摘要:

ANALYST, MAY 1993, VOL. 118 48 1 Intercomparison of Methods for the Determination of Vitamins in Foods Part 2." Water-soluble Vitamins Peter C. H. Hollman and Jean H. Slangen DLO-State Institute for Quality Control of Agricultural Products (RIKILT-DLO), Bornsesteeg 45, NL-6708 PD Wageningen, The Netherlands Peter J. Wagstaffe and Uta Faure Commission of the European Communities, Community Bureau of Reference (BCR), Rue de la Loi 200, 5- 1049 Brussels, Belgium David A. T. Southgate and Paul M. Finglas AFRC Institute of Food Research, Norwich Laboratory, Norwich Research Park, Colney Lane, Norwich, UK NR4 7UA An intercomparison of methods involving 18 European laboratories was organized to assess the state-of-the-art of vitamin determination in foods. Each laboratory received identical samples of dry food reference material (homogeneous powders, milk powder, pork muscle and haricot vert beans), which have recently been certified for major dietary components and elements.Each laboratory was requested to perform the analyses by its own routine methods. The results for water-soluble vitamins are reported. The reproducibility for the determination of vitamin B, in milk powder, pork muscle and haricot vert beans with high-performance liquid chromatography (HPLC), fluorimetric and microbiological methods was good, with the relative standard deviation of reproducibility (RSDReprod) ranging from 11 to 18%. Differences between laboratories for the determination of the vitamin B2 content of milk powder, pork muscle and haricot vert beans determined using HPLC and microbiological methods were very high, with RSDReprod ranging from 28 to 74%.The extraction and hydrolysis procedures were probably the most important sources of variation. For vitamin B6 various HPLC and microbiological methods were used. The variation in the results for vitamin B6 was high, except in milk powder. The RSDReprod ranged from 18 to 51 %. A major part of this variability was due to differences in the extraction and hydrolysis procedures and problems with the identification of the vitamin B6 vitamers by HPLC. Variation in the results for niacin obtained with the microbiological methods in milk powder, pork muscle and haricot vert beans, was small; RSDReprod = 9-15%. The variation in the results reported for vitamin C in haricot vert beans was high, with an RSDReprod of 15%.Participants used HPLC, f I u ori met ry and i nd o p h en o I ti trat i o n . Keywords: Intercomparison; food; vitamin B,, B2, B6 and C; niacin; reference material Even in affluent societies vitamin deficiencies can occur. Elderly people, especially, run the risk of insufficient intake of, among others, vitamins B1, B2, B6 and C. Other groups at risk for hypovitaminosis of one or more of these vitamins are infants, pregnant women and adolescents. Governments try to prevent these deficiencies by legislation involving fortified products, food labelling and education. However, it is evident that accurate methods of analysis for these vitamins are indispensable. Only a very limited amount of information is available on the reproducibility of the determination of water-soluble vitamins in foods between laboratories using their own routine methods.' As part of a Community Bureau of Reference (Commission of the European Communities) programme on the improvement of the quality of vitamin determination in food, an intercomparison of methods for fat- and water- soluble vitamins in foods was planned.We now report on the results of this intercomparison of methods for the determination of water-soluble vitamins in foods. Eighteen laboratories with experience of vitamin determination in food participated in this study. Participants were encouraged to apply the routine methods used in their * For Part 1 of this series see ref. 2. respective laboratories.? A previous paper dealt with fat- soluble vitamins.2 Experimental Protocol and Materials The choice of methods was left to the participants and was only subject to the requirements of achieving the best level of accuracy.Three dry foods certified reference materials (CRMs) in the form of homogeneous powders, whole milk powder (not enriched, CRM 380), freeze-dried pork muscle (CRM 384) and dried haricot vert beans (CRM 383) were selected for this intercomparison. Both water- and fat-soluble vitamins were studied together in one intercomparison. The protocol, materials and statistical evaluation used have been described in detail previously.2 In order to assess between-sample variation (RSD) of the multivitamin mixture, five bottles were randomly chosen and analysed for a number of water-soluble vitamins in one laboratory. The variation between the samples was rather t A report with detailed data may be obtained from Peter C.H. Hollman, DLO-State Institute for Quality Control of Agri- cultural Products (RIKILT-DLO). Bornsesteeg 45, NL-6708 PD, Wageningen, The Netherlands.482 ANALYST, MAY 1993, VOL. 118 small (RSD = 0.6-1.5%, no data for vitamin C are available) compared with the analytical variation (RSDrepe,,) in the laboratories (Table 1) and thus the multivitamin mixture can be regarded as homogeneous. Results Laboratories carried out the analyses within a period of 3 months, with instruction to store the samples at 4 "C until use. Multivitamin Reference Mixture The precision data achieved for the water-soluble vitamins in this intercomparison are summarized in Table 1.The relative standard deviation of reproducibility ( RSDReprod) achievable was calculated for the determinations in the multivitamin mixture, using the empirical equation of Horwitz.3 It rep- resents the reproducibility that can be expected when all laboratories use the same rigidly defined standardized methods. The actual RSDReprod was 2-3 times larger than the predicted RSDReprod. As discussed previously, inhomogeneity of the samples can be ruled out as an explanation of this high variability.2 This variation being larger than expected, could point to differences in calibration procedures between the laboratories. However, similar to the fat-soluble vitamins, correction of the data for the level found in the multivitamin mixture by each laboratory [RSDR(,,,) in Table 11 did not cause a decrease in the variation.So again, possible effects on the precision from differences in calibration of the vitamin standards could not be revealed in this intercomparison. The moderate results for water-soluble vitamins in the standard mixture were probably caused by inadequate execution of the extraction procedures prescribed in the protocol. Vitamin B1 Reproducibility of the determination of vitamin B1 in milk 11-18% (Table 2) compares well with the R S D R ~ ~ ~ ~ ~ Of t 1 W o predicted by the Honvitz equation.3 The RSDs between powder, pork muscle and haricot vert beans, RSDReprod - - Table 1 Summary of the variation* in the results for water-soluble vitamins in the multivitamin mixture, and the effect of the correction on the reproducibility Haricot vert beans Multivitamin mixture Milk powder Pork muscle RSD, RSDR RSD, RSDR RSDR(corr) RSDR RSDR(,orr, RSDR RSDR(co,r) Vitamin (Yo) (Yo) (%I (Yo 1 (Yo 1 (%) (%) (%) (%) Vitamin B2 5.0 7.9 4.3 28 27 74 73 35 33 Vitamin B6 4.0 8.8 4.9 Niacin 5.5 10 3.8 9.2 10 Vitamin C 3.9 8.9 2.6 the equation of Horwit~.~ Vitamin B1 3.8 8.8 4.8 17 17 11 8 18 19 - - - - - - - - - - - 15 13 - - - * RSDr, RSDrepeat; RSDR, RSDReprod; RSDR(corr, , RSDReprod of corrected results; and RSD,, predicted RSDReprod according to Table 2 Summary of the results* of the intercomparison (expressed as mg per 100 g of dry mass) Number of Mean RSD, RSDR RSD, (range) (% 1 (% 1 (%) Ratio Vitamin laboratories Vitamin B1- Milk powder Pork muscle Haricot vert beans Vitamin B2- Milk powder Pork muscle Haricot vert beans Vitamin Bg- Milk powder Pork muscle Haricot vert beans Niacin- Milk powder Pork muscle Haricot vert beans Vitamin C- Haricot vert beans 9 10 10 13 11 12 6 7 5 7 6 7 10 0.306t (0.227-0.385) 3.11+ (2.63-3.63) 0.213t (0.160-0.294) 1.44 0.502 (0.083-1.153) 0.318 (0.175-0.564) 0.339 (0.289-0.381) 1.31 (0.74-1.85) 0.199 (0.065-0.313) 0.831 (0.742-0.898) (0.795-2.54) 24.25 1.71 ( 19.53-27.26) (1.32-2.17) 15.55 (12.94-20.45) 4.8 3.7 4.6 7.0 4.2 6.0 7.4 5.6 4.5 5.3 3.1 4.5 5.4 17 11 18 28 74 35 18 35 51 9 11 15 15 14 10 15 11 13 13 14 11 15 12 7 10 6 0.7 1 .o 1.2 2.6 5.9 2.6 1.3 3.2 3.5 0.8 1.5 1.5 1.3 * Mean = mean of means of the laboratories; RSD, = RSDrepeat; RSDR = RSDReprod; RSD, = predicted RSDReprod according to the t Expressed as mg of thiamin chloride.hydrochloride per 100 g of dry mass.equation of Horwit~;~ and Ratio = RSDR,,,d/predicted RSDReprod.ANALYST, MAY 1993, VOL. 118 483 - 5 ~ ~~ Table 3 Extraction and hydrolysis methods used for the determination of vitamin B1 and B2 Extraction and hydrolysis HPLCpost- \ \ HPLC pre-column-column : -€ I Laboratory 1 2 3 4 5 7 8 9 11 12 15 16 17 Vitamin B1 15 rnin at 120 "C with 0.15 rnol dm-3 HC1 1 h at 45 "C with takadiastase 15 rnin at 121 "C with 0.3 rnol dm-3 HCI 3 hat 48 "C wth takadiastase 15 rnin at 121 "C with 0.05 rnol dm-3 H2SO4 3 h at 45 "C with takadiastase - 30 rnin at 100 "C with 0.1 rnol dm-3 HC1 2 h at 50 "C with takadiastase 5 rnin at 100°C with buffer pH 4.5 16 h at 37 "C with amyloglucosidase 20 rnin at 100 "C with 0.1 rnol dm-3 HCI 1 h at 4&50 "C with takadiastase - 30 rnin at 121 "C with 0.1 rnol dm-3 H2S04 18 h at 45 "C with takadiastase 20 min at 121 "C with 0.1 rnol dm-3 HCl 2-3 h at 45 "C with takadiastase-papaine 30 min at 121 "C with 0.1 rnol dm-3 H2SO4 15 h at 45 "C with takadiastase 30 rnin at 121 "C with 0.1 rnol dm-3 HCI 18 h at 37 "C with takadiastase-phosphatase 30 rnin at 121 "C with 0.1 mol dm-3 HCl 18 hat 45 "C with takadiastase Vitamin B2 See vitamin B1 See vitamin B1 15 rnin at 121 "C with 0.05 mol dm-3 H2SO4 No dephosphorylation Extract with 4 rnol dm-3 urea-formic acid (12%) Clean-up with silica gel RP18 See vitamin B1 (1 + 1) 30 rnin at 121 "C with 0.1 mol dm-3 HCI No dephosphorylation 20 rnin at 100 "C with 0.1 rnol dm-3 HCI 2-3 h at 38 "C with takadiastase 30 rnin at 100 "C with 0.1 rnol dm-3 H2SO4 1.5 h at 45 "C with amylase-trypsin See vitamin B1 See vitamin B1 See vitamin B1 30 rnin at 121 "C with 0.1 rnol dm-3 HCI No dephosphorylation See vitamin B1 laboratories obtained in an intercomparison of high-perfor- mance liquid chromatography (HPLC) methods for vitamin B1 in fortified breakfast cereals, were much higher.1 The RSDs were 47 and 34% for cereal samples containing 1.18 and 2.49 mg of thiamin per 100 g, respectively.Most of the participants extracted vitamin B1 by autoclav- ing, or boiling in acid (laboratories 5 and 8) followed by enzymic hydrolysis of the phosphorylated thiamin (Table 3). Takadiastase or mixtures of takadiastase and papaine (labora- tory 12), or takadiastase and phosphatase (laboratory 16) were used.Duration of the enzymic hydrolysis varied con- siderably, between 1 and 18h. Laboratory 7 used a mild extraction, boiling with a buffer of pH4.5, followed by enzymic hydrolysis (overnight) with amyloglucosidase. Two different types of HPLC methods were used. Labora- tories 1, 3 and 11 separated thiamin on the HPLC column, followed by a post-column reaction of thiamin to thiochrome, which was measured by fluorescence detection. Laboratories 5 , 12, 15 and 17 converted thiamin to thiochrome prior to injection into the system and thus thiochrome was chromato- graphed. Laboratories 7 and 8 did not use an HPLC method but applied a fluorimetric procedure (referred to here as thiochrome methods). In these methods thiamin is converted to thiochrome and is measured fluorimetrically after extrac- tion into butan-2-01. Only laboratory 16 used a microbiolog- ical assay.One laboratory omitted the dephosphorylation step and consequently found very low results. In the Youden rank sum test2 this laboratory had an extreme score and so the results of this laboratory (not shown) were not included. One laboratory used ultraviolet (UV) detection after separation by HPLC. The results of this laboratory were rejected for all samples, because the chromatograms showed a broad back- ground of peaks interfering with thiamin. Comparison of the results of the microbiological assay (laboratory 16) with the results of the other methods, showed that this assay gave high results for all samples (Fig.1, milk powder is shown as an example). However, no clear-cut conclusions could be made because only one laboratory used a microbiological assay. The thiochrome methods also showed a tendency to produce high 0.40 tn E $ 0.35 P, 0 2 k 0.30 CT \ F r rn c 0.25 .- Y s T € € € Microbiologica \ \ Thiochrome Fig. 1 Results of individual laboratories for vitamin B1 in milk powder (mg of thiamin chloride-hydrochloride per 100 g of dry mass). Data represent the mean & standard deviation of at least three separate determinations for each laboratory results, but only with milk powder and pork muscle. It was possibly because these methods are less specific than HPLC methods that they gave higher results. No effects of the different HPLC methods (separation as thiamin followed by post-column derivatization, or pre-column derivatization and separation as thiochrome) were apparent. The variation in extraction procedure used (Table 3) did not seem to affect the results.However, the low results of laboratory 5 in the three food samples were probably caused by insufficient acid hydrolysis at 100 "C combined with a relatively short enzymic hydrolysis. In summary, the results for vitamin B1 in milk powder, pork muscle and haricot vert beans obtained with different methods based on HPLC, fluorimetry and microbiology, agreed well between laboratories. The extraction and hydrolysis condi- tions were not critical in these materials. Ultraviolet detection after reversed-phase separation of thiamin is not suitable for the determination of vitamin B1 in these foods.484 In 2 L- m ANALYST, MAY 1993, VOL.118 HPLC : Microbiological € Vitamin B2 Differences between laboratories for the determination of vitamin B2 content found in milk powder, pork muscle and haricot vert beans are very high, with RSDReprod ranging from 28 to 74% (Table 2). For collaborative studies using stan- dardized methods an RSDReprod of only 13% is predicted by the Horwitz equation.3 Extraction by autoclaving or boiling with acid was the extraction method of choice for all of the participants, except for laboratory 4 (Table 3). Laboratories using HPLC mostly applied an enzymic dephosphorylation step with takadiastase (laboratories 1,2,5,11,12,15 and 17). The HPLC methods all involved reversed-phase columns and fluorescence detection, except for laboratory 2, which used UV detection (Table 5 ) .Three laboratories applied microbiological assays. These laboratories differed in the choice of the micro-organisms used: laboratories 7 and 8 used Lactobacillus casei, and laboratory 16 used Enterococcus falcalis (also known as Streptococcus zymogenes) . Laboratory 16 reported extremely high values in all samples (Fig. 2, pork muscle is shown as an example). However, laboratories 7 and 8 also used microbiological methods, but with a different micro-organism, and found values close to the results of laboratories 11 and 15. As laboratory 16 did not apply an enzymic dephosphorylation step, flavin mononucleo- tide (FMN) is expected to be present in the extracts. However, the response of both micro-organisms to riboflavin and FMN is equal, so no explanation is apparent for the high values of laboratory 16.Laboratory 2 did not obtain adequate chroma- tographic resolution for the haricot vert beans. Because of the Table 4 HPLC and other methods used for the determination of vitamin B1 (for extraction methods see Table 3 ) Laboratory Principle HPLC methods, separation as thiamin- 1 Column 8-Si-10,lO pm, Radial-Pak module Eluent Detection Fluorescence, 367/418 nm Eluent Detection Fluorescence, 3601450 nm 11 Column Hypersil-ODs, 5 pm, 250 X 4.6 mm Eluent 0.05 mol dm-3 phosphate buffer pH 7.4- ethanol (300 + 110) 3 Column p-Bondapak CIS, 10 pm, 2.50 x 4.6 mm H20-methanol-acetic acid (735 + 250 + 15), ion-pair Pic-B6 5 mmol dm-3 0.054 mol dm-3 phosphate buffer pH 3.5- methanol (70 + 30), ion-pair heptane- S03Na (7 mmol dm-3), tetraethyl- ammonium chloride (7 mmol dm-3) Detection Fluorescence, 368/420 nm HPLC methods, separation as thiochrome- 5 Column Spherisorb-ODS2,lO pm, 250 x 5.0mm Eluent H20-methanol (60 + 40) Detection Fluorescence, 375/435 nm Eluent H20-methanol (65 + 3.5) Detection Fluorescence, 36.5/435 nm Eluent Methanol-butan-2-01-acetonitrile Detection Fluorescence, 370/420 nm 17 Column Cromasil CIS, 5 pm, 250 x 4.6 mm Eluent Detection Fluorescence, 375/440 nm 12 Column p-Bondapak CL8, 10 pm, 300 x 3.9 mm 15 Column Nucleosil-120 CIS, 5 pm, 250 X 4.0 mm (SO + 10 + 10) H20 + acetic acid, pH 4.5-methanol (40 + 40) Other methods- 7 Thiochrome method Clean-up with Biorex 70 ion-exchange column Fluorescence detection, 365/435 nm Fluorescence detection, 36.5/440 nm Lactobacillus viridescence, ATCC 12706 8 Thiochrome method 16 Microbiological assay background of interfering peaks, a reliable determination of the content was not possible. Results of laboratory 9 are very low because the enzymes used, amylase and trypsin, did not have phosphatase activity.Tn Fig. 2 two groups of results are noticeable: high results reported by laboratories 11 and 15 and laboratories 7,8 and 16 1.5 I . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . m I . . . . . . . . . . . . . . . * . .. . I . . . . . . . . . . . . . . . 5 9 12 3 17 I 11 15 7 8 16 Labcode Fig. 2 Results of individual laboratories for vitamin B2 in pork muscle (mg of riboflavin per 100 g of dry mass).Data represent the mean f standard deviation of at least three separate determinations for each laboratory Table 5 HPLC and microbiological methods used for the determina- tion of vitamin B2 (for extraction methods see Table 3) Laboratory HPLCmethods- Principle 1 2 3 4 5 9 11 12 15 17 Column Hypersil-ODs, 5 pm, 125 x 4.6 mm Eluent H20-methanol (80 +20) Detection Fluorescence, filter FSA 404/FSA 115 Column p-Bondapak CI8, 10 pm, 300 x 3.9 mm Eluent H20-methanol-acetic acid (68.5 + 31 + O S ) , ion-pair hexaneS03Na-heptaneS03Na ( 5 mmol dm-3) Detection UV, 254 nm Column Eluent Detection Fluorescence, 440/530 nm Column Eluent Detection Fluorescence, 450/530 nm Column Eluent H20-methanol (60 + 40) Detection Fluorescence, 450/540 nm Column Eluent Buffer (NH4)2HP04-methanol-1 ,4-dioxane Detection Fluorescence, 453/521 nm Column Eluent p-Bondapak CIS, 10 pm, 250 x 4.6 mm H20-methanol-acetic acid (735 + 250 + 15), ion-pair Pic-B6 5 mmol dm-3 Supelcosil LC-18,3 pm, 75 X 4.6 mm Buffer KH2P04 (100 mmol dm-”), pH 2.9- acetonitrile (85 + 15) Spherisorb ODS2,lO pm, 250 x 5.0 mm LiChrosorb RP-18,s pm, 250 x 4.0 mm (650 + 350 + 10) Hypersil-ODs, 5 pm, 250 x 4.6 mm 0.054 mol dm-3 phosphate buffer pH 3.5- methanol (70 + 30), ion-pair heptane- S03Na (7 mmol dm-3), tetraethyl- ammonium chloride (7 mmol dm-3) Detection Fluorescence, 468/520 nm Column Eluent H20-methanol (SO + 50) Detection Fluorescence, 450/510 nm Column Eluent (65 + 35) Detection Fluorescence, 467/525 nm Column Eluent Detection Fluorescence, 440/520 nm Apex ODs, 3 pm, 150 x 4.6 mm LiChrospher 100 RP-18,5 pm, 125 x 4.0 mm Buffer sodium acetate pH 5-methanol Shandon ODS Hypersil, 5 pm, 250 x 4.6 mm Buffer acetate pH 4.5-methanol (3 + 2) Microbiological methods- 7 Lactobacillus casei, ATCC 7469 S Lactobacillus casei, ATCC 7469 16 Enterococcus falcalis, ATCC 10100ANALYST, MAY 1993.VOL. 118 485 using microbiological methods, and low results reported by the other laboratories. For the microbiological methods the hydrolysis step is merely important to liberate riboflavin and its phosphorylated forms from protein combinations.4 Dephosphorylation is not required because the micro-organ- isms used respond equally to riboflavin and its phosphorylated forms. On the other hand, the HPLC methods used in this intercomparison (except laboratory 4), only determine free riboflavin, so low results might indicate insufficient dephos- phorylation.From a comparison of the extraction procedures (Table 3), it is evident that laboratories 11 and 15 used the most rigorous procedures, which involved acid as well as enzymic hydrolysis. Laboratory 11 examined a few combina- tions of different autoclave and enzymic hydrolysis periods (Table 6). It showed that 15 min autoclaving and 3 h enzyme treatment was insufficient for both pork muscle and haricot vert beans. Also hydrolysis of pork muscle is more difficult than hydrolysis of haricot vert beans, which corresponds with a larger RSDReprod in this trial. These data indicate the need for rigorous hydrolysis procedures. Takadiastase was used by most of the participants; however, the experience of labora- tory 1 shows that the performance of takadiastase varies between different suppliers.These enzyme preparations are mixtures of enzymes with different substrate specificities and activities. Specifications are difficult to obtain.5.6 It is concluded that variability in the results for vitamin B2 in milk powder, pork muscle and haricot vert beans was high. In this intercomparison participants used HPLC and microbio- logical methods. The extraction and hydrolysis procedures are probably the most important sources of variation between laboratories. Table 6 Effect of acid and enzymic hydrolysis duration on the vitamin B2 content measured in one laboratory (No. 11). Results are expressed as a percentage of the highest value obtained. Acid hydrolysis at 121 "C with 0.1 mol dm-3 H2S04; enzymic hydrolysis at 4.5 "C with takadiastase Acid hydrolysis Enzymic hydrolysis 15 min 30 min Product 3 h 34 48 Pork muscle 18h 100 99 Pork muscle 78 91 Haricot vert beans 91 100 Haricot vert beans Vitamin B6 There is a rather large variability in total vitamin B6 values reported in milk powder, pork muscle and haricot vert beans; RSDReprod ranged from 18 to 51% (Table 2), whereas the empirical equation of Honvitz3 predicted an RSDReprod of Two laboratories, 8 and 16, applied microbiological assays with different micro-organisms (Table 7).Laboratories 1, 11 and 12 used reversed-phase HPLC with fluorescence detec- tion, and separated pyridoxine (PN), pyridoxamine (PM) and pyridoxal (PL) after dephosphorylation (Table 7).Laboratory 15 also separated PL, PN and PM, but did not apply enzymic dephosphorylation. Extraction and chromatographic condi- tions, together with the choice of excitation and emission wavelengths and the pH in the cell of the detector, ensured that the phosphorylated vitamers were included in the results.7 Laboratory 18 used gradient elution and was able to separate the phosphorylated vitamin isomers as well. There- fore, a mild extraction procedure, preventing dephosphoryla- tion was applied. One laboratory constantly reported the lowest results (not shown), because only PN was determined. Therefore, these results were not included. The results from laboratories 8 and 16 using microbiological methods do not agree very well (Fig.3, pork muscle is shown as an example). However, laboratory 8 only analysed one food <IS%. 2.0 v) 2 > -0 0" 1.5 0 a, n t-- L F FJ -a 1.0 rn tz .- 4- 5 0.5 Fig. 3 a I i ..................................................... . ........................ . ..... r. HPLC ________+_ Microbiological I I I I 1 I I 12 18 1 15 11 8 16 La bcode Results of individual laboratories for vitamin B6 in pork m&le (mg of pyridoxine per 100 g of dry mass). Data represent the mean k standard deviation of at least three separate determinations for each laboratory Table 7 HPLC and microbiological methods used for the determination of vitamin B6 Principle Laboratory Extraction and hydrolysis HPLC methods- 1 Trichloroacetic acid 5% 1 h at 45 "C with takadiastase 11 Trichloracetic acid 5% 18 h at 37 "C with p-glucosidase or phosphatase 12 10 min at 121 "C with 0.1 rnol dm-3 2 h at 45 "C with amyloglucosidase 30 min at 120 "C with 0.1 mol dm-3 H2S03 15 HzSOj 18 Perchloric acid (0.5 rnol dm-3) at 0 "C Microbiological assays- 8 16 1 h at 100 "C with 1 .O rnol dm-3 HCl 5 h at 121 "C with 0.44 mol dm-3 H2S03 Column Eluent Detection Fluorescence, 333/375 nm Column Eluent Hypersil-ODs, 5 pm, 125 x 4.6 mm Buffer KH2P04 (0.1 mol dm-3) pH 2.1-ethanol (? + 30), ion-pair Pic-B8 LiChrospher RP-18,S yrn, 37.5 X 4.0 mm Buffer KH2P04 (45 mmol dm-3) pH 2.6-acetonitrile (91.5 + 85), ion-pair heptaneS03Na (7.5 mmol dm-3), tetraethylammonium chloride (7.5 mmol dm-3) Detection Fluorescence, 32Y38.5 nm Column Eluent Detection Fluorescence, 290/395 nm Column Eluent Detection Fluorescence.290/39.5 nm Column Eluent Detection Fluorescence, 340/400 nm Spherisorb ODS2-5 ym, 250 x 4.6 mm H2S04 (0.04 rnol dm-3)-methanol(99 + 1) Nucleosill20,S ym, 250 x 4.0 mm H2S04 (0.04 mol dm-3)-methanol (gradient) LiChrosorb RP-18,s pm, 125 X 4.0 mm Phosphate buffer (0.03 rnol dm-3) pH 2.7-octaneS03 (4 mmol dm-")- methanol (gradient) Neurospora sitophila, ATCC 9276 Sacchuromyces carlsbergensis, ATCC 9080486 ANALYST, MAY 1993, VOL. 118 sample, i . e . , pork muscle. The results from laboratory 16 were high in each food sample. Table 8 gives a summary of the results of the laboratories using HPLC methods to separate the different vitamin B6 vitamers. Contrary to the other laboratories, laboratory 12 did not find PN in the haricot vert beans and pork muscle, and did not find PL in the haricot vert beans.This accounts for the relatively low results for total vitamin B6 reported by laboratory 12. Laboratory 1 did not detect any PM in pork muscle, but instead found a relatively high level of PN. Only laboratories 15 and 18 found PN in milk powder. Laboratory 11 used P-glucosidase to dephosphorylate the vitamin B6 vitamers. These results are reported in Tables 2 and 8, and Fig. 3. In addition, laboratory 11 also performed analyses in the haricot vert beans, using acid phosphatase instead of (3-glucosidase. In this case the PN content was only 0.02 mg per 100 g. It follows that the major part of PN in haricot vert beans is bound as pyridoxine-(3-glucoside. Two problem areas can be recognized.Firstly, the identifi- cation of the different vitamin B6 vitamers proves to be difficult, e.g., PN in milk powder. Consequently, total vitamin B6 values as determined by HPLC with summation of the individual vitamers, will be erroneous. Secondly, the enzymic hydrolysis in relation to pyridoxine-6-glucoside in haricot vert beans. As was shown, phosphatase did not hydrolyse these glucosides, whereas f%-glucosidase obviously liberated PN. Takadiastase probably hydrolyses only a fraction of these glucosides. As the bioavailability of these glucosides is less than the bioavailability of PN, it is important to know whether or not these glucosides are included. In summary, the variation between laboratories using various HPLC and microbiological methods in the results for vitamin B6 was high.The extraction and hydrolysis procedures for the vitamin B6 vitamers need to be studied with special attention to the pyridoxine-P-glucoside in vegetable products. Identification of the vitamin B6 vitamers by HPLC proved to be a source of error. Niacin Results for niacin agreed well, with the RSDReprod ranging from 9 to 15% (Table 2), and are close to or better than the values predicted by using the Horwitz equation.3 All of the laboratories using the microbiological methods applied an identical micro-organism, Lactobacillus plantarum (Table 9). Extraction procedures were based on acid hydroly- sis using autoclave heating. One laboratory used HPLC with UV detection after autoclave extraction for the determination of niacin in pork muscle and haricot vert beans.However, difficulties with chromatographic resolution were apparent, so no reliable data could be reported. These data are not included. It is concluded that because of the similarity in methods, results for niacin obtained with microbiological methods agreed well. It is striking that in this intercomparison none of the laboratories (except one laboratory with poor results) used HPLC. Vitamin C The differences between the laboratories for the determina- tion of the vitamin C content found in haricot vert beans are rather high, RSDReprod = 15% (Table 2). These data compare favourably with the results obtained in an intercomparison of HPLC methods for ascorbic acid in breakfast cereals.1 The RSDs between the laboratories were 23 and 19% for cereals containing 20.2 and 41.0 mg of ascorbic acid per lOOg, respectively.Most of the participants used HPLC methods with fluores- cence detection (Table 10). Fluorescence detection requires oxidation of ascorbic acid to dehydroascorbic acid and a reaction with o-phenylenediamine to form a fluorescent quinoxaline. Laboratories 1, 3 and 11 applied enzymic oxidation of the ascorbic acid, whereas laboratory 15 oxidized ascorbic acid with activated carbon. These laboratories thus determined total vitamin C, i.e., the sum of ascorbic and dehydroascorbic acid. However, laboratories 7 and 14, also using HPLC, applied UV detection and consequently were only able to determine ascorbic acid, because dehydroascorbic acid is a non-absorbing compound in the UV region.Laboratories 10, 12 and 16 used fluorimetric methods, also measuring the quinoxaline compound. Laboratory 2 used the classical dichlorophenolindophenol titration to determine ascorbic acid. This method is not as specific as HPLC, because of possible interferences from other reducing substances present in the sample extract. Results of laboratory 2 are rather high, probably because of the lack of specificity of the indophenol titration (Fig. 4). Laboratory 9 reported extremely high values (not shown) in the haricot vert beans. This proved to be because of problems with the chromato- Table 8 Results for individual vitamin B6 vitamers determined by HPLC Pyridoxal Pyridoxine (% Of total B6) (% Of total B6) Laboratory MP* PMt HV+ MP PM HV 1 78 69 30 0 31 27 11 78 76 38 0 7 36 12 69 53 0 0 0 0 15 63 76 54 16 5 14 18 60 59 - 12 14 - * MP, milk powder.t PM, pork muscle. * HV, haricot vert beans. Pyridoxamine (% Of total B6) Total vitamin B6 (mg of pyridoxine per 100 g dry mass) MP PM HV 22 0 43 22 17 26 31 47 100 21 19 32 28 27 - MP PM HV 0.38 1.38 0.16 0.36 1.85 0.24 0.29 0.74 0.07 0.38 1.64 0.22 0.25 0.98 - Table 9 Microbiological methods used for the determination of niacin Laboratory Extraction 1 3 6 7 8 11 16 15 min at 120 "C with 1 mol dm-3 HCl 30 rnin at 121 "C with 0.5 mol dm-3 H2S04 30 rnin at 121 "C with 1 mol dm-3 H2S04 15 rnin at 121 "C with acetic acid-acetate-KCN 15 rnin at 120 "C with 2 mol dm-3 HCl 15 rnin at 121 "C with 1 mol dm-3 HCl - Micro-organism Lactobacillus plantarum, ATCC 8014 Lactobacillus plantarurn, ATCC 8014 Lactobacillus plantarum, ATCC 8014 Lactobacillusplantarum, ATCC 8014 Lactobacillusplantarurn, ATCC 8014 Lactobacillusplantarurn, ATCC 8014 Lactobacillusplantarurn, ATCC 8014ANALYST, MAY 1993, VOL.118 487 ~ ~~ Table 10 Methods used for the determination of vitamin C Laboratory Extraction Principle HPLC metliods, separation us quinoxaline derivutive- 1 Trichloroacctic acid ( 5 % ) Column Hypersil ODs, 3 pm, 125 x 4.6 mm Oxidation with ascorbatc oxidase Eluent Buffer phosphate (0.8 mol dm-3) pH 7.8-mcthanol(80 + 15) 3 Metaphosphoric acid (1 %) Oxidation with ascorbate oxidase 11 Metaphosphoric acid ( I YO) Oxidation with ascorbate oxidase 15 Mctaphosphoric acid-acctie acid Oxidation with activated carbon HPLC methods, sepurution us uscorhic acid- 7 Metaphosphoric acid Detection Column Eluent Detection Column Eluent Detection Column Eluent Dctcction Column Fluore&&, 3671418 nm SNCV-18 RCM cartridge, 4 pm, 125 x 3.9 mm Buffer phosphate (0.08 mol dm-3) pW 7.8-methanol Fluorescence, 355/425 nm Chromsphcr RP-l8,5 pm, 100 x 3.0mm Buffer phosphate (0.08 mol drn-’) pH 7.8-methanol Fluorescencc, 355/425 nm Spherisorb ODs, 10 pm.250 x 4.0 mm Buffer acetate pH 5.2-methanol (1 + 1 ) Fluorescence, 3501430 nm RP-18,5 um, 250 X 4.0 mm 80 + 20) 915 + 85) . . Eluent Detection UV, 254 nm Eluent Detection UV, 254 nm Elucnt Detection UV, 248 nm Buffer acetate (2 mol dm-”), ion-pair tetrahcxylammonium bromide 9 Metaphosphoric acid-acetic acid Column LiChrosorb RP-18,5 ym, 250 X 4.0 mm H20(triethylamine, HOAc)-methanol(830 + lSO), ion-pair heptancS03 14 Metaphosphoric acid-acetic acid Column Partisil PS, S pm, 250 X 4.6 mm H20-methanol-acetic acid (744 + 250 + 5.6) Otli or rnetliod- 2 Mctaphosphoric acid-acetic acid Titration with 2,6-dichlorophenolindophcnol 10 Metaphosphoric acid, 5% Microfluorimetric method, continuous flow 12 Metaphosphoric acid-acetic acid Microfluorimetric method Oxidation with activated carbon, dcrivatization with o-phenylcnediaminc 16 Metaphosphoric acid-acetic acid Microfluorimetric method, continuous flow Oxidation with activalcd carbon.derivatization with o-phcnylenediaminc 25 v) m E >. -0 0 D 20 : L al a E“ 15 Y E, C .- 5 10 __~ / HPLC-Flu d H P L C - U V t Fluorimetric I lndophenol T tit ration + I L € .... . ........._ .... ......_. n.. .............. 9. ... . I i 1 1 I I I I I I l l 1 3 11 15 7 14 10 12 16 2 Labcode Fig. 4 Results of individual laboratories for vitamin C in haricot vert bcans (mg per 100 g of dry mass). Data represent the mean k standard deviation of at least threc separate determinations for each laboratory graphic resolution; the wrong peak was identified as vitamin C. These results were rejected. On comparison of the results of laboratorics 7 and 14, who only determined ascorbic acid, no differences with the results of methods determining total vitamin C are evident. So, probably the dehydroascorbic acid content of the haricot vcrt bcans was low. In summary, results with HPLC and fluorimetric methods, excluding the indophenol titration method, showed a good agreement in the determination of vitamin C in haricot vert beans.Discussion A summary of the results of this intercomparison on the methods for the determination of water-soluble vitamins in foods is given in Tablc 2. Although homogeneous materials were involved, it can be argued that homogeneity of most of the vitamins (except for vitamin C) in the haricot vcrt beans, and of niacin and vitamin B6 in the other two foods was not demonstrated beforehand. As participants carried out sep- arate determinations in at least two different sachets, inho- mogeneity will be reflected in the analytical variation of the laboratories (RSD,,,,,,). By comparing RSD,,,,,, and RSDKeprod for these vitamins (Table 2), it can be concluded that the possible inhomogeneity was not an important factor.The reproducibility (RSDReprod) shown gives an impression of the state-of-the-art of water-soluble vitamin analyses, performed by experienced food laboratories. These labora- tories applied their own routine mcthods. Similar to the fat-soluble vitamins2 the ratio of the RSDReprod found in this intercomparison and the predicted RSDKeprod was calculatcd. If different procedures used by different laboratories do not have a strong influence on the results, the ratio will be close to 1. This proved to be true for vitamin B I , niacin and vitamin C, and for vitamin B6 only in milk powder. As these data will be useful for laboratories, vitamin B,, niacin and vitamin C have been issued as indicative values in these food reference materials, which have been certified for major dietary components and elements.8 In contrast to the fat-soluble vitamins, a wider range of methods were used.Only 7 out of the 10 laboratories used HPLC for the determination of vitamin B,. Three laboratories used a microbiological assay for thc determination of vitamin B2, as opposed to 10 participants using HPLC. The microbio- logical results did not agrec very well with HPLC. Inadequate extraction and/or hydrolysis procedures were probably the most important cause of the large differences between laboratories for vitamin B2. Results for vitamin B6 were discrepant. Tdentification of the different vitamcrs and the extraction and hydrolysis procedures were judged to be the major sources of variation. Optimization of the extraction and hydrolysis procedures will be an important step in the improvement of methods for the determination of vitamins B2 and B6.The similarity in values reported for niacin was caused488 ANALYST, MAY 1993, VOL. 118 by all of the laboratories using a microbiological assay with an identical micro-organism. Results for vitamin C did not agree well. The HPLC and fluorimetric methods and the dichloro- phenolindophenol titration were used. As expected, the dichlorophenolindophenol titration was found to give the highest results. The present intercomparison failed to identify the role of different calibration procedures, because of unfamiliarity with the extraction procedures prescribed for the multivitamin mixture. The skilful participation of the following laboratories is gratefully acknowledged: Instituto del Frio, lnstituto del Fermentaciones Industriales CSIC, Madrid, Spain; Universi- tat-Gesamthochschule Paderborn, Paderborn, Germany; Laboratory of the Government Chemist , Teddington, Mid- dlesex, UK; The National Food Agency of Denmark, Soborg, Denmark; Schweizerisches Vitaminin-Institut, Basel, Swit- zerland; State Institute for Quality Control of Agricultural Products (RIKILT) Wageningen, The Netherlands; Unilever Research, Bedford, UK; VTT Food Research Laboratory, Espoo, Finland; TNO-CIVO Institutes, Zeist, The Nether- lands; Bundesforschungsanstalt fur Ernahrung, Stuttgart, Germany; Leatherhead Food R. A., Leatherhead, Surrey, UK; University College Cork, Cork, Ireland; Federal Dairy Research Institute, Liebefeld-Bern, Switzerland; Swedish National Food Administration, Uppsala, Sweden; Produits Roche, Fontenay sous Bois, France; Food Inspection Service, Maastricht, The. Netherlands; AFRC Institute of Food Research, Norwich, UK. References 1 2 Nicolson, 1. A . , Macrae, R., and Richardson, D. P., Analyst, 1984, 109, 267. Hollman, P. C. H.. Slangen, J . H . , Wagstaffe. P. J . , Faure, U., Southgate, D. A. T., and Finglas, P. M., Analyst, 1993, 118, 463. Horwitz, W., Anal. Chern.. 1982, 54, 67A. Methods of Vitamin Assay, cd. Myer F., Interscicncc Publish- ers, New York, 3rd cdn., 1966, p. 150. MacBridc, 11. E., and Wyatt, C. J . , J. Food Sci., 1983,48,748. Schrijver, J . , Die Nahrung, 1987, 31, 1045. Bognar, A., 2. Lehensm. Unters. Forsch., 1485, 181, 200. Hollman, P. C. H., Boenke, A . , and Wagstaffe, P. J . , Fresenius' J. Anal. Chern., 1993, 345. 174. NorE-Ref. 2 is to Part 1 of this series. 3 4 5 6 7 8 Puper 21061 72 D Received November 19, 1992 Accepted Junuary 18, 1993

