) and atomic emission (Na and K) spectrometry after digestion of the sample with HN03 in domestic and analytical microwave ovens in sealed poly(tetrafluoroethy1ene) reactors. Microwave-oven mineralization allowed accurate determination as tested against certified milks. The results were also in good agreement with those obtained after precipitation with trichloroacetic acid in certified and real liquid milk samples.Low recoveries for Ca in whole milk were found, owing to matrix effects, and the method of standard additions was necessary to avoid them. Mean recoveries determined for one of the milk reference materials studied were 99.3, 100, 100.5 and 100.3% for Ca, Mg, Na and K, respectively. The method gave relative standard deviations of 1-2% for the four elements. The limits of detection were 63,15, 18 and 72 pg 1-1 for Ca, Mg, Na and K, respectively. The method is very practical for routine laboratory analyses of relatively large numbers of samples. Keywords: Calcium, magnesium, sodium and Potassium determination; flame atomic spectrometry; milk; microwave-oven digestion Introduction Although a variety of instrumental methods (fluorimetry, X-ray fluorescence, neutron activation, inductively coupled plasma atomic emission spectrometry) have been used in recent years to determine the different mineral elements in milk, flame atomic spectrometry is perhaps the most prevalent because it is sufficiently sensitive for the macro-elements and easy to operate, having successfully replaced the conventional spectrophotome tric methods.Sample preparation represents an important stage in analysis and some publications have been devoted to this subject.l.2 For milk, preparation of the solution for spectro- photometric measurement is of paramount importance, and this varies depending on the authors: some simply dilute the milk with de-ionized water;3 another method involves dry- ashing the sample4 preceded by a preliminary heating step to evaporate the milk to dryness;Sfj and one of the most common procedures is that involving precipitation of the milk proteins with trichloroacetic acid (TCA) -7-10 Both TCA precipitation and dry incineration methods have the drawback of requiring different reagents and/or long operating times.Two developments are apparent in the sample preparation procedures used or recommended over many years: first, the use of sealed pressure vessels to accelerate sample digestion and minimize contamination and loss of volatile elements; and second, the use of microwave radiation to assist in digestion. * To whom correspondence should be addressed. Recently, rapid, safe and efficient acid decomposition methods based on the use of microwave ovens have been proposed for the determination of different elements in a variety of matrices and materials including foods.llJ2 In milk and dairy products, only a few reports have been published involving microwave digestion, more specifically trace ele- ments in infant formula powdered milk13 and cheese,l4 and none of them studied the major elements contained in milk.The purpose of this work was to establish operating conditions for the determination of Ca and Mg by atomic absorption spectrometry (AAS) and Na and K by atomic emission spectrometry (AES) in milk, using samples prepared by digestion in domestic and analytical microwave ovens. The sample preparation procedure was compared with that of precipitation with TCA . Experimental Apparatus The determination of K and Na was carried out by AES using a Model 5100 PC atomic absorption spectrometer (Perkin- Elmer, Norwalk, CT, USA).Ca and Mg were determined by AAS using the same instrument with a multi-element (Ca, Mg, Zn) hollow-cathode lamp. The elements under study were determined using an air-acetylene flame and the recommended values for the instrumental parameters are given in Table 1. Reagents High-purity water with a metered resistivity of 18 Mi2 cm was used to prepare all samples and standards. All reagents used were of the highest purity available and at least of analytical- reagent grade. HN03, 65% m/v. Suprapur grade (Merck). H202, Perhydrol. Suprapur grade (Merck). Ca, K, Na and Mg standard