摘要:

ANALYST, SEPTEMBER 1986, VOL. 111 1045 Spectrophotometric Study of the Iron(ll1) - Micellar Medium F. Hernandez Hernandez, J. Medina Escriche, R. Marin Saez and Analytical Chemistry, University College of Castellon, University Spain Morin Complex in a M. C. Roig Barreda of Valencia, 12004 Castellon de la Plana, A spectrophotometric study of the iron(ll1) - morin complex in the presence of several surfactants (non-ionic, cationic and anionic) at different concentrations is reported. The study was performed at pH 4, and all the surfactants used (except a non-ionic surfactant of the ethylene oxide - propylene oxide condensate group) solubilised the complex. However, maximum sensitisation could be obtained only by using non-ionic surfactants. A subsequent spectrophotometric study in the presence of the non-ionic surfactant Nemol K-I 030 (polyoxyethylenated nonylphenol) was carried out.The green - brown complex formed in micellar medium showed maximum absorption at 417 nm. Using a concentration of 8.95 x 1 0 - 6 ~ of Fe(lll), the optimum conditions were 1.18 x 10-4 M (0.004%) morin, 1.25 x 10-2 M (lo/o) Nemol K-1030 and pH 3.7, adjusted with acetic acid - acetate buffer. No variations in absorbance were observed between 15 and 35 "C. The stoicheiometry of the Fe(lll) - morin complex in the presence of 1% of Nemol K-1030 was calculated by the molar ratio, continuous variations and slope ratio methods. A ratio of Fe(lll) to morin of 1 : 4 was found in all instances. The mean conditional formation constant was found to be 3.90 x 102*.Using a morin concentration of O.O04%, the calibration graph was linear up to 1.4 p.p.m. of Fe(lll), the molar absorptivity of the complex being 6.33 x l o 4 I mol-'cm-1. The detection and quantification limits were found to be 0.0038 and 0.01 2 p.p.m. of Fe(lll), respectively. A study in the presence of different ions showed that the most important interferences are due to Zr(lV), AI(III), Ti(lV), and V(IV) and V(V). Keywords: Spectrophotometry; iron(ll1) - morin complex; surfactant Malatl.2 found that the colour reaction between tin(1V) and pyrocatechol violet is greatly sensitised by adding gelatin. Since then, a number of spectrophotometric methods have been proposed in which different surfactants are used in conjunction with metallochromic reagents.3 Thus, visible absorption methods for the determination of metal ions are improved by the use of suitable surfactants.One improvement that has been reported from some of these procedures is due almost solely to the solubilising ability of micellar systems. Some metal ions react with an appropriate chelometric indicator or ligand to form binary metal - chelate complexes that are not soluble in water. Hence the complex must be extracted into a suitable organic solvent prior to measurement of the absorbance. However, in some instances it has been reported that the addition of a surfactant to the aqueous system renders the metal complex water soluble. This is due to the formation of an aqueous micellar system that solubilises the metal - chelate complex.4 The metal complexes of 8-hydroxyquinoline75 thiazolylazonaphthol,6 thiazolylazo- dimethylaminophenol,7 dithizonates,8 xylidyl blue9 and pyridylazonaphthol1° have been thus solubilised.Non-ionic micellar systems are typically employed in these method^,^ although in some instances cationic surfactants can also be used as solubilising agents for water-insoluble complexes.3 The coloured complexes formed in micellar media are characterised by high molar absorptivities (sometimes greater than 105 1 mol-1 cm-1) and high stability over a wide pH range, and usually by a large bathochromic shift caused by the addition of surfactants to the binary complex formed in water.''