ISSN:0003-2654    

DOI:10.1039/AN9931800481

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 489 Recovery of Sulfadimidine From Pig Feeds Ian M. Barwick and Peter Warwick Department of Chemistry, Loughborough University of Technology, Loughborough, Leicestershire, UK LEI? 3TU Neil T. Crosby Laboratory of the Government Chemist, Queen's Road, Teddington, Middlesex, UK TWI I OLY The poor analytical recovery of sulfadimidine from medicated pig feeds was investigated by high-performance liquid chromatography (HPLC) using unlabelled and 14C-labelled sulfadimidine. In addition, autoradiographic studies of 1%-labelled sulfadimidine binding in feeds was investigated. The HPLC studies using unlabelled sulfadimidine showed that there was an inverse relationship between feed moisture content and sulfadimidine recovery. Feeds with high moisture content (18% m/m) gave lower recoveries than both control feeds (12% m/m) and feeds with low moisture content (5% m/m).Studies using HPLC with both ultraviolet and radioactivity detection showed no evidence for degradation products of sulfadimidine. Autoradiographic studies showed that the non-recovered drug was dispersed throughout the feed. A hypothesis is proposed to account for the poor recoveries. Keywords: Sulfadimidine determination; animal feed; [W]-sulfadimidine; autoradiograph y Medicinal additives are routinely incorporated into medicated animal feeds as prophylactics or as growth promoters. The nature and levels of inclusion of such compounds must be legally declared by the feed manufacturer' and therefore checks are necessary to ensure that these declarations are correct.A problem frequently encountered with this type of analysis is that medicinal additives are only partially recovered €rom the feed, particularly after storage. It has been suggested that the poor recoveries are due to degradation of the drug or binding of the drug to feed constituents. Additionally, many feeds are supplied in pelleted form and have been subjected to elevated moisture levels, temperatures and pressures during the production process; these conditions have been thought to accentuate any degradation or binding tendency. Sulfadimidine (SDM) [4-amino-N-(4,6-dimethyl-2-pyrimi- dinyl)benzenesulfonamide] is a widely used antibacterial agent included in pig feeds, and several methods have been reported for its determination.'-" Several workers47 have shown that the recovery of SDM from feeds is inversely proportional to the time it is in contact with the feed, but many8-1(' have not taken this into account when validating their methods and have used only freshly spiked samples. Problems for the regulatory analyst occur when SDM levels are found to be below the permitted lower tolerance, as this could simply be due to a poor analytical recovery of the drug and not the result of an accidental low level of drug addition by the manufacturers.Conversely, an over addition of drug by the manufacturer might pass undetected owing to incomplete drug extraction. In this work, the causes of poor SDM recoveries by the novel use of 14C-labelled SDM were investigated. Experimental and Results Reagents As specified by Conway.s bury, Buckinghamshirc, UK).MBq (Sigma, Poole, Dorset, UK). national, Amersham, Buckinghamshire, U K). Wigston, Leicestershire, UK). Liquid scintillant. Ecoscint A (National Diagnostics, Ayles- Labelled SDM. [Phenyl-ring-UL-1~C]-sulfatlirnidinc, 10 Autoradiography film. Hyperfilm (%Max (Amersham Inter- Pig feed. Ultra Finisher Meal (Dalgety Agriculture, South Apparatus High-performance liquid chromatographic system. Philips (Cambridge, UK) PU 4021 multi-channel detector; Philips PU 4100 liquid chromatograph fitted with a Rheodyne Model 7125 injection valve (Fisons, Loughborough, Leicestershire, UK) and 20 p1 sample loop (Fisons); Philips P 3202 computer with PU 6003 diode-array software and PU 6000 integration software. High-performance liquid chromatographic columns. Apex Octadecyl, 5 pm, 250 x 4.6 mm i.d.(Fisons); Apex Octadecyl, 5 pm, 50 x 4.6 mm i.d. guard column (Fisons). Fraction collector. RediFrac (Pharmacia, Milton Keynes, Buekinghamshire, UK). Liquid scintillation counter. LKB Wallac 1215 Rackbeta (Pharmacia). Sample tubes. Falcon poly(propy1ene) disposable sample tubes (50 ml) (Northern Media, Nottingham, UK). Scintillation vial inserts. Zinsser Analytic (Scientific Indus- tries International, Loughborough, Leicestershire, UK). Methods The method used throughout this paper is that described by Conways and is based on the extraction of SDM from the feed with aqueous acetonitrile, followed by cation-exchange clean- up and analysis of the resulting solution with a reversed-phase high-performance liquid chromatography (HPLC) system.During our work, various dosing methods to introduce SDM into the feed were investigated to ascertain their suitability. These methods included dosing a bulk amount of feed with a premix of SDM in calcium carbonate, and dosing both a bulk amount of feed and individual 2 g portions with a solution of SDM in methanol. It was found that dosing bulk amounts of feed with either a premix or a solution introduced errors due to uneven distribution of the SDM throughout the bulk, even after significant mixing times (6 h). The most satisfactory dosing method was found to be the addition of a solution of SDM in methanol to small portions of feed. However, this method of dosing has disadvantages, as indicated under Discussion. To facilitate the preparation of large numbers of dosed feed samples for use in analytical studies, 2 g feed samples were weighed into 50 ml poly(propy1ene) sample tubes.Conways specified that 10 g of feed containing 100 pg g-1 of SDM were extracted with 100 ml of extraction solution. To ensure that490 ANALYST, MAY 1993, VOL. 118 Table 1 HPLC and radiochemical results showing decrease in SDM rccovcry from feed with time SDM recovery (%) Samplc NO. 1 2 3 4 5 6 Contact time/ d 0.04 0.04 7 7 16 16 UV detection 94 I 00 73 70 60 55 (+4%) Radiochemical detection (-tS%) 90 93 85 79 68 69 ~ -~ Table 2 Analytical method mass balance Sample No. Contact time (d) 1 2 3 4 5 6 0 0 7 7 1 6 1 6 Anulyticwl S ! N ~ : C ( ~ S % O ) - Extraction efficiency (YO)* 92 88 85 79 68 69 Retention of SDM on cation- exchangc column (%))t 97 96 96 96 90 88 Activity rcmovcd by washing column Activity rcmovcd by washing column KccoveryofSDMfromcolumn(%)a 98 96 81 86 88 83 * Extraction efficiency is the amount of activity found in the extraction solution as a pcrcentage of the total amount of activity added to feed..F Rctention is the percentage of the extracted activity retained on the cation-exchange column calculated by measuring the activity of the load solution after passing through the column. :i: Pcrccntagc of activity retained on the cation-exchange column removcd by washing with either water o r methanol. Percentage of the retained activity on the cation-exchange column that was rcmovcd by elution with 25 ml of HPLC mobile phase. with water (YO)$ 0 0 0 0 0 0 with methanol ( % ) 3 0 0 0 0 0 0 the ratio of the mass of feed used in the analysis and the volume of extraction solution remained constant when using 2 g feed samples, 20 ml of extraction solution were used.To ensure that the amounts of SDM injected onto the column were the same as those used by Conway, dosing levels were between 300 and 500 pg g- I . Preparation of [W]-SDM Dosing Solution Both [1JC]-SDM (10 MBq) and carrier SDM (21.5 mg) were transferred into a 25 ml calibrated flask, dissolved in methanol and diluted to volume with methanol. This was the labelled SDM dosing solution and had a specific activity o f 401 kBq ml-1 and an SDM concentration of 0.96 mg ml-l. Analytical Recovery of '4C-labelled SDM From Pig Feed Amounts of 2 g of commercially obtained pig feed were accurately weighed into each of six sample tubes (1-6) and to each was added 0.5 ml of the 1T-labelled SDM dosing solution to give SDM dosing levels of 232-237 pg g-1.The tubes were then capped and mixed using a vortex mixer for 10 s to allow interaction between the dosing solution and the feed. The drug was then extracted with solvent from tubes 1 and 2 after a drug-feed contact time of 1 h. After a drug-feed contact time of 7 d the drug was extracted from tubes 3 and 4 and after 16 d the drug was extracted from tubes 5 and 6. The SDM recoveries were measured by HPLC using ultraviolet (UV) detection and also radiochemical detection by fraction collection and liquid scintillation counting. The results in Table 1 show that SDM recoveries decreased with time.In order to ensure that the decrease in activity did not occur during the extraction, clean-up and chromatographic stages of the analysis, the experimental specific activities of the solutions were compared with theoretical specific activities, which enabled a mass balance study to be carried out. Hence all the activity extracted could be rnonitorcd to ensure that no Fig. 1 [ l-'C]-SDM autoradiograph of feed sample following drug extraction after 16 d drug-feed contact timeANALYST, MAY 1993. VOL. 118 49 1 losses occurred during the subsequent processing stages. The results are shown in Table 2. Autoradiography Studies In order to investigate the distribution of the non-extracted [“TI-SDM, autoradiography studies were initiated. After the feed solid had been filtered off from the extraction solution, the filter-paper and the solid were carefully dried and Sellotape strips were then used to remove small portions of the feed from the filtcr-paper, which were then fixed onto microscope slides.This method was found to be much easier than trying to embed the feed solid in wax and taking sections using a microtome. Measurement of the amount of activity associated with a particular sample due to the blackening of the film was not possible owing to the uneven thicknesses of the particles on the slides. The feed solids from sample 5 were transferred onto slides and all were then exposed to the autoradiographic film for 4 d, after which time they were developed. The results shown in Fig. 1 indicate that SDM, after the extraction process, was generally distributed on all the feed particles.Dosing Individual Feed Components With “V-labelled SDM To investigate further whether the non-extracted [ “C]-SDM was generally distributed on the feed Constituents, individual Table 3 Recovery of [WI-SDM from constituents (%) Contact timeid Constituent Wheat Fishmeal Riccbran Barley HiPro soya Calcium carbonate Pull fat soya Malt culms Meat and bone meal Wheatfced 0 85 83 87 93 97 91 90 94 96 90 7 76 80 80 87 100 95 90 79 91 73 16 75 89 75 72 I00 92 73 69 71 59 constituents were dosed with the l-‘C-labelled SDM solution and then extractcd at time intervals. Feed constituents (wheat, barley, meat and bone meal, HiPro soya, ricebran, wheatfced, malt culms, fishmeal and full fat soya) were obtained from a local mill.Calcium carbonate was also chosen as it is used in some laboratories to prepare an SDM premix. Eight samples of each constituent (0.1 g) were accurately weighed into scintillation vial insets and 20 1-11 of thc 1T-labelled SDM were added to each vial, Duplicate samples from each constituent batch were then extracted using 70% aqueous acetonitrile by shaking mechanically for 1 h, after which time the samples werc centrifuged at 6000g for 1 min to separate the extraction solution from the solid. Samples of 20 1-11 of the cxtraction solutions were then taken and placed in scintillation vials containing 4.2 ml of Ecoscint. Following mixing, all samples were counted in the 14C channel of the liquid scintillation counter.The activity values were quench corrected and background subtracted. Calculations of the extraction effi- ciencies were made by dividing the experimentally found specific activity by the theoretical specific activity. The extraction process was repcatcd after 7, 14 and 25 d contact times between SDM and the feed and the results are given in Table 3. Samples extracted after 7 d contact time were then prepared for autoradiography as described previously. Each sample was contacted with the film for 10 d. The results of the autoradio- graphy are shown in Fig. 2. Table 3 and Fig. 2 show that calcium carbonate does not sorb [l4C)-SDM and that the soya-based materials have little affinity for [I4C]-SDM. The former phenomenon can be explained by the lack of porosity of the calcium carbonate and the latter by the high oil content of the soya-based materials, which prevented penetration of the water into the pores.Effect of Storage Conditions on SDM Recovery The feed moisture content of a feed already dosed with SDM was modified by placing samples in cithcr a high-humidity, water-saturated environment or a low-humidity, dry environ- ment, in the following way. Amounts of 2 g of commercial Fig. 2 [I4C]-SDM autoradiograph of feed constituents following drug extraction after 7 d drug-feed contact time492 - - - ANALYST, MAY 1993. VOL. 118 - 60 L- 2 > 40 8 2 20 6.6 I I 98 6.4 6.2 6.0 5.8 5.6 4- w c 0 2 .- 0 73 9 5.4 5.2 I I I 10 20 30 40 Tim e/d Fig. 3 Moisturc and B. SDM rccovery SDM recovery from low moisture contcnt feed samples.A, I 8 0 I I 10 10 20 30 Ti me/d 40 Fig. 4 Moisturc and B, SDM recovery SDM rccovery from high moisturc content feed samples. A , Table 4 Effect of storage conditions and oven drying on SDM recovery SDM rccovery (%) Storage in Contact timc/d Drying time/d silica gcl conditions Storagc in water-saturated 1 1 7 66 54 25 5 54 39 37 11 48 91 non-medicated pig feed were accurately weighed into each of 40 sample tubes (labelled 1-40). To each tube was added 0.5 ml of a solution of SDM in methanol to give an average dosing level of 300 yg g-1. Mixing of the methanol dosing solution with the feed samples was as described previously. Tubes 1-20 were then placed uncapped in an air-tight box containing 500 g of self-indicating silica gel to absorb moisture. Samples 21-40 were placed in another air-tight box containing 500 ml of water to provide a water-saturated environment.Both air-tight boxes were placed inside polystyrene insulation containers, which were stored in an air-conditioned room maintained at 20 k 3 "C. After 11,25 and 37 d in their respective environments, four samples were removed from each box. Two samples were analysed to determine SDM recovery and the results are shown in Figs. 3 and 4. The remaining two samples were dried to constant mass in an oven maintained at 105 "C to determine the feed moisture content. After oven drying, the samples were analysed for SDM recovery to determine the effect of oven drying on drug recovery and the results are given in Table 4. The results showed that with the samplcs of high moisture content the longer the drying time the greater was the recovery of SDM.Conversely, the samples of low moisture content were adversely affected by prolonged drying, with SDM recoveries showing a decrease with increasing drying time. The above study was repeated with the following modifica- tions to the method. To ensure that the relative humidities in 80 - I I 0 10 Contact time/d 20 Fig. 5 *, 75%; 0, 33%; and ., 0% SDM recovery as a function of storage humidity, 0, Control; the environments remained constant, saturated salt solutions, as reviewed by Young,l2 were used. Additionally, the interval between analyses was reduced to 3 d to ascertain whether the initial change in SDM recovery after dosing was gradual over the entire period of the study or whether the recovery changcd sharply to become more constant subsequently.Portions of 2 g of feed were weighed into each of 112 sample tubes. Mixing of the methanol dosing solution with the feed samples was as described previously. Samples 1-28 were placed in an air-tight box on their own to scrve as controls. Samples 29-56 were placed in contact with a saturated solution of sodium chloride (75% relative humidity), samples 57-84 were placed in contact with silica gel and samples 85-1 12 were placed in contact with a saturated solution of magnesium chloride (33% relative humidity). Samples were removed at 3 d intervals, with two samples being used in the respective moisture determinations and two samples being analysed for SDM recoveries. The SDM recoveries are shown in Fig.5. The results for the samples stored at 33% relative humidity are incomplete because some samples were contaminated with the saturated magnesium chloride solution. Effect of Modifying Feed Moisture Content Before Dosing With SDM As the results from the study above had shown an invcrse relationship between feed moisture content and SDM recovery, it was decided to investigate whether modifying the fced moisture content before dosing would also show such a relationship. Amounts of 2 g of a non-medicated feed were accurately weighed into each of 60 sample tubes (labelled 1-60). Samples 1-20 and 2 1 4 0 were placed in air-tight boxes containing silica gel and water, respectively, as in the previous study. Samples 41-60 were placed in an empty air-tight box to serve as controls.All three boxes were placed inside polystyrene insulation containers and placed in an air-conditioned room as before, for 6 d. Representative samples were taken from the respective environments after 6 d and each sample was dried for 6 h at 105 "C. From a previously measured feed moisture content of 12% m/m, it was found that samples 1-20 had decreased to 5% m/m, samples 2 1 4 0 had increased to 18% m/m and samples 41-60 remained unchanged. All samples were then dosed with 0.5 ml of SDM in methanol solution and the procedure from the previous study was followed. Follow- ing removal from the fume hood, the sample tubes were capped then stored in an air-conditioned room and protected from light. Four samples were removed after 0,7,28 and 42 d of drug-feed contact time and two samples were analysed for SDM.The results are shown in Fig. 6. The remaining two samples from each environment were dried for 6 h at 105 "C to determine the feed moisture content, then stored for a further week and analysed for SDM. The results are shown in Fig. 7. The fact that in feeds of high moisture content the SDM recovery decreased compared with feeds of low moisture content led us to the consideration that the increased moistureANALYST, MAY 1093, VOL. 118 493 loo A - i A B v) 40 - C I I I I I 10 20 30 40 20 1 ' 0 Contact time/d SDM recovery as a function of fecd moisture content. A. Fig. 6 Low moisture; B, control; and C, high moisture g 5 0 t 40 t 0 10 20 30 Contact time/d SDM recovery from feed samples after drying.A , Low Fig. 7 moisture; B, control; and C, high moisture content might accelerate hydrolysis of SDM. To investigate this, the stability of a saturated aqueous solution of SDM was determined. Stability of Aqueous SDM Solutions The SDM (65.7 mg) was weighed into a 100 ml calibrated flask and 80 ml of HPLC-grade water were added. The flask was stoppered, shaken and placed in an ultrasonic bath for 2 min to aid dissolution. On removal from the ultrasonic bath, the solution was diluted to volume and filtered through a 0.45 pm filter and aliquots were transferred into two 30 ml glass storage bottles, labelled 1 and 2. Bottle 1 was wrapped in aluminium foil to protect it from light and bottle 2 was left exposed to light. Samples of each solution were taken after 0,1,8 and 62 d and injected onto the HPLC column to determine the concentration of SDM in the solution.The results are given in Table 5 and indicate that SDM was stable in solution over the period of the study. Sorption of SDM on Feed The experimental results showed that binding of SDM to feed constituents was occurring and that drug degradation was not in evidence. In order to investigate whether there was a limited number of binding sites in the feed, an experiment was set up to investigate the effects of varying the amount of SDM added to the feed. Sample tubes labelled A 1-16, R 1-16 and C 1-16 containing feed (2 g) were prepared and the previously described dosing protocol was followed using known dilutions of a solution of SDM in methanol. The dosing levels ranged from 11 to 500 pg g-1.Control samples were dosed with 0.5 ml of methanol. Analyses were then performed at 7,28 and 42 d on the sets of Table 5 Stability of saturated aqueous solution of SDM Average SDM peak area at timc/d RSD Sample 0 I 8 62 Avcragc (%) Light excluded 1042.8 1077.2 1029.9 1036.7 1046.7 2.0 Light exposed 1071.3 1095.4 1041.6 1068. I 1069.1 2.1 400 I 7 7 300 0 0 5 3 5 200 0 .+ 5 100 0 100 200 300 400 500 SDM added/pg g - l Sorption study. A, 7; B, 28; and C, 42 d Fig. 8 samples A , B and C, respectively. Fig. 8 shows that the higher the SDM dose, the higher was the percentage SDM recovery and that the recovery decreased with drug-feed contact time. Effect of Reducing the Particle Size of a Feed on SDM Recovery The results from the sorption study suggested that SDM was sorbing to the feed particles.To investigate this further, the specific surface area of a feed was increased by reducing the particle size of the feed in a grinder. The reduction in particle size was measured using sieve analysis and by laser diffraction in a Malvern 2600c laser particle sizer. Although the particle size range (10-100 pm) did not change during the grinding, the particle size was reduced, as was evidenced by the observation that in the original feed 50% of the particles were <350 pm whereas after grinding 50% of the particles were <150 pm. Recovery studies were then performed on both the feed and reduced particle size feed. An amount of feed (about 10 g) was ground for 1 min using a domestic coffee grinder.Three 2 g portions were accurately weighed into sample tubes (1-3). The same feed but unground was used to weigh out a further three 2 g samples (4-6). Dosing the samples with a solution of SDM in methanol was performed to give SDM levels of 662-695 yg g- 1 . The samples were then stored in a polystyrene insulation box and left for 18 d to allow interaction between the SDM and the feed. The recoveries from the two feed samples were found to be the same. If it is assumed that n o alteration of the surface groups took place during the grinding process, then it could be deduced that SDM is not predominantly confined to the feed particle surfaces. Alternatively, if there is an excess of sorption sites on the surface of the particles, size reduction would not necessarily effect an increased surface sorption. Discussion Considering the moisture content of the feed, calculations concerning the maximum amount of SDM that can dissolve in the moisture in the feed were made using the following assumptions: (i) t h e nominal feed moisture content is 12% m/m; (ii) all the water in the feed is capable of dissolving SDM; and (iii) the maximum solubility of SDM is about 400 pg g-1 (determined in the stability study).If a feed contains SDM at a level of 100 pg g-1, then in an aged feed approximately 50% can dissolve in the water present in the feed. During some extractions, low recoveries of 50% were observed.494 ANALYST, MAY 1993, VOL. 118 Extending this further, a hypothesis is proposed whereby dissolution of the SDM by moisture in the feed results in solutions being formed that could penetrate into the feed matrix, either by absorption or diffusion. Scanning electron microscopic examination of the feed constituents showed that all exhibit a degree of porosity.If drug dissolution and diffusion into the feed particles occurs, then thc non-extract- ablc SDM could be explained by penetration of the solution deep into the pores of the feed particles. This hypothesis may be tested by using altcrnative extraction solvents. This work is in progress. The moisture content of the feed was altered from its original value of 12% to artificial values of 5 and 18% in order to provide samples of feed with different moisture contents. The samplcs of high moisture content could dissolve more SDM than the corresponding samples of low moisture content, which thcrcforc allowed morc of the SDM present to be transferred into the pores of the feed.I n the dosing method in this study a solution of SDM in methanol was used. Although this method is not the same as that used in industry, it did allow accurate addition of SDM to the samples. Onc feature noted about the methanolic dosed samples was that the decrease in drug recoveries was larger in magnitude and occurred at a fastcr rate than with commer- cially manufactured samples. This might be explained by the methanol not being removed completely after dosing and by the increased solubility of SDM in methanol compared with water. Bettcr recoveries from commcrcial samples could also result from (a) overage, i . e ., adding more than the declared content, ( h ) addition of SDM in the solid state, resulting in increased time for dissolution from the premix into the feed moisture, or ( c ) the use of a granular form of SDM. The feed moisture content-SDM recovery correlation could explain why pelleted feeds give lower recoveries than non- pcllcted feeds owing to the higher moisturc, temperature and pressures experienced by the feed in the conditioning or ripening stage of the production process immediately bcfore extrusion. Conclusions These studies have shown an inverse relationship between the feed moisture content and the recovery of SDM. The SDM was found to be stable in aqueous solution and degradation products were not observed after it had been in contact with feed. A hypothesis has been proposed to account for the extraction behaviour of SDM from feed and also t o account for the fact that pelleted fceds give lower recoveries than the same feed before pelleting. The authors thank the Laboratory of the Government Chemist for funding this research (Grant No. EMRC/35) and A. Arafa at Dalgety Agriculture, South Wigston, Leicestcr- shire, for kindly supplying the feeds. References 1 Stututory Instrunzrnt 1985, No. 1533, The Medicines (Medicated Animal Feeding Stuffs) Regulations, HM Stationery Otficc, London, 1985. Horwitz, W., J . Assoc. Off. And. Chem.. 1981, 64, 104. Horwitz, W., J . Assoc. Off. And. Clwm., 1981. 64, 814. Holder, C. L., Thompson, H . C., Jr., and Bowman, M. C.. J . CIirornutogr. Sci., 1981, 19. 625. Conway, B. 0. B., Analysr, 1988, 113, 1397. Analytical Methods Committee, Annlysr, 1992, 117. 817. Munns, R. K.. and Roybal, E. J . , J . As~oc. Off. And. Chc.m., 1982, 65, 1048. Blanchflower, W. J . . and Rice, D. A . , J . A~soc. Off Anal. Chem., 1988, 71, 302. McGary, E. D., Anuly~t. 1986, 111, 1341. Schwarz, D. P., J . AJSOC. Off Anal. Chem., 1985, 68, 214. Stringham, R . W., Mundell, E. C., and Smallidge, R. L., J . Assoc. Off. Anal. Chem., 1982, 65, 823. Young, J . F . , J . Appl. Chem.. 1967, 17. 241. 2 3 4 5 6 7 8 9 10 11 12 Paper 21055.525 Received October 19, 1992 Accepted Notiemher 23, 1992