solutions for spectrophotometry , 1 k 0.002 g 1-1 (Panreac).Ca, Mg, Na and K standard Titrisol solutions, 1 k 0.002 g 1-1 (Merck). Lac13 solution, 5% mlv. Prepared by mixing lanthanum oxide (Phaxe) (5.86 g), 5 ml of distilled water, 25 ml of concentrated HCl (Merck) and diluting to 100 ml with distilled water. Trichloroacetic acid, 24% mlv aqueous solution. Standard Solutions Two groups of standard solutions were prepared. First, to measure the elements by the TCA deproteinization proce- dure, the composition described by Juarez and Martinez- Castrog was used. Second, to determine major cations by the108 Analyst, January 1995, Vol. 120 microwave-oven digestion method, two standard solutions were used with Ca, Mg, Na and K contents of 2.0,0.2,3.0 and 1.0 and 5.0,0.5,7.0 and 2.5 mg 1-1, respectively. Lanthanum was added to all standards at 500 pg ml-1.Samples Milk samples utilized for repeatability experiments were purchased in the market and corresponded to three types of UHT milk with different fat contents: 3.6% for whole milk, 1.8% for semi-skim milk and 0.1% for skim milk. During the course of the assays, samples were stored in refrigerated conditions. Reference Material Validation of the method presented in this study was performed by using two reference materials: one from the National Institute of Standards and Technology (NIST), Standard Reference Material (SRM) of 1549 Non Fat Milk Powder, and the other from the Community Bureau of Reference, BCR-63 Skim Milk Powder. Amounts of 5 g of these powders were dissolved in water and diluted to 50 ml in order to reconstitute the liquid milk for subsequent use in the different assays.Procedures Sample preparation by de-proteinization with the TCA method was described in previous papers.8.9 For the microwave-oven digestion method, different reagent volumes an conditions (times and power settings of the microwave oven) were tested in order to establish the experimental parameters of the recommended procedures as described below. For sample preparation using a domestic microwave oven, six reactors were taken and filled with 1 ml of milk sample and 2 ml of concentrated HN03. The reactors were sealed and placed at outer positions inside a lidded plastic container, which was placed on the revolving plate. The alternative digestion mixture consisted of 2 ml of HN03 and 2 ml of H202 for the same amount of milk sample (1 ml).The six samples were then irradiated according to the following programme: 4 min at 583 W followed immediately by 8 min at 247 W. After irradiation, the reactors were set in an ice-bath to cool before opening. Appropriate dilutions must be carried out for sample measurements in the linear range of each of the elements to be determined. The procedure using the analytical microwave oven was similar to that described above for the domestic microwave. The six reactors were placed in the carousel at alternate locations, leaving a gap between every two reactors so that the radiation would be evenly distributed among them. Lan- thanum was added to the solution measured at 500 yg ml-1 as a releasing agent to break up the Ca3(P0& in all the samples.Table 1 Instrumental conditions for measurement of Ca, Mg, K and Na by flame atomic spectrometry Flame characteristics/ Lamp 1 min-1 Wave- Slit width/ intensity*/ Element lengthhm nm mA Air Acetylene Ca 422.7 0.7 15 10 2.0 285.2 0.7 15 10 3.8 K 766.5 0.4 - 10 2.0 Mg Na 589.0 0.4 - 10 2.0 * Multi-element (Ca, Mg, Zn) Intensitron (Perkin-Elmer). Microwave Ovens and Reactors Two types of microwave oven were used for the assays. The domestic microwave oven was a Texet Model-112, programm- able for time and microwave power in three discrete steps with nine power settings (ranging from 112 to 650 W) and equipped with a revolving plate. For safety reasons the domestic microwave was placed in an isolated area of the laboratory.The magnetron frequency was 2450 MHz. The poly- (tetrafluoroethylene) (PTFE) vessels used for the solutions were laboratory-made and had a volume of approximately 100 ml with a 10 mm wall thickness and tight-fitting screw-cap lids. A commercially available analytical microwave oven, Model