-12 Usually the metal - chelate complexes formed in micellar systems are much more stable than those formed in the absence of micelles.4 Iron has been determined spectrophotometrically using various reagents, and in the presence of different surfactants.The determination of iron with the complexing agent 2-bromo-4,5-dihydroxyazobenzene-4'-sulphonate (BDAS), using the cationic surfactant cetylpyridinium chloride (CPC) has been pr0posed.1~ Using the complexing agent Eriochrome Cyanine R, iron has been-determined in the presence of the cationic surfactants cetyltrimethylammonium chloride (CTAC)14 and tridodecylethylammonium bromide ,I5 and in the presence of the non-ionic surfactant polyoxyethylene sorbinate lauryl ester (POESLE) .I6 Analogously, Chromazu- rol S has been used for the determination of iron using the cationic surfactant CTAC17 and zephiramine .18 In these instances, the surfactant produces a bathochromic shift in the absorption maximum and an increase in the molar absorptiv- ity.However, the determination of iron with morin has hardly been described. In 1966, a method for the spectrophotometric determination of iron(II1) by solvent extraction of the iron - morin chelate was proposed; the complex was extracted with isoamyl alcohol and the absorbance was measured at 500 or 600 nm at pH 4. The method was linear between 0.1 and 0.7 yg ml-1 of Fe(III).19 In the same year, Elbeih and Abou-Elnaga20 proposed a photometric determination of iron (111) after paper chromatographic separation; the absorbance of the morin complex was measured in an ethanolic medium at 433 nm and at an optimum pH of 5. Beer's law was obeyed in the range 0-1.5 pg ml-1.Paletskite and Finkelshteinaite21 studied the reaction of iron with morin and quercetin, and proposed a stoicheiometry of 1 : 2 for the iron - morin and 1 : 1 for the iron - quercetin complex. In this work, the influence of different kinds of surfactant on the absorbance of the iron - morin complex was studied. A subsequent spectrophotometric study of the iron - morin system in the presence of the non-ionic surfactant Nemol K-1030 (polyoxyethylenated nonylphenol) was made in order to find the optimum conditions for the determination of iron, and the analytical characteristics of the method. The stoi- cheiometry and conditional formation constant of the solubi- lised complex were determined by different methods. Experimental Apparatus Absorption measurements and spectra were obtained with a Shimadzu UV-240 spectrophotometer with automatic record- ing; 1.00-cm quartz cells were used.1046 0.4 Q 0.2 ANALYST, SEPTEMBER 1986, VOL.111 - - A Termotronic S-389 thermostat was used to control the temperature to k0.5 "C. pH was measured using a Crison Digilab 517 pH meter (pH k 0.001). A conventional stalag- mometer was used for surface tension measurements. Reagents Analytical-reagent grade chemicals were used and the water used to prepare solutions was distilled and de-ionised. Fe(lll) stock solution, 1000 p.p.m. Obtained by dissolving an ampoule of J. T. Baker iron standard solution (1.000 k 0.002 g of Fe; substrate FeC13) in 11 of water. Working solutions were prepared freshly by appropriate dilution of the stock solution with 1% HCl.Morin solution. Prepared by dissolving 0.1 g of the reagent (Merck) in 100 ml of absolute ethanol. Nernol K-1030 solution, 10% . Prepared by dissolving 10 g of surfactant (polyoxyethylenated nonylphenol; hydro- phobic - lipophilic balance, HLB = 14.1; Mass6 y Carol)22 in 100 ml of water. Buffer solution, p H 3.7. Prepared by mixing 100 ml of acetic acid (Panreac) with 16.406 g of anhydrous sodium acetate (Merck) and diluting to 1 1 with water. Procedure An aliquot of the Fe(II1) standard solution (containing up to 70 pg of metal) was transferred into a 50-ml beaker and 5 ml of Nemol K-1030 solution were added. The pH was adjusted to 3.7 by adding 5 ml of acetic acid - acetate buffer solution, then 2 ml of morin solution were added and the solution was diluted with water to 50 ml.It was mixed well and the absorbance was measured after 15 min at 417 nm at 25 "C. Under these conditions the absorbance was