ISSN:0003-2654    

DOI:10.1039/AN9931800489

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 495 Coupled Gas Chromatography-Atomic Absorption Spectrometry Using Semi-enclosed Tubular Atomizers: Theoretical Analysis of Detector Performance Characteristics Douglas C. Baxter and Wolfgang Frech Department of Analytical Chemistry, University of Ume5, S-90 I 87 UmeA, Sweden For the determination of specific organometallic compounds in various environmental samples, coupled gas chromatography-atomic absorption spectrometry (GC-AAS) has proved to be a useful instrumental combination. However, to enable the optimum dimensions and operating conditions of the atomizer to be identified requires consideration of the theoretical aspects of GC-AAS detector performance. In this work, expressions are derived to describe the relative values of the detector sensitivity (peak height), and chromatographic efficiency and resolution in terms of the ratio between the residence time of the analyte in the atomizer and the standard deviation of the (assumed) Gaussian peak eluting from the GC system (q+).The greatest detector sensitivity is obtained at high tR/oG values, but at the expense of degraded separation capabilities. Recommendations are given for the design of GC-AAS systems providing the best possible performance characteristics. A detector response model is also presented for predicting characteristic mass values (peak height and peak area) provided by GC-AAS systems and verified using literature data and experimental results. Keywords: Gas chromatography; atomic absorption spectrometry; graphite and quartz tube atomizers; detector response model and performance characteristics; speciation Organometallic species are ubiquitously distributed in the environment as a result of both natural and anthropogenic processes.1.2 These species are often more toxic at a given concentration than the corresponding inorganic compounds or metal ionsl-3 (although there are exceptions, such as arsenic), exhibiting a greater potential for bioaccumulation.Thus, there is a current impetus for developing suitable analytical procedures to determine the concentrations of specific organo- metallic compounds (speciation) in every conceivable type of environmental sample , for regulatory and monitoring pur- poses. One of the most attractive and widely used analytical approaches to this type of speciation problem is to couple the separatory power of gas chromatography (GC) with the metal-specific detection capabilities of atomic absorption spectrometry (AAS) .4 The required instrumental components are readily available and numerous articles and reviews provide the necessary details on interfacing, as well as practical information concerning sample collection, storage and preparation.1+9 Various detector designs have been employed, the most significant, with respect to detection limits, being electrothermal (ET) atomizers of the graphite tube type, and flame- or electrically-heated quartz tube (QT) atomi~ers.l+~,~ However, despite the diversity of detector designs studied, optimum atomization cell dimensions and operating conditions have not yet been identified, that may, at least partly, be attributed to the fact that the theoretical aspects of gas chromatographic detection by AAS have not been considered.Here, expresssions are derived to describe the relative change in detector sensitivity, chromatographic efficiency and resolution depending on experimental parameters of atomizer dimensions and operating conditions. A GC-AAS detector response model is then developed by combining these expressions with the theoretical relationships proposed by L'vov and co-workers1*,11 that are used to calculate charac- teristic mass values in conventional ETAAS. The reliability of the model is confirmed by experiment and by using literature data,12-14 and recommendations for optimum GC-AAS performance are given. Theory Assumptions In deriving a model to describe the atomic absorption detector performance characteristics, the following assumptions have been made: (i) all organometallic species injected into the gas chromato- graph are quantitatively eluted, without decomposition or transformation on the column, and transferred to the detector cell (ET or QT atomizer); (ii) on entering the detector cell all solutes are completely and instantaneously atomized, and uniformly distributed over the atomizer cross-section; (iii) the shape of the chromatographic peak entering the detector cell is described by the Gaussian distribution model15 (see Fig.l), the standard deviation of which is denoted by (sG(s); (iv) as the solute band passes through the atomizer, the peak shape is altered by the time constant of the detector cell, TR (s),*6 also known as the residence time of analyte atoms in the detector -3 0 3 TirnelaG Fig.1 Gaussian distribution of the solute concentration eluting from a gas chromatographic system assumed in the detector response model. The width of the Peak at half height is given as 00.5. Broken lines indicate the integration limits for the area under the curve for a detector residence time 'l;R = 2oC, where oc is the standard deviation of the peak496 ANALYST, MAY 1993, VOL. 118 I tR Time- Fig. 2 Ex onentially modified Gaussian peak shape model. The full width at ll% of the Peak maximum, O O . ~ , is equal to A + B. The asymmetry factor is given by the ratio B/A cell.loJ1 Removal of these atoms from the detector cell is described by an exponential loss function in terms of tR.Thus the shape of the chromatographic peak following the AAS detector is described by an exponentially modified Gaussian (EMG) function1622 (see Fig. 2); ( v ) the spectrometer readout system does not distort the recorded signal shape. This requires that the time constant of the readout system is less than 10% of the temporal half-width of any detected peak,16 a requirement that is readily fulfilled by all modern AAS instruments; (vi) the gas flow rate from the gas chromatograph entering the detector cell is constant through- out the chromatographic run. See point (vi) in the section on 'limitations of the model' for further discussion of this assumption; (vii) the shape of the absorption line profile is the same as that obtained for the analyte in an argon atmosphere, irrespective of the vapour composition.While this assumption introduces some errors in the 'detector response model' discussed below, it is convenient because most of the relevant spectroscopic parameters are only available for the binary analyte-argon systems.10Jl Calculation of the Standard Deviation of the Gaussian Component of the Peak, oG For a Gaussian peak, the standard deviation is simply related to the full width at half maximum, OJ~.~(S)*~ OG = 00.5 (8 In 2)-0.5 (1) However, for any non-Gaussian profile, this relationship is invalid.lG21 In the case of skewed peak shapes such as described by the EMG function,the total variance or second statistical moment of the peak M2 (s2), is given by the sum Calculation of M2 is described elsewhere,18 and of tR in the following section.Unfortunately, statistical moment calcula- tions are severely impaired by baseline noise in experimental chromatograms,l8720 limiting the applicability of eqn. (2) in deriving 06. Foley and Dorsey20 recommend the use of the following empirical equation to calculate 06 from the chro- matographic peak, which is derived from measurements made on computer generated data f@2 = OG2 + ZR2 (2) 00.1 t3.27 (BIA) + 1.21 OG = (3) where 00.1 is the full width at 10% of the peak maximum and B/A is an asymmetry factor illustrated in Fig. 2. Provided that the parameters and B/A can be determined with sufficient accuracy, the maximum error in oG calculated from eqn. (3) is about 2%. Calculation of the Atomic Absorption Detector Time Constant, ZR When coupled to a gas chromatograph, there are two mechanisms leading to the loss of vapours from the atomiza- tion cell of the atomic absorption detector.The first is convection resulting from the carrier gas flow through the column and transferring the solute into the atomization cell. The time constant for convective analyte losses, zc(s), is related to the ratio of the cell volume, V (cm3), and the gas flow rate through the detector, F (cm3 s-1). Expansion of the gas passing from the interface region at Tint (K) to the atomizer where a vapour phase temperature of Tgas (K) is reached enhances the effective flow rate, thus23 (4) The proportionality constant, K , is a function of the flow pattern through the detector and was assumed to be unity here, although this is strictly only valid under laminar flow conditions at low gas flow rates.Diffusion is the second loss mechanism. The diffusion coefficient, D (cm2 s-I), for a binary gas mixture is tempera- ture dependent10.11 with Do (cm2 s-1) the value at standard temperature and pressure and n the dimensionless gas combination factor having values in the range 1.6 to 2.11 For more complex gas mixtures, the binary diffusion coefficients, Di, of the analyte atoms in each of the j major vapour phase components can be summed as follows24 i xi - 1 D = ( iZ16) (6) where xi is the fraction of component i in the mixture. Theoretical values of Do and n are available for most of the atomic species commonly determined by atomic absorp- tion.11.25 The time constant for diffusional loss, tD (s), is given by the expression23.25 l(cm) being the depth of the absorbing layer, tacitly assumed to be equivalent to the length of the atomization cell.Equations (4) and (7) may be combined to yield the overall atomic residence time in the detector volume, TR (s)23 ZR = ( l / t D + l/tc)-l (8) Evaluation of Detector Response In order to evaluate the performance of a GC-AAS detector with respect to sensitivity and chromatographic efficiency and resolution, the terms 06 and tR must be known. The Gaussian standard deviation component may be calculated from eqn. (3) following experimental measurements, and tR is evaluated using eqns. (4)-(8) as appropriate. Relative values for the sensitivity, efficiency and resolution for a GC-AAS system can all be expressed in terms of CJG and 'cR, as discussed below.It is well-known that the sensitivity in AAS is dependent on zR (among other factors)23,25 and hence the flow rate through the detector..Considering the shape of a Gaussian chromato- graphic peak with a standard deviation of 06 (see Fig. 1) entering the detector, the relative sensitivity, Srel, will be a function of the ratio z&G. With increasing values of this ratio the fraction of the analyte contained in the detector volume at the peak, and hence Srel, will approach unity. Let X be a point on the curve shown in Fig. 1, which represents the normal distribution standardized by dividing by UG and thus XEN(0,1).26 Srel is given by finding the probability that Xlies between the limits - tR/(20G) and + t,/(20G), i.e.,ANALYST, MAY 1993, VOL. 118 497 Srel = P [- tR/(20G) < x < + TR/(hG)] (9) Srel can then be obtained from standard statistical tables of the normal distribution.26 For the example shown in Fig.1, TR/DG = 2, and the area under the curve for P(- 1 < X < + 1) is equal to 0.6826. Thus Srel for a detector yielding a tR value of 2oG will be = 68% of that of a detector having an infinite analyte residence time. One measure of the efficiency of a specific chromatographic system is the effective theoretical plate number, expressed here in terms appropriate for an EMG peak shape’s-20 where tM and tR are the retention times (s) of a non-retained sample component and the analyte species of interest, respectively, and tR’ is the adjusted residence time.l5 The relative efficiency, Nrel, compared to an ideal Gaussian peak is given by This expression is identical to that for the ‘relative system efficiency’ proposed by Foley and Dorsey20 as a chromato- graphic figure of merit for characterizing both ideal (Gaus- sian) and skewed peaks.Resolution may be defined as15 0.5 At R= (12) ( oG2 + tR2) + ( oG2 + TR2)20’s where At (s) is the retention time difference between two adjacent EMG peaks (subscripts 1 and 2) of variance c~G2 + tR2. Assuming that the variances of the two peaks are equal, the resolution relative to that for a pair of Gaussian peaks may be expressed as Note that relationships (1 1) and (13) are discussed in the first section under ‘Results and Discussion’ (and illustrated in Fig.4) - Detector Response Model In ETAAS, the characteristic mass is frequently used as a measure of sensitivity.27 The characteristic mass is defined as the absolute mass of analyte yielding a peak height absorbance of 0.0044 (m’,, pg) or, more commonly, a peak area of 0.0044 s (m’,, pg). Expressions have been derived enabling the calculation of theoretical characteristic masses on the basis of fundamental physical and spectroscopic parameters. 10~1,28729 Comparisons of experimental and theoretical m’, values have demonstrated good agreement , generally confirming the validity of the model. 10,11730 However, the correlation between experimental and theoretical m’, values has been less than satisfactory. This discrepancy can be attributed to the fact that the model for calculating mlP assumes that the entire analyte mass is present in the vapour phase in the atomizer at the peak, a condition which cannot be readily fulfilled by most ETAAS instruments.10,11323328729 In GC-AAS, the analyte is introduced in gaseous form as a solute band with an assumed Gaussian distribution. The fraction of analyte present in the atomizer volume at the peak is therefore a function of tR/oG, the ratio of the removal and supply functions, and the theoretical value of mp for a GC-AAS detector can be evaluated mp = m’df(tR/oG) = m’,/Srel (14) with m’, representing the peak height characteristic mass value calculated for conventional ETAAS conditions on the stipulation that all atoms in the sample are simultaneously present in the atomizer at the peak absorbance.23.28.29 Several models have been used to describe the development of transient absorbance signals in ETAAS when the analyte is vaporized from a gradually heated graphite tube atomizer surface ( e .g . , refs. 23,25 and 31). In this case the temperature- dependent processes of both the rates of atom formation (supply) and dissipation must be accounted for, resulting in a complex convolution integra1.23.31 The situation in GC-AAS is considerably simpler, as the atomizer is operated at constant temperature and the rates of atom supply and removal are well-defined. Calculation of the peak area characteristic mass using the model of L’vov and co-workers10.11 assumes that diffusion is the only analyte loss mechanism and hence the residence time of atoms in the detector volume, z ’ ~ , is as described by eqn.(7). In GC-AAS, the eluent gas from the GC flows continuously through the detector, in effect reducing the residence time of analyte atoms in accordance with eqn. (8). Therefore the presence of a convective flow will increase the characteristic mass for the GC-AAS detector m, 1 m’, T’DI‘cR (15) where m’, and t l D are the theoretical values calculated for the atomization cell vapour phase temperature and dimensions, but assuming gas stop conditions and a pure Ar atmosphere for which data on the diffusion coefficient are available.11 Eqn. (15) shows that m, is directly and inversely related to the detector time constant. In other words the area of a GC-AAS peak increases in direct proportion to tR.Experimental Literature Data Sets To test the applicability of the GC-AAS detector response model, several data sets from the literaturel2-14 have been analysed, with the relevant experimental conditions being summarized in Table 1. Other than providing the necessary experimental details, the references from which the data sets shown in Table 1 were taken fulfilled two further criteria. First, chromatograms were presented allowing peak width (w) values to be estimated. Second, detector sensitivity data were provided as peak height responses per ng of analyte, S, (ng-1). In two cases,12.13 peak height sensitivities were expressed in terms of the signal observed on a chart recorder (mm ng-I), these values being recalculated using a conversion factor reported in ref.13 to units of peak absorbance per ng. As no integrated absorbance data were given, the corre- sponding sensitivities, S, (s ng-I), were estimated as22 S, = 0.753 S, ~00.25 (16) where is the full width (s) at 25% peak height. Foley22 has shown that this empirical equation yields accurate area values providing that the peak shape is of Gaussian or exponentially modified Gaussian form. Sufficiently accurate measurements on chromatograms found in the literature could not be made for calculating UG according to eqn. (3). The following approximate relationship was used instead This one experimental parameter LO^,^) expression was derived from measurements of w0.5 made on simulated chromatographic peaks generated at known values of oG and tR (see Fig.3), using the computer programme for evaluating the EMG function described by Foley and Dorsey.21 The maximum error in calculating oG using eqn. (17) is estimated to be 10%. Use of eqn. (3) is preferable for estimating oG as it provides more accurate results20 than eqn. (17). However, eqn. (17) is useful when the chromatograms are not recorded with (sG = [c00.52/(8 In 2) - TR1.587]o’5 (17)498 ANALYST, MAY 1993, VOL. 118 Table 1 Experimental data on the GC-AAS systems used to test the detector response model Element Wavelengt h/nm Interface temperature/'C Camer gas flow rate/cm3 min-1 Auxiliary gas flow rate/cm3 min-1 Atomization cell Dimensions (i.d. X length)/cm Set ternperature/'C Ref. 12 Pb 283.3 200 110 (Ar) Not used Perkin-Elmer HGA 74 0.59 x 2.8 (graphite tube) 2000 Ref.13 Pb 283.3 140 40 ( Ar) Perkin-Elmer MHS-10 accessory 1.0 X 16.5 (quartz tube) 950 110 0 3 2 ) Ref. 14 Hg 100 (N2) 253.6 Ambient Not used Not specified 0.7 x 18.0 (quartz tube) Ambient sufficiently high resolution for reliable measurements of the necessary parameters, and B/A, required by eqn. (3). Instrumentation To test the effect of the dimensions of the detector cell on the signal, experiments were made using both a graphite tube atomizer (ETAAS system) and a QT atomizer interfaced to a gas chromatograph. A Varian Model 3300 GC oven (Varian AB, Solna, Sweden) equipped with a fused-silica, bonded phase, megabore capillary column (15 m x 0.53 mm i.d., 1.5 pm non-polar DB-1 coating; J & W Scientific, Rancho Cordova, CA, USA) and a silanized, splithplitless injection liner was used.The column outlet was attached to a 50 cm length of deactivated, fused-silica, transfer line (0.25 mm i.d. , J & W Scientific) using a quartz, press-fit connector (J & W Scientific). The transfer line passed out of the left wall of the GC oven and through a heated interface zone maintained at 180°C. A Swagelok T-union fitted with graphite ferrules was used to connect the transfer line to the atomizer. Auxiliary gas was introduced through the side arm of the T-union. For the GC-ETAAS system, a laboratory-constructed graphite tube atomizer was employed,30 incorporating an integrated contact (IC) cuvette, 25 mm long and of 5 mm square cross-section (equivalent radius 2.82 mm) manufac- tured of RWO grade, high-density graphite, and coated with pyrolytic graphite (Ringsdorff-Werke, Bonn, Germany).An uncoated graphite tube (RWO, 4 cm long, 3 mm o.d., 1 mm i.d., Ringsdorff-Werke) was fixed in the enlarged injection hole of the IC cuvette, the other end being connected to the Swagelok T-union. The transfer line from the GC terminated at the upper inside wall of the IC cuvette. Argon (500 cm3 min-1) was used to purge the atomizer housing. For the GC-quartz tube AAS system, a quartz T-tube (15.0 cm x 6.9 mm i d . , Wicklunds Glasinstrument, Stockholm, Sweden) enclosed in an electrically-heated oven32 provided with a NiCr-Ni thermoelement connected to a thermoregula- tor (Staticor Model SYC 2, Corei, France) for temperature control was used. The stem of the quartz tube was connected to the Swagelok T-union, and the transfer line from the GC terminated in the centre of the tube cross-section.The atomizers were installed in a spectrometer system based on a Varian AA-6 monochromator complete with an H2 lamp (Varian) for background correction.30 A Perkin-Elmer (Uberlingen, Germany) Pb Intensitron hollow cathode lamp operating at a current of 5 mA was used as the light source. Data acquisition and processing were facilitated using soft- ware obtained from B. Radziuk (Perkin-Elmer). Procedure Solutions containing tetramethyllead and tetraethyllead in hexane ('distilled-in-glass' quality; Burdick & Jackson, Mus- keyon, MI, USA) were prepared from standards donated by Y. Thomassen (National Institute of Occupational Health, Oslo, Norway).Sample aliquots of 1.0-1.5 mm2 were injected into the GC system at an inlet temperature of 150 "C. The GC oven was initially maintained at 50°C for 1 min and then temperature programmed to 150°C at a rate of 30°C min-1. Other relevant parameters are given in Table 2. In the GC-ETAAS system, an auxiliary gas flow of argon was introduced through the Swagelok T-union to prevent the vapours eluting from the capillary transfer line from diffusing back into the graphite interface region. In the GC-quartz tube AAS system, hydrogen was used as the auxiliary gas, both to prevent back diffusion into the stem of the quartz T-tube and to aid in the atomization of the organolead species.7 Gas phase temperatures in the GC-ETAAS system were determined spectroscopically using the two-line (283.3 nm and 360.8 nm lead lines) atomic absorption method,33 and those for the GC-quartz tube AAS system were estimated on the basis of data reported in refs.34 and 35. Calculations Calculations of theoretical characteristic masses, mlP and mlo, and T ' ~ for conventional ETAAS conditions were made using the computer program CHMASS29 (Eos & Temis, University of Ume5, Sweden). The necessary experimental data required by the program (atomizer dimensions and gas phase tempera- tures) were taken from the references cited in Table 1. Characteristic mass data produced by the program were then converted into values relevant to GC-AAS detectors accord- ing to eqns. (14) and (15). Simulated EMG peaks were generated using a computer program written in BASIC based on that described by Foley and Dorsey.21 Results and Discussion Effect of Detector Time Constant on GC-AAS Performance Characteristics Fig.3 illustrates the effect of the detector time constant, t R , on the shape and size of GC-AAS peaks. These profiles were computer generated using the EMG model for various values of the ratio TR/OG. The program described by Foley and Dorsey21 was modified to include the effect of linearly increasing peak area obtained in GC-AAS as TR/oG increases. Other important effects of increasing detector time constant are the shift in the position of the peak maximum and the greater peak broadening and asymmetry, as discussed else- where. 16-20 Peak broadening has important implications for the chro- matographic performance of the system.16-20 Fig.4 depicts the effect of the ratio -cR/ac on the relative values of detector peak height sensitivity and chromatographic efficiency and resolu- tion. Here it can be seen that GC-AAS detector designs exhibiting high relative sensitivities (high values of the ratio qJaG) severely compromise the observed separation capabili- ties. In this respect, GC-AAS detectors are far from ideal as good detection power can only be obtained at the expense of reduced chromatographic efficiency and resolution. It is also worth noting that when the detector is operating under conditions of low relative sensitivity, variations in tR/oG will result in significant changes in Srel (see Fig. 4). Thus the peak height sensitivity will decrease during an isothermal chromatographic run due to increasing peak widths (oG) forANALYST, MAY 1993, VOL.118 499 > 1.0 4- .- > rn a, .- u .- 5 4- $ 0.5 4- U a, .- w - a, a 0 - 2 0 2 4 6 8 1 0 Ti me/aG Fig. 3 Effect of T~ on the shape and size of GC-AAS peaks generated using a computer program for evaluating the exponentially modified Gaussian function, based on that of Foley and Dorsey.21 Profiles generated at T ~ / ~ . G ratios of A, 1.0; B, 2.0; C, 3.0; D, 4.0; E, 5.0; and F, 6.0. Retention time (tR) of a Gaussian peak is 0 oG 100 0 I I 0 2 4 6 tR/OG Fig. 4 Variation of the relative values of (S) detector peak height sensitivity, (N) chromatographic efficiency and (R) resolution with the ratio tR/oG, calculated using eqns. (9), (11) and (13), respectively Table 2 Experimental data on the GC-AAS systems used to test the detector response model.Analyte element was lead and the wavelength was 283.3 nm Quartz tube ETAAS AAS Interface temperature/’C 180 180 Carrier gas flow rate/cm3 min- 1 18 (He) 18 (He) Atomization cell Graphite tube Quartz tube Auxiliary gas flow rate/cm3 min-1 4 (Ar) 15.5 (H2) Dimension (i.d. x length)/cm 0.564 x 2.5 0.69 x 15.0 Gas phase temperaturePC 980 980 later eluting species. For conditions of high initial Srel, variations in tR/oG will be of lesser importance.23 Evaluation of the Detector Response Model In an attempt to test the validity of the proposed detector response model, characteristic mass data have been calcu- lated, appropriate for the operating conditions used in three applications of GC-AAS found in the literature,*2-14 and compared with experimental data.The results are compiled in Tables 3-5. For the gas chromatographic separation of tetraalkyllead species, de Jonghe et a1.12 utilized a packed column and temperature programming. The set atomization temperature in the ETAAS detector was optimized with respect to the peak height response for the five Pb compounds studied (see Table 3). Although the optimum set temperature was 2O0O0C, the actual vapour phase temperature is likely to be considerably lower as a result of the continuous flow of cool eluent gas (110 cm3 Ar min-1) through the atomizer. Thus a vapour phase temperature of 2000 K in the atomizer was assumed in the detector response model. As can be seen in Table 3, there is good correlation between calculated and experimental charac- teristic mass values for all compounds studied.Average values of the ratio of calculated to experimental characteristic masses for mp and m, are 62 and 60%, respectively. These may be compared to the ratio of m, values found by L’vov11 of 78% in conventional ETAAS. Thus a large part of the discrepancy between characteristic mass values seen in Table 3 is attribu- table to errors in the theoretical mp and m, data calculated using the model and the fundamental physical and spectro- scopic parameters of L’vov and co-workers.10Jf It is unlikely that the differences can be ascribed to poor atomization efficiency ( E ’ ~ ) for Pb species under the conditions used, as calculated values have been shown to be fairly constant over a wide temperature range (1200-2700 K).30 In a continuation of the work on Pb compounds, Chakra- borti et al.13 replaced the ET atomizer with a flame-heated QT atomizer.Again, the gas chromatograph was equipped with a packed column and temperature programming was employed for the separation of the alkyllead species. Although no information relating to the atomization temperature was reported, a compilation of data for similar flame-heated QT atomizers has been made in ref. 34. Furthermore, Welz and Melcher3S have measured spectroscopic temperatures in QT atomizers under conditions comparable to those used by Chakraborti et al.,13 and as a result of these data a vapour phase temperature of 1100 K was assumed in the calculations. Table 4 shows that the model accurately predicts both mp and m, values (on average within 10 and 6%, respectively), although the experimental data are consistently lower than calculated. This finding is in contrast to the results shown in Table 3 and discussed above, but is probably due to uncertainties in some of the parameters used in the calcula- tions [such as the temperature or errors in assumption ( 4 1 .Otherwise it would be expected that the same degree of relative error should be present in both data sets, as almost identical chromatographic conditions and the same Pb species were used in these two studies.12,13 It is encouraging to note that the model satisfactorily predicts the trend of increasing mp values as the relative molecular mass of the tetraalkyllead species increases (see Tables 3 and 4).This effect is due to greater band broadening in the chromatographic system for the later eluting species, and it is obvious that temperature programming is particularly important in GC-AAS to avoid large changes in peak height sensitivity for compounds of differing volatilities (cf. Fig. 4). Considering the results of Tables 3 and 4 it is obvious that the larger absorption volume provided by quartz tube AAS13 compared with ETAAS12 is advantageous with respect to both peak height23 (in accordance with Fig. 4) and peak area sensitivity. The poor relative sensitivity of ETAAS when used as a GC detector is a result of the high carrier-gas flow rate required in applications with a packed column, the small volume of the graphite tube, and the high temperature required for efficient atomization.12 These factors lead to short analyte residence times in the atomizer, and hence low relative sensitivity.As a final example, Table 5 shows the results of the comparison of model and experimental data for the speciation of Hg compounds by GC-quartz tube AAS. Gui-bin et aZ.I4 utilized a twin-packed column system, one stationary phase being used to separate the dialkylmercury species under temperature-programmed conditions, the other for the isothermal separation of the alkylmercury chlorides. After elution from the gas chromatograph, the Hg species were decomposed in a stainless-steel pyrolyser maintained at a temperature of 7OO0C, and the Hg vapour was swept into an unheated QT absorption cell.14 In the calculations, the temperature of the vapour phase in both the interface and the detector was assumed to be 500 K.500 ANALYST, MAY 1993, VOL.118 Table 3 Comparison of calculated and experimental12 characteristic masses for the determination of Pb compounds by GC-ETAAS mdpg molPi3 Compound* W ~ ~ . ~ / S OG/S TR/oG Srel (Y) Sdng-1 SJs ng-1 calc. exp. calc. exp . Me4Pb 4.8 2.034 0.0368 1.47 0.0146 0.0905 201 30 1 34 49 Me3EtPb 6.8 2.885 0.0260 1.04 0.0098 0.0779 285 449 34 57 Me2Et2Pb 7.2 3.055 0.0245 0.98 0.0094 0.0815 302 468 34 54 MeEt3Pb 7.2 3.055 0.0245 0.98 0.0081 0.0703 302 543 34 63 Et4Pb 9.6 4.074 0.0184 0.74 0.0066 0.0752 400 667 34 59 Calculation parameters: Conventional ETAAS- T = 2000 K TD’ =0.318~ mpl = 2.96 pg D(Ar) = 3.083 cm2 s-l m,’ = 7.86 pg GC- E TA A S- TI,, = 473 K T,,, =2000K TD = 0.318 s D(Ar) = 3.083 cm2s-1 TC = 0.098 s TR = 0.075 s * Me = CH3 (methyl); Et = C2H5 (ethyl).Table 4 Comparison of calculated and experimental13 characteristic masses for the determination of Pb compounds by GC-quartz tube AAS mdpg molpg Compound* w0.& a& TR/oG Srel (YO) Sp/ng-l So/s ng-1 calc. exp . calc. exp. Me4Pb 9.53 3.74 0.467 18.5 0.096 0.885 53 46 5.7 5.0 Me3EtPb 11.43 4.60 0.380 15.1 0.079 0.893 65 56 5.7 4.9 Me2Et2Pb 11.71 4.72 0.370 14.7 0.074 0.830 67 60 5.7 5.3 MeEt3Pb 13.07 5.33 0.328 13.0 0.061 0.748 76 72 5.7 5.9 Et,Pb 13.61 5.57 0.313 12.5 0.058 0.742 79 75 5.7 5.9 Calculation parameters: Conventional ETAAS- T = 1100 K TD‘ =33.161 s “P; = 9.89 pg D(Ar) = 1.026cm2s-’ rn, = 0.30 pg GC-quartz tube AAS- Tint = 413 K T,,, = 1100K TD = 16.950 s TC = 1.946 s t R = 1.746s D(Ar + H2) = 2.008 cm2s -1 * Me = CH3 (methyl); Et = C2H5 (ethyl).Table 5 Comparison of calculated and experimental14 characteristic masses for the determination of Hg compounds by GC-quartz tube AAS mplpg mdpg Compound* W ~ . ~ / S OG/S TR/oG Srel (Yo) Sdng-1 Soh “8-1 calc. exp . calc. exp . Et2Hg 12.9 4.58 0.875 34.0 0.0400 0.286 223 110 19 15 MeHgCl 19.7 7.81 0.513 19.5 0.0083 0.091 389 530 19 48 Calculation parameters: Conventional ETA AS- T Me2Hg 12.0 4.11 0.975 38.1 0.0314 0.209 199 140 19 21 EtHgCl 25.7 10.49 0.381 15.3 0.0065 0.093 496 680 19 47 = 500 K TD‘ = 111.01 s mp’ = 75.9 pg D(Ar) = 0.365 cm2 s-l m,’ = 0.68 pg TD = 111.01 s D(N2) = 0.365 cm2 s-1 GC-quartz tube AAS- Tint = 500 K T,,, = 500K TC ~ 4 .1 5 6 ~ tR = 4 . 0 6 s * Me = CH3 (methyl); Et = C2H5 (ethyl). The most obvious deviation between predicted and experimental data in Table 5 occurs in the m, values for the alkylmercury chlorides. This is probably a result of incomplete elution of these species from the column. O’Reilly36 has shown that alkylmercury chlorides have a tendency to undergo partial on-column decomposition and anion-exchange reac- tions, probably at active sites on the solid support. While these problems can be alleviated temporarily by ‘passivation’ with concentrated organic solutions of HgC12,36 this approach could not be used by Qui-bin et aZ.14 because of the Hg specific nature of the GC-quartz-tube AAS detector. It should be noted that O’Reilly36 used an electron capture detector, which responds to the chloride moiety in the solute species, in his studies.Limitations of the Model Some of the comparisons discussed above highlight the need for well-designed experiments under carefully controlled conditions to rigorously evaluate the detector response model. A few such experiments are discussed in the following section. Nevertheless, some limitations of the model can be identified, as follows: (2) the most obvious limitation lies in the calculation of mrP and m’, from theoretical relationships. However, companson of predicted and experimental m’, data for conventional ETAAS shows that the model of L’vov and co-workers*0,11 is accurate to within 30% for the elements of interest (As, Ge, Hg, Pb, Sb, Se and Sn) in such speciation studies.10+11,30 As the model for predicting mlP differs only inANALYST, MAY 1993, VOL.118 501 the exclusion of the atomic residence time,29 then consider- able confidence can be placed in the reliability of the calculations; (ii) although a Gaussian solute distribution entering the detector is assumed, chromatographic peaks might exhibit a more complex shape, for example due to dead volume in the interface region leading to an EMG input function, which means that the predicted peak height may be overestimated. The advantage of the Gaussian distribution function is that the input peak shape may be characterized by a single parameter, oG, which simplifies the model. The true form of the input peak from the gas chromatograph may be experimentally determined using a high auxiliary gas flow rate such that z ~ / o G is less than 0.1.Under such conditions the input peak shape will be largely independent of ~R;16 (iii) an experimentally derived parameter, CJG, is required. It seems unlikely that oG could be predicted with sufficient accuracy from theoretical relationships to avoid this requirement; (iv) as noted above for the alkylmercury chlorides, complete elution, and perhaps also quantitative transfer to the detector, may not always be realized in practice. However, these problems relate more to deficiencies in the instrumental system and conditions used, rather than to the model; (v) it is possible that certain species are not efficiently atomized under certain conditions, as for, e . g . , Se, Sn and Ge in conventional ETAAS systems.30 Indeed attempts made to model the response of three GC-AAS systems to Sn species37-39 resulted in calculated characteristic mass values (both mp and m,) which were between 4 and 10 times lower than the experimen- tal data.This may be interpreted as indicating poor atomiza- tion efficiency for Sn compounds introduced into both ET37 and QT38739 atomizers; (vi) in temperature programmed GC, the assumption of a constant flow rate entering the detector from the gas chromatograph will not be valid for systems providing a constant gas inlet pressure. For such systems the outlet flow rate, F, (cm3 min-I), varies with column temperat- ure, T(K), in the following manner40 where F,,T, represents the outlet gas flow rate at the starting temperature, T,(K), of the temperature program.The exponent 1.7 arises from a linear dependence of gaseous expansion on temperature and a 0.7 power dependence of the gas viscosity on temperature.40 Thus compounds eluting at higher temperatures would be expected to reside in the 100 s - c 50 .- m 0 u c atomizer for longer times due to the reduction in the flow rate, i.e., the time constant for convective loss increases, see eqn. (4). However, at the same time, the lower gas flow will reduce the rate of cooling of the vapour phase in the atomizer. Hence the vapour phase temperature will increase as the flow rate decreases, such that the diffusional loss rate will increase, eqn. (5). (Note that a change in temperature will also alter the absorption coefficient,29 further complicating matters).There- fore the reduction in the outlet flow rate as the column temperature increases causes two opposing effects, which are assumed to cancel each other out. Further work is required to check the validity of this assumption, or to derive suitable expressions for modifying the detector response model to account for these effects. However, the agreement between theory and experiment shown in Tables 3-5 is sufficiently good to use the assumption of a constant flow rate as a first approximation. Model Calculations To illustrate the importance of parameters affecting the detector sensitivity, model calculations have been performed, the results of which are summarized in Fig. 5. For the calculations, the dimensions of the three atomizersl2-14 were used (see Table l), OG was assumed to be 1.0 s (as may be obtained using capillary columns) and the interface temperat- ure 500 K, and D, and n in eqn.(5) were assigned values of 0.3 cm2 s-l and 1.95, respectively. The latter two values were chosen to represent an element having a high diffusion coefficient, with a large temperature dependency.” Such conditions would be observed using a carrier gas such as H2 or He to obtain the highest chromatographic efficiency,ls and H2 as auxiliary gas to improve the atomization efficiency.7J3135 The analyte containment at the peak was calculated using eqns. (4,5,7-9), and the relative peak heights and peak areas were computed as and 50 - 25 Hrel = $( :)2 re1 Are,= (:)2 T’R 80 - 2 40 20 80 140 20 80 140 20 80 140 Gas flow rate/cm3 min-1 Gas flow rate/cm3 min-1 Gas flow rate/cm3 min-1 Fig.5 Model calculations showing the effect of experimental parameters on ( a ) analyte containment in the atomizer volume at the chromato- graphic peak, ( b ) relative peak height defined by eqn. (19), and (c) relative peak area defined by eqn. (20). Calculations assume (76 = 1.0 s and atomization temperatures of: A, 900 K; B, 1200 K; and C, 1500 K. Atomizer dimensions are as given in Table 1 from de Jonghe eta1.12 (HGA, dashed lines at bottom of figures), Chakraborti et ~ 2 1 . ~ 3 (MHS-10, broken lines) and Gui-bin et ~ 2 1 . 1 ~ (quartz tube, solid lines)502 ANALYST, MAY 1993, VOL. 118 respectively, the primes indicating that these values are for the HGA at the highest temperature and gas flow rate (where appropriate).As can be seen in Fig. 5(a), the QT atomizer of the MHS-10 system is most effective in terms of analyte containment at the peak maximum. This is a result of the large volume of the atomizer tube. However, as seen in Fig. 5(b) and (c), the other QT atomizer, designed by Gui-bin et aZ.,14 provides the greatest sensitivity for both peak-height and peak-area measurements under all the conditions considered. The most important design feature of this QT atomizer is its smaller diameter, which results in a more concentrated atomic vapour and hence greater sensitivity,25 despite slightly poorer analyte containment characteristics. These results show that detector sensitivity can be improved by an order of magnitude by substituting a QT atomizer for a graphite tube atomizer.Some improvements can also be made by reducing the operating temperature (provided that the atomization efficiency is not impaired) and the gas flow rate. I 1 20 80 140 Gas flow rate/cm3 min 1 Fig. 6 Model calculations showing the effect of the standard deviation of the Gaussian band entering the atomizer on analyte containment at the peak. Calculations based on the quartz tube dimensions from ref. 14 and atomization temperatures of: A, 900 K; B, 1200 K; C, 1500 K. Solid lines OG = 0.5 s, broken lines OG = 2.0 s and dashed lines at bottom of figure oG = 8.0 s Fig. 6 shows the effect of CIG on analyte containment at the peak for the QT atomizer of Gui-bin et aZ.14 These results demonstrate the importance of chromatographic systems providing narrow Gaussian peaks for detector sensitivity. Values of (JG less than 1.0 s can be achieved using capillary columns,15 whereas packed columns are less efficient with (JG between about 2.0 and 10.0 s for the examples in Tables 3-5.Thus, order of magnitude improvements in detector sensitiv- ity may be readily achieved by using capillary column GC-AAS systems, as indicated in Fig. 6. In combination, capillary columns and QT atomizers may therefore provide improvements in sensitivity of as much as 100 times compared with packed column-graphite tube atomizer equipped GC- AAS instruments. Indeed, considering the experimental peak height sensitivities ( Sp values) for tetraalkyllead species in Tables 3 and 6, it is seen that improvements of about 50 times may be realized in practise by substituting an open tubular column GC-quartz tube AAS system (Table 6, this work) for 0.2 0.15 0.1 0.05 0, g o 4 80 120 160 200 240 280 n 0.3 a 0.25 0.2 0.15 0.1 0.05 0 I 1 I 1 I ~ ~~~~ 80 120 160 200 240 280 Ti me/s Fig.7 Chromatograms obtained using (a) GC-ETAAS and (b) GC-quartz tube AAS systems described in Table 2. For (a), 1.0 mm3 injection of 3 mg dm-3 tetramethyllead (first peak) and 3.7 mg dm-3 tetraethyllead in hexane. For (b), 1.3 mm3 injection of 0.33 mg dm-3 tetramethyllead and 0.4 mg dm-3 tetraethyllead in hexane Table 6 Comparison of calculated and experimental characteristic masses for the determination of Pb compounds by GC-ETAAS and GC-quartz tube AAS using the same GC system* Compoundt aGYs t&C; Srel (%) Sp/ng-l S,/s ng-1 calc.exp . calc. exp. Me4Pb 0.842 0.226 9.00 0.069 0.233 32.1 63.8 15.2 18.9 Et,Pb 0.853 0.223 8.88 0.059 0.230 32.5 75.0 15.2 19.2 Calculation parameters: Conventional ETAAS- T = 1250K (2.875) (84.94) (0.627) (2.767) (5.10) (7.02) (1.79) (1.59) (2.838) (84.40) (0.584) (2.933) (5.13) (7.53) (1.79) (1 S O ) to’ = 2.89 pg (4.33 pg) D(Ar) = 1.298cm2s-1 m,’ = 4.81 pg (0.20 pg) = 1250K z, = 0.274s(7.221 s) D(He + Ar) = 2.856cm2s-1 z, [D(He + H2) = 3.895 cm2s-I] ZR = 0.190s (2.421 s) = 0.602 s (21.662 s) m,,‘ GC-ETAAS (GC-quartz tube AAS)- T,,, = 453 K Tgas = 0.618 s (3.641 s) * Values in parentheses are for the GC-quartz tube AAS system. t Me = CH3 (methyl); Et = C2H5 (ethyl). * Calculated according to eqn. (3).20ANALYST, MAY 1993, VOL. 118 the packed column GC-ETAAS system used by de Jonghe et aZ.12 (Table 3).Table 6 summarizes results obtained for the determination of two organolead species, confirming the expected improve- ments in detector sensitivity obtained using a longer atomiza- tion cell (quartz tube instead of graphite tube, Table 2). The experimental conditions were optimized to maximize the peak height responses using both the GC-ETAAS and GC-quartz tube AAS systems. Typical chromatograms are shown in Fig. 7. As seen in Table 6, there is better agreement between calculated and experimental rn, values than for rnp values, particularly for the GC-ETAAS system. This is a result of an additional contribution to peak tailing arising from the interface, which is not accounted for in the model. In the GC-quartz tube AAS system, band broadening in the interface is less important because of the much larger detector time constant (2.42 s compared to 0.19 s for the GC-ETAAS instrument).Otherwise, the improvements shown in Table 6 are similar to the results reported in Tables 3 and 4, which also compare GC-ETAAS and GC-quartz tube AAS systems, and emphasize the benefits conferred by the latter type of coupled instrumental arrangement. Conclusions In constructing a GC-AAS system for the speciation of metal or metalloid containing compounds, two important points should be considered to obtain maximum detector sensitivity. The first concerns the choice of the atomization cell. Larger volume detector cells as provided by QT atomizers, are to be preferred over graphite tube ET atomizers, as they permit better analyte containment at the peak and hence greater sensitivity (Fig.5). In this respect, the quartz tube design used by Gui-bin et aZ.14 appears to be close to the optimum (Table 1 and Fig. 5). Unfortunately, detectors conferring high sensitiv- ity cause major contributions to band broadening (Fig. 3) and chromatographic efficiency and resolution suffers (Fig. 4). When improved resolution is required, it is a simple matter to reduce dispersion in the detector (ie., reduce the analyte residence time, tR) by means of an increased auxiliary gas flow. Thus the greatest operational flexibility will be obtained for QT atomizer-equipped GC-AAS systems. However, an increased total gas flow through the detector cell will reduce the vapour temperature and may thus reduce the atomization efficiency for some analyte species.A disadvantage of the QT atomizer is the limited temperature range, which reduces the number of elements that can be determined compared with the graphite tube ET atomizer. The second important design consideration lies in the selection of the chromatographic column. Open tubular columns provide higher resolution and sharper, more concen- trated bands and may be operated at lower carrier gas flow rates than packed columns,15 and so tR is increased and oG decreased simultaneously. This will improve sensitivity (Fig. 6), which should translate into better detection limits as confirmed by the results of Anderson et aZ.41 Thus capillary columns are to be recommended for GC-AAS applications. Both experimental (Table 6) and literature (Tables 3-5) data generally validate the detector response model, which could be useful in assessing the atomization efficiency (&IA) obtained and optimizing GC-AAS systems.The model could also be extended to other techniques involving gaseous sample introduction, such as hydride generation.35 Alternatively, if is known, the integrity of analyte species transport to the atomizer can be evaluated. This work was financially supported by the Swedish Centre for Environmental Research in UmeA and the Swedish Natural Sciences Research Council. We also thank B. Hutsch (Rings- dorff-Werke, Bonn, Germany) for supplying graphite parts, 503 Dr. B. Radziuk (Perkin-Elmer, Uberlingen, Germany) for modifying the software used, Dr.Y. Thomassen (National Institute of Occupational Health, Oslo, Norway) and M. Johansson (University of Umeti) for technical assistance. 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 31 32 33 34 35 36 References Donard, 0. F. X., and Pinel, R., in Environmental Analysis using Chromatography Interfaced with Atomic Spectroscopy, eds. Harrison, R. M., and Rapsomanikis, S., Ellis Horwood, Chichester, 1989, ch. 7, pp. 189-222. Chau, Y. K . , Analyst, 1992, 117, 571. Morrison, G. M . P., in Trace Element Speciation: Analytical Methods and Problems, ed. Batley, G. E., CRC Press, Boca Raton, 1989, ch. 2, pp. 2541. Ebdon, L., Hill, S., and Ward, R. W., Analyst, 1986,111,1113. Ebdon, L., Ward, R. W., and Leathard, D. A., Analyst, 1982, 107, 129.Nygren, O., J. Anal. At. Spectrom., 1987, 2, 801. Forsyth, D. S., Anal. Chem., 1987, 59, 1742. Batley, G. E., ed., Trace Element Speciation: Analytical Methods and Problems, CRC Press, Boca Raton, 1989. Harrison, R. M., and Rapsomanikis, S., eds., Environmental Analysis using Chromatography Interfaced with Atomic Spectro- scopy, Ellis Horwood, Chichester, 1989. L’vov, B. V., Nikolaev, V. G., Norman, E. A. Polzik, L. K., and Mojica, M., Spectrochim. Acta, Part B , 1986, 41, 1043. L‘vov, B. V., Spectrochim. Acta, Part B , 1990,45,633. de Jonghe, W., Chakraborti, D., and Adams, F., Anal. Chim. Acta, 1980, 115, 89. Chakraborti, D., de Jonghe, W. R. A., van Mol, W. E., van Cleuvenbergen, R. J. A., and Adams, F. C., Anal. Chem., 1984,56, 2692. Gui-bin, J., Zhe-ming, N., Shun-rong, W., and Heng-bin, H., J. Anal. At. Spectrom., 1989,4,315. Lee, M. L., Yang, F. J., and Bartle, K. D., Open Tubular Column Gas Chromatography. Theory and Practice, Wiley, New York, 1984, ch. 2, pp. 1449. McWilliam, I. G., and Bolton, H. C., Anal. Chem., 1969, 41, 1755. Esser, R. J. E., Fresenius’ 2. Anal. Chem., 1968,236, 59. Kirkland, J. J., Yan, W. W., Stoklosa, H. J., and Dilks, C. H., J. Chromatogr. Sci., 1977, 15, 303. Barber, W. E., and Carr, P. W., Anal. Chem., 1981,53, 1939. Foley, J. P., and Dorsey, J. G., Anal. Chem., 1983, 55, 730. Foley, J. P., and Dorsey, J. G., J. Chromatogr. Sci., 1984, 22, 40. Foley, J. P., Anal. Chem., 1987,59, 1984. van den Broek, W. M. G. T., and De Galan, L., Anal. Chem., 1977,49, 2176. Marrero, T. R., and Mason, E. A., J. Phys. Chem., Ref. Data, 1972, 1, 3. L’vov, B. V., Atomic Absorption Spectrochemical Analysis, Hilger , London, 1970. Box, G. E. P., Hunter, W. G., and Hunter, J. S., Statistics for Experimenters, Wiley, New York, 1978, ch. 2, pp. 21-56 and Appendix A, p. 630. Slavin, W., and Carnrick, G. R., Spectrochim. Acta, Part B, 1984, 39,271. Sturgeon, R. E., and Berman, S. S., Anal. Chem., 1983, 55, 190. Berglund, M., and Baxter, D. C., J. Anal. At. Spectrom., 1992, 7, 461. Frech, W., and Baxter, D. C., Spectrochim. Acta, Part B, 1990, 45, 867. Paveri-Fontana, S. L., Tessari, G., and Torsi, G., Anal. Chem., 1974,46, 1032. Cedergren, A., Frech, W., Lundberg, E., and Persson, J.-A., Anal. Chim. Acta, 1981, 128, 1. Siemer, D. D., Lundberg, E., and Frech, W., Appl. Spectrosc., 1984,38, 389. Radojevic, M., in Environmental Analysis using Chromato- graphy Interfaced with Atomic Spectroscopy, eds. Harrison, R. M., and Rapsomanikis, S., Ellis Horwood, Chichester, 1989, ch. 8, pp. 223-257. Welz, B., and Melcher, M., Analyst, 1983, 108, 213. O’Reilly, J. E., J. Chromatogr., 1982, 238, 433.504 ANALYST, MAY 1993, VOL. 118 37 Hodge, V. F., Siedel, S. L., and Goldberg, E. D., Anal. Chem., 1979, 51, 1256. 38 Burns, D. T., Glockling, F., and Harriot, M., Analyst, 1981, 106, 921. 39 Donard, 0. F. X., Rapsomanikis, S., and Weber, J. H., Anal. Chem., 1986,58, 772. Paper 2105272E 40 Sibley, E . M., Eon, C., and Karger, B. L., J. Chromatogr. Sci., Received October I , 1992 1973, 11, 309. Accepted January 27, 1993 41 Anderson, K. Nilsson, C.-A., and Nygren, O., Scand. J. Work Environ. Health, 1984, 10, 51.