MDS-2000 (CEM, Indian Trail, NC, USA) with a magnetron frequency of 2455 MHz and 650 W output power, was also used. This apparatus was equipped with pressure and temperature control systems and a carousel for 12 vessels. Lined digestion vessels (100 ml) made from perfluoroalkoxy (PFA)-Teflon were obtained from CEM. The vessels can be used at temperatures up to 250 "C and a maximum pressure of 1400 kPa. Venting of the digestion or reaction products is controlled with a proprietary sealing and vent stem (a thin fluoroplastic rupture foil is inserted between the body and the cap to act as a seal) and the vessel can easily be tightened or opened by hand.Results and Discussion Irradiation Efficiency For domestic ovens that are not specifically designed for analytical purposes it is important to note that different positions in the oven cavity are not identical in terms of microwave irradiation.15 Thus a prior study of the irradiation efficiency at different positions on the revolving plate was necessary. In order to carry out this study, 61 beakers (28 mm in diameter), each containing 25 ml of distilled water, were placed inside the cavity of the microwave oven on the revolving plate, covering its entire surface in nine energetic- ally different positions.The oven was operated at maximum power for 30 min, then the loss of water from each beaker was measured and related to the level of microwave power absorbed, taking the maximum loss of water as equivalent to a power absorption level of 100%. Fig. 1 shows the microwave distribution on the plate, from which it was concluded that the central position is the least energetic and outer positions received more microwave power. During the assays, the reactors were placed at outer positions on the revolving plate to ensure maximum irradiation of the solutions. The power distribution depends on the amounts of materials (samples and reagents) employed, because the PTFE vessels are transparent to microwave radiation.16 In the assays six vessels were used simultaneously.Digestion Conditions As the basic aim of this study was to arrive at the mildest possible conditions of mineralization with the smallest pos- sible number and lowest concentrations of reagents and with safety margins sufficient to prevent deterioration of both the microwave oven and the reactors, the experiments were designed as described below. At the same time, these conditions had to be adequate for the preparation of a solution that would meet analytical requirements for subsequent measurement by atomic spectrometry. The temperature and especially the pressure inside the reactor are limiting factors in the digestion of different matrices in closed vessels.17 In order to ascertain the pressure and temperature conditions inside the reactors in the domestic microwave oven, the digestion conditions described underAnalyst, January 1995, Vol.120 109 Procedures for the domestic microwave oven were reproduced in the analytical microwave oven. The evolution of both parameters is illustrated in Fig. 2. The temperature increased exponentially during the first moments of digestion up to 100 "C, thereafter rising more slowly up to 140 "C for the final part of the first stage (4 min) of high-power irradiation (583 W). During the second mineralization stage, the temperature dropped to 125 "C, then remained constant. During the first stage, the pressure increased gradually up to 235 kPa (4 min), settling down to 150-160 kPa for the remaining 8 min of the programme. The mineralization temperature and pressure ensure oxidation of the sample because the microwave energy 1 I I I ' 140 0 5 10 13 Radiudcm Fig.1 the revolving plate. Distribution of microwave radiation within the oven cavity on 140 300 120 0 e, 250 100 5 g 200 80 g I-" m \ 5 150 60 $ 100 40 CL 20 50 0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 Timehin Fig. 2 Pressure (A) and temperature (B) evolution during the programmed digestion in the microwave oven (4 min at 583 W and 8 min at 235 W). Tabie 2 Mean values and standard deviations for Ca, Mg, Na and K content (mg 1-1) in whole milk as determined by flame atomic spectrometry following mineralization