stable for at least 24 h. For each measurement a parallel reagent blank without Fe(II1) was run. Results and Discussion The absorbance of the Fe(II1) - morin complex in the presence of several surfactants at different concentrations (between 0.04 and 5%) was measured. The non-ionic surfac- tants used were Nemol K-1030 (polyoxyethylenated nonyl- phenol; Mass6 y Carol), Triton X-100 (polyoxyethylene p-tert-octylphenol; Panreac) , Genapol PF-20 (ethylene oxide - propylene oxide condensate; Hoechst) and Genapol SE-070 (polyoxyethylenated fatty alcohol; Hoechst) . The anionic surfactant Humectante OZB (linear alkylbenzenesul- phonate; Ciba-Geigy) and the cationic surfactants CPB (cetylpyridinium bromide; Sigma), CTAB (cetyltrimethylam- monium bromide; Merck) and zephiramine (tetradecyl- dimethylbenzylammonium chloride; Pharmaceuticals Inc.) were also used.Table 1. Influence of different surfactants on the Fe(II1) - morin complex. Fe, 0.5 p.p.m.; surfactant, 0.1% m/V; pH, 4; morin, 0.004%. Surfact ant hm,x./nm Absorbance Non-ionic: NemolK-1030 . . . . . . TritonX-100 . . . . . . GenapolSE-070 . . . . . . GenapolPF-20 . . . . . . HumectanteOZB . . . . . . CPB . . . . . . . . . . CTAB . . . . . . . . . . Zephiramine . . . . . . Anionic: Cationic: 417 0.549 417 0.535 417 0.560 - - 415 0.228 429 0.180 428 0.170 429 0.180 The study of the influence of surfactants was made at the recommended19 pH of 4.At this pH, the brownish black iron(II1) - morin complex was insoluble in aqueous media. At concentrations L 0.1%, all surfactants used solubilised the complex, except the non-ionic surfactant Genapol PF-20, with which the Fe(II1) - morin complex precipitated at all concen- trations tested. The absorbance of the solubilised complex was stable for at least for 2 h. Table 1 gives the absorbance values measured at the absorption maximum of the binary complex, using a surfactant concentration of 0.1%. It can be seen that maximum absorbances were obtained in the presence of non-ionic surfactants, the absorption maximum being 417 nm. The anionic surfactant Humectante OZB also solubilised the complex, the absorption maximum being 415 nm.The cationic surfactants tested, in addition to solubilising the complex, produced a bathochromic shift of 11-12 nm (from 417 nm to 428-429 nm) in the absorption maximum. The non-ionic surfactant Nemol IS-1030 was used in subsequent experiments in order to find the optimum con- ditions and analytical characteristics of the system. Effect of pH The influence of pH was studied over the range 1-10, adjusted by means of HC1 and NaOH solutions. Fig. 1 shows the relationship between absorbance and pH; it can be seen that the maximum absorbance is obtained at a pH between 2 and 4.5, the maximum absorption appearing at 417 nm. At pHs > 4.5, a decrease in absorbance was observed, accompanied by a bathochromic shift, whereas at pH 9 the absorption maximum occurred at 443 nm. On the other hand, the absorbance values were stable with time except at pH < 2, where the absorbances became unstable with time.Therefore, the optimum pH range was 2-4.5, where the absorbance was maximum and stable. In the optimum pH range a study with two different buffer solutions was made: a 1 M HC1 - 2 M potassium hydrogen phthalate buffer solution was chosen to adjust the pH between 2 and 4 and a 2 M acetic acid - 2 M sodium acetate buffer solution was used to adjust the pH in the range 3.4-4.5. The range of minimum error corresponded to the interval between pH 2.5 and 4, with maximum variations in the absorbance of about 5%. No differences in absorbance were observed when using the HCl - phthalate or acetic acid - acetate solution, so either buffer can be used to adjust the pH over the optimum range.A pH of 3.7, adjusted with acetic acid - acetate buffer solution as indicated in the general procedure, was chosen as the optimum value. n c . I I I I I 0 2 4 6 8 PH Fig. 1. Influence of pH on the absorbance