ISSN:0003-2654    

DOI:10.1039/AN9931800495

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 505 Determination of Dissolved Copper, Nickel and Cadmium in Natural Waters by High-performance Liquid Chromatography Sean Comber Water Research Centre, Medmenham, P.O. Box 16, Marlow, Buckinghamshire, UK SL7 2HD A novel technique has been developed t o determine dissolved copper, nickel and cadmium in natural waters, and in particular estuarine and sea-waters, by using high-performance liquid chromatography (HPLC). The main performance criteria were that the method was sensitive, reproducible and interference free. This has been achieved by developing a system whereby the trace metals were initially preconcentrated on a bonded-silica (C,) guard column coated with an ammonium pyrrolidin-I-yldithioformate (ammonium pyrrol id ine dit h ioca r ba mate)-cety ltri met h y la m mon iu m brom ide ion pair.This selectively removed the trace metals from the sample matrix, thereby excluding potential interferents. After preconcentration the metal-dithiocarbamate complexes were flushed with a modified organic mobile phase from the pre-column onto a C18 reversed-phase analytical column for separation. The complexes were detected by means of a variable-wavelength ultraviolet detector. Performance data were obtained, yielding limits of detection of better than 0.5 pg 1-1 for copper, cadmium and nickel, for 10 ml sample volumes. The precision was better than 5% relative standard deviation for concentrations in the range 0.510 pg 1-1. Once optimized the technique was evaluated by analysing estuarine samples with a variety of salinities and comparing the data with those obtained by an accepted reliable method: chelation-solvent extraction-atomic absorption spectrometry.Results showed that there was no significant difference in the data obtained by the two methods. A single calibration graph could be used for samples of all salinities by using the HPLC method. Keywords : High -performance liquid chroma tog rap h y; trace metals; natural waters; dith ioca rbama tes; preconcentration The use of liquid chromatography in the analysis for trace metals is a rapidly expanding subject mainly because of its sensitivity, low cost and multi-element detection capacity. It is also one of the few techniques that offers the possibility of unattended operation. Over the last decade many different systems have been developed, implementing a variety of procedures.Methods have involved, for example, the separa- tion of metal chelates by chelating organic acids, with post-column derivatization and spectrophotometric detection with use of dithizone,l 4-(2-pyridylazo)resorcinol (PAR)2,3 and 8-hydroxyquinoline-5-sulfonate .4 Another system that has received a great deal of attention is the formation of metal-dithiocarbamate complexes, with separation by reversed-phase high-performance liquid chromatography (HPLC) on columns and use of organic-based mobile phases.5-9 The dithiocarbamate can be incorporated in the mobile phase, and the chelates formed ‘on-column’, thereby dispensing with the need for post-column derivatization. Detection is carried out by monitoring the ultraviolet (UV) absorption at wavelengths typically between 254 and 300 nm.Samples are generally introduced via an injection loop of up to 100 pl, allowing detection limits in the region of 50 pg 1-1 to be achieved for trace metals. Such a technique has been successfully used for determining trace metals in effluent from a plating works.778 Trace metals are present in natural waters at low pgl-1 concentrations, therefore necessitating preconcentration before analysis by liquid chromatography. There are few methods currently available for trace-level enrichment. Examples have included enrichment on a cation-exchange pre-c0lumn~2.3 and adsorption of a dithiocarbamate ion pair onto a cl8 pre-column that chelated the metals from a sample. 10 Metal-dithiocarbamate complexes can also be pre-formed ‘off-line’, then concentrated on a CIS pre-column, before elution onto the analytical column.11 None of these techniques, however, has been applied to sea-water analysis, mainly because of severe interference from the matrix, which contains g 1-1 amounts of calcium, sodium and magnesium.Currently, the only successful methods, based on liquid chromatography, for the determination of trace metals in saline waters involve preconcentration on a chelating pre- column (typically containing Chelex-100, an iminodiacetate- bonded resin), followed by washing with ammonium acetate solution to remove any residual major ions. The metals are then eluted from the resin with dilute nitric acid and are collected on a strong cation-exchange column, which is flushed with ammonium nitrate solution, before separation on the analytical column with use of a pyridinedicarboxylic acid (PDCA) as eluent.Detection is by post-column derivatization with PAR, with the complexes being monitored at 520 nm.12.13 Although the method has been successfully used for the analysis of estuarine and sea-waters, the lack of a trace-metal specific preconcentration medium that also allows elution of the metals, without the need for acid, makes the three-column system rather elaborate, hence requiring a large amount of sample manipulation, leading to high blanks and poor sample throughput. Another drawback is that sensitivity is poor for cadmium and lead owing to competition between the PDCA eluent and the PAR spectrophotometric reagent for complexation of the elements.As a result, the objective of this work was to develop a simple preconcentration system that would allow the deter- mination of dissolved copper, nickel and cadmium in natural waters, including complex matrices such as sea-water. The method would have to be simple and inexpensive and allow determination of the metals at sub-ppb levels. Experimental Equipment and Reagents The liquid chromatography system consisted of the following components: a liquid chromatograph [Varian (Palo Alto, CA, USA) 9010 Star System tertiary solvent delivery system]; an analytical column [Zorbax ODs, c18 column, 5 pm silica (25 cm X 4.6 mm i.d.) (Thames Chromatography, Maiden- head, Berkshire, UK); or Spherisorb ODS2, C18, 5 pm silica (25 cm X 4.6 mm i d .) (Thames Chromatography)]; a precon- centrating column (2.5 cm x 3 mm i.d. column hand-packed with C2 bonded silica [from an Alltech (Arlington Heights, IL,506 .) Sample in cu 1 ANALYST, MAY 1993, VOL. 118 a) 1 I Loop or preconcentrator column Waste Fig. 1 Schematic diagram of the HPLC system Table 1 System parameters for liquid chromatography Sample volume 0.5-15 ml Preconcentration-pump flow rate 1.5 ml min-1 Mobile-phase flow rate 1.0 ml min-1 Detector sensitivity 0.01-0.05 a.u.f.s. Detector wavelength 260 nm Cd; 430 nm Cu; 330 nm Ni Cd b) USA) Maxi-clean cartridge]} ; a preconcentrating pump [Dionex QIC pump (Sunnyvale, CA, USA)]; and a detector (Varian 9050 Star System programmable UV/visible detec- tor). A diagrammatic representation of the HPLC system is shown in Fig.1, with typical running parameters being displayed in Table 1. All glassware was soaked in 5% v/v nitric acid for at least 24 h and washed with de-ionized water before use. Reagents were of analytical-reagent grade or better. Metal standards were prepared by dilution, with de-ionized water, of Merck (Darmstadt, Germany) SpectrosoL (loo0 ppm) stock solu- tions to a working range. The mobile phase consisted of: acetonitrile (68%); buffer (32%); plus 10 mmol 1-1 cetyltrimethylammonium bromide [cetrimide (CTAB) (Merck)] for the Zorbax column or plus 15 mmol l-1 CTAB for the Spherisorb column. Acetonitrile and methanol were of HPLC grade (Rath- burn, Walkerburn, Peeblesshire, UK), and the buffer was prepared from 50 mmol l-1 sodium acetate and 20 mmol l-1 sodium nitrate (Merck), adjusted to pH 6.0 with dilute hydrochloric acid.The dithiocarbamate-cetrimide ion pair was prepared by dissolving sufficient ammonium pyrrolidin-l- yldithioformate [ammonium pyrrolidine dithiocarbamate (APDC)] (Sigma, St. Louis, MO, USA) and CTAB in de-ionized water to yield a solution that was 2 mmol 1-1 in each compound. This solution was mixed 80 + 20 with the mobile phase buffer and was stable for over 1 week, provided that it was kept in the dark when not in use. The enrichment column was custom built and was dry- packed with C2-bonded silica obtained from an Alltech Maxi-Clean cartridge. The large particle diameter prevented back-pressures building up in the system. Preconcentration Procedure After pumping the mobile phase through the column for approximately 30 min to achieve equilibration, 1 ml of the ion-pair reagent (80% of 2mmoll-1 each of APDC and CTAB plus 20% of mobile phase buffer) was pumped onto the enrichment column, which was filled with C2-bonded silica.This was followed by the sample (0.5-15 ml, pH 4.5-5.5) and 2 ml of de-ionized water. The valve was then switched to allow back-flushing of the metal chelates onto the CI8 analytical 0 10 20 Time/min Fig. 2 Sea-water (5 ml) (a) before and (b) after the addition of 50 ng of Ni and Cu, and 30 ng of Cd, using a Spherisorb ODS2 analytical column column by the mobile phase. The chelates were detected (peak-area mode) at their optimum wavelengths. By this method, excellent sensitivity and separation of the metals were achieved (Fig.2), with use of either the Zorbax or Spherisorb ODS analytical column. Before the start of the analysis the analytical column was flushed with mobile phase overnight to ensure column stabilization. Once conditioned, the analytical columns whether Spherisorb or Zorbax , perfor- med similarly. Owing to matrix elimination in the preconcen- tration stage, the analytical column should provide an undiminished life-span. It is possible, therefore , that the pre-column could become contaminated and suffer a loss of performance. We have, however, yet to experience this, but, as a precaution, the medium is changed every few hundred samples, depending on the matrix to be analysed. The analysis of three standards thereafter conditions the system. The preserved standards and samples (preserved by adding concentrated nitric acid to a concentration of 2 ml 1-1) were adjusted to the correct pH by first neutralizing the acid with an appropriate volume of concentrated ammonia solution and then adding 10% v/v of the mobile-phase buffer to allow for any variations in the sample matrix.Results Effect of Salinity on Sensitivity In order to analyse estuarine water samples routinely it is essential that only one calibration is required for samples of all salinities. In order to assess the effect of salinity on the sensitivity of the system, 5 ml samples of estuarine waters (collected from the Humber estuary, UK) of varying salinity were analysed in triplicate before and after spiking with 10 pg 1-1 each of copper, cadmium and nickel.The relative response of each metal (expressed as a percentage) with respect to that of de-ionized water (peak-area measurement) is displayed in Table 2. It can be seen that there is no significant variation in sensitivity for any of the metals over aANALYST, MAY 1993, VOL. 118 507 range of salinities from river water to coastal water of approximately 35%0 (Fig. 3). the standard deviation was calculated by analysing replicates of water spiked with very low levels of cadmium to produce a signal. Results show that, for a 10 ml preconcentration, limits of detection of better than 0.5 yg 1-1 can be achieved for cadmium, copper and nickel, with a relative standard devia- tion of less than 5%. Method Comparison Test In order to validate the HPLC method, two experiments were carried out; initially two commercial saline Aquacheck (Water were Effect of Sample Volume on Sensitivity An experiment was undertaken to assess the sensitivity of the system for copper, nickel and cadmium when various volumes of water were preconcentrated.Samples of between 1 and 20 ml of clean sea-water were spiked with 50 ng of the metals and then analysed in triplicate. The sensitivities, expressed as responses relative to that recorded for a 1 ml preconcentra- tion, are displayed in Fig. 4. The data show that, for the preconcentration column used, a 10 ml sample volume affords the best compromise between sensitivity and recovery. Greater volumes result in a signifi- cant reduction in trace-metal recovery, particularly for nickel, from the enrichment column.The use of a larger enrichment column could allow the loading of greater sample volumes, Research Centre, -Medmenham, UK) check samples .- CI thereby improving limits of detection. .- Limits of Detection The limit of detection and reproducibility were assessed by performing repeat analyses of 10 ml samples of clean sea- water and de-ionized water blanks, adjusted to pH 4.5 with dilute hydrochloric acid, and of sea-water spikes (10 ml) containing 10 pg 1-1 each of copper, nickel and cadmium. Six replicates were analysed, together with six blanks, and the element were calculated on 4.65 times the standard deviation W results are recorded in Table 3. The limits of detection of each 0 5 10 15 20 Sample volu me/ml of the blank. Because the cadmium blank was undetectable, Fig.4 Variation of sensitivity with increasing sample volume (with respect to a 1 ml sample volume). A, 100% recovery; B, Cu; C, Ni; and D, Cd Table 2 Variation of sensitivity with salinity Table 3 Performance data for the technique Recovery of metals (%) Salinity ("A) Ni SD* Cu SD Cd SD 0 100.0 4.5 100.0 2.2 100.0 2.6 0 107.8 4.0 96.8 4.7 94.4 6.6 0 102.8 6.0 105.1 5.6 89.7 3.4 Ni c u Cd to. 13 0.05 t0.13 0.04 5 .oo 0.04 0.8 0.19 De-ionized water Thames River (UK) Boothferry (Hull, UK) Blacktoft (Humber, Hessle (Humber) Easington (North Sea) Selsey (English Channel) UK) DIW* blank/pg I-' Standard deviatiodpg 1-* Sea-water blank/pg 1-1 Standard deviatiodpg 1-l Sea-water + 50 ng/pg 1-1 Standard deviatiodpg 1-1 RSDt (% ) Limit of detectiordpg 1-1 0.89 0.11 0.97 0.08 5.97 0.18 3.0 0.38 1.08 0.09 1.36 0.07 6.36 0.13 2.0 0.35 9 106.8 4.0 101.5 4.8 98.4 4.6 23.3 96.0 1.5 103.2 2.2 101.3 4.8 35.5 98.6 3.4 102.5 1.7 94.5 2.9 34.4 99.8 4.8 101.5 5.3 96.4 2.8 * DIW = De-ionized water.t RSD = Relative standard deviation. * SD = Standard deviation (n = 5). Ni cu Ni Jl Ni CU -4 L I I 0 10 200 10 20 0 10 20 0 10 20 0 10 200 10 20 Tirnelmin Fig. 3 Samples (5 ml) of (a) and (b) Thames River water; (c) and (d) Humber estuary water ( 9 O h salinity); and (e) and cf) North Sea coastal water with and without spikes of 10 pg 1-1 of Ni, Cu and Cd508 ANALYST, MAY 1993, VOL. 118 Table 4 Comparison of HPLC data with those for Aquacheck saline samples Ni f Cu rt Cd f Distribution 36- Mean of all HPLC datdpg 1-l 3.10 0.25 2.80 0.42 1.45 0.25 Mean of all datdpg 1-1 20.4 1.55 0.90 HPLC datdyg 1-1 21.06 0.50 1.70 0.25 1.05 0.22 laboratories/pg 1-1 3.39 3.17 1.29 Distribution 40- Table 5 Comparison of test data* for HPLC versus solvent extraction- AAS AAS Nickel- Cadogan Canvey Island Tilbury Copper- Cadogan Canvey Island Tilbury Cadmium- Cadogan Canvey Island Tilbury * In yg 1-1.HPLC -I- 11.17 0.77 4.57 0.50 6.58 0.82 7.37 0.94 4.49 0.75 8.14 0.89 0.21 0.05 0.27 0.10 0.30 0.10 In-house 10.11 5.09 7.49 6.48 4.47 7.72 0.20 0.19 0.23 0.00 0.42 0.37 0.60 0.52 0.66 0.025 0.022 0.021 External 10.28 5.26 7.38 6.72 5.83 8.31 0.22 0.21 0.29 analysed and the results were compared with the means obtained from all other participating laboratories (Table 4). A second experiment involved analysing three Thames estuary (UK) samples (Table 5) of different salinities and comparing the results with solvent extraction-atomic absorption spec- trometry (AAS) data (in-house and external).l4 Errors are quoted as confidence intervals (p = 0.05).Excellent agreement was obtained for the analysis of the Aquacheck marine samples. Data for the Thames estuary samples were also good, not only in comparison with those of the in-house solvent extraction-AAS method, but also with those of the external analysis. Discussion The complexity of sea-water matrices has hampered the analysis for trace metals by liquid chromatography.15 Attempts to alleviate the problem , although successful ,12J3 require a three-column system and are, therefore, compli- cated, subject to contamination and insensitive towards lead and cadmium.Of the various reported methods, preconcentration by an APDC ion pair adsorbed onto a solid medium offers the best approach to detect successfully trace metals in complex matrices on a routine basis. To date, however, there has been no method available to determine trace metals in complex matrices such as sea-water. In an attempt to discover a suitable medium that would reproducibly adsorb the APDC ion pair, but show minimal affinity for dissolved organic material, a range of different commercially available sorbent packings was investigated. These included CI8, diol-, phenyl- and C2-bonded silicas and poly(viny1 dibenzene) resins XAD-2 and XAD-4. Of these, C2 was found to afford optimum sensitivity while being selective against organic and inorganic interferents present in natural water samples.Cetrimide was added to the mobile phase to prevent unreacted silanol groups from interacting with the metal-dithiocarbamate complexes and adversely affecting the reproducibility of the method. The addition of the CTAB also resulted in a significant increase in the retention time of the cadmium complex, which moved the peak away from the shoulder of the APDC peak, thereby improving its repeatabil- ity and sensitivity. Cadmium has often been considered a difficult metal to determine, owing to the relative instability of the cadmium-dithiocarbamate complexlIJ6 and to the fact that it appears to be present as the anionic complex Cd(DTC)3-.7 Cetrimide forms a neutral ion pair with this chelate and therefore allows it to be chromatographed in a reproducible manner.It was found that analytical columns of varying carbon coverage could be used, provided that the CTAB concentra- tion was adjusted to optimize copper and cadmium peak resolution. For example, for the Zorbax column with 17% carbon coverage, 10 mmol l-1 CTAB in the mobile phase was sufficient to yield baseline resolution of the peaks. For the Spherisorb column, however, with 12.5% carbon coverage, and therefore less retention power, 15 mmol l-1 was required to yield the same degree of separation. Each new column also had to be conditioned by pumping through the appropriate mobile phase (0.5 ml min-1) overnight to obtain a stable and sensitive response for cadmium in particular.The system revealed no interferences from the sea-water matrix, with authentic estuarine samples of salinities ranging from 0 to 35Y4 yielding no significant variation in recovery. This allows a single calibration to be used with samples of all salinities encountered in a river-estuarine-coastal water environment. For maximum sensitivity, the pH of the samples was adjusted to 4.5, with detection at the optimized wavelengths of 260 nm for cadmium, 330 nm for nickel and 430 nm for copper. Detection limits for a 10 ml sample of sea-water were 0.19 pg 1-1 for cadmium, 0.38 pg 1-1 for nickel and 0.35 Fg 1-1 for copper, which is comparable to those obtained by solvent extraction-AAS, currently the most popular method for saline water analysis. The comparison test between the two techniques revealed no significant differ- ences in the results obtained for samples collected across a range of salinities in the Thames estuary. Although a wide variety of trace metals form complexes with APDC, only cadmium, nickel and copper could be determined by the proposed method.Zinc only forms a weak complex of poor stability. Under suitable conditions (i.e., fine tuning of eluent composition and possibly a change of preconcentration column packing material, to obtain a signal for mercury) , mercury dithiocarbamate complexes elute late in the chromatogram, yielding broad peaks and poor sensitiv- ity. Although it is possible to obtain speciation data for chromium, complexation of chromium(m) and chromium(vr) with dithiocarbamates is slow and requires heating.Lead does form a stable complex, but with the optimized system, it was found to elute on the shoulder of the reagent peak, making the determination imprecise. The addition of APDC to the mobile phase made the peak sensitive and repeatable, with a retention time shorter than that of nickel; its performance is currently being assessed. Cobalt and bismuth also formed dithiocarbamate complexes, but failed to be detected by using the quoted method; under suitable conditions they could also be included in the analytical suite. Although the system is constructed from stainless steel throughout, the blanks are small; indeed they seem to be on a par with those recorded for a non-metal ~ystem.12~13 The blanks were further reduced and sensitivity was improved by the use of titanium frits in the enrichment and analytical columns, as titanium contains copper and nickel impurities at much lower levels than those of the stainless-steel frits.This had two effects: first, the frits in the columns have been shown to contribute over 90% of the stainless-steel contact surface area for the mobile phase,l7 resulting in any excess of APDC reagent leaching nickel and copper from the frits, yielding increased blank signals. Second, nickel and copper have been shown to displace cadmium from the dithiocarbamate chelate _ _ _ _ _ . ~ - .ANALYST, MAY 1993, VOL. 118 509 because they possess greater complexation stability constants; use of titanium therefore results in a more sensitive and repeatable cadmium signal.Conclusions An HPLC-based method has been developed and optimized to allow the detection of copper, nickel and cadmium in natural waters, and in particular saline waters, which have in the past exhibited too much interference for analysis by HPLC. A simple preconcentration procedure was used, involving the adsorption of a dithiocarbamate-CTAB ion pair onto a small trace-enrichment column containing a C2-bonded silica, before loading the sample. This allowed sample volumes of typically 10 ml to be concentrated, leading to excellent detection limits (Cd 0.19 pg 1-1; Ni 0.38 pg 1-1; Cu 0.35 pg 1-I), well within the Environmental Quality Standards set for these elements. Such performance for copper and nickel compared favourably with that obtained by the more labour-intensive solvent-extraction-electrothermal AAS method,18 with no significant differences in data being encountered when estuarine samples were analysed by both techniques.No interference was observed when analysing sea-water over a range of salinities, hence allowing a single calibration to be used for all saline samples. The system displayed several advantages over the existing liquid-chromatography technique developed for the determi- nation of trace metals in saline waters; namely, the simplicity of the system, the enhanced sensitivity for cadmium and the fact that a post-column derivatization step is unnecessary. The fact that complexation takes place ‘on-column’ is advan- tageous in that sample manipulation is greatly reduced, leading to improved precision and lower blank levels.This work was carried out under contract to the National Rivers Authority as part of their research and development programme. The project leader was Dr. Ian Fox of NRA Thames Region. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 References Jones, P., Hobbs, P. J., and Ebdon, L., Anal. Chim. Acta, 1983, 149, 39. Cassidy, R. M., and Elchuk, S., J. Chromatogr. Sci., 1981, 19, 503. Cassidy, R. M., and Elchuk, S., Anal. Chem., 1982, 54, 727. Jones, P., Barron, K., and Ebdon, L., Anal. Proc., 1985, 22, 373. Ge, H., and Wallace, G. G., Anal. Chem., 1988,60, 830. Smith, R. M., and Yankey, L. E., Analyst, 1982,. 107, 744. Bond, A. M., and Wallace, G. G., Anal. Chem., 1984,56,2085. Bond, A. M., and Majewski, T., Anal. Chem., 1989,61, 1494. Timerbaev, A. R., Petrukhin, 0. M., and Zolotov, Yu. A., Fresenius’ Z. Anal. Chem., 1987, 327,87. Irth, H., de Jong, G. J., Brinkman, U. A. Th., and Frei, R.W., Anal. Chem., 1987, 59, 98. King, J. N., and Fritz, J. S., Anal. Chem., 1987, 59, 703. Siriraks, A., and Kingston, H. M., Anal. Chem., 1990,62,1185. Comber, S . , J. Znst. Water Environ. Manage., 1991, 5 , 158. Apte, S. C., and Gum, A. M., Anal. Chim. Acta, 1987, 193, 147. Apte, S. C., and Cowling, S. J., An Evaluation of Zon Chromatography for the Analysis of Trace Metals in Natural Waters, Water Research Technical Report PRU 2051-M, Water Research Centre, Medmenham, 1989. Hutchins, S. R., Haddad, P. R., and Dilli, S., J. Chromatogr., 1982,252, 185. Young, T. S., and Carr, P., Talanta, 1981,28,411. Methods for the Examination of Waters of Associated Materials: The Determination of Twelve Trace Metals in Marine and Other Waters by Voltammetry or AAS, HM Stationery Office, London, 1987, pp. 101-104. Paper 2105774C Received October 29, 1992 Accepted January 15, 1993