in a domestic microwave oven in the presence (A) and absence (B) of H202. Each result is the mean of eight determinations Digestion Ca Mg K Na A 1008 f 18 101 f 1 1628 f 19 650 k 9 B 1008 f 20 100 f 1 1622 f 13 643 f 7 penetrates the sample in depth and so causes heating throughout the liquid rather than at the surface only. When microwaves are absorbed by an acid, they cause heating by two mechanisms: dipole rotation and ionic conduction, which can occur simultaneously.16 The acid used in digestion was HN03 in view of its high oxidizing capacity and the solubility of nitrates in aqueous media.It was decided not to use HCI in view of its weaker oxidizing capacity and the high pressure that it produces, and perchloric acid was rejected in order to obviate problems with corrosive fumes and its explosive potential,l5318 which could damage the reactors. As an alternative to HC104, H202 was used in combination with HN03 in order to improve the digestion efficiency and achieve a clearer post-digestion solution, which would make for more accurate atomic spectrometric measurement.The samples mineralized with- out H202 were yellowish, whereas those mineralized with H202 provided a clear solution. Table 2 gives the results for the elements studied for mineralization of samples under identical conditions with and without added H202 for the remainder of the study. Matrix Effects At the outset of the study, the final atomic spectrometric determination of the four elements was carried out by external calibration through extrapolation from the straight lines determined using solutions of known concentration. In this way the results given in Tables 2 and 3 were arrived at.However, on attempting to determine Ca in whole milk samples in order to establish the precision of the procedure (as has been noted, the certified milks assayed were reconstituted powdered skim milks), the concentrations of this element were found systematically to be lower than those determined by TCA de-proteinization. The decrease with respect to the values as measured by the method of Brooks et a1.8 were in excess of 15%. This loss did not appear to affect the other three elements, however. Longer microwave digestion times (20-25 min) produced no improvement. It was thought that addition of H202 combined with HN03 to the digestion mixture might enhance the destruction of organic matter and hence the release of Ca; however, as the results noted previously (Table 2) showed, the presence of H202 did not alter the Ca concentration as determined in whole milk.To remove the matrix effects, in the determination of Ca in whole milk, it was decided to perform the final atomic spectrometric measurement by standard additions because the conditions could not be optimized for accurate measurement by straight calibration. The values found for Ca by the standard additions, method are close to those determined by TCA de-proteinization (Table 4), and therefore this pro- cedure was adopted to determine Ca in repeatability tests using whole milk. ~ Table 3 Certified and experimental mean (k standard deviation) values for Ca, Mg, Na and K (mg 1-1) in reconstituted certified milks BCR-63 and SRM-1549. Measurement performed by flame atomic spectrometry following precipitation with trichloroacetic acid (TCA) or digestion with domestic (A) or analytical (B) microwave oven.The values shown are the means of five determinations Sample Method Calcium Magnesium Potassium Sodium BCR-63 Certified values* 1260 f 30 112 f 3 1780 f 60 457 f 16 TCA 1292 f 18 110 f 3 1736 f 22 460 k 14 Microwave A 1261 f 22 112 f 2 1780 f 10 446 f 14 Microwave B 1261 f 32 113 k 2 1782+13 4 4 9 f 3 SRM-1549 Certified values* 1300 k 50 120 f 3 1690 ? 