of the Fe(II1) - morin system in the presence of 0.1% of Nemol K-1030. Fe, 0.5 p.p.m.; morin, 0.004%. A, h = maximum absorption for each pH (pH d 4.5, 417 nm; pH 5,429 nm; pH 643,435; pH 9,443 nm). B , h = 417 nmANALYST, SEPTEMBER 1986, VOL. 113 1047 Effect of Ethanol Concentration The influence of the concentration of ethanol on the absor- bance of the Fe(II1) - morin system in the presence of 0.1% Nemol K-1030 was studied. The maximum absorbance was stable up to ethanol concentrations of 10%; at higher ethanol concentrations the absorbance of the system gradually de- creased, probably because ethanol destroys the micelles.A concentration of 4% was chosen in subsequent experi- ments. This concentration was obtained by adding 2 ml of 0.1% morin solution in absolute ethanol, according to the general procedure. Effect of Morin Concentration The effect of the concentration of morin at fixed concen- trations of Fe(II1) (0.5 p.p.m.), ethanol (8%) and Nemol K-1030 (0.1%) was studied. The results obtained are shown in Fig. 2. It can be seen that the absorbance was maximum and stable for morin concentrations in the range 0.001-0.005%. This study at different concentrations of morin was repeated using a higher concentration of surfactant, in order to avoid the decrease in absorbance observed at high concentrations of morin.Fixed concentrations of Fe (0.5 p.p.m.), ethanol (8%) and surfactant (1%) were used and the pH was adjusted to 3.7 with acetic acid - acetate buffer. Fig. 2 shows that when the surfactant concentration was increased the absorbance was constant up to 0.02% of morin. At higher morin concen- trations a small decrease in absorbance was observed. Effect of Surfactant Concentration The effect of the concentration of Nemol K-1030 on the absorbance of the Fe(II1) - morin system was studied, main- taining fixed concentrations of Fe (0.5 p.p.m.) and morin (0.004%) and a pH of 3.7. Solubilisation of the Fe(II1) - morin complex was observed only at surfactant concentrations higher than 0.03%, which is 0.6 I 1 0-4 t b '11 0.2 Morin, % Fig.2. Influence of morin concentration on the absorbance of Fe(II1) - morin system in the presence of Nemol K-1030. Fe, 0.05 p.p.m.; pH, 3.7; h, 417 nm. A, 1% Nemol K-1030; B, 0.1% Nemol K-1030 Table 2. Stoicheiometry and conditional formation constant of the Fe(II1) - morin complex in a micellar medium (1% Nemol K-1030). pH, 3.7; A, 417 nm Method M : L Conditional constant Molar ratio: cM constant* . . . . . . 1 : 4.08 1.00 x 1023 cL constantt . . . . . . 1 : 3.94 1.20 x 1022 Continuous variations$ . , 1 : 4.00 5.16 x 1021 Slope ratio . . . . . . 1 : 3.92 - * cM = 8.95 X t cL = 5.91 x 10-5 M. $ Total concentration = 5.37 X 10-5 M. M. higher than the critical micellar concentration (CMC) of Nemol K-1030, calculated as indicated below.Thus, below the surfactant CMC, precipitation or turbidity was observed, which is in accordance with other w0rkers.~J3 At concen- trations between 0.03 and 0.1% , the absorbance gradually increased with increasing surfactant Concentration. For con- centrations higher than 0.1% the absorbance was maximum and stable over a wide range of surfactant concentrations (between 0.1 and 5%). A concentration of 1% mlV (1.25 X 10-2 M) was chosen for subsequent experiments. Determination of CMC In order to establish whether the solubilisation of the complex and the enhanced absorbance are due to a micellar phenome- non, it is necessary to know the surfactant concentration at which the micelles are formed (CMC). The CMC of Nemol K-1030 in water, buffered at pH 3.7 with acetic acid - acetate buffer, was 1.04 X 10-4 M (0.008%) , determined by surface tension measurements.The presence of 1.18 x 10-4 M morin slightly decreased the CMC to 0.75 X The decrease in the CMC in the presence of dyes seems to be caused by the formation of mixed micelles of dye ions and surfactant at well below the CMC.z4 In these instances, it is said that the