ISSN:0003-2654    

DOI:10.1039/AN9931800505

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 511 Ultraviolet Spectrophotometric, Solvent Extraction, Hig h-pedormance Liquid Chromatographic and Polarographic Study of the H+/K+ Adenosine Triphosphatase Inhibitor (SK&F 95601) and Its Acid-catalysed Degradation Products Edward O'Kane, Venkataraman N. Ramachandran, Stephen McClean and W. Franklin Smyth" Department of Applied Physical Sciences, University of Ulster, Coleraine, Co. Derry, UK BT52 ISA pK, values of 3.3, 5.35, 9.0 and 12.0 have been established for the H+/K+ adenosine triphosphatase inhibitor SK&F 95601 (1) with optimum solvent extractability into ethyl acetate in the pH range 7.5-8.5. Following solvent or solid-phase extraction of 1 X 10-6 mol 1-1 1 from urine, a recovery of 92.3% was calculated using quantification by high-performance liquid chromatography (HPLC) with ultraviolet detection at 300 nm.The complex degradation of 1 at pH 2 can be followed by HPLC and differential-pulse polarography, which results in the detection of D1 (the sulfenamide) and D2 (the sulfide) in a 10 min time scale. Differential-pulse polarography was also used in situ to follow the reaction between a thiol (2-mercaptoethanol) and 1 in 1 x 10-2 mol I-' HCI as a simulation of its believed reaction in vivo. Keywords: H+IK+ adenosine triphosphatase inhibitor; benzimidazole sulfoxide; high-performance liquid chromatography; differential-pulse polarograph y; degradation products Cimetidine, discovered by Smith, Kline & French (SK&F) (Philadelphia, PA, USA), was the first of the H2-receptor antagonists that revolutionized the treatment of ulcers.1 Compounds of this class block the action of histamine in stimulating gastric acid secretion.More recently, a new class of compounds exemplified by omeprazole have been shown to be effective in ulcer treatment. These compounds act by irreversible inhibition of the H+/K+ adenosine triphosphatase (ATPase) enzyme, which is part of the 'proton pump' located in the parietal cell of the stomach wa11.2 Omeprazole and other benzimidazole sulfoxides such as SK&F 95601 (1) undergo an acid-catalysed rearrangement in vitro to form sulfenamides (such as 2 in the case of SK&F 95601), which can react with thiols.3-5 It is believed that a similar mechanism occurs in vivo and that reaction of the sulfenamide with a thiol residue on the enzyme results in the irreversible inhibition observed.I M OCH3 (2) This paper is concerned with determination of the pK, valoes of 1 by ultraviolet (UV) spectrophotometry , postulation of the ionic forms of 1 existing in aqueous solutions over the pH range 0-14, followed by a study of the solvent extraction of these ionic forms of 1 into ethyl acetate. This results in the design of an optimum method for solvent extraction of 1 from an aqueous sample. The complex degradation of 1 in acidic media is monitored by UV spectrophotometry, solvent extraction, * To whom correspondence should be addressed. high-performance liquid chromatography (HPLC) and dif- ferential-pulse polarography (DPP) , with some of the products of the degradation being subjected to separation and identifica- tion procedures.The DPP technique was also used in situ to follow the reaction between a thiol (i. e., 2-mercaptoethanol) and 1 as a simulation of its believed reactions in vivo. The work is completed by the development of an analytical method, based on HPLC, for the determination of 1 and its degradation products in a biological matrix such as' urine. A related benzimidazole sulfoxide and its sulfone metabolite have been determined in serum and plasma by HPLC-UV detection.6 Experimental A Perkin-Elmer (Norwalk, CT, USA) Lambda 5 instrument was used in the UV spectrophotometric studies. A 30 pl aliquot of a 5 x 10-3 moll-1 stock solution of 1, prepared in acetonitrile-water (50 + 50), was added to 5 ml of the particular aqueous buffer before recording of the UV spec- trum.The pH of the buffer was measured at room tempera- ture by means of a Philips (Eindhoven, The Netherlands) PW 9410 pH meter and HCI was used to cover the pH range 0-3, acetate buffer the range 3-6, tris(hydroxymethy1)methyl- amine (Tris) buffer the range 6-11 and various concentrations of NaOH were used for pH > 11. Ultraviolet spectrophoto- metric scans were also carried out on a degraded solution of 1 at a selection of pH values. Solvent extraction was carried out between 5 ml of the aqueous buffer containing 1 or a degraded solution of 1 and 5 ml of ethyl acetate. The aqueous layer and the organic layer could then be examined by UV spectrophotometry, using the Perkin- Elmer Lambda 5 instrument, before and after extraction.The degradation of 1 was monitored by HPLC as follows. A 5 x 10-3 moll-1 stock solution of 1 was made up in acetonitrile- water (50 + 50). Thiswasdilutedten timeswith 1 x 10-2 mol 1-1 HCI, and the degradation was thus initiated at pH 2. At certain times in this longer-term kinetic study, i.e., over a 5 d period, 50 p1 were sampled, mixed with 50 p1 of acetonitrile, and 20 p1 were injected onto a CIS Novapak (Waters Millipore, Milford, MA, USA) column for HPLC, with a mobile phase of acetonitrile-water (50 + 50). The operating pressure was 9000 kPa (90 bar) and a flow rate of 5 ml min-1 was used. For the shorter-term kinetic study (i.e., over a 10 min period), at t = 0,lO plofa5 x 10-3 moll-lstocksolutionoflweredispersedin 100 pl of pH 2 solution, and at time t, 100 pl of acetonitrile were5 12 ANALYST, MAY 1993, VOL.118 added to the whole sample before injection of 20 yl onto the CI8 Novapak column. The degradation of 1 was monitored by DPP by means of a Metrohm 646 VA Processor (Metrohm, Herisau, Switzerland). The mercury drop size was 0.40 mm2, the voltage step for the sweep 8 mV and the time step for the measurement interval 500 ms. To a20 ml solution of 1 x 10-2 moll-1 HCI were added 200 yl of a 1 X 10-3 moll-' solution of 1 in methanol. The time from this addition, i.e. , the start of the acid-catalysed hydrolysis to the appearance of the D1 peak at approximately -735 mV, was noted and this was designated the particular reaction time t. Scans were initiated every minute for the first 30 min, then every 30 min for the next 2 h.During the remainder of the 5 d study, scans were initiated every 3-4 h. The reaction of 2-mercaptoethanol with 1 was carried out as follows. To 20 ml of 1 x 10-2 moll-' HCI were added 200 yl of 1 x 10-3 moll-1 2-mercaptoethanol, which was prepared by dissolving 0.08 g of the thiol in 100 ml of methanol. The thiol solution was scanned first, followed by addition of 200 yl of 1 x 10-3 mol 1-1 1 and then recording was carried out of the subsequent reaction every 1.5 rnin for the first 10 rnin and then every 8 rnin for the remainder of the 30 min study. Compound D2 was isolated and identified as follows. Compound 1 (200 mg) was suspended in distilled water (50 ml) in a 500 ml stoppered conical flask, and dilute HCI (pH 2, 225 ml) was added in portions, with vigorous shaking to dissolve the solid material during 40 min.By this time the colour of the reaction mixture had turned to dark brown. The flask was covered with aluminium foil and kept in the dark at room temperature for 5 d. During this period, small samples of the reaction mixture were removed to monitor by HPLC the formation of D2 and other degradation products. After 5 d, the pH of the reaction mixture was brought to 7 by the addition of small amounts of solid Na2C03. A brown solid separated out. The reaction flask was refrigerated for 2 h to maximize the precipitation of the solid material. The brown solid was filtered off, washed with water and dried in a desiccator (yield: 60 mg). Thin-layer chromatographic examination on a reversed- phase (CI8 silica gel) plate, with acetonitrile-water (60 + 40) as the mobile phase, indicated that the brown solid was a mixture with spots at RF = 0.60,0.50,0.4O-O.48,0.36,0.30 and0.25 with the major compound at RF 0.40-0.48.Separation on a preparative scale on the reversed-phase plates with the same mobile phase was then carried out. Elution was carried out twice to effect better separation. Bands at RF (after second run) of 0.85,0.75,0.65-0.70,0.55-0.65,0.40-0.50,0.35 and 0.30 were removed and the organic components were eluted with chloroform-methanol(80 + 20). Each fraction was submitted to HPLC, and the bands at RF = 0.65-0.70 and 0.55-0.65 were found to be the same and to correspond to D2 (20 mg). Compound D2 had a melting-point of 185 "C with decompo- sition.Fast-atom bombardment mass spectrometry of D2 yielded an MH+ peak at mlz 391 with fragment ions at mlz values of 355, 243, 212 and 180. This would suggest a sulfide structure 3 with a relative molecular mass of 390. H (3) Compound 1 was determined in spiked urine as follows. First, 1 was extracted from the biological matrix by either solvent extraction or by use of a Sep-Pak cartridge. Solvent extractionof 300 yl of a 1 x 10-3 moll-1 stock solution of 1 in 3 ml of urine with5 mlof ethyl acetatewascarriedout, followed by washingof the organic layer with 5 ml of water, rotary evaporation of the organic layer to dryness, and dissolution of the residue in 3 ml of acetonitrile-water (50 + 50) for injection and HPLC determina- tion. The same procedure was adopted for solvent extraction of the degraded solution of 1 spiked in urine. As an alternative to solvent extraction of 1 and of the degraded solution of 1 from urine, the use of Sep-Pak cartridges was evaluated as an alternative technique. A Sep-Pak CI8 cartridge was connected to a vacuum pump and conditioned with 2 ml of methanol, followed by 2 ml of sodium acetate buffer, pH 7-8.Urine (3 ml) was spiked with 300 yl of a 1 x 10-3 moll- 1 solution of 1 and the sample was mixed thoroughly. This spiked urine sample was drawn slowly through the cartridge. The cartridge was then washed with 1 ml of sodium acetate buffer solution (pH 7-8) and then dried for 5 rnin under full vacuum. By using 5 ml of ethyl acetate, 1 was eluted and the solvent was rotary- evaporated. The residue was reconstituted in 3 ml of acetonit- rile-water (50 + 50) for injection and HPLC determination.The recovery of 1 was evaluated in the concentration range 1 X 10-5-1 x 10-6 mol 1-1. The same procedure was adopted for Sep-Pak extraction of the degraded solution of 1 from urine. Results and Discussion Determination of pK, Values; Sites of Protonation and Deprotonation, and Solvent Extraction Studies The UV spectrophotometric behaviour of 3 x 10-5 moll-' solutions of 1 was studied in the pH range 0-14. In the pH range 4.5-14, the molecule was stable and three pK, values were determined at pK3 = 5.35, pK4 = 9.0 and pK5 = 12.0 with use of the Henderson-Hasselbalch equation as shown in Table 1. A further pK, value at pK2 = 3.3 was determined at 300 nm by immediate measurement of the absorbance when the stock solution of 1 was added to the appropriate buffer solutions.This is illustrated in Fig. 1. The pK1 value was unobtainable for the parent drug 1 owing to its rapid decomposition at pH d 2. Assignment of these five pK, values to particular locations in the molecular structure of 1 is as follows. (i) pK1 to the morpholino-nitrogen atom. It should be noted that the literature value of 2-chloroaniline at 25 "C is 2.65.7 (ii) pK2 of 3.3 to the pyridine-N atom. It should be noted that the literature pK, value for 3-chloropyridine at 25 "C is 2.84.7 This also cross-correlates with pK, data reported in the paper by Ife et aZ.8 where the pyridine pK, value of the related benzimidazole sulfoxide drug, omeprazole , has a pK, value of 3.98. Omeprazole has two CH3 substituents and one OCH3 substituent in the pyridine ring at positions 3, 5 and 4, Table 1 UV spectrophotometric behaviour of 1 over the pH range &14 and resulting pK, values.(s) = shoulder in spectrum Absorbance for a 3 x 10-5 moll-' Ionic form of 1 L,x/nm solution of 1 Notes * * - - * - * pKi=-* HSA3+ H4A'+ - P K ~ = 3.3+ pK3 = 5.35 H2A+ 297 0.9 pK4 = 9.0t HA 303 0.9 H3A2+ 300 0.75 333(s) 0.53 pK, = 12.06 280(s) 0.74 380(s) 0.83 A- 304 1 .0 * Data unobtainable owing to decomposition of I. + Determined at 300nm immediately after addition of 1 to an * Absorbance measured at 280 nm. 9 Absorbance measured at 304 nm. appropriate buffer solution.ANALYST, MAY 1993, VOL. 118 513 1.10 1.05 0 < 1 .oo 0.95 2.0 4.0 6.0 PH Fig.1 Variation of absorbance at 300 nm with pH for 3 x 10-5 moll-1 1 respectively, with no electron-withdrawing substituents such as C1 in the case of 1, so it is not surprising to find that the latter molecule has a lower pyridine pK, value than that of omeprazole. This pK, value also cross-correlates with polaro- graphic data9 in the form of an Eg (half-wave potential) versus pH plot for a model benzimidazole sulfoxide 4. (4) A change of slope of this plot, corresponding to a pK, value, occurs at pH = pK, = 3.7. {iii) pK3 of 5.35 to the azomethine-N atom in benzimidaz- ole. It should be noted that the literature values for benzimidazole and 2-phenylbenzimidazo1e are 5.53 and 5.23, respectively.' (iv) pK4 of 9.0 to the sulfoxide-0 atom. The sulfoxide-0 atom in chlorpromazine sulfoxide (5) has a reported pK, value of 7.0.10 0 t CI (5) It is assumed, therefore that the sulfoxide group in 5 is a stronger acid than the sulfoxide group in 1 owing to greater resonance delocalization through the two benzene rings in 5 and through the effect of the electron-withdrawing C1 substituent , also in 5.( v ) pKs of 12.0 to loss of a proton from the N-H group of the benzimidazole. Literature values for benzimidazole and 2-phenylbenzimidazole are 13.211 and 11.91,7 respectively, and Johansson and Persson12 have determined a pK, value of 10.0 by titration with 0.01 mol 1-1 NaOH in aqueous 40% ethanol for a related model compound that does not possess the electron- donating methoxy function in the benzimidazole ring system. Table 2 UV spectrophotometric behaviour of degraded solutions of 1 over the pH range 0-14 and resulting pK, values.(s) = shoulder in spectrum I,,, of degraded Absorbance of solutions of l/nm degraded solutions of 1 Notes 333 1.6 280(s) 330 335 303 297 340(s) 0.45 1.1 0.6 pK3 = 6.40 0.63 1 .o 0.65 Solutions (3 x 10-5 moll-') of 1 at acidic pH values in the range 0-4.5 were then left to degrade for 48 h. The HPLC study of the degradation of 1 at acidic pH values, which is the subject of a later section of this paper, shows that after a 2 d period, there are three UV-active decomposition products, D1, Dz and D3. The UV spectra of the degraded solutions were recorded and pK, values of pK1 = 0.9 and pK2 = 3.75 were measured by using the Henderson-Hasselbalch equa- tion, as shown in Table 2.The UV spectral behaviour of degraded solutions of 1 at pH > 4.5 was studied by first allowing a 5 x 10-3 mol 1-1 stock solution of 1 to degrade at pH 2 for a 48 h period and by then adding 30 p1 aliquots of the resulting degraded solution to 5 ml of the appropriate buffer. These solutions were then subjected to UV spectropho- tometry. Application of the Henderson-Hasselbalch equation yielded a pK3 value of 6.40, as shown in Table2. The degraded solution of 1 appeared to be unstable in alkaline media of pH > 8 in that the UV spectral characteristics of the species at this pH, as given in Table 2, changed to a single peak at 285 nm with an absorbance of 1.35. This UV spectral change does not, therefore, correspond to an acid-base equilibrium and has not been ascribed a pK, value.These three pK, values can be used to some effect in detailing sites of protonation/deprotonation of the mixture of DI, D2 and D3 that exists in the degraded solutions of 1. (i) The pK1 of 0.9 is likely to refer to the morpholino-N atom. A similar value would, therefore, be expected for the parent molecule 1. (ii) A pK2 of 3.75 could refer to the 'aniline-type' N atom linked directly to the S atom in 2. For reference, 4-methyl- thioaniline has a literature pK, value of 4.35.7 Tt is also possible that this pK2 value of 3.75 could refer to a pyridine-N atom in a structure such as 1. (iii) A pK3 of 6.4 is likely to refer to a benzimidazole azomethine-N atom. It should be noted that this renders the benzimidazole azomethine-N atom as a marginally weaker acid than the corresponding N atom in 1.Solvent-extraction studies were then carried out to help confirm the existence of the proposed acid-base equilibria for 1 and the degraded solution of 1. A 5 ml solution of 1, at a particular concentration (3 x 10-5 mol 1-1) and at the appropriate pH, was vigorously shaken with an equal volume of ethyl acetate for 30s, and the ethyl acetate layer was subjected to UV spectrophotometry. Compound 1 in ethyl acetate has a A,,, of 296 nm with a pronounced shoulder at 300 nm. The results for 1, depicted in Fig. 2, suggest that the H2A-t species (i.e., with the S -+ 0 group protonated) is optimally extracted in the pH range 7.5-8.5. The species HA and H3A2+ are less efficiently extracted, as shown. Solvent extraction of the degraded solutions of 1 at selected pH values provided the data reported in Table3.The UV spectral behaviour of the degraded solution of 1 , extracted with ethyl acetate, yieldedasinglepeakath,,,of297 nm,butwithno accompanying shoulder as with 1. Optimum extraction of the species giving rise to the 297 nm peak is achieved at 9.28 > pH > 7. Solvent extraction at pH 11.92 yielded species with A,,,514 ANALYST, MAY 1993, VOL. 118 1.1 0.9 0.7 0 B 0.5 0.3 0.1 I I I I 6.0 8.0 10.0 12.0 PH Fig. 2 Variation of absorbance of 1 at 300nm in ethyl acetate, following solvent extraction from aqueous solutions of various pH values Table 3 Variation of absorbance of degraded solutions of 1 at 297 nm in ethyl acetate, following solvent extraction from aqueous solutions of various pH values pH of aqueous layer O* 2.3* 4.4Y 6.98+ 9.287 11.92t Absorbance of 297 nm peakinethylactate 0.01 0.31 0.56 1.0 1.1 0.6 * For the preparation of these solutions, solutions of 1 were left at the appropriate pH for 48 h.t For the preparation of these solutions, a stock solution of 1 was left at pH 2 for 48 h and then aliquots were made up to the appropriate pH before extraction. 1 ~ -Time Fig. 3 Degradation of 1 at pH 2 as monitored by HPLC at 300 nm; 0.2 V, 0.16 A detector settings. Initial concentration of 1 on-column 2.5 x mol 1-1; mobile phase, acetonitrile-water (50 + 50). (a) Initial concentration of 1 on-column without degradation; (b) after 3 rnin degradation; and (c) after 7 min 18 s degradation. (Retention times: 1, 3.1; D1, 2.15; and D2, 4.8 min) 292 nm and a shoulder at 330 nm.This could be due to alkaline degradation of the D1-D2-D3 mixture or extraction of species at this pH that is/are not extracted at acidic and neutral pH. Solvent extraction is, therefore, not very efficient at separating the D1-D2-D3 mixture, so the technique of HPLC was used in order to study more effectively the acid-catalysed degradation of 1. Study of Acid-catalysed Degradation of 1 by HPLC By using the shorter-term kinetic study method described under Experimental, the degradation of 1 was rapid at pH 2, 1 min H / - Time Fig. 4 Degradation of 1 in longer-term kinetic study as monitored by HPLC at 300 nm; 0.5 V, 0.01 A detector settings. (a) After 27 rnin degradation; (b) after 277 rnin degradation; (c) after 2 d degradation; and ( d ) after 5 d degradation.(Retention time: D3, 2.7 min) yielding two decomposition products D1 and D2 in the first 7 min, as shown in Fig. 3. The peak height of D1 then remained constant for the longer-term kinetic study, i.e., a 5 d period, and the peak due to 1 disappeared within 30 min. The peak height due to D2 continued to increase (Fig. 4) until 2 d had elapsed and then remained constant for the remainder of the 5 d study. The peak for D3 appeared after 2 d had elapsed and this also remained relatively constant in height for the remainder of the 5 d study. The separate UV absorption spectra of D1, D2 and D3 were then obtained by varying the detection wavelength for D1, D2 and D3 as they emerged from the HPLC column, following a constant20 plinjectionofamixtureof50 plof a2 dolddegraded solution of 1 and 50 pl of acetonitrile.CompoundDl gave rise to the long-wavelength absorption at 350nm as observed by Senn-Bilfinger et al. ,5 which is consistent with the almost planar structure of sulfenamide 2 as compared with the parent compound 1 and others in which the S atom is not involved in a ring system ( e . g . , the product of the reaction of a thiol with 1 or2 to form 6). In addition, D1 yielded a h,,, at 270 nm. (6) The UV/visible spectra of D2 and D3 were similar in that they yielded A,,, values at 300 and 290 nm, respectively. Study of the Acid-catalysed Degradation of 1 by DPP The kinetic study of 1 x 10-5 moll-1 1 in 1 x 10-2 moll-' HC1 was monitored by DPP, as shown in Fig.5 . The parent compound (1) yieldsa cathodicpeakat -545 mV, which rapidly decays and is replaced by a new cathodic peak at approximately -495 mV (only one composite peak is ever observed). Its peak height reaches a constant value by about 10 min and remains so for the remainder of the 5 d study. This peak can be confirmed as the Dz of the HPLC-UV study as a pure sample of D2 yielded a DPP cathodic peak at -486 mV. A new peak is also rapidly formed at approximately -735 mV and also reaches a constant peak height by about 10 rnin and maintains this for approxi- mately 1 h. For the remainder of the study, its peak height continues to decrease, reaching a steady value after about 2 dANALYST, MAY 1993, VOL. 118 560 490 420 350 280 210 140 70 P a 400 350 300 250 200 150 100 50 0 -160 -480 - 800 I -a93 m~ -493 mV -495 mV -891 mV ? -733 mV -160 -480 -800 EImV Fig. 5 DPP study of acid-catalysed degradation of 1 x 10-5 mol 1-1 1 in 1 X 10-2 moll-l HCl after A, 1 rnin 15 s; B, 6 rnin 53 s; C, 46 rnin 39 s; D, 6 h; E, 47 h 30 min; and F, 120 h 515 and maintaining this until the end of the 5 d study. A third cathodic peak is observed at -491 mV after 2 d and its height remains essentially constant for the remainder of the study.This kinetic study was repeated for 1 x 10-4 moll-1 1, again in 1 x 10-2 moll-' HCl. Degradation peaks at -495 mV (Dz) and -735 mV were again observed, except that both the -495 and -735 mV peaks remained constant in height after 10 min degradationuntil theendofthe5 dstudy. Inaddition, after48 h, two further DPP cathodic peaks appeared at -249 and -913 mV.Thelatterpeakislikely to be analogous tothe peak at -891 mV observed in the DPP study of the degradation of 1 x 10-5 moll-1 1. It is, therefore, obvious that the over-all kinetics of the acid-catalysed degradation of 1 is influenced by the initial concentration of 1. The D1 degradation product from the HPLC-UV study is very likely to be analogous to the -735 mV DPP peak, and the D3 product could be analogous to either the -249 or -913 mV DPP peak observed in the above study of the acid-catalysed degradation of 1 x 10-4 mol 1-1 1. From the HPLC-UV and DPP studies, D1 would appear to be the sulfenamide 2, with D2 and D3 being degradation products where the S atom is not involved in a ring system. The cathodic peak of the parent compound (1) at -545 mVis as a result of reductive fission of the C-S bond, resulting in a four-electron process to yield a substituted pyridine and the appropriate thiol.9 The D2 peak at -495mV appears to be proportionately smaller than the peak for the parent molecule (1) and could, therefore, be attributed to a two-electron process.The sulfenamide or D1 peak at -735 mV is, by comparison with those of 1 and D2, rather large; a catalytic process with D1 as the catalyst is suspected, with the unshared electron pair(s) in the sulfenamide 2 acquiring protons before reduction and catalytic generation of H2. This process could occur with 2 in the adsorbed state on mercury (as with other drug substances, e.g. , cephalothin), or in homogeneous solution.The DPP technique was also found to be a useful instrumental tool for monitoring in situ the reaction between a thiol (z.e., 2-mercaptoethanol) and 1 asasimulation ofits believedreaction in vivo. Fig. 6(a) shows the differential-pulse polarogram for 1 x 10-5 mol1-~2-mercaptoethanol in 1 x 10-2 moll-1 HC1, with a peak at -116 mV corresponding to reduction of the product of the reaction of the thiol with mercury at 0 V, i.e., HO-CH2-CH2SHg. Compound 1 is then added at a concentra- tion of 1 x 10-5 moll-'. The already observed degradation pattern occurs, i.e. , early production of D1 at approximately -730 mV and D2 at approximately -490 mV, with the parent drug reducing at -544 mV and disappearing from solution within 10 min. What is most noticeable is that the thiol signal essentially disappears within this same time scale, with no new peaks being generated.This suggests that a product of the degradation of 1 in acidic solution of pH 2 is reacting with the thiol, but at a site in the molecule that does not contribute to the electrochemical signal responsible for either D1 or D2. If the drug 1 is degraded first, during 10 min in 1 x 10-2 moll-' HCl, to theD1-D2mixture and then the thiolisadded, the thiolpeakis hardly in evidence for the first scan after thiol addition, and the D1 and D2 peaks are again unaltered. A sample of D2, of concentration approximately 1 x 10-5 mol 1-1, prepared as previously stated, was then added to 1 x 10-5 moll-' 2-mercaptoethanol in 1 x 10-2mol1-1 HCl and, in this instance, the thiol and D2 peaks were unaltered within 20 min.This suggests that the thiol reacts with D1 rather than with D2, i.e., a simulation of the believed reaction in vivo of the sulfenamide (1) with a thiol residue on the enzyme, resulting in its inhibition. Further, as the sulfenamide (2, D1) reacts with thiols, such as 2-mercaptoethanol, to form disulfides (6),5 it is not surprising to observe electrochemical disappearance of the thiol signal, with the D1 signal remaining unaltered. This would also suggest that the N lone pair(s) that idare responsible for the believed516 ANALYST, MAY 1993, VOL. 118 It is, therefore, possible to separate a mixture of 1, D1 and D2 contained in a biological matrix such as urine by either Sep-Pak extraction or solvent extraction with ethyl acetate before determination of 1 and D2 by HPLC, as detailed in this paper, with a limit of detection of the order of 1 x 10-6 moll-' for both compounds.. 350 300 250 4 200 c 150 100 50 0 ' I I I -160 -480 - 800 EImV Fig. 6 Monitoring of reaction of 1 x 10-5 mol 1-1 2-mercaptoethanol with 1 X 10-5 moll-' 1 by DPP in 1 x 10-2 moll-' HCl. A, 1 X 10-5 moll-' 2-mercaptoethanol in 1 x 10-2 mol 1-1 HCI; B, as in A with 1 X 10-5 moll-' 1 added, t = 59 s; C, as in B at t = 3 rnin 29 s; D, as in B at t = 10 min 15 s; and E, as in B at t = 32 min 44 s catalytic hydrogen wave of D, idare those on the benzimidazole azomethine-N atom and the morpholino-N atom. If catalytic hydrogen evolution took place in the adsorbed state, molecules 2 and 6 would be expected to be orientated with the 'S end' of the molecules away from the negatively charged mercury surface and the positively charged N atom, together with benzimidazole azomethine-N and morpholino-N held towards the mercury surface.HPLC Determination of 1 and Its Degradation Products in a Biological Matrix (i.e., Urine) The benzimidazole sulfoxide 1 can be determined by HPLC- UV by using the experimental conditions outlined earlier in this paper. A plot of peak area versus concentration is linear in the concentration range 1 x 10-5-1 x 10-6moll-1, with a correlation coefficient of 0.9982. The limit of detection (i.e., the concentration of 1 that produces a peak twice the standard deviation of the background signal) was found to be 5 X 10-7 moll-1 for 1 in acetonitrile-water (50 + 50).When urine was spiked with 1 at a concentration of 1 x 10-5 moll-1, the average recovery for five samples was 95.7% with use of Sep-Pak cartridges. At the 1 x 10-6 moll-1 level, the recovery was 92.3%. Solvent extraction with ethyl acetate was similarly efficient. When urine was spiked with the degradation mixture at a concentration level corresponding to 1 x 10-5mol1-1 1 before degradation, the positively charged sulfenamide D1 (2) was not extracted by either a Sep-Pak cartridge or by solvent extraction with ethyl acetate, but the believed sulfide D2 was extracted with 91y0 recovery using the Sep-Pak cartridge (80% recovery by solvent extraction with ethyl acetate). Conclusion Ultraviolet spectrophotometry and solvent extraction have been used to investigate the acid-base equilibria of the H+/K+ ATPase inhibitor (SK&F 95601) (1) and its acid-catalysed degradation products.This has resulted in &postulation ef the ionic forms of 1 existing in aqueous solution over the pH range 0-14 and corresponding pK, values. Limited information on the chemical nature of the degradation products can 'be derived by this approach, although optimum conditions for the solvent extraction of these drug species from aqueous solution can be obtained. The techniques of HPLC and DPP can be used to monitor, with significantly greater selectivity, the complex acid-cataly- sed degradation of 1. At pH 2 (1 x 10-2 moll-1 HCI) and at concentrations in the range 1 x 10-5-5 x 10-4 mol 1-1 1, the degradation products D1 (the sulfenamide) and D2 (the sulfide) are produced within 10 min, with other degradation products such as D3 being observed over a longer time scale in the 5 d kinetic study.The DPP technique was also used in situ to monitor the reaction between a thiol (i.e., 2-mercaptoethanol) and 1 as a simulation of its believed reaction in vivo. Both 1, through its sulfenamide degradation product 2, and 2 react witli 2-mercap- toethanol, resulting in rapid disappearance of the thiol signal and maintenance of the sulfenamide DI signal within a 30 min period. This is because the electrochemical signal for D1 is probably due to catalytic hydrogen evolution in the adsorbed state with the 'S end' of the molecule orientated away from, and the positively charged N atom orientated to, the mercury surface. The authors thank SmithKline Beecham Pharmaceuticals for provision of drug samples and support of this project:' 1 2 3 4 5 6 7 8 9 10 11 12 References Reuben, B. G., and Wittcoff, H. A., in Pharmuceuticul Chemicals in Perspective, Wiley-Interscience, New York, 1989, p. 284. Clissold, S. P., and Campoli-Richards, D. M., Drugs, 1986,32, 15; and references cited therein. Klemm. K., J . Chem. SOC., Chem. Commun., 1986, 125. Lundberg, P., J. Med. Chem., 1986,29, 1327. Senn-Bilfinger, J., Kruger, V., Sturm, E., Figala, V., Klemm, K., Kohl, B., Rainer, G., Schaefler, H., Blake, T. J., Darkin, D. W., Ife, R. J., Leach, C. A., Mitchell, R. C., Pepper,'E. S., Salter, C. J . , Viney, N. J., Huttner, G., and Zsolnai, L-. ,J. Org. Chem., 1987, 52,4573, 4582. Huber, R., Muller, W., Banks, M. C., Rogers, S. J., Norwood, P. C., and Doyle, E., J. Chromatogr., 1990. 529, 389. + CRC Handbook of Chemistry and Physics, ed. Weast, R. C., CRC Press, Boca Raton, FL, 56th edn., 1975-1976, p. D-147. Ife, R. J., Dyke, C. A., Keeling, D. J., Meenan, E., Meeson, M. L., Parsons, M. E., Price, C. A., Theobald, C J . , and Underwood, A. H., J. Med. Chern., 1989,32, 1970. Johansson, B.-L., Ph.D. Thesis, Uppsala UniverSity, Beckett, A. H., Essien, E. E., and Smyth, W:F., J . !Pharm. Pharmacol., 1974, 26, 399. Gilchrist, T. L., in Heterocyclic Chemistry, Pitman, London, 1985, p. 187. Johansson, B. L., and Persson. B.. Anal. Chim. Acta, 1978,102. 121. Paper 21061 66J Received November 19, $992 Accepted December 21, 1992