30 497 f 10 TCA 1298 f 16 119 k 4 1683 k 28 496 k 13 Microwave A 1306 f 16 120 f 1 1692 f 17 500 f 6 Microwave B 1308 f 12 123 k 1 1690 k 19 499 f 4 * With 95% confidence limits.110 Analyst, January 1995, Vol. 120 In order to confirm the extent to which fat content influences Ca determination, microwave digestion was perfor- med on commercial milks with different fat contents (whole, semi-skim and skim milk); Ca was determined by external calibration and the results were compared with those obtained by precipitation with TCA.The results are given in Table 5. In skim milk as in the certified milks, the Ca levels as determined by external calibration were similar for both procedures. In semi-skim milk, the Ca concentration was found to be slightly lower after mineralization in the microwave oven when determined by external calibration, whereas in whole milk this difference was marked. These results confirm the noticeable influence of the matrix on the analysis of whole milk. Association of two thirds of the Ca with the casein micelles in the presence of fat19 may be the reason behind this marked matrix effect on the determination of Ca, because organic materials are not totally decomposed to C02 and water.This effect was not apparent (Table 5) with Mg (only one third of which is bound to the caseins and whose over-all concentration is of a lower order of magnitude than that of Ca) or Na or K, both of which are present almost entirely in the soluble phase.20 Precision of the Method The repeatability of the procedure was studied by carrying out eight replicate assays on a single sample of whole milk prepared by mineralization in analytical and domestic micro- wave ovens and also be precipitation with TCA. The eight assays were performed in two lots in each microwave oven. The mean values and standard deviations are given in Table 4.~ ~~ ~ ~~~~~~~~~ Table 4 Precision in determination of Ca, Mg, Na and K (mg I-l) in whole milk (mean values f standard deviations) mineralized by domestic (A) and analytical microwave (B) oven digestion and by precipitation with trichloroacetic acid (TCA). Each result is the mean of eight determinations Method Calcium Magnesium Potassium Sodium TCA 1198k23 98-t.2 1535 f 21 573 k 21 Microwave A 1205 f 15 97 f 1 1518 f 16 578 k 7 Microwave B 1206 f 19 98 f 1 1531 f 16 573 f 11 Table 5 Determination of Ca and Mg (mg 1-1) by atomic absorption spectrometry (quantified by external calibration) in whole, semi-skim and skim milk (mean value f standard deviation) following minerali- zation by domestic microwave oven or precipitation with trichlo- roacetic acid (TCA).Each result is the mean of eight determinations C a 1 c i u m Magnesium Milk TCA Microwave TCA Microwave Whole 1198k23 982k 14 9 8 f 2 97 k 1 Semi-skim 1115 f 14 1105 k 18 106 f 2 109 f 1 Skim 1264f6 1261 -1-23 1 1 2 f 2 112f 1 The variability of the TCA and microwave methods are not statistically different (P d 0.05). The relative standard deviations (s,) for the four elements were between 1 and 2% (Ca 1.2 and 1.6%; Mg 1%; K 1.1%; Na 1.2 and 1.9%). In all instances, the values found for samples prepared by digestion in the two ovens were lower than for those prepared by TCA precipitation (Ca 1.9%; Mg 2%; K 1.4%; Na 3.7). These results were comparable to those reported by Gaines et aZ.4 for Ca (1.1 % ) in a dry-ashing procedure.Accuracy In order to assess the accuracy of the procedure, in addition to examining BCR-63 and SRM-1549 certified samples, percen- tage recoveries were determined after addition of known amounts of Ca, Mg, Na and K to the latter. Table 3 gives the contents of the elements studied in the certified samples prepared both by digestion in the two experimental microwave ovens and by precipitation with TCA. For SRM-1549, the results following the two mineralization procedures were within the range of certified values, those of samples prepared by microwave being slightly higher. The results after mineralization in the two different types of microwave oven were comparable. For BCR-63 the results were within the certified range, although the results for Na were closer to the certified value in samples prepared by precipitation with TCA.However, the concentration of Ca as determined by the latter procedure was slightly above the certified value. The spread of the results, with the exception of Ca in BCR-63, was found to be smaller when samples were mineralized with either of the microwave ovens. Table 6 shows the recovery study in which 1 ml of SRM-1549 reconstituted with 1 ml of standard solution was spiked with 1.0, 0.1, 0.5 and 1.5 g 1-1 of Ca, Mg, Na