dye induces micelle formation. M (0.006%). Effect of Temperature The effect of temperature on the absorbance of the Fe(II1) - morin complex in the presence of 1% Nemol K-1030 was studied. The study was performed at temperatures between 15 and 50 "C. No differences in the maximum absorption were observed when the temperature was varied between 15 and 35 "C. However, below 25 "C the development of the colour was slower and it was necessary to wait at least 30 min before measuring the absorbance.Above 25 "C, maximum absor- bance was obtained in 5 min. Moreover, at temperatures higher than 35 "C the absorbance gradually decreased with increasing temperature until it reached 50 "C, which was the maximum temperature employed. Determination of Stoicheiometry and Formation Constant The composition of the Fe(II1) - morin complex in the presence of 1% Nemol K-1030 was calculated by the molar ratio, continuous variations and slope ratio methods (Fig. 3). A molar ratio of Fe(II1) to morin of 1:4 was found by all methods. The values obtained for the stoicheiometry and conditional formation constant are given in Table 2. A mean value of 3.90 x 1022 for the conditional constant was obtained.The stoicheiometry of 1 : 4 does not coincide with the 1 : 2 proposed by Paletskite and Finkelshteinaite21 for the Fe(II1) - morin complex in the absence of surfactant. This difference can be explained by the micellar system producing a novel reaction medium in which the stoicheiometry can be affected.4 This is in concordance with the work of other authors, indicating that the micellar medium seems to favour elimina- tion of water (or hydroxide ligands) from the chelate giving rise to an increase in the coordination number of the cati0n.~5 Analytical Characteristics Linearity range The calibration graph showed that the complex obeys Beer's law up to 1.4 yg ml-1 of Fe(II1). The apparent molar absorptivity was calculated to be 6.33 X lo4 1 mol-1 cm-1 at 417 nm.ANALYST, SEPTEMBER 1986, VOL.111 1048 0.6 0.4 q 0.2 0 10 20 30 Morin- Fe I I I 0.2 0.4 Fe - morin ( b ) 0.8 \r I\ I \ 0.6 q 0.4 0.2 0 0.4 0.8 XFe Fig. 3. Determination of stoicheiometry and conditional formation constant of the Fe(II1) - morin complex in micellar medium (Nemol K-1030). Nemol K-1030, 1%; pH, 3.7; h, 417 nm. (a) Molar ratio method. A, cFe constant = 8.95 x 10-6 M; B, cmOrin constant = 1.18 x 10-4 M. (b) Continuous variations method for Fe to morin ratio. Total concentration, 5.37 x 10-5 M A Ringbom plot showed that the zone of minimum spectrophotometric error corresponded to the interval 0.2-0.8 p.p.m. of Fe(II1). Precision The relative standard deviation (s,), evaluated from eleven independent determinations of 0.02 p.p.m. of Fe(III), was 5.9%.When the concentration of metal was increased to 0.2, 0.6 and 1.2 p.p.m., s, was 0.54, 0.24% and O.l8%, respec- tively. Detection and quantification limits The detection limit, obtained from the sensitivity of the calibration graph and for 3sb [sb = standard deviation of a blank without Fe(III), n = 113, was found to be 0.0038 p.p.m. of Fe(II1). The quantification limit, obtained for 10sb, was 0.0128 p.p.m. of Fe(II1). Interferences The effect of 30 ions on the absorbance of a solution containing 0.2 p.p.m. of Fe(II1) was studied. The tolerance in Table 3. Tolerance to foreign ions. Fe, 0.2 p.p.m.; morin, 0.004%; Nemol K-1030, 1%; pH, 3.7; h, 417 nm a- . . HC03- F- . . N03- . . HP04*- so42- . . Si032- Al(II1) Ba(I1) . . .Be(II) . . Ca(I1) . . Cd(I1) Cr(II1) Cr(V1) Co(I1) Cu(I1) Hg(II) Mg(II) K(1) , .Mn(I1) Mn(VI1) Na(1) . . Ni(I1) . . Pb(I1) . . Sn(I1) . . Ti(1V) V(1V). , V(V) . . Zn(I1) Zr(1V) Limiting Ion concentration, p.p. m. . . . . . . > 100 . . . . . . > 100 . . . . . . 20 . . . . . . > 100 . . . . . . > 100 . . . . . . >lo0 20 . . . . . . 0.005 . . . . . . >loo . . . . . . 1 . . . . . . > 100 . . . . . . >loo . . . . . . 10 . . . . . . 20 . . . . . . 0.5 . . . . . . 