ISSN:0003-2654    

DOI:10.1039/AN9931800511

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 5 17 6,7-Dimethoxy-I -methyl-2-oxo-l ,2-di hydroquinoxalin-3-ylpropiono- hydrazide as a Fluorescence Derivatization Reagent for Aldehydes in High-performance Liquid Chromatography Tetsuharu Iwata, Tsuyoshi Hirose, Masaru Nakamura and Masatoshi Yamaguchi" Faculty of Pharmaceutical Sciences, Fukuoka University, Nanakuma, Jonan-ku, Fukuoka 814-0 I , Japan 6,7-Dimet hoxy-I -methyl-2-oxo-l,2-di hyd roqui noxa Ii n-3-yl propiono hyd razide has been developed as a highly sensitive and selective fluorescence derivatization reagent for aliphatic and aromatic aldehydes. The reaction conditions were optimized with cyclohexanecarbaldehyde, 2-naphthaldehyde and heptanal. The reagent reacted with the aldehydes in the presence of trichloroacetic acid at 60°C to produce the corresponding fluorescent derivatives, which were separated on a reversed-phase column, L-column ODs, by isocratic elution with 70% v/v aqueous methanol as eluent.The derivatives were detected spectrofluorimetrically at 442 nm with excitation at 362 nm. The detection limits (signal-to-noise ratio = 3) for the aldehydes are 13-18 fmol for an injection volume of 10 pI. Keywords: 6,7- Dimethoxy- 1 -meth yl-2-0x0- 1,2-dihydroquinoxalin-3-ylpropiono h ydrazide; aldehyde; fluorescence derivatization reagent; high-performance liquid chromatography Various fluorescence derivatization reagents having a hydra- zino group have been reported for the determination of aldehydes and ketones by high-performance liquid chromato- graphy (HPLC) , e. g., 5-hydrazino-N, N-dimethylnaph- thalene-1-sulfonamide,' 4-hydrazino-7-nitro-2,173-benzoxa- diazole ,* 4-aminosulfon yl-7-hy drazino-2 , 173-benozoxadia- zole , 4-( N , N-dimethylaminosulfony1)-7-hydrazino-2,1,3-ben- zoxadiazole and 4-hydrazino-7-nitro-2 , 173-benzoxadiazole hydrazine,3 fluoren-9-yl-methoxycarbonylhydrazine4 and anthracence-1- and 2-carbohydrazides.5 In previous research, 6,7-dimethoxy-l-methyl-2-oxo-1,2- dihydroquinoxalin-3-ylpropionohydrazide (DMEQ-hydra- zide; Table 1) was developed as a fluorescence derivatization reagent for carboxylic acids;677 the reagent reacted with carboxylic acids in aqueous solution in the presence of pyridine and l-ethyl-3-(3-dimethylaminopropyl)carbodi- imide.In the present work, the reaction of DMEQ-hydrazide with aldehydes and ketones was studied.Consequently, it was found that DMEQ-hydrazide also reacted fairly selectively with aliphatic and aromatic aldehydes in the presence of trichloroacetic acid (TCA) under conditions different from those for carboxylic acids to give highly fluorescent hydra- zones. Hence, a highly sensitive method for the determination of aldehydes using DMEQ-hydrazide , based on HPLC with fluorescence detection , has been developed. Cyclohexane- carbaldehyde, 2-naphthaldehyde and heptanal were used as model compounds to establish reaction conditions suitable for a more general method. Experimental Reagents and Solutions All chemicals were of analytical-reagent grade, unless stated otherwise. Distilled, de-ionized water was used throughout. The DMEQ-hydrazide was prepared as described previously;6 it is now commercially available from Wako Pure Chemicals (Tokyo, Japan).Apparatus Uncorrected fluorescence spectra and intensities were measured with a Hitachi (Tokyo, Japan) 650-60 spectroflu- orimeter using a 10 x 10 mm quartz cell; spectral bandwidths * To whom correspondence should be addressed. of 5 nm were used in both the excitation and emission monochromators. Proton nuclear magnetic resonance ('H NMR) spectra were recorded on a Hitachi R-90 spectrometer at 90 MHz in [*H7]-dimethylformamide using tetramethyl- silane as an internal standard (abbreviations used: s, singlet; t, triplet; and m, multiplet). Electron-impact mass spectra (MS) were taken with a JEOL (Palo Alto, CA, USA) DX-300 spectrometer. The pH was measured with a Hitachi-Horiba (Tokyo, Japan) M-7 pH meter at about 25°C.Uncorrected melting-points were measured with a Yazawa (Tokyo, Japan) me1 ting-poin t apparatus. A Hitachi L-6200 high-performance liquid chromatograph, equipped with a high-pressure sample injector (20 pl loop) and a Jasco (Tokyo, Japan) FP-210 spectromonitor fitted with a 12 p1 flow cell, was used. It was operated at an excitation wavelength of 362 nm and an emission wavelength of 442 nm. The column was an L-column ODS (250 X 4.6 mm i.d., particle size, 5 pm; Chemicals Inspection and Testing Insti- tute, Tokyo, Japan). The column temperature was ambient (20-27 "C) . Synthesis of Fluorescence Reaction Product From Heptanal The DMEQ-hydrazide (7.7 g, 25 mmol) and heptanal (2.9 g, 25 mmol) were dissolved in 200 ml of methanol and the mixture was heated at 60 "C for 1 h.The reaction mixture was evaporated to dryness under reduced pressure. The residue, dissolved in 5 ml of chloroform, was chromatographed on a silica gel 60 column (25 X 2.7 cm i.d.; about 120 g, 7&230 mesh; Japan Merck, Tokyo, Japan) with ethyl acetate-hexane (2 + 1). The main fraction was evaporated to dryness under reduced pressure and the residue was recrystallized from 70% v/v aqueous methanol to give colourless needles (yield Table 1 Structures of DMEQ-hydrazide and product 1* CH3 cH30J-J30 CH30 CH2C H 2-R * DMEQ-hydrazide: R = CONHNH,; product 1: R = CONHN= CH-(CH2)5CH3.518 ANALYST, MAY 1993, VOL. 118 4 0 6 12 Time/m i n Fig. 1 Chromatogram of DMEQ derivatives of aldehydes.A portion (100 pl) of a standard mixture of the aldehydes (1.0 nmol ml-1 each) was treated according to the described procedure. Peaks: 1, cyclo- hexanecarbaldehyde; 2, 2-naphthaldehyde; 3, heptanal; and 4, reagent blank components 1.0 g, m.p. 149-151 "C). MS: mlz 402 (M+, base peak). NMR ([2H7]-Me2NCHO): 6 0.7-1.15 (3 H, m, CH3), 1.2-1.7 [lo H, m, (CH2)5], 2.63 (2 H, t, J = 7 Hz, CH2), 3.13 (2 H, t, J = 7 Hz, CH2), 3.69 (3 H , s, N-CH3), 3.90 (3 H, s, 0-CH3) 4.04 (3 H , s, 0-CH3), 7.04 (1 H, s, aromatic H), 7.23 (1 H, s, aromatic H), 7.45 (1 H, t, J = 5 Hz, N=CH). (Found: C, 62.91; H, 7.71; N, 13.69. Calc. for C2]H3"N404: C, 62.69; H , 7.46; N, 13.93%.) Derivatization Procedure To 100 pl of a test solution of the aldehydes in N,N-dimethyl- formamide were added 100 pl of 5 mmol l-1 DMEQ-hydra- zide in N,N-dimethylformamide and 50 pl of 20% TCA in water.The mixture was heated at 60°C for 15 min. The reaction mixture (10 pl) was injected into the chromatograph. For the reagent blank, 100 pl of N,N-dimethylformamide in place of 100 pl of the test solution were subjected to the same procedure. Results and Discussion HPLC and Derivatization Conditions The simultaneous separation of DMEQ derivatives of cyclohexanecarbaldehyde, 2-naphthaldehyde and heptanal was studied on several reversed-phase columns, viz. , TSK gel ODS 120-T and L-column ODs, with aqueous methanol and aqueous acetonitrile. The best separation was achieved on L-column ODS with 70% v/v aqueous methanol in the mobile phase. Fig. 1 shows a typical chromatogram obtained for the aldehyde derivatives.All the peaks were completely separated and eluted within 12 min. The derivatization conditions were examined using a mixture of three aldehydes (1.0 nmol ml-1 each). The most intense peaks were obtained at concentrations greater than about 4.0mmoll-1 of the reagent solution for all the aldehydes; 5.0 mmol l-1 was used in the procedure. Among the solvents examined for the derivatization reaction (methanol, ethanol, acetonitrile, N, N-dimethylformamide and dimethyl sulfoxide) , the use of N, N-dimethylforrnamide resulted in the most intense peaks in the chromatogram and DMEQ-hydrazide was found to dissolve easily in this solvent; N, N-dimethylformamide was, therefore, used in the proce- 100 U .- S 3 F E -F 4- .- Y al 50 rn 2 L 4- 0 al 4- 0" 0 15 30 45 Time/m in Fig.2 Effect of reaction temperature and time on the peak height of heptanal. Portions (100 pl) of heptanal (1 nmol ml-I) were treated according to the procedure, except for temperatures. Temperature: 1 , 40; 2, 50; 3, 60; 4, 70; and 5, 100°C 300 400 500 Wavelengthhm Fig. 3 Fluorescence excitation and emission spectra of the fluor- escent product from heptanal in: 1, methanol; 2, acetonitrile; and 3. water dure. Trichloroacetic acid was used to facilitate the derivatiza- tion of aldehydes with the reagent. Maximum and constant peak heights could be achieved at TCA concentrations greater than 1.0% for cyclohexanecarbaldehyde and heptanal and 10% for 2-naphthaldehyde. Therefore, 20% was selected for the simultaneous derivatization of the three aldehydes.The derivatization reaction of aldehydes with the reagent occurred more rapidly with increasing reaction temperature. This effect is shown in Fig. 2 for heptanal. At 60 "C, the peak heights for all the aldehydes were almost maximum after heating for 10 min. However, at 100"C, the peak heights decreased on prolonged heating. Hence, heating for 15 min at 60 "C was employed in the procedure. The DMEQ derivatives in the final mixture were stable for at least 12 h in daylight at room temperature.ANALYST, MAY 1993, VOL. 118 5 19 Table 2 Retention times and detection limits (signal-to-noise ratio = 3) for DMEQ derivatives of aldehydes Mobile phase composition: Detection aqueous Retention limit/fmol methanol Compound time/min per 10 pl (% v/v) Heptanal 9.4 Octanal 14.0 Decanal 34.8 Dodecanal 11.4 Cyclohexanecarbaldehyde 7.4 Benzaldehyde 18.6 Vanillin 20.2 Isovanillin 41.4 2-Naph thaldeh yde 8.2 Heptan-3-one 13.4 Hexan-2-one 10.0 Acetophenone 8.4 Aldosterone 7.1 D-Glucose 27.0 D-Galactose 25.2 D-Xylose 34.6 Benzophenone 20.0 17 24 55 15 13 25 27 53 18 3 750 45 1 446 18 750 225 4 120 1150 860 * Mobile phase: 10% v/v aqueous acetontirile.70 70 70 85 70 50 50 50 70 60 60 60 60 60 * * * - - - Fluorescent Derivatives in the Determination of Aldehydes and Their Fluorescence Properties The fluorescent reaction products obtained in the determina- tion of aldehydes were examined by using heptanal. As hydrazino groups react with carbonyl compounds in acidic solution to give hydrazones,1.5 the reaction product from heptanal should be the corresponding DMEQ-hydrazone derivative, which was confirmed as product 1 (in Table 1) by the analytical data described under Experimental.The fluorescence properties of the product in methanol, acetonitrile and water, which have been widely used as mobile phases in reversed-phase liquid chromatography, were exam- ined (Fig. 3). The fluorescence excitation (maximum, 362 nm) and emission (maximum, 442 nm) spectra of the product in methanol were almost identical with those in acetonitrile and water. The maxima in aqueous methanol and acetonitrile were independent of the concentration of water. On the other hand, the fluorescence intensities in aqueous methanol were higher than those in aqueous acetonitrile at any concentration. The fluorescence intensity in aqueous methanol and aqueous acetonitrile was maximum at a water concentration of The fluorescence excitation (maximum, 362 nm) and emis- sion (maximum, 442 nm) spectra of the product in 70% v/v aqueous methanol were almost identical with those of the eluates from the peaks for cyclohexanecarbaldehyde, 2-naph- thaldehyde and heptanal.When product 1 dissolved in N, N-dimethylformamide was applied directly onto the HPLC column used in this work, a single peak having exactly the same retention time as that of peak 3 in Fig. 1 appeared in the chromatogram. These results indicated that product 1 is the main product in the determination of heptanal with DMEQ- hydrazide. The products from the other aldehydes might be predicted to be the corresponding hydrazones. The yield of the fluorescent derivative from heptanal under the conditions used here was found to be 39.1% by comparing 1O-9OYo.the value of the peak height for heptanal with that of product 1. Precision, Calibration Graphs and Detection Limits The within-day precision was established by repeated determi- nations using a standard mixture of the three aldehydes (1 .O nmol ml-1 each). The relative standard deviations did not exceed 2.5% for all the aldehydes examined (n = 10 in each instance). The between-day precision was obtained by per- forming the analyses (n = 3, each day) for 7 d using the same standard mixture of the three aldehydes (1 .O nmol ml-1 each) stored at -40 "C. The relative standard deviations did not exceed 5.5% for all the aldehydes examined.The relationship between the peak height and the amount of the individual aldehyde was linear up to at least 30pmol per injection volume (10 pl); the linear correlation coefficients were 0.908 or better for all the aldehydes tested. The limits of detection for the aldehydes were 13-18fmol per injection volume (10 pl) at a signal-to-noise ratio of 3 (Table 2). Reaction of DMEQ-hydrazide With Other Compounds The reactivity of DMEQ-hydrazide with other carbonyl compounds was studied under the recommended conditions (Table 2). Many aliphatic and aromatic aldehydes readily reacted with DMEQ-hydrazide to form intensely fluorescent DMEQ derivatives. The relatively low sensitivities of aldo- sterone and sugars may be due to the formation of hemi- acetals.On the other hand, the reactivity of ketones with DMEQ-hydrazide was poor compared with that of aldehydes. The reactivity of ketones could not be accelerated even by prolonged heating and/or heating at higher temperatures. a-Keto acids (pyruvic, a-ketoglutaric and phenylpyruvic) gave no peaks under the HPLC and derivatization conditions employed. Other substances such as fatty acids, a-amino acids, alcohols, amines, phenols and thiol compounds gave no fluorescent derivatives under these conditions. This suggests that the proposed method is fairly selective for aliphatic and aromatic aldehydes. The reagent permits the sensitive and fairly selective derivatization of various aliphatic and aromatic aldehydes under moderate conditions. Therefore, the reagent should be useful for the detection of aliphatic and aromatic aldehydes of biological importance at femtomole levels by HPLC. The application of the reagent to the determination of some drugs that have an aliphatic aldehyde in their molecule is continuing. References 1 Chayen, R., Duir, R., Gould, S., and Harrel, A., Anal. Biochem., 1971,42,283. 2 Gubitz, G., Wintersteiger, R., and Frei, R. W., J . Liq. Chromatogr., 1984,7, 389. 3 Uzu, S., Kanda, S., Imai, K., Nakashima, K., and Akiyama, S., Analyst, 1990, 115, 1477. 4 Zhang, R., Cao, Y., and Hearn, W. M., Anal. Biochem., 1991, 195, 160. 5 Goto, J., Saisho, Y., and Nambara, T., Anal. Sci., 1989,5,399. 6 Yamaguchi, M., Iwata, T., Inoue, K., Hara, S., and Nakamura, M., Analyst, 1990, 115, 1363. 7 Iwata, T., Inoue, K., Yamaguchi, M., and Nakamura, M., Biomed. Chromatogr., 1992, 6 , 120. Paper 21061 05 H Received November 17, 1992 Accepted January 25, 1993