and K, respectively. The mean recoveries after digestion in the domestic micro- wave oven ranged from 99.3% for Ca to 100.5% for Na. These results are closer to 100% than those reported by JuArez and Martinez-Castrog using the TCA de-proteinization method for Na (101.9%) and Mg (101.4%) and comparable to those for Ca and K (99.4% and 99.9%, respectively).Using dry mineralization in a furnace at 550 "C, Gaines et aZ.4 achieved Ca recoveries between 101% and 102%. It may be concluded that milk digestion with HN03 (and H202) in PTFE reactors using microwave heating for AAS and AES provides an efficient alternative to dry mineraliza- tion or de-proteinization methods. The limits of detection were 63 (Ca), 15 (Mg), 18 (Na) and 72 (K) pg 1-1. The precision and accuracy of the proposed method are similar to those of the techniques cited and are appropriate for the determination of Na, Ca, Mg and K in real samples. It also has the added advantages of simplicity in the digestion procedure and above all rapidity of execution: six milk samples can be digested in only 12 min. The possibility, here confirmed, of performing digestion in such affordable apparatus as a Table 6 Recovery of Ca, Mg, Na and K added to reconstituted SRM-1549. The sample was digested with the domestic microwave oven and measured by flame atomic spectrometry.Each result is the mean of five determinations Certified Amount Total Amount values* added/ calculated found?/ Element mg 1-1 mg 1-1 mg 1-1 mg 1-1 Recovery (%) Calcium 1300 f 50 lo00 2300 k 50 2284 f 32 99.3 Potassium 1690 zk 30 1500 3190 f 30 3200 f 37 100.3 Sodium 497 f 10 500 997 f 10 1002 f 19 100.5 Magnesium 120 k 3 100 220 f 3 220 _+ 1 100.0 * With 95% confidence limits. t Mean value f standard deviation.Analyst, January 1995, Vol. 120 111 domestic microwave oven means that this procedure can easily be applied to routine determinations.The authors acknowledge financial support for this research project (ALI93-0005-CP) from the Comision Interministerial de Ciencia y Technologia (Interministerial Commission for Science and Technology). They also thank G. Guerrero for his assistance with the spectrometric measurements and they are grateful to Drs. R. Montoro and M. de la Guardia for providing advice on different aspects of this work. References Blake, C. J., Sci. Tech. Surv., 1980, 122. Sulcek, Z., and Povondra, P., Methods of Decomposition in Inorganic Analysis, CRC Press, Boca Raton, FL, 1989. Rebmann, H., and Hoth, H. J., Milchwissenschaft, 1971, 26, 411. Gaines, T. P., West, J. W., and McAllister, J. F., J. Sci. Food Agric., 1990, 51, 207. Murthy, G. K., and Rhea, U., J. Dairy Sci., 1967,50, 313. Cerbulis, J., and Farrell, J. M., Jr., J. Dairy Sci., 1976,59,589. International Dairy Federation, Determination of the Calcium Content of Milk, International Standard, FIL-IDF 36: 1966, 1966. Brooks, I. B., Luster, G. A., and Easterly, D. G., At. Absorpt. Newsl., 1970, 9, 93. 9 10 11 12 13 14 15 16 17 18 19 20 Juarez, M., and Martinez-Castro, I., Rev. Agroquim. Tecnol. Aliment., 1979, 19,45. Zucchetti, S . , and Contarini, G., At. Spectrosc., 1993, 14, 60. Kimber, G. M., and Kokot, S., Trends Anal. Chem., 1990, 9, 203. Kuss. H.-M., Fresenius' Z . Anal. Chem.. 1992, 343, 788. Burguera, M., Burguera, J . L., Garaboto, A. M., and Alarcbn, 0. M., Quim. Anal., 1987, 6, 427. Tattersall, P., Lab. Pract., 1986, 35, 95. Ybafiez, N., Cervera, M. L., Montoro, R., and de la Guardia, M., J . Anal. At. Spectrom., 1991, 6, 379. Neas, E. D., and Collins, M. J., in Introduction to Microwave Sample Preparation. Theory and Practice, ed. Kingston, H. M., and Jassie, L. B., American Chemical Society, Washington, Kingston, H. M., and Jassie, L. B., Anal. Chem., 1986, 58, 2534. Matusiewicz, H., Sturgeon, R. E., and Berman, S. S., J. Anal. At. Spectrom., 1989, 4, 323. Holt, C., Yearbook, Hannah Research Institute, Ayr, 1989, pp. Holt, C., in Developments in Dairy Chemistry-3, ed. Fox, P. F., Elsevier Applied Science, London, 1985, pp. 143-215. DC, 1988, pp. 7-32. 51-55. Paper 4/02-5671 Received May 3, 1994 Accepted July 4, 1994

ISSN:0003-2654    

DOI:10.1039/AN9952000107

出版商:RSC    

年代:1995

数据来源: RSC