0.2 . . . . . . >loo . . . . . . > 100 . . . . . . 20 . . . . . . > 100 . . . . . . 2 . . . . . . >loo . . . . . . 10 . . . . . . 1 . . . . . . 2 . . . . . . 0.02 . . . . . . 0.1 . . . . . . 0.1 . . . . . . 5 . . . . . . 0.006 . . . . . . Ratio, cion/cFe >500 >500 100 >500 >500 >500 100 0.025 >500 5 >500 >500 50 100 2.5 1 >500 >500 100 >500 10 >500 50 5 10 0.1 0.5 0.5 25 0.03 the measurements of absorbance was taken to be twice the relative standard deviation of the method (1.08%), i.e., an interference is tolerated if its effect on the absorbance signal is less than twice the relative standard deviation.Table 3 shows the limiting concentrations and ratios to Fe(II1) tolerated for the different ions studied. It can be considered that none of the anions studied produce interfer- ence. The high tolerance for F- (20 p.p.m.), which generally interferes in other methods for the determination of Fe(III), is emphasised. Al(III), Zr(1V) and Ti(IV) produce strong interferences and must be absent if Fe(II1) is to be deter- mined; their complexation with morin has been extensively described.26 V(IV), V(V), Cu(I1) and Cr(V1) can be tolerated at concentrations up to 0.1, 0.2 and 0.5 p.p.m.Be(I1) and Pb(I1) can be present in an excess of up to five times the Fe(II1) in the sample. References 1. 2. 3. Malat, M., Fresenius 2. Anal. Chem., 1962, 187, 404. Malat, M., Fresenius Z. Anal. Chem., 1964, 201,262. Ueno, K., “MTP International Review of Science, Physical Chemistry, Series 1, Volume 13, Analytical Chemistry Part 2, Organic Reagents”, 1973, pp. 43-69. 4. Hinze, W. L., in Mittal K. L., Editor, “Solution Chemistry of Surfactants”, Volume 1, Plenum Press, New York, 1979, pp. 79-127. 5. Shijo, Y., and Takeuchi, T., Bunseki Kagaku, 1967, 16,51. 6. Watanabe, H., and Matsunaga, H., Bunseki Kagaku, 1976,25, 35. 7. Ishii, H., and Watanabe, H., Bunseki Kagaku, 1977, 26, 86. 8. Watanabe, H., and Miura, J., Bunseki Kagaku, 1977,26,196. 9. Watanabe, H., and Tanaka, H., Bunseki Kagaku, 1977, 26, 635. 10. Watanabe, H., and Sakai, Y., Bunseki Kagaku, 1974,23,396. 11. Chernova, R. K., Zh. Anal. Khim., 1977,32, 1477. 12. Tikhonov, V. N., Zh. Anal. Khim., 1977,32, 1435.ANALYST, SEPTEMBER 1986, VOL. 111 1049 13. 14. 15. 16. 17. 18. 19. 20. 21. Wakamatsu, Y., and Otomo, M., Anal. Chim. Acta., 1975,79, 322. Shijo, Y . , and Takauchi, T., Bunseki Kagaku, 1971,20,908. Shijo, Y., Bull. Chem. SOC. Jpn., 1975,48, 2793. Shijo, Y., and Takeuchi, T., Bunseki Kagaku, 1971, 20, 987. Shijo, Y., and Takeuchi, T., Bunseki Kagaku, 1968,17, 1519. Horiachi, Y., and Nishida, H., Bunseki Kagaku, 1968,17,756. Kohara, H., Ueno, K., and Ishibashi, N., Bunseki Kagaku 1966, 15, 1252. Elbeih, I. I. M., and Abou-Elnaga, M. A., Chemist-Analyst, 1966, 55, 43. Paletskite, V., and Finkelshteinaite, M., Zh. Anal. Khim., 1969, 24, 1550. 22. 23. 24. 25. 26. Carrih, J. L., de la Guardia, M., and Medina, J., Quim. Anal., 1983,2, 271. Marczenko, Z., and Jarosz, M., Analyst, 1982, 107, 1431. Garcia Alonso, J. I., Diaz Garcia, M. E., and Sanz Medel, A,, Talanta, 1984,31, 361. Sanz Medel, A., and Garcia Alonso, J. I., Anal. Chim. Acta, 1984, 165, 159. Hernandez Hernandez, F., and Medina Escriche, J., Quim. Anal., 1986, 5 , 1. Paper A6124 Received January 23rd, 1986 Accepted March I9th, I986

ISSN:0003-2654    

DOI:10.1039/AN9861101045

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST SEPTEMBER 1986 VOL. 111 1051 Spectrophotometric and High-performance Liquid Chromatographic Determination of the Kinetics and Mechanisms of Hydrolysis, lsomerisation and Cyclisation of Both E and Z Isomers of 2-{ [(2-Amino-5-chlorophenyl)phenylmethylene]amino}acetamide Maurice Bernard Fleury" and Sabine Letellier Laboratoire de Chimie Analytique (UA-CNRS No. 484) UER des Sciences Pharmaceutiques et Biologiques 4 Avenue de I'Observatoire 75270 Paris Cedex 06 France and Jean-Pierre Porziemsky and Bernard Mompon Laboratoires d'Etudes et de Recherches Synthelabo 3 1 Avenue Paul Vaillant