ISSN:0003-2654    

DOI:10.1039/AN9931800517

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 521 Determination of Zirconium and Molybdenum With 4,5-Dihydroxybenzene-1,3-disulfonic Acid Disodium Salt by Ion-pair Reversed-phase High-performance Liquid Chromatography Suh-Jen Jane Tsai and Hsiao-Tzu Yan Department of Applied Chemistry, Providence University, Taichung Hsien, Taiwan, Republic of China 43309 An analytical process based on reversed-phase high-performance liquid chromatography, with ultraviolet/visible detection at 31 5 nm, was studied. 4,5-Di hydroxybenzene-I ,3-disulfonic acid disodium salt (Tiron) was used as chelating agent. Optimum conditions for the determination of zirconium and molybdenum ions, including the concentration of the buffer, the pH of the chelating medium and the mobile phase, the concentration of Tiron, the time of colour development and the composition of the mobile phase, were investigated. Tiron chelates were eluted within 15 min, with methanol-water (63 + 37) as the mobile phase, which contained acetate buffer (1.5 x 10-3 mol 1-1, pH 4.0), Tiron (1.5 x 10-3 mol 1-1) and tetrabutylammonium bromide (3 x 10-2 mol 1-1).The separation was performed with a CI8 reversed-phase column. The proposed method was applied to the determination of Zrlv and MoV1 in a relatively complex substance, nickel-base alloy. The precision and accuracy of the method were evaluated. The experimental values obtained from the calibration graphs and by the standard additions method were compared with the certified values. There was good agreement between these values. With five consecutive injections of 100 pl of sample solution, containing 200 ng each of ZrIV and MoV1, the recoveries obtained for Zr and Mo were 97 and 106%, respectively.The detection limits (signal-to-noise ratio = 3) were 9.0 and 3.6 ppb for ZrtV-Tiron and MoV1-Tiron, respectively. Keywords: Zirconium and molybdenum determination; Tiron; tetrabutylammonium bromide; ion-pair reversed-phase high-performance liquid chromatography; spectrophotometry Reversed-phase high-performance liquid chromatography (HPLC) has often been used in the trace-level determination of metal ions.14 By chelating with highly absorptive chromo- phores, metal ions in environmental samples or in certain other materials have been determined by this technique.5-9 The chelating agents usually play an important role.In addition to having a strong chelating ability, the correspond- ing chelates should be stable and have strong absorption in the ultraviolet/visible (UV/VIS) region to ensure a sensitive detection. S-Hydroxyquinoline,'OJ* dithiocarbamic acids,12,13 P-diketonesl4 and 4-(2-pyridylazo)resorcinol (PAR)15-*7 meet with the above requirements. 4,5-Dihydroxybenzene-1,3-disulfonic acid disodium salt (Tiron) has been known as an effective chelating agent for the spectrophotometric determination of titanium18-22 and iron.23 It forms water-soluble chelates with numerous metal ions. In spite of its unique properties, only limited attention has been paid to Tiron as a chelating agent in reversed-phase HPLC.24 Several studies on the use of ion-pairing reagents for attaining satisfactory separations of metal ions have been reported.25-32 By careful control of the retention of analytes by the ion-pairing reagents in the mobile phase, the analytes can be well separated.Alkylsulfonic acids and quaternary ammonium hydroxides are the ion-pairing reagents usually used for neutralizing cationic and anionic species, respectively. Of the ion-pairing reagents, such as octanesulfonic acid, dibutylamine, tetra- methylammonium chloride and tetrabutylammonium bro- mide (TBAB), TBAB has been the one most often used. The aim of this work was to investigate the spectral characteristics and chromatographic behaviour of Zr'"-Tiron and Mo"'-Tiron in order to develop a methodology for the determination of metal chelates by ion-pair reversed-phase HPLC, with UV/VIS detection. As Zr'" and Mo"' form anionic chelates with Tiron, TBAB was included in the mobile phase to enhance the hydrophobicity to the anions.Silica columns with a bonded organic phase (C18) were used. Factors that determined the performance of the proposed method have been extensively studied. These included the choice of buffer, the pH of the medium, the constituents of the mobile phase, the concentrations of Tiron in complex-forming reactions and in the mobile phase, the concentration of TBAB in the mobile phase, the stability of the metal chelates, and the optimum flow rate of the mobile phase. The precision and accuracy of the proposed method have also been evaluated. Experimental Apparatus The pH measurements were carried out with a PHM 85 precision pH meter and a combined electrode (Radiometer, Copenhagen, Denmark).The absorption spectra were obtained with a Shimadzu (Kyoto, Japan) UV-260 UVNIS recording spectrophotometer, equipped with 1 cm silica cells. The chromatographic system consisted of a Waters (Milford, MA, USA) 600 E system controller, a Waters U6K universal liquid chromatography injector (2 ml injection loop), a Waters 740 data module and a Waters Lambda-Max Model 481 LC spectrophotometer. A Waters pBondapak C18 column (No. 27324, 300 X 3.9 mm i d . ) , coupled with a Waters guard column (No. 88141), was used. A Barnstead (Boston, MA, USA) Nanopure I1 system was used for water purification. Reagents and Solutions Atomic absorption standard solutions of Zr (1030 pg ml-1 of Zr in 5% HCI) and Mo (995 pg ml-1 of Mo in € 3 2 0 ) were products of Aldrich (Milwaukee, WI, USA).Tiron was a product of Merck (Darmstadt, Germany). Stock solutions of the metal ions were prepared from the nitrates or chlorides. The nickel-base alloy used, 211X11221D, was a certified reference material and was obtained from MBH Analytical (Barnet, Hertfordshire, UK) . Acetonitrile and methanol were of liquid-chromatography grade and were obtained from Mallinckrodt (St. Louis, MO, USA). Procedure A sample solution (10 ml) was prepared as follows. After 1 ml of Tiron solution (1.5 X 10-2 moll-1) had been added to 3 ml522 of acetic acid-sodium acetate buffer solution (0.1 moll-1, pH 4.0), 50 p1 of metal solution were added. The resulting sample solution was diluted to 10 ml in a calibrated flask with distilled, de-ionized water (DDW).The pre-column chelation of the metal ions was completed after 10 min. An aliquot of the sample solution (usually 100 pI) was injected into the HPLC system. The mobile phase was prepared by placing measured volumes of methanol (315 nm) and DDW in a 500 ml calibrated flask; 50 ml of Tiron solution (1.5 X 10-2 mol 1-1) and 4.9805 g of TBAB were added to give a final concentra- tion of 3 x mol 1-1 TBAB, then 7.5 ml of acetate buffer solution were added and the solution was diluted to 500 ml with DDW. The sample solution and mobile phase were passed through a 0.45 pm filter before use. The Tiron chelates were detected at 315 nm. Analysis of Nickel-base Alloys An accurately weighed amount of 211X11221D alloy (0.14.2 g) was treated with a 2 ml mixture of HCI and HN03 (4 + 1).The alloy solution was heated gently until the solid was completely decomposed. The excess of acid was removed by heating the sample solution to near dryness and final solution was prepared by diluting the residual material to 5 ml with 5% HCI. The corresponding blank solutions were prepared similarly, but without the addition of any alloy. Results and Discussion Spectrophotometric Studies In order to choose a suitable ligand for the trace-level determination of Zr and Mo by HPLC, several complex- forming reagents, which included PAR, diantipyrylmethane (D APM) , 2-( 5-bromo-2-pyridylazo)-5-( diet hy1amino)phenol- (2,s-BrPADAP) and Tiron, were studied. The reagent DAPM did not form a chelate with Z P .Although both PAR and 2,5-BrPADAP formed chelates with ZrIV and MoV1, the injection of such chelates into a C18 column did not yield any chromatographic peak. This might be due to the decompo- sition of the metal chelates during the chromatographic separation or to the retention of the chelates on the column. Tiron rapidly formed highly absorptive, water-soluble chel- ates with both ZrIv and MoV1. Consequently, any excessive sample handling, such as in the extraction process, could be avoided and interference from extraction solvents and irrepro- ducibility between samples could be reduced to a minimum. Therefore, further studies were performed with Tiron. The chelating reactions were performed in a variety of buffer systems based on citrate, oxalate, tartrate and acetate.Fig. 1 shows the absorption spectra of metal chelates in either citrate, oxalate, tartrate or acetate buffer solutions. Clearly, the sensitivity would be much greater were the analysis to be performed in an acetate buffer solution. The pH dependency of the absorption spectra is shown in Fig. 2. Under the given conditions, Zr chelates yielded the maximum absorption at pH 4.0 whereas Mo chelates yielded the best sensitivity at pH 6.0. The above observations imply that the absorption characteristics were highly susceptible to the buffer system and also to the pH of the sample solution. Buffer concentra- tion was one of the important factors that needed to be considered. According to Fig. 3, the optimum acetate buffer concentration was 3 x 10-2 rnol 1-1 for both ZrKV and MoV1 chelates.The completeness of the colour development was also dependent on the development time. The absorption of Mo chelates reached a maximum at 15 min, after which the absorption decreased gradually, owing to the decomposition of the metal chelates. Zirconium chelates showed maximum absorption after 5 min. There was only a slight deviation in the absorbance during 24 h. Therefore, 15 min was chosen as the colour-development time. The maximum absorption of 0.50 8 + a 0 0.25 0 3 ANALYST, MAY 1993, VOL. i! !A ! I ! I ! ! 500 300 Wavelengthlnm 18 500 Fig. 1 Absorption s ectra of metal-Tiron chelates at pH 4.0 in various buffers. (a) %Iv; and (b) MoV1 chelate. cmetalr 2 x 10-5 rnol 1-I; and C T , ~ ~ " , 1.5 x rnol I-'. A , acetate; T, tartrate; 0, oxalate; and C, citrate 1.5 1 .o 8 e $ 2 0 0.5 0 4.0 r.I To ! ! ! ! 200 .-.- ! ! ! I I 6.0 il\ ! - ! ! I I \ 400 300 Wavelengthlnm 500 Fig. 2 Effect of pH on the absorption spectra of metal-Tiron chelates in 3.0 x lo-* mol1-l.acetate buffer. (a) Zrlv chelates; and (b) MoV1 chelates. cmetal, 4 x 10-5 rnol 1-1; and cTiron, 1.5 X 10-3 mol I-* 1.4 1.2 1 0, 0.8 2 0.6 G a 0.4 0.2 t I I I I I 1 Buffer concentrationIl0-2 rnol 1-1 Fig. 3 Effect of the acetate buffer concentration on the absorption of metal chelates (pH 4.0). A, ZrIV chelate; and B, MoV1 chelate. cZr. 5 x 10-5 rnol 1-l; cM0, 2 X low5 mol 1-1; and C T ~ ~ ~ ~ , 1.5 X 10-3 rnol I-' 0 1 2 3 4ANALYST, MAY 1993, VOL. 118 523 the metal chelates was obtained when the concentrations of Tiron and metal ions were kept at 1.5 x 10-3 and 5.0 x 10-4 moll-1, respectively.This indicated that the concentration of Tiron only needed to be three times that of the metal. Further increase in the Tiron concentration did not result in any noticeable enhancement in the absorption of metal chelates. Both the continuous-variation and molar-ratio methods were applied to establish the stoichiometry of the chelates formed under slightly acidic conditions (pH 4.0, acetate buffer). Job plots of the continuous variation clearly showed the composition of the chelates was 1:3 and 1 : 2 for ZrIV-Tiron and MoV1-Tiron chelates, respectively. These metal-Tiron ratios were consistent with the results obtained by the molar-ratio method. The molar absorptivity of the ZrIV chelate was 3.4 x 104 1 mol-1 cm-1 at A, 309 nm and that of the MoV1 chelate was 4.5 x 104 1 mol-1 cm-1 at A,,, 252 nm.The following chromatographic studies were carried out, with monitoring at 315 nm in order to avoid the strong absorption of acetate buffer, which started at 250 nm. The molar absorptivities of ZrIV and MoV1 chelates were 3.1 X 104 and 1.9 x 104 1 mol-1 cm-1, respectively, at 315 nm. Chromatographic Studies The integrity of the metal chelates was susceptible to the pH of the solution. Therefore, the separations were performed under acidic conditions. Fig. 4 shows the chromatograms for chelates at various pH values. Both the retention time and the shape of the peaks were susceptible to the pH of the mobile I 6.998 (a’ I 6.838 I Mo 1 6.888 0 12.50 25.00 Tirnehin Fig.4 Representative chromatograms of Tiron chelates at various pH values. (a) pH 4.0; (b) pH 5.0; and (c) pH 6.0. Mobile phase: methanol-water (63 + 37); caCetate buffer, 1.5 x moll-l; cmetal, 6.0 x 10-5 mol 1-1; cTiron, 1.5 X 10-3 mol 1-1; and cTBAB, 3.0 X loF2 mol 1-1. Flow rate, 0.7 ml min-1; 315 nm; att, 256 phase. Increase in the pH not only prolonged the retention time, but also seriously broadened the chromatographic peak for ZrIV chelates. This was partly due to the variation of the chelating ability of Tiron and the over-all charges on the metal chelates as the pH of the mobile phase changed. There was no detectable chromatographic signal attributed to Zr*”-Tiron in either tartrate, oxalate or citrate buffer solutions. In fact, tartrate was one of the strongest masking agents for Zr.11 The selectivity and sensitivity of the chromatographic analysis were highly susceptible to the content of organic modifier in the mobile phase. Many chromatographic separa- tions have been performed with buffered aqueous acetonit- rile, tetrahydrofuran or methanol.In this work, several different isocratic mobile phases, acetonitrile-water, acetonit- rile-methanol-water and methanol-water, were tested. Of these, methanol-water afforded the best resolution. Fig. 5 shows the dependency of the capacity factor, k’, on the methanol content. The elution strength increased as the methanol content increased. This resulted in a shortening of the retention time for Tiron chelates. Zirconium chelates were more sensitive to the variation in methanol content than were Mo chelates.This implied that Zr chelates were less polar and bonded more strongly to the non-polar group of C18. Hence, they were more sensitive to the elution strength. As Mo chelates were more polar than Zr chelates, the elution strength had less effect on the capacity factor of Mo chelates. Although better separation would be achieved with methanol -water (60 + 40), the peak shape for Zr chelates was greatly flattened with this mobile phase. Chromatographic signals for Zr chelates could not be found when the methanol content was less than 60%. Poor resolution was encountered if the methanol content was higher than 65%. Optimum selectivity was achieved with a mobile phase of methanol-water (63 + The separation of the chelates was accomplished by ion-pair reversed-phase HPLC.The chromatographic peaks for metal chelates could not be resolved from each other when elution was effected without the addition of either the chelating agent Tiron or the ion-pairing reagent TBAB, as shown in Fig. 6(a). The inclusion of the chelating agent in the mobile phase prevented the decomposition of chelates during the separation process. However, the chromatographic peaks for Tiron and the metal chelates could still not be resolved, as shown in Fig. 6(b). Although a peak for Zr chelates appeared when TBAB was added to the mobile phase [Fig. 6(c)], better results were achieved with the presence of Tiron and TBAB in the mobile phase. Hence, the Tiron chelates of ZrIV and MoV1 were well resolved [Fig.6(d)]. Identical chromatographic characteristics were observed in the work of Yamada and Hattori,*4 where the peak for 37). MoV1-Tiron could only be obtained under similar 6 elution I I I 50 58 66 74 MeOH (YO) Fig. 5 Effect of the percentage of methanol in the mobile phase (pH 4.0) on k‘. A, Zrl” chelate; and B, MoV1 chelate. Separation was performed with the same operating conditions given in Fig. 4524 ANALYST, MAY 1993, VOL. 118 conditions. The distinct chromatographic behaviour of these metal chelates on a CI8 column was partly due to the differences in their dissociation kinetics. Apparently, both Tiron and TBAB were required in order to obtain a satisfactory separation of metal chelates. t 0.02 C 0) v) .- 12.52 0.01 I i ( C ) Zr I 0.02 , 12.52 0.01 Time/mi n 40 12.51 (4 !r 12.51 Fig.6 Typical chromatograms obtained with various constituents in the eluent. (a) Without Tiron or TBAB; (b) with 1.5 x 10-3 rnol 1-1 Tiron; (c) with 3 x mol 1-' TBAB; and (d) with 1.5 x 10-3 moll-' Tiron and 3 X mol 1-I TBAB. The operating conditions given in Fig. 4 were followed, except, att 32 The retention of Zr-Tiron and Mo-Tiron chelates was examined as a function of Tiron and TBAB concentrations in the mobile phase. The results are shown in Fig. 7. The rapid increase in the capacity factors for these chelates with increasing concentrations of TBAB indicated that the ion pairing was the predominant retention process at low TBAB concentrations. It reached a maximum when the concentra- tions were 3.0 x 10-2 and 2.0 X 10-2 mol 1-1 for the Zr and Mo chelates, respectively, as shown in Fig.7 ( a ) . The capacity factor remained nearly constant for both chelates with TBAB concentrations higher than those given above. This phenorne- non implied that chelates acted as neutral species when sufficient amounts of ion-pairing agents were present .27 Therefore, the stationary-phase adsorption of ion-paired species, anionic metal chelates and TBA+ on the CI8 surface dominated the retention of the chelates. As Tiron was required in the mobile phase to prevent the dissociation of MoV1 chelates, the retention of these chelates was more susceptible to the concentration of Tiron. Satisfactory peak resolution was observed when 1.5 x 10-3 moll-' Tiron was added to the mobile phase, as shown in Fig. sr 2 - J 0 2 4 6 [TBAB]/10-2 rnol 1 I I , 1 ' 0.8 1.6 2.4 3.2 [Tironl/lO 3 rnol I 1 Fig.7 Effect of the concentration of TBAB and Tiron in the eluent (pH 4.0) on k'. (a) k' versus TBAB concentration. ( b ) k' versus Tiron concentration. Other operating conditions are given in Fig. 4. A, Zr'" chelate; and B, Mo"' chelate 6 4 L 2 0 0.4 0.8 1.2 1.6 Flow rate/ml min-1 Plots of k' versus flow rate, pH 4.0. The operating conditions Fig. 8 are given in Fig. 4ANALYST, MAY 1993, VOL. 118 525 7(b). No further improvement in the separation of the chelates was achieved with higher concentrations of Tiron. The chromatographic performance was also susceptible to the flow rate of the mobile phase. Fig. 8 depicts the retention of the chelates as a function of the flow rate. The capacity factors for Zr-Tiron and Mo-Tiron decreased as the flow rate was increased. Chelates would be retained longer in the column were the separation to be performed at a slower flow rate.This would eventually broaden the chelate peaks, and a poor resolution would result. Consequently, 0.7 ml min-1 was chosen as the optimum flow rate. In ion-pair reversed-phase HPLC separation, anionic metal chelates are retained on the column, presumably through both ion-exchange and adsorption reactions? The content of the organic modifier and the eluting strength of the mobile phases were the key factors determining the predominant reaction. However, both reactions could take place under certain conditions. With the optimum operating conditions establi- shed above, ZP-Tiron and MoVi-Tiron were successfully separated.The representative chromatograms are reproduced in Fig. 9. Peaks for Tiron appeared before 7.0 min. The retention times for Mo-Tiron and Zr-Tiron were 9.4 and 12.8 min, respectively. Apparently, Zr-Tiron reacted more strongly with the stationary phase than did Mo-Tiron. Table 1 presents the regression analysis data and detection limits for Tiron chelates. Linear calibration graphs were obtained for ZrlV-Tiron and MoVi-Tiron concentrations in the range 0.5241.2 and 0.48-51.7 ppm, respectively. Good linearity of the calibration graphs was obtained over approximately two orders of magnitude for these chelates. The linear regression coefficient was 0.9998 for both chelates. The detection limits (signal-to-noise ratio = 3) were 9.0 and 3.6 ppb for ZriV-Tiron and MoV1-Tiron, respectively.The relative standard deviation of the retention times of these peaks under the same 0.01 L 12.51 0 12.50 D n d Timehi n Fig. 9 Typical chromatograms of Tiron chelate determination. (a) Blank solution and (b) chelates of ZrIV and MoIV. Mobile phase: methanol-water (63 + 37); caCetare buffer, 1.5 x 10-3 mol I-', pH 4.0; cmet,l, 6.0 X lop5 moll-I; +iron, 1.5 X mol I-[; and cTBAB, 3.0 X 10-2 mol I-'. Flow rate. 0.7 ml min-1; 315 nm; and att 32 experimental conditions was about 2%, whereas that of the peak-area measurements was less than 10% (five replicate injections). This work demonstrated that a successful separa- tion and determination of trace amounts of ZriV and Mo"' could be achieved by ion-pair reversed-phase HPLC on an octadecyl-bonded silica column with careful control of the amount of Tiron and TBAB in the mobile phase.Interference Studies Numerous metal ions, which included those of TiIV, Crvl, NbV, TaV, Mg", Ni", Zn", Fell', CO", Cull, Call and Seiv, were examined for their possible interference effects. The effect of foreign metal ions on the peak area and resolution of 100 ng ml-1 of Zr and Mo was studied. These metal ions, with amounts injected of up to 10.0 mg ml-1, did not yield any chromatographic peak under the conditions cited in Table 1. Also, the presence of these metal ions in solutions of MoIV did not cause any noticeable deviation (less than 10%) in the peak area for the MoVL chelates. The same effect was observed with Zr, except for Nb and Ta, which reduced the peak area for the ZrIV chelates.The tolerance level for NbV or TaV was 1.0 mg ml-1. Although 10.0 mg ml-1 of Vv did not yield any chromatographic peak, it decreased the peak area for Zr chelates and increased that for Mo chelates. Tungsten(v1) yielded strong absorption signals, which overlapped those of Zriv and MoV1. The tolerance level for Fell1 was 1.0 mg ml-1. With higher concentrations, a broad chromatographic peak, which could not be resolved from those for Zr and Mo chelates, appeared. The addition of tartrate or ethylenediam- inetetraacetic acid (EDTA) to the sample solutions greatly decreased the peak area for Zr-Tiron. Hence, these two compounds were not suitable for use as masking agents to eliminate the interference from foreign ions.Another interesting phenomenon was also observed. As shown in Fig. 9, Zriv and MoVi chelates were well separated when these species were present at compatible concentra- tions. However, serious overlapping occurred when the concentration ratio was greater than 100, as shown in Fig. 10. Further investigations were performed to clarify the relation- ship betwen the resolution and the molar ratio of MoV1 and Z P . The results are shown in Fig. 11. An excellent resolution of 2.9 was attained when the Mo-to-Zr molar ratio was 1 : 1; however, the resolution dropped rapidly to 1.1 as the molar ratio increased to 100 : 1. The resolution eventually decreased to only 0.4, with a molar ratio of 200 : 1. Application In order to verify the applicability of the proposed chromato- graphic method, a relatively complex substance, nickel-base alloy (211W11221D), was analysed.This certified reference material is known to have the following composition (all values in mg g-1): Si, 0.40; Mn, 2.50; Cr, 143.0; Co, 12.6; Mo, 37.2; Nb, 32.3; Ti, 15.2; Al, 63.5; Cu, 0.60; Fe, 2.90; Zr, 1.10; Table 1 Regression analysis data and the detection limits for Tiron chelates of ZrIV and MeV' Parameter Zrl" chelate MeV' chelate Linear range (ppm) 0.52-41.2 0.48-51.7 Correlation coefficient 0.9998 0.9998 b* 8.8 x 10-3 -1.2 x 10-3 k* (l/slope) 6.6 x 10-4 4.4 x Detection limit? (ppb) 9.0 3.6 Sensitivity (slope) 1.51 x 103 2.29 x 103 * Amount of metal, ng = k x peak area + 6 ; b = -(intercept X k ) ; t Signal-to-noise ratio = 3. k = l/slope.526 t - m C 01 5 ANALYST, MAY 1993, VOL.118 a) t - m 01 5 Zr 0.02 12.52 0.01 12.51 0.02 12.52 Time/min Fig. 12 Typical chromatograms of MoV1-Tiron chelate and Z P - Tiron chelate in a nickel-base alloy, 211W11221D. (a) Blank solution; ( b ) MoV'-Tiron chelate; and ( c ) ZrIv-Tiron chelate. The operating conditions are as given in Fig. 9 ir (c) , 1 I 12.50 25.00 Ti me/m in 12.50 0 Table 2 Results of the analysis of 211W11221D nickel-base alloy Fig. 10 Typical chromatograms of different molar ratios of MoV1 : ZrIV. ( a ) Mo : Zr = 150 : 1 and ( b ) Mo : Zr = 200 : 1. Separation conditions as given in Fig. 9 Item Zr (a) Certified value- Concentration in solidlmg g-1 1.10 Amount in 100 pYng 220 ( b ) Determined by the standard additions method- Concentration in solid/mgg-' Amount in 100 pYng Concentration in solidlmg g-1 Amount in 100 pllng 1.05 * 0.04 210 f 8 1.02 * 0.03 204 k 6 (c) Determined by the calibration graph- Recovery* (%) 97 Mo 37.2 149 36.8 * 1.4 147 f 6 36.2 * 1.0 145 f 4 106 A I * The recovery was evaluated with the addition of 200 ng each of standard Zr or Mo to 100 pl of sample solution.l t respectively, and those determined by the standard additions method are 1.05 k 0.04 and 36.8 k 1.4 mg g-1 for Zr and Mo, respectively. Clearly, there is good agreement between the certified values and the experimental results. With five consecutive injections of 100 pl of sample solution containing 200 ng each of Zrtv and MeV', the recoveries obtained were 97 and 106% , respectively. The over-all analytical results for the nickel-base alloy are presented in Table 2.As mentioned in the preceding section, the tolerance level for Nb", TaV or Felrl was 1.0 mg ml-1, corresponding to 100 ng ml-1 of Zrvl and MoV1. Hence the tolerance levels for these metals could be up to 104 times the concentration of Zr and Mo. Therefore, there was no interference from FelI1, NbV or Fe"'. 0.4 0 40 80 120 160 200 Molar ratio of Mo : Zr Fig. 11 Resolution versus molar ratio of MoV'-Tiron chelate to ZriV-Tiron chelate and Ta, 2.60. As there was 30 times more Mo than Zr in the alloy, separation was poor, as expected. Fortunately, the appearance of an MoV1-Tiron peak could be effectively controlled by the presence of Tiron in the mobile phase, as mentioned earlier. Fig. 12 shows the chromatograms for a nickel-base alloy solution and for the corresponding blank solution, Fig.12(a). Fig. 12(6) is a typical chromatogram obtained with Tiron in the mobile phase; a peak for MoV1-Tiron is clearly visible. Without Tiron in the mobile phase, only the Zr peak can be detected, Fig. 12(c). The precision and accuracy of the proposed method were evaluated. The experimental values obtained from the calibra- tion graphs and by the standard additions method were compared with the certified values. The certified values for Zr and Mo in 211W11221D nickel-base alloy are 1.10 and 37.2 mg g-1 , respectively. Those determined from the calibration graphs are 1.02 k 0.03 and 36.2 k 1.0 mg g-1 for Zr and Mo, Conclusions The spectral characteristics and chromatographic behaviour of ZP-Tiron and MoV1-Tiron have been studied; Zr and Mo formed anionic chelates with Tiron.Therefore, these two species could be separated on a C18 column with Tiron and TBAB in the mobile phase. Although poor separation would result if the molar ratio of Mo to Zr was >lo0 : 1, these two chelates have been determined selectively with careful control of the amount of Tiron in the mobile phase. The proposed ion-pair reversed-phase HPLC method has been successfullyANALYST, MAY 1993, VOL. 118 527 applied to the determination of Zr and Mo in a certified reference material, viz., a nickel-base alloy. The financial support of this work by a grant from the National Science Council acknowledged. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Willeford, B. 61. of the Republic of China is gratefully References R., and Veening, H., J.Chromatogr., 1982,251, O’Laughlin, J. W., J. Liq. Chromatogr., 1984, 7, 127. Nickless, G., J. Chromatogr., 1985, 313, 129. Steinbrech, B., J. Liq. Chromatogr., 1987, 10, 1. Shofstahl, J. H., and Hardy, J. K., J. Chromatogr. Sci., 1990, 28,225. Nagaosa, Y., Kawabe, H., and Bond, A. M., Anal. Chem., 1991,63,28. Timerbaev, A. R., Petrukhin, 0. M., Alimarin, I. P. and Bol’shova, T. A., Talanta, 1991, 38, 467. Lin, C.-S., and Zhang, X.-S., Analyst, 1987, 112, 1659. Qiping, L., Huashan, Z., and Jieke, C., Talanta, 1991,38,669. Wenclawiak, B., Fresenius’ 2. Anal. Chem., 1981, 308, 120. Karcher, B. D., and Kruel, I. S., J. Chromatogr. Sci., 1987,25, 472. Bond, A . M., and Wallace, G. G., Anal. Chem., 1982,54,1706. Yu, J. J., and Wai, C. M., Anal. Chem., 1991, 63, 842. Huber, J. F. K., and Kraak, J. C., Anal. Chem., 1972,44,1554. Ruter, J., Fislage, U. P., and Neidhart, B., Chromatographia, 1984, 19, 62. 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Noffsinger, J. B., and Danielson, N. D., J. Liq. Chromatogr., 1986,9, 2165. Tsai, S. J., and Lee, Y., Analyst, 1991, 116,615. Okac, A., and Sommer, L., Anal. Chim. Acta, 1956, 15,345. Yoe, J. H., and Armstrong, A. R., Anal. Chem., 1947,19.100. Koch, S . , Ackerman, G., and Winkler, G., Talanta, 1979, 26, 821. Koch, S., Ackerman, G., and Scholze, V., Talanta, 1981, 28, 915. Tsai, S. J., and Hwang, H., J. Chin. Chem. SOC. (Taipei), 1989, 36, 187. Yoe, J. H., and Jones, A. L., Ind. Eng. Chem., 1944, 16, 111. Yamada, H., and Hattori, T., J. Chromatogr., 1985,320,403. Hoshino, H., and Yotsuyanagi, T., Anal. Chem., 1985,57,625. Connor, M., O’Shea, T., and Smyth, M. R., Anal. Chim. Acta, 1989,224, 65. Uehara, N., Kanbayashi, M., Hoshino, H., and Yotsuyanagi, T., Talanta, 1989,36, 1031. Zhang, X., Wang, M., and Cheng, J., Anal. Chim. Acta, 1990, 237, 311. Mulero, O., Nelson, D. A., Archer, V. S., Miknis, G., Beckett, J. R., and McLean, H. L., Talanta, 1990,37, 381. Hoshino, H., Nakano, K., and Yotsuyanagi, T., Analyst, 1990, 115, 133. Kaneko, E., Hoshino, H., Yotsuyanagi, T., Gunji, N., Sato, M., Kikuta, T., and Yuasa, M., Anal. Chem., 1991, 63,2219. Xu, X., Zhang, H., Zhang, C., and Cheng, J., Anal. Chem., 1991, 63, 2532. Paper 210541 7E Received October 9, 1992 Accepted December 3, I992

ISSN:0003-2654    

DOI:10.1039/AN9931800521

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 529 Separation of Chromium and Nickel Using Extraction Chromatography. Application to the Analysis of Alloys Ricardo 0. Crubellati and Ariel G. Ledesma Departamento Quimica Analitica, Comision Nacional de Energia Atomica, Libertador 8250, 1429 Buenos Aires, Argentina Extraction chromatography using Kel-F as support and acetylacetone (AA) as stationary phase has been employed to achieve the chemical separation of Cr and Ni. Critical studies were made on several factors affecting the distribution of each metal, such as the retention efficiency at 25 and 80 "C and the pH of the solution. It was observed that the concentration of AA in CHCl3 played an important role in the chromatographic procedure. Application of the procedure to the determination of Ni in alloys is reported.Keywords: Nickel; chromium; extraction chromatography; alloy Nickel is used extensively as an alloying element in ferrous and non-ferrous alloys. Generally it is accompanied by elements such as Fe, Cu, Co and Cr. The most outstanding character- istic of Ni is the ability to impart to steel increased hardness, strength, toughness and resistance to corrosion without sacrifice of ductility. It is clear that its determination is very important. Extrac- tion chromatography using Kel-F as support and tributyl phosphate (TBP) as stationary phase has been employed to achieve chemical separations.1.2 It is possible to elute Nil1 with HCI as mobile phase, but in the same fraction other elements would be present and it would be impossible to carry out a correct complexometric titration.However, these elements may be separated using ion-exchange chromatography.3.4 Nickel(i1) and Crrll have similar chemical behaviour, especially with regard to their extraction coefficients with several ligands and organic solvents.5 In this paper, the separation of these two analytes by extraction chromatography with acetyl- acetone (AA) as extractant is reported. Experimental Reagents Kel-F (a trade name of polytrifluorochloroethylene) mould- ingpowder. Purchased from 3M, St Paul, MN, USA; 45-80 mesh powder was used. Ion-exchange resin. The strongly basic anion-exchanger Dowex 1-X8 (chloride form; 100-200 mesh) was used for the separation experiments. Standard solutions. Exactly weighed amounts of Fell1, Co", Cu'l, Nil' and Crlll as chlorides were dissolved in 1 moll-' HCI to give solutions containing about 1 mg ml-1 of the element.The concentration of each solution was standardized by titration with ethylenediaminetetraacetic acid (EDTA). Acetylacetone, TBP and CHC13 were all of a chemically pure grade. Reagent-grade EDTA disodium salt was used without further purification. Column Preparation Glass chromatographic columns, 1.3 cm in diameter and 30 cm long were used. The first column consisted of two beds. The lower bed was filled with Dowex 1-X8 (7.0 g). The resin was poured in small aliquots into the chromatographic column containing water. The bed was gently pressed with a glass rod after each addition until a bed length of about 10 cm was obtained. A glass wool plug was used to separate the beds.The upper bed was filled with Kel-F-TBP.Kel-F powder (4.5 g) was placed in an Erlenmeyer flask fitted with a stopper, and then 15 ml of TBP were added. The stoppered flask was shaken gently by hand and was then left to stand for 20 min. The slurry was then poured into the column taking care that the bed was homogeneous. Its length was also about 10 cm. A small glass wool plug was again added. About 25 ml of 8 mol 1-1 HCI were passed through the column to condition it and to remove any contaminant. To prepare the second column, Kel-F powder (9.0 g) was placed in an Erlenmeyer flask fitted with a stopper, and then 30 ml of AA-CHCI3 (1 + 2) were added. The stoppered flask was shaken gently by hand and then left to stand for 60 min. The slurry was then poured into the column, again taking care that the bed was homogeneous.The Kel-F-AA-CHCI3 bed length was about 20 cm and was sealed with a small glass wool plug. About 25 ml of 0.5 moll-' NH40Ac solution, adjusted to a pH of 5.0, were passed through the column to condition it and to remove any contaminant. Procedure About 0.4 g of the material (exactly weighed) was transferred into a 150 ml Teflon beaker and dissolved in 10 ml of a mixture of HCl-HN03-HzO (1 + 1 + 1). The mixture was evaporated until oxides of nitrogen had been completely removed. With some non-ferrous alloys it was necessary to add a few drops of HF to obtain complete dissolution. Once dissolved, the solution was cooled and made up to 25 ml with water. An exactly measured portion, containing between 0.1 and 2 mg of Ni, was transferred into a 150 ml glass beaker and evaporated to near dryness. The residue was dissolved in 2-3 ml of 8 mol 1-1 HCI.The solution was passed through the first chromatographic column at a rate of about 0.5 ml min-1, followed by about 40 ml of the eluting solution (8 moll-1 HCI) at a flow rate of about 0.8 ml min-1. The effluent (about 30 ml) was collected. The Fell', Co" and CuL1 ions were retained on the column while Nil1 and C P passed through. The retained elements can be eluted from the column with 4 mol 1-1 HCI (Co), 2.5 moll-1 HC1 (Cu) and 0.1 mol 1-1 HCI (Fe).6 The effluent, containing Ni and Cr, was evaporated to a syrupy consistency. This residue was dissolved in 50 ml of water, and the pH adjusted to 5.0 with CH3C02H or ammonia solution.A 3 ml aliquot of AA was added, and the solution was heated gently for 90 min. After cooling the solution 20 rnl of 2 moll-' NH40Ac were added and the pH of the solution was re-adjusted to 5.0. The solution was then passed through the second chromatographic column at a rate of about 0.5 ml min-1. The eluting solution (30 ml of 0.5 moll-' NH~OAC, adjusted to a pH of 5.0) was then passed through at a flow rate of about 1.0 ml min-1. The effluent (about 100 ml) was collected. Chromium was retained and could be eluted from the column with 40 ml of ethanol. The total separation scheme is shown in Fig. 1.530 ANALYST, MAY 1993, VOL. 118 1- Sample (HCI: 8 mol 1-1) B HCI:4 mol I-' Ni + Cr: (pH = 5.01 HCI:2.5 mol 1-1 HCI : 0.1 mol 1-1 C NH~OAC: 0.5 mol 1-1 (pH = 5.0) Fe t Ni Cr 1""" Fig.1 Scheme of the separation. A, Kel-F-TBP; B, anion-exchange resin; and C, Kel-F-AA-CHC13 About 7 ml of HN03 were added to the effluent containing Ni and the solution was evaporated to near dryness. The residue was dissolved in 10 ml of water. About 3 ml of HC104 were then added and the solution was heated to near dryness. The residue was dissolved in 150 ml of water and the pH was adjusted to 9.0 with ammonia solution. The solution was titrated with 0.01 moll-1 EDTA solution using murexide as a metallochromic indicator. Results and Discussion Chromatographic Behaviour of Metal Ions The first column combines the advantages of two powerful techniques: extraction chromatography and ion-exchange chromatography.This gives a better separation of Fe, Co and Cu, as Cu and Co are not retained in the upper part of the column (extraction chromatography) while Fe is retained. Nickel and Cr are present in the same fraction and the purpose of the second column is their separation. Nearly all known ligands form equally stable complexes with these two elements making it very difficult to find a suitable extractant to achieve the separation. Acetylacetone forms complexes with Crfi17,8 and with Ni"9 that may be charged or neutral. A negatively charged complex would not be extracted. To achieve Ni extraction, it would be necessary to add a neutral non-hydrogen-bonding complexing agent which would give a neutral complex.'* The Cr complex is more stable than the Ni complex and hence, by varying the pH and the type of anion present in the solution, it would be possible to extract the chromium complex but not the nickel complex. Note that the objective is to maintain the Cr as a neutral complex and the Ni as a charged complex.The behaviour of divalent elements with P-diketones is well known.llJ2 If a divalent metallic cation, M2+, is present in a large excess of acetate ions in a buffer solution, competition may occur between the extractant and the acetate for the metal ion. The pH plays a very important role in the extraction and the extraction coefficient (D) depends on both the chelating agent and the hydrogen ion concentration. It is necessary, for our study, to find a pH value where D is sufficiently low to avoid extraction of the Ni complex.For the Cr complex, with a trivalent metallic ion, the situation is Table 1 Distribution of metal acetylacetonates Retention (%) pH TI0C 25 80 25 80 25 80 25 80 25 80 25 80 2 3 4 5 6 7 Cr Ni 0 0 60.0 0 0 0 68.8 0 0 0.7 81 .O 0.7 0 0.8 99.8 0.8 0 25.1 99.8 25.0 0 33.0 99.7 32.9 Table 2 Analytical data for Ni in certified samples Reference material BCS-CRM* 335 BCS-CRM 337 BCS-CRM 334 BCS-CRM 342 NIST SRM? 867 NIST SRM 865 Type of alloy Austenitic Austenitic Austenitic Ferritic Incoloy, 825 Inconel, 625 Stainless Steel Stainless Steel Stainless Steel Stainless Steel Certified (%) 9.47 9.52 20.6 2.16 45.3 59.5 Found (70 1 9.50 9.48 20.8 2.19 43.8 59.8 * BCS-CRM: British Chemical Standard Certified Reference 1- NIST SRM: National Institute of Standards and Technology Material.Standard Reference Material. different and it is possible to find a pH value and a chelating agent concentration where D is favourable for the Cr complex and unfavourable for the Ni complex. The kinetics of complex formation must also be considered. It is well known that the extraction of CrlI1 from aqueous solutions with AA is very slow but the rate can be enhanced by heating the solution prior to the extraction.13-15 The same effect occurs in extraction chromatography. For Ni-AA complexes no problem occurs with the kinetic aspect. Generally, the behaviour of each metal ion in reversed- phase chromatography is in good agreement with that predicted from the distribution ratio in the batch extraction. Several tests were carried out to verify this.To measure the retention of each metal ion in a column, about a 50 ml aliquot of a solution of the metal ion was passed through the column and then eluted with the corresponding eluting solution (about 50 mi). The metal retained in the column was then removed with a suitable eluting solution or by washing the stationary phase with ethanol. The retention (%) was calculated from the amounts of the metal ion in the column and in the effluent. The results are shown in Table 1. After heating the solution in the presence of AA prior to the chromatography, CrlIl has almost constant retention over a wide range of pH values, whereas Nil1 has approximately zero retention at pH 4 . 0 and the retention increases above this value. Therefore, the pH is critical in achieving a good separation between Cr and Ni.Atomic absorption spectrometry was used to evaluate Cr and Ni retentions. The concentration of AA in CHC13 is known to play an important role in the chromatographic procedure; the opti- mum concentration was 1 : 2. The Ni contents of some ferrous and non-ferrous alloys were determined by the above method. These results are listed in Table 2. The data show satisfactory agreement with the certified values. A study of the Cr : Ni ratio was also made to find the optimum value for separation and determinatM. It isANALYST, MAY 1993, VOL. 118 531 possible to work with a ratio up to 50 : 1 without any mutual interference. In conclusion, all the procedures, including separation with AA and titration with EDTA, can be used successfully with inorganic samples.Titration with EDTA was used to evaluate Ni in preference to other techniques, such as atomic absorp- tion spectrometry, because good precision and accuracy were obtained, especially for samples with higher Ni concentra- tions. Chromatographic separation methods are emphasized because of their ability to achieve clean separations on complicated samples. These methods can replace other tedious conventional methods, such as gravimetry, with improved accuracy and precision. References 1 Cerrai, E., and Testa, C., J . Chromatogr., 1962, 9, 216. 2 O’Laughlin, J. W., and Banks, Ch. V., Anal. Chem., 1964,36, 1222. 3 Kraus, K., and Moore, G., J . Phys. Chem., 1954, 58, 11. 4 Fritz, J. S., Pure Appl. Chem., 1977,49, 1547. 5 Strelow, F. W. E., and Eloff, C., Anal. Chem., 1971, 43, 870. 6 Kraus, K., and Moore, G., J. Am. Chem. SOC., 1953,75,1460. 7 McKaveney, J. P., and Freiser, H., Anal. Chem., 1958, 30, 1965. 8 Honjo, T., Honda, Y., Matsumoto, T., Honda, R., and Kiba, T., Bull. Chem. SOC. Jpn., 1977,50, 3051. 9 Beinrohr, E., Analyst, 1985, 110, 1317. 10 Headridge, J. B . , and Richardson, J . , Analyst, 1969, 94, 968. 11 Koshimura, H., and Okubo, T., Polyhedron, 1983, 2, 645. 12 Akaza, I., Bull. Chem. SOC. Jpn., 1966,39,971. 13 Stary, J., and Hladsky, E., Anal. Chim. Acta, 1963, 28, 227. 14 Sekine, T . , and Inagaki, H., J. Inorg. Nucl. Chem., 1980, 42, 115. 15 Kido, H., Bull. Chem. SOC. Jpn., 1980,53, 82. Paper 2104058A Received July 29, 1992 Accepted October 14, 1992

ISSN:0003-2654    

DOI:10.1039/AN9931800529

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 533 Determination of Aluminium in Different Tissues of the Rat by Atomic Absorption Spectrometry With Electrothermal Atomization Aleksandar Radunovic and Michael W. 6. Bradbury Physiology Group, Biomedical Sciences Division, King’s College London, Strand, London, UK WC2R 2LS H. Trevor Delves Clinical Biochemistry, Southampton General Hospital, Tremona Road, Southampton, UK SO9 4XY Atomic absorption spectrometry with electrothermal atomization was used for the determination of aluminium in brain, liver, spleen, kidney cortex, skeletal muscle and bone of the rat following digestion by nitric acid and in serum following simple dilution and in situoxygen ashing. The method of standard additions in the presence of a chemical modifier, ammonium dihydrogenphosphate, was essential for bone tissues.The detection limits ranged from 3 to 58 ng per gram of wet mass of tissue and were 4-19 times lower than the observed physiological levels of aluminium. The between-day precision for serum was 8.9% at a mean concentration of 6.8 pg 1-1 and 2.4% at a mean concentration of 125.3 pg I-’. Additionally, repeated analyses of National Institute of Standards and Technology Standard Reference Material 1577b Bovine Liver gave a relative standard deviation of 12.2% (mean concentration = 0.8 pg 9-1). Of the tissues studied, bone had at least ten times higher levels of aluminium than others (0.959 k 0.322 pg g-1). The aluminium concentration in cerebellum (0.073 k 0.043 pg g-1) was approximately twice that in the cerebral hemisphere (0.034 f 0.009 Keywords: Aluminium determination; rat tissue; electrothermal atomic absorption spectrometry; L’vov platform; chemical modification c1g 9-l). Aluminium is known to be toxic to the central nervous system and its role in causing dialysis encephalopathy is well established.1 Other aluminium-related clinical disorders that can occur in patients undergoing long-term haemodialysis include dialysis osteodystrophy2 and microcytic hypochromic anaemia unrelated to iron deficiency.3 Hence, monitoring aluminium concentrations in patients who are at potential risk of aluminium toxicity is of major importance. In addition, a possible link between aluminium and Alzheimer’s disease and other neurodegenerative disorders4 suggests the importance of studying the mechanism of aluminium transport across the blood-brain barrier in an experimental animal. Electrothermal atomic absorption spectrometry (ETAAS) is the method of choice for the determination of aluminium in clinical specimens.5 It is also currently the most important analytical method applied in research studies, as the only usable aluminium radioactive isotope, 26AI with a half-life of 7.2 x 105 years, is difficult to obtain and extremely expensive.Electrothermal atomic absorption spectrometry is a reliable and sensitive method, and it is also claimed that atomization from a L’vov platform with chemical modification and Zeeman-effect background correction should be able to overcome all matrix interferences and thus permit aqueous calibration graphs to be used for analysis.6 However, there are different reports on sample pre-treatment ,’-lo the use of various chemical modifiers,11.1* altering the time and temper- ature settings of the furnace programmesl3.14 and the use of an aqueous calibration graph’s or the method of standard additions16J7 for analysis.As a result, published reference intervals for aluminium in tissues vary widely, indicating the difficulties of measuring this element, and the optimum procedure for its determination is not apparent. For tissues, a digestion or ashing procedure is necessary and also the concentration of potentially interfering substances is different from that in serum and may be much greater, e.g., in bone. We report here a study of the determination of aluminium in different tissues of the rat using an ETAAS method that is simple, sensitive, accurate and includes minimum sample pre-treatment to avoid contamination.Experimental Instrumentation A Model 5000 atomic absorption spectrometer was used, equipped with a Zeeman-effect background corrector, a Model 500 heated graphite atomizer (programmable), an AS40 autosampler and a Model 10 data system (all from Perkin-Elmer, Nonvalk, CT, USA). Pyrolytic graphite coated graphite tubes with L’vov platforms were used. Prior to the read stage, the baseline was adjusted (-4 s) and the recorder switched on (-2 and -1 s for methods 1 and 2, respectively). The instrument settings and furnace programmes are given in Table 1. Reagents Concentrated nitric acid (AnalaR), concentrated Triton X-100, ammonium dihydrogenphosphate ( AnalaR) and alu- minium nitrate standard solution for atomic absorption spectrometry (SpectrosoL grade) were purchased from BDH (Poole, Dorset, UK).Chelex-100 chelating resin (analytical- reagent grade) was obtained from Bio-Rad Laboratories (Richmond, CA, USA). Quality control (QC) bovine serum samples were obtained from The Robens Institute (University of Surrey, Guildford, Surrey, UK). Low- and high-QC samples were designed to have aluminium concentrations in the ranges 3.9-11.1 and 106.6-130.0 pg 1-1, respectively. Standard Reference Material (SRM) 1577b Bovine Liver was obtained from National Institute of Standards and Technology (NIST; Gaithersburg, MD, USA). Methods Standard solutions Working standard solutions were prepared by appropriate dilution of 1 mg ml-1 aluminium nitrate stock standard solution for atomic absorption spectrometry with de-ionized water.A linear calibration graph was obtained using 0,20,40,534 ANALYST, MAY 1993, VOL. 118 Table 1 Instrument settings and furnace programmes for determining aluminium by ETAAS Furnace Time/s Internal gas temperature/ flow rate/ Step "C Ramp Hold mlmin-l Method I - 1 2 3 4 5 6 7 8 9 120 140 550 800 1200 20 2650 2700 20 1 40 1 30 1 1 0 1 1 40 1 20 1 20 20 4 3 15 300 300 300 300 10 0 50 300 50* Method 2- 1 250 20 10 300 2 1200 1 20 300 3 20 1 20 10 4 2650 0 3 0 5 2700 1 3 50 6 20 1 15 300 Wavelength 309.3 nm Rollover 0.30 A Spectral bandwidth 0.7 nm Integration time 3.0 s Signal mode peak area Standard gas argon Lamp current 20 mA Sample volume 10 p1 * Alternative gas: oxygen.60, 80, 100 and 120 pg 1-1 working standard solutions of Al. Trace element-free ammonium phosphate was obtained by passing it through a Chelex-100 column. De-ionized water was obtained by passing distilled water through a Millipore Milli-Q water purification system (Millipore; Milford, MA, USA). All glass and plasticware were soaked overnight in 10% nitric acid and rinsed four times with de-ionized water before use. A solution of 0.1% Triton X-100 in 0.5% nitric acid was used to rinse the exterior of the autosampler probe. Samples Female Wistar rats, body mass 200-230 g, were anaesthetized by intraperitoneal injection of a combination of 0.6 ml kg-1 of FentanyVFluanisone (Hypnorm) (Janssen Pharmaceutical, Grove, Oxford, UK) and 1.5 mg kg-1 of diazepam (Valium) (Roche Products, Welwyn Garden City, Hertfordshire, UK).Neither of the anaesthetic preparations contained aluminium. Blood samples were collected by cannulating the abdominal aorta, after anaesthesia. The rats were then killed by decapitation and samples of 200-600 mg were removed and placed in pre-weighed, clear quartz conical flasks (25 ml) (TSL Quadrant, Harlow, Essex, UK), which were immediately re- weighed and covered with Parafilm M (American, Green- wich, CT, USA) and stored at -20 "C for subsequent analysis. Arterial blood was allowed to clot. After centrifugation at 4OOOg for 15 min, serum was separated and stored in aluminium-free polystyrene tubes (2 ml) (Teklab, Sacriston, Durham, UK) at -20°C. Digestion The following method was used to digest all tissue samples except bone.A 1 ml volume of concentrated nitric acid and 1 ml of de-ionized water were added to tissues already in clear quartz conical flasks. The flasks were electrically heated on a hot-plate set to a nominal temperature of 250 "C until dry. The digests were then diluted to 2 ml with de-ionized water and stored in aluminium-free polystyrene tubes at 4 "C. After this treatment, digestion was considered to be complete, even for the larger samples of cerebral hemispheres and liver where a few fine fat droplets were present. Fatty residues do not take up significant amounts of aluminium.16 Table 2 Comparison of slopes of calibration graphs (n = 3) Slope * SE/10-4 Sample Aslpg-l Aqueous 7.66 f 0.09 Serum 8.12 f 0.26 Liver 7.92 f 0.26 Spleen 8.05 f 0.23 Kidney cortex 7.65 f 0.18 Skeletal muscle 7.49 f 0.13 Brain 8.10 f 0.13 .B . . , % . . . . . . . . . . . . . . . . . . . . . . . ,?b.\> .\.. ..... . . -,J' ,.::. '. ' . --+&;--- ~ ~ . : ~ y - : ! . : : * : . . . ......... .:I.::<;.. ._. .;.; ............... ----. ...... . . . . . I . - 0.5 s ........................ If-.. - Time - Fig. 1 Absorbance signals for (a) 0, (b) 40, ( c ) 80 and ( d ) 120 pg I-' of aluminium standard solutions added to digested bone tissue and aqueous solution. A, Aqueous solution; A', background for aqueous solution; B, digested bone solution; and B' , background for digested bone solution Bone tissues were treated with 1.5 ml of nitric acid overnight. The following day, the bone digest solutions were electrically heated on a hot-plate set to a nominal temperature of 250 "C.Digestion was taken to the stage where the solution became clear yellow. The digests were then diluted to 10 ml with de-ionized water and stored at 4°C. Blanks were taken through the same digestion procedure. Serum, digested tissue samples, blanks and standards were diluted with an equal volume of a 1% solution of the chemical modifier, ammonium dihydrogenphosphate. If necessary, the sample digest volume was reduced or diluted with de-ionized water so as to fall within the linear range of the calibration graph (less than 0.3 absorbance). Serum and other tissue samples were analysed by method 1 and method 2, respectively, as described in Table 1. All analyses were performed in triplicate, including blank measurements to control contamination.The slopes of the calibration graphs were tested for statistical significance using the two-tailed t-test. Results are presented as means k standard deviations (s), unless stated otherwise. iANALYST, MAY 1993, VOL. 118 535 Results and Discussion Instrument Settings An oxygen atmosphere was used for efficient oxidation of serum samples.5 An internal gas flow of 50 ml min-1 of oxygen at 550 "C gives good elimination of carbonaceous residues within the graphite tube furnace. Argon was used as the internal gas for all other stages, including desorption of oxygen from the graphite. 0.075 I I Chemical Modifier Several workers have used Mg(N03)2 for chemical modifica- tion.11,12 However, Fellows18 has shown that NH4H2P04 is preferable to Mg(N03)2 for measuring aluminium in bone and sera by ETAAS.Therefore, we used 1% NH4H2P04 solution, which permits ashing of aluminium up to 1400 "C without any loss. In addition, it removes chloride from NaCl to form NH4CI, which is volatilized at low temperature.5 Calibration The linearity of the calibration graph was verified. The linear correlation coefficient ( r ) was in the range 0.9984.999. Detection Limit The detection limit (3s of blank measurements) ranged from 3 to 58 ng of aluminium per gram of wet mass of tissue samples. The lower detection limit was obtained with a smaller amount of tissue, e.g., kidney cortex, or when a greater dilution factor was used, e . g . , bone (see Table 3). The detection limit for serum varied on a day-to-day basis from 0.2 to 0.7 pg 1-1 for favourable and unfavourable days, respectively.Precision The between-day precision was monitored with low and high QC samples obtained from bovine serum. Our values for low- and high-QC samples of 6.8 and 125.3 pg 1-1, respectively, were well within the designed range. Relative standard deviations (RSDs) obtained for low- and high-QC samples were 8.9 and 2.4%, respectively, demonstrating acceptable precision for our studies. Additionally, the between-day precision was determined using NIST SRM 1577b Bovine Liver. Ten analyses of SRM 1577b over 5 d (usually two analyses per day) gave an RSD of 12.2% for a mean concentration of 0.8 pg g-1. 0.050 1 0.025 ul . 8 m g o v) fi 4 0.090 a) +- L m +- 0.060 0.030 0 20 40 60 80 100 120 Al urn i ni urn concentration/pg I - 1 Fig.2 Aqueous and digested bone tissue calibration graphs for aluminium with 1% NH4H2P04 chemical modifier performed at different dilutions, i.e., sample + standard + modifier of 1 + 1 + 4 or 1 + 1 + 6. A, Aqueous (1 + 1 + 4); B, bone (1 + 1 + 4); C, aqueous (1 + 1 + 6); and D, bone (1 + 1 + 6). * p <0.02, *** p <0.001 for difference between aqueous and digested bone calibration graph slopes in each instance Table 3 Aluminium concentration in different issues of the control rat plus associated detection limits (DL) (n = 8). Units are pg per gram of wet mass except for serum, pg 1-1 Tissue Serum Cerebral hemisphere Cerebellum Liver Spleen Kidney cortex Skeletal muscle Bone Ratio: Mean f s DL* mean/DL 5.700 f 1.950 0.034 f 0.009 0.073 +- 0.043 0,057 +_ 0.024 0.055 f 0.037 0.104 f 0,061 0.060 f 0.016 0.959 f 0.322 * DL data are 3 s of blank measurements.0.430 0.005 0.01 1 0.003 0.007 0.024 0.008 0.058 13.2 7.6 6.7 18.7 7.6 4.3 7.4 16.4 Effects of Sample Matrix Table 2 gives results for the mean slopes of the calibration graphs for different tissues, except bone, obtained by the method of standard additions. The mean slopes between aqueous solution and each tissue, except bone, are not significantly different. These results show that at least under our conditions the method of standard additions is not necessary for the determination of aluminium in serum or in diluted digests of liver, spleen, kidney cortex, skeletal muscle and brain. Analysis of Bone Samples Fig.1 shows absorbance signals obtained for aluminium by using the method of standard additions in preparing the calibration graphs for the aqueous solution and digested bone solution. The small background signals from the bone matrix are easily resolved from the analyte signal and present no problem. However, both the appearance times and peak times for aluminium atomized from the digested bone are delayed relative to the signals from the aqueous solution. This lack of temporal coincidence indicates that accurate analysis of bone samples requires either matrix-matched standards or the method of standard additions. The slopes of the calibration graphs for the aqueous solution and the digested bone solution differ significantly (Fig. 2). Fig. 2(b) shows the calibration graphs for aqueous and digested bone solutions at dilutions of 1 + 5.These slopes correspond to the absorbance data shown in Fig. 1. The difference between the slopes for the aqueous solution [slope k standard error (SE) = (5.33 f 0.11) x 10-41 and the digested bone solution [slope 2 SE = (4.37 k 0.05) x 10-41 is significant (p <0.02). Fig. 2(a) shows calibration graphs for aqueous and digested solutions of different bone at dilutions of 1 + 7. For these measurements increased amounts of ammonium dihy- drogenphosphate were used in the dilutions. Whereas the slope for the digested bone solution increased [slope k SE = (4.81 k 0.07) x 10-41, the slope for the aqueous solution decreased [slope f SE = (3.55 k 0.07) X 10-41 and their difference was even more significant (p <0.001).The slopes of the calibration graphs were very different for various dilutions536 ANALYST, MAY 1993, VOL. 118 of the bone digest. Such various dilutions are, of course, necessary for bone with very different aluminium contents, e.g., in experimental or in clinical conditions. Hence it follows that the method of standard additions is more practical for use than the method of matrix-matched standards. The analytical sensitivity of the described procedures permitted the detection of low levels of aluminium in all the tissues investigated. The detection limits which ranged from 3 ng per gram of wet mass for liver up to 58 ng per gram of wet mass for bone were 4-19 times lower than observed physio- logical levels of aluminium.The precision was sufficient to detect differences in aluminium content in different tissues and in two different regions of the brain (Table 3). Aluminium tends to accumulate in bone with at least ten times higher levels than those in other tissues. The approxi- mate two-fold increase in the aluminium content of the cerebellum over that in the cerebral hemisphere has been noted by Slanina et al. ,* although they reported a three times lower concentration for brain and bone. The latter discre- pancy can be explained by possible differences in aluminium concentrations of food diet and drinking water. In addition, various strains of animals may have different levels of aluminium. Van Ginkel et a1.16 used female Wistar rats in their study and found similar values of aluminium in brain to those found here.The fact that the aluminium concentration in cerebellum is higher than in other brain regions may be compared with the more rapid uptake of 59Fe into the cerebellum.19 It is possible that a similar mechanism is involved in the transport of both metals into the brain. The lack of a certified reference material made it difficult to obtain an independent assessment of the accuracy of the reported methods. However, our value for the non-certified NIST SRM 1577b Bovine Liver of 0.80 k 0.10 pg g-1 agrees well with earlier observations from this laboratory by Suchak,20 who found 0.95 k 0.10 and 1.2 k 0.07 pg g-1 for two separate studies of the related material NIST SRM 1577a. All of these values are much lower than the non-certified mean value of 3 pg g-1.It is possible that this higher value can be ascribed to contamination. The accuracy of the method for measuring aluminium in serum, developed for human sera by Fellows,'* has been demonstrated over many years in an external quality assess- ment programme (The Robens Institute). It should also be pointed out that the methods reported here minimized contamination, which was confirmed by the low blank values obtained. The lack of background during the atomization of the sample also favoured low detection limits. Conclusion The methods for determining aluminium in both serum and various body tissues have been successfully validated. The method of standard additions has been applied to serum and to tissues digested with nitric acid. Under our conditions, the slope of the calibration graph is the same for standards in water and in the presence of tissue digests, except for bone, where there are significant differences.Hence we have confirmed that the method of standard additions is not necessary for serum or soft tissue digests. For bone, it is essential to allow for interferences, preferably by using the method of standard additions. Our values for aluminium concentration in rat brain are within the range of other recent analyses of this tissue,sv16 i.e., 0.013-0.041 mg per kilogram of wet mass of cerebral cortex or hemisphere. Data presented by Van Ginkel et aZ.16 indicate that the values for aluminium in rat brain have fallen by two orders of magnitude since the mid-1960s. A similar reduction was shown for aluminium in serum by Stewart.21 It is likely that both sets of values have now reached a steady level of aluminium.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 References Alfrey, A. C., LeGendre, G. R . , andKaehny, W. B . , N . Engl. J . Med., 1976, 294, 184. Ott, S. M., Maloney, N. A . , Coburn, J . W., Alfrey. A. C . , and Sherrard, D . J . , N. Engl. J. Med., 1982, 307,709. McGonigle, R. J. S., and Parsons, V., Nephron, 1985, 39, 1. Editorial, Lancet, 1989, i, 82. Delves, H. T., Biochim. Clin., 1989, 13, 113. Slavin, W., J. Anal. At. Spectrom., 1986, 1, 281. D'Haese, P. E., Van de Vyver, F. L., de Wolff, F. A . , and De Broe, M. E . , Clin. Chem. (Winston-Salem) 1985, 31, 24. Slanina, P., Falkeborn, Y . , Frech, W., and Cedergren, A . , Food Chem. Toxicol., 1984.22, 391. Boumen, A. A . , Platenkamp. A . J., and Posma, F. D . , Ann. Clin. Biochem., 1986, 23,97. Pierson, K. B., and Evenson, M. A . , Anal. Chem., 1986, 58, 1774. Bettinelli, M., Baroni, U . , Fontana, F., and Poisetti, P., Analyst, 1985, 110, 19. Leung, F. Y., and Henderson, A . R., Clin. Chem. (Winston- Salem), 1982, 28, 2139. Redfield, D. A . , and Frech, W., J . Anal. At. Spectrom., 1989,4, 685. Hewitt, C. D., Winborne, K . , Margrey, D . , Nicholson, J. R. P . , Savory, M. G . , Savory, J . , and Wills, M. R . , Clin. Chem. (Winston-Salem), 1990, 36, 1466. Brown, S . , Bertholf, R. L., Wills, M. R . , and Savory, J., Clin. Chem. (Winston-Salem), 1984, 30, 1216. Van Ginkel, M. F., Van der Voet, G. B . , and de Wolff, F. A . , Clin. Chem. (Winston-Salem), 1990, 36, 658. Morris, C. M., Candy, J . M., Oakley, A . E., Taylor, G. A . , Mountfort, S., Bishop, H., Ward, M. C., Bloxham, C. A . , and Edwardson, J. A . , J. Neurol. Sci., 1989, 94, 295. Fellows, C. S., Ph.D. Thesis, University of Southampton, 1992. Ueda, F . , Raja, K. B . , Simpson, R. J . , Trowbridge, I . S., and Bradbury, M. W. B . , J . Neurochem., 1993, 60, 106. Suchak, B . , M.Phil. Thesis, University of Southampton, 1992. Stewart, W. K., in Aluminium in Food and the Environment, eds. Massey, R., and Taylor, D . , Royal Society of Chemistry, London, 1988, p. 6. Paper 2f04294K Received August 10, I992 Accepted November 12, 1992

ISSN:0003-2654    

DOI:10.1039/AN9931800533

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, MAY 1993, VOL. 118 537 Determination of Trace Amounts of Heavy Metals in Alumina by Electrothermal Atomic Absorption Spectrometry After Diethyldithiocarbamate Extraction and Matrix Precipitation Masataka Hiraide, Takaaki Uchida and Hiroshi Kawaguchi Department of Materials Science and Engineering, Nagoya University, Chikusa-ku, Nagoya 464, Japan Milligram amounts of alumina powder were decomposed with 4 rnol 1-1 sulfuric acid and the pH of the solution was adjusted to 3 with 4.5 rnol 1-1 ammonia solution. After adding sodium diethyldithiocarbamate (DDTC) and chloroform, the pH of the solution was further increased to 7 with 1 rnol 1-1 ammonia solution, while the solution was mixed vigorously. Impurities present in the alumina sample, such as iron, copper, lead and cadmium, were quantitatively extracted into the chloroform, leaving the aluminium matrix in the aqueous phase as a bulky hydroxide precipitate.The DDTC and chloroform should be added before the formation of aluminium hydroxide, otherwise significant amounts of the desired heavy metals are lost because of coprecipitation. The heavy metals in the chloroform layer were directly determined by electrothermal atomic absorption spectrometry. Keywords: Alumina; trace metal determination; chelate extraction with diethyldithiocarbamate; cop recip ita tion; electro t h e rm a I a tom ic absorption spectrometry High-purity aluminas have varied fields of application because of their outstanding mechanical, chemical, optical and thermal characteristics. The determination of impurities is essential in order to control their quality and to investigate the synergistic action and correlation of the impurities.Direct analysis by inductively coupled plasma atomic emission spectrometry (ICP-AES) after fusion' or acid decomposition2 of samples causes a decrease in signal intensity and an increase in the background. This makes determinations difficult for impurities at sub-ppm levels. Injection of alumina suspensions for electrothermal atomic absorption spectrometry (ETAAS)3 needs accurate back- ground correction and the tedious preparation of standard alumina suspensions. Fusion of alumina followed by copre- cipitation of impurities with zirconium hydroxide4 results in high blank values for iron and copper and an increase in the background in ICP-AES because of the presence of zirco- nium.Extraction of impurities with dithizone and ammonium pyrrolidin-1-yldithioformate (ammonium pyrrolidine dithio- carbamate)s requires large amounts of ammonium tartrate to mask the precipitation of aluminium, which can cause appreciable contamination. Although removal of the aluminium matrix as the hydroxide is probably the most convenient separation technique, serious losses of trace amounts of heavy metals could occur during precipitation. In this work, therefore, the matrix precipitation is combined with diethyldithiocarbamate (DDTC)-chloro- form extraction. Losses from coprecipitation are successfully overcome by adding both DDTC and chloroform before the formation of the matrix precipitate. The proposed method is simple and rapid and is applicable to the AAS determination of heavy metals at ng g-1 levels in high-purity aluminas.Experimental Apparatus and Reagents A Seiko I & E SAS-760 atomic absorption spectrometer equipped with an SAS-705V electrothermal atomizer was used for the determination of heavy metals under the following conditions: wavelengths, iron 248.3, copper 324.8, lead 217.0 and cadmium 228.8 nm; drying at 120 "C for 15 s; decomposition at 450 "C for 15 s; atomization at 2400 "C (iron), 2300 "C (copper, lead) and 2200 "C (cadmium) for 3 s; hollow cathode lamp current, 10 mA (except Cd, 5 mA); and argon gas flow rate, 2 1 min-1. A DDTC solution (10 mg ml- 1) was prepared by dissolving DDTC in water. Sulfuric acid and ammonia solution (Kata- yama Chemicals, Osaka, Japan) were of special grade for the determination of toxic heavy metals.Procedure Matrix precipitation Ammonia solution (1 and 4.5 mol 1-1) was added to 50 ml of sample solution (pH 3) containing 100 mg of aluminium, 500 ng of cadmium and 1500 ng each of iron(nr), copper(r1) and lead, to adjust the pH to different values. The aluminium hydroxide formed was separated by centrifugation at 1000 g for 10 min. The supernatant was analysed by ETAAS for the measurement of coprecipitation losses of trace heavy metals. D D TC extraction A 1 ml volume of DDTC solution was added to 50 ml of sample solution (pH 3) containing 500 ng of cadmium and 1500 ng each of iron(iri), copper(I1) and lead. The pH of the solution was adjusted to different values with 1 and 4.5 moll-' ammonia solution. After shaking mechanically with 5 ml of chloroform for 5 min, the aqueous phase was collected and analysed by ETAAS to obtain the percentage extraction. Matrix precipitation and DDTC extraction A sample solution (50 ml, pH 3) containing 100 mg of aluminium, 50 ng of cadmium and 150 ng each of iron(m), copper(I1) and lead was treated in different ways.For example, the pH of the sample was first adjusted to 7 with 1 and 4.5 rnol 1-1 ammonia solution to form aluminium hydroxide, then 1 ml of DDTC solution and 5 ml of chloroform were added and extraction was carried out by shaking for 5 min. The organic phase was separated and analysed by ETAAS for the heavy metals. Similarly, matrix precipitation and DDTC extraction were conducted for other samples, but the order of the addition of the reagents was changed.These procedures are shown schematically in Fig. 1. Recommended Procedure for the Analysis of Alumina Place a 250-350 mg alumina sample and 10 ml of 4 mol 1-1 sulfuric acid in a 20 ml pressure decomposition vessel and heat at 220 "C for 18 h to decompose the sample completely. After538 - --* for - ANALYST, MAY 1993, VOL. 118 Analysis of organic phase by ETAAS I Fig. 1 Matrix preci itation and DDTC-chloroform extraction by procedures 1 (-), 8 (------+) and 3 (e) 50v 0 4 6 8 ,i 4 6 8 PH Fig. 2 Coprecipitation of iron (0), copper (a), lead (A) and cadmium (B) with aluminium hydroxide cooling the solution to room temperature, transfer a one fifth aliquot into a Pyrex glass bottle (25 mm in diameter x 55 mm tall) and adjust the pH of the solution to about 3 with 4 mi of 4.5 moll-1 ammonia solution.Add 0.2 ml of DDTC solution and 1 ml of chloroform, then increase the pH to about 7 with 2.5 ml of 1 moll-1 ammonia solution while mixing vigorously with a vibrating mixer (at approximately 25 Hz). Allow the mixture to stand for 1 min to obtain the clear organic phase at the bottom of the bottle (the flocculent precipitate of aluminium hydroxide remains completely in the aqueous phase). Collect the chloroform with the aid of a pipette and transfer a 10 pl aliquot into the graphite tube for the determination of heavy metals by ETAAS. Construct calibration graphs by taking 5 ml volumes of solutions containing iron(m), copper(n), lead and cadmium and extracting the metals at pH 7 with 0.2 ml of DDTC solution and 1 ml of chloroform.Results and Discussion Losses of Heavy Metals Due to Matrix Precipitation Aluminium hydroxide was precipitated at different pH values from synthetic sample solutions containing aluminium and trace amounts of heavy metals, as described under Matrix precipitation. A bulky precipitate of aluminium hydroxide was formed at about pH 4, collecting significant amounts of iron, copper and lead as shown in fig. 2. All heavy metals of interest were virtually completely coprecipitated at pH 7-8. Therefore, it is difficult to use the matrix precipitation directly for the removal of aluminium. Combined Use of DDTC Extraction and Matrix Precipitation Chelate extraction with DDTC is the most typical extraction system, where DDTC forms stable chelates with various heavy metals but does not react with aluminium.@ Diethyldithio- carbamate is also inexpensive and readily available.The extraction behaviour in the absence of aluminium was first examined (see under DDTC extraction). As shown in Fig. 3, copper and lead are completely extracted into chloroform over the pH range 4-8. The simultaneous extraction of heavy -0 C 50 v) c i- i - i d 0 1 3 5 7 9 PH Fig. 3 DDTC-chloroform extraction of iron (0), copper (a), lead (A) and cadmium (B) Table 1 Extraction of heavy metals with 1 ml of DDTC solution and 5 ml of chloroform Heavy metals extracted into CHC13 (%) Sample Containing/ Procedure Procedure Procedure (50 ml, pH 3) ng l* 2’ 3* Iron(ii1) 150 45-55 28-35 93-97 Copper(i1) 150 56-89 5&79 97-106 Cadmium( 11) 50 82-97 73-88 99-108 Lead( 11) 150 68-73 63-67 100-108 Aluminium(Iri) 1 x 108 - - - * For explanation of procedures 1-3, see Fig.1. metals including iron and cadmium can be achieved at neutral pH, where aluminium is precipitated as the hydroxide. The synthetic sample solutions containing aluminium and trace heavy metals were treated in three different ways as shown in Fig. 1. The extraction recoveries are summarized in Table 1, which indicates that the important factor is the time when DDTC and chloroform are added to the samples. When heavy metals had been coprecipitated with aluminium hydrox- ide, quantitative extraction was not possible (see procedure 1). About 50% of the iron, 1045% of the copper and 30% of the lead were still retained in the precipitate even after shaking with DDTC-chloroform.Cadmium, however, was extracted in >80% yields, although most of the cadmium had been coprecipitated at pH 7. The addition of DDTC before the formation of the matrix precipitate was not effective in reducing the coprecipitation losses (see procedure 2). This indicates that both species of metal ions and DDTC complexes can be coprecipitated with aluminium hydroxide. Procedure 3 is the best approach, where both DDTC and chloroform are added to the sample solution before the matrix precipitation. The heavy metals are almost completely extracted into the chloroform, leaving the aluminium hydroxide in the aqueous phase. Analysis of Alumina Based on the above findings, a simple separation technique was developed (see the recommended procedure).Alumina samples were decomposed under pressure with sulfuric acid.* The pH of the solution was increased to about 3 because DDTC is unstable in strongly acidic solutions.6 Samples of commercial alumina powder (99.9% purity, particle size 2-3 pm) were analysed for iron, copper, lead and cadmium. Table 2 shows reasonable agreement (within 10%) of the results. The maximum deviations of the, calibration graphs (ng in 1 ml of chloroform) were 4 for iron, <2 for copper and (1 for lead and cadmium. The blank valuesANALYST, MAY 1993, VOL. 118 539 Table 2 Analysis of alumina powder (99.9% purity) Foundng Amount Aliquot Sample takenlmg taken Iron Copper Lead Cadmium y- Alumina 269 115 220 16 2 ND* 115 260 15 2 ND 324 115 580 28 1 ND 115 700 39 2 ND a-Alumina 283 115 150 13 21 ND 115t 360 52 57 ND * Not detected.t 200 ng of iron(ii1) and 40 ng each of copper(i1) and lead(I1) were added. Concentration in samplelng mg- 1 Iron Copper Lead Cadmium 4.1 0.30 0.04 - 4.8 0.28 0.04 - 9.0 0.43 0.02 - 10.8 0.60 0.03 - 2.7 0.23 0.37 - 2.8 0.21 0.30 - through the whole procedure were less than 5 ng for iron and zero for copper, lead and cadmium. Iron was found to be heterogeneously distributed in the y-alumina. The proposed separation technique could also be applicable to the determination of other impurities (e.g., titanium, chromium, manganese, cobalt, nickel, zinc) in alumina samples. References 1 Ishizuka, T., Uwamino, Y., Tsuge, A., and Kamiyanagi, T., Anal. Chim. Acta, 1984, 161, 285. 2 Morikawa, H., Iida, Y., Ishizuka, T., and Yokota, F., Bunseki Kagaku, 1986,35, 636. 3 Slovak, Z., and Docekal, B., Anal. Chim. Acta, 1981,129,263. 4 Harada, Y., and Kurata, N., Bunseki Kagaku, 1986,35,641. 5 Koch, 0. G., Mikrochim. Acta, 1958, I, 92. 6 Sandell, E. B., and Onishi, H., Photometric Determination of Traces of Metals: General Aspects (Part I), Wiley, New York, 1978, p. 812. 7 Minczewski, J., Chwastowska, J., and Dybczynski, R., Separa- tion and Preconcentration Methods in Inorganic Trace Analysis, Ellis Horwood, Chichester, 1982, p. 97. 8 Mizuike, A., Enrichment Techniques for Inorganic Trace Analysis, Springer, Berlin, 1983, p. 30. Paper 2105876 F Received November 3, 1992 Accepted December 14, 1992

ISSN:0003-2654    

DOI:10.1039/AN9931800537

出版商:RSC    

年代:1993

数据来源: RSC