Couturier 92220 Bagneux, France The behaviour of 2-{ [(2-amino-5-chlorophenyl)phenylmethylene]amino}acetamide in aqueous solution was investigated in the pH range 1.0-8.0 with respect to the €or Zconfiguration around the imine bond.The € isomer yields kinetic data that obey the kinetic law for successive first-order reactions. The first step consists of the isometric transformation of the € to the Z isomer. The second step implies a parallel reaction mechanism i.e. either a hydrolysis yielding 2-amino-5-chlorobenzophenone or an intramolecular ring closure reaction yielding desmethyldiazepam. This parallel route constitutes a major reaction at pH values below the pK of the immonium - imine equilibrium as demonstrated by UV spectrophotometry and high perform a n ce I i q u id c h ro mat o g ra p h y . Keywords 2- {[(2-Amino-5-chlorophen yl)phen ylmeth ylene]amino}acetamide; E - Z isomerism; hydrolysis and c yclisa tion; high -perform an ce liquid c h ro ma tog rap h y; s pectro p h o to m e tr y The important role of the inhibitory neurotransmitter glycine in neuropsychiatric disorders is gaining increasing support as a result of the experimental evidence accumulated over the last few years.l One possible approach to the correction of dysfunctions due to a deficiency of this neurotransmitter could be to increase the glycine concentration in the central nervous system (CNS).This line of thinking has already been applied to another neurotransmitter GABA and has led to the development of the anti-epileptic drug progabide.2.3 The over-all lipophilicity of drugs in which glycinamide is reversibly attached to a very lipophilic benzophenone is sufficient to enable these mole-cules to pass the blood - brain barrier.Certain substitutions on the benzophenone moiety had to be envisaged such as the introduction of a functional group at the 2 position in order to improve the stability of the imine functional group.4 Kinetic studies concerning the hydrolysis of 2-methoxy and 2-hydroxy derivatives have been reported.5-8 The previous investigations indicate that (1) for the 2-methoxy compound, the hydrolysis rate follows a kinetic law of the type v = k,,,[substrate] with kexp = kH+[H20][H+] where kexp is the experimental rate constant and kH+ is the H+-catalysed reaction rate constant; (2) for the various substituted 2-hydroxy compounds the kinetic law is of the above type in acidic media whereas in neutral solutions the predominant reaction is the addition of a water molecule to the quinonoid tautomer; (3) in basic media (pH 2 lo) the hydrolysis of 2-methoxy derivatives is uncatalysed and proceeds very slowly and the 2-hydroxy derivatives are rapidly hydrolysed according to a kinetic law of the type v = kexp[substrate] with kexp = koH- [OH-] where koH- is the OH-catalysed reaction rate constant.Regardless of the acidity of the medium the rate of hydrolysis is controlled essentially by steric factors.8 The polarity of the imine bond also needs to be considered.' This work deals with the 2-amino derivatives A and B (Scheme 1). The hydrolysis rate of the azomethine bond yielding compound C was investigated in relation to the E or 2 configuration around the imine bond and the conformation of * To whom correspondence should be addressed.the molecules. In addition in acidic media the 2 isomer A is involved in an intramolecular ring-closure reaction yielding desmethyldiazepam D. Experimental Materials General procedures for the preparation of the benzylidene derivatives A and B have been previously reprted.4 The treatment of 2-amino-5-chlorobenzophenone C with ethanolamine at 135 "C gave 2-{ [ (2-amino-5-chloro-phenyl)phenylmethylene]amino}ethanol which was dis-solved in a solution of glycinamide hydrochloride in methanol, heated at 60 "C for 6 h and then poured into water. The product was extracted with chloroform and the extracts were washed with water dried and evaporated. The NMR spec-trum showed that this residue was a mixture of Z (A) and E (B) isomers (65 + 35).H 4' A Z isomer C R = H E R = COCHzNHz B E isomer H D Scheme 1052 ANALYST SEPTEMBER 1986 VOL. 111 The residue was washed with diethyl ether; the ether-insoluble part gave A (48%) by recrystallisation from methanol and the ether solution evaporated to dryness and recrystallised from ethanol gave B (28%). (2)-2- { [ (2-Amino-5-chlorophenyl)phenylmethylene]amino} acetamide ( A ) . Map. 180 “C (decomp.). Found C 62,55; H, 4.88; N 14.48; C1 12.45%. C15H14C1N30 requires C 62.61; H 4.90; N 14.60; C1 12.32%. ymax. (CHC13) 3510 and 3390 (NH,) 1690 (C=O) and 1625 cm-1 (C=N); A,,,. (MeOH) 300 (shoulder) 248 and 206 nm 2.25 x 103 ~ 2 4 8 2.40 x 104 and &206 3.86 x lo4); aH (200 MHz; CDC13) 3.56 (2H br s, 5.82 and 7.30 (lH br s and l H br s CONH2) 6.72 (lH d J 9 Hz 3-H) 6.87 (lH d J 2.5 Hz 6-H) 7.20 (lH dd J 9 and 2.5 Hz 4-H) 7.52-7.35 (3H m 3’-H 4’-H and 5’-H) and 7.72 p.p.m.(2H m 2’-H and 6’-H); bC (50.32 MHz; CDCI3) 56.4 (2C; C-2’ and C-6’) 128.8 (2C; C-3’ and C-5’) 130.3 (C-4), 131.5 (C-4’) 137.3 (C-1’) 141.4 (C-2) 167.6 (C-7) and 173.5 p.p.m. (C-10). (E)-2- { [ (2-Amino-5-chlorophenyl)phenylmethylene]amino} acetamide ( B ) . M.p. 143-144 “C. Found C 62.50; H 4.98; N, 14.64; C1 12.33%. C15H14C1N30 requires C 62.61; H 4.90; N 14.60; C1 12.32%. ymax. (CHC13) 3510 and 3390 (free NH2 of amide) 3470 (free NH of aniline) 3270 (bonded NH of aniline) 1690 (C=O) and 1610 cm-1 (C=N); A,,, (MeOH) 365 260 (shoulder) 232 and 204 nm (&365 6.15 X lo3 &26() 7 X lo3 ~ 2 3 2 2.88 X lo4 and ~ 2 0 4 2.9 x lo4); (200 MHz; CDC13) 3.99 (2H s 9-H) 6.40 and 6.49 (lH br s and l H br s, CONH2),6.56(2H,brs,2-NH2),6.67(1H,d,J8.8 Hz,3-H), 6.78 (1H J 2.4 Hz 6-H) 7.16-7.03 (3H m 2’-H 6’-H and 4-H) and 7.54-7.43 p.p.m.(3H m 3’-H 4’-H and 5’-H); 6~ 120.5 (C-5) 127.1 (2C; C-2‘ and C-6’) 129.2 (2C; C-3’ and 147.7 (C-2) 173.2 (C-10) and 174.3 p.p.m. (C-7). 2-NH2),3.96(1H d,J18H~,9-H),4.07(1H,d,J18H~,9-H), (C-9) 117.4 (C-3) 122.7 (C-1) 123.6 (C-5) 127.3 (C-6) 128.2 (50.32 MHz; CDC13) 57.0 (C-9) 118.1 (C-3) 120.5 (C-1), C-5’) 129.2 (C-4’) 131.4 (C-4) 132.7 (C-6) 135.7 (C-l’), Configuration and Conformation of A and B The IR and UV spectra of A and B were of particular interest. In the IR spectrum of A no absorption band characteristic of the bonded NH aniline group was observed in dilute solution, but B showed an absorption band at 3270 cm-1 that was indicative of an internal hydrogen bridge between one of the aniline hydrogens and the imine nitrogen.The UV absorption spectra of A and B were different; B exhibited a supplementary band at h = 365 nm (MeOH &365 6.15 X lO3) suggesting the existence of a tautomeric equilibrium in a neutral medium (HA The existence of the hydrogen bridge in B (Scheme 1) forces the side chain into an E configuration in which the imino substituent is anti to the substituted benzene ring and compound A occurs in the 2 configuration. The NMR spectra of A and B were also consistent with the proposed configura-tion. Hence the difference of 3 p.p.m. in the IH NMR signal of the NH2-aniline (3.56 in A compared with 6.56 p.p.m.in B) confirmed the hydrogen bonding in B and the absence of a hydrogen bridge in A. Changes in the shifts of aromatic protons and in the 13C chemical shifts gave information concerning the conjugative interactions between the phenyl rings and the imine bond. The two phenyl groups are connected to an sp2-hybridised carbon atom and the tendency of the phenyl ring towards co-planarity (a gain in delocalisation energy) is counteracted by steric interactions mainly between 6-H and 2’-H (6’-H) and between 2’-H (6’-H) and the methylene group of the side-chain. If one phenyl group assumes a position in the nodal plane of the imine bond the other will be moved further out of this plane. HB).5 This type of steric hindrance has been studied by means of molecular models spectroscopic methods (UV IR NMR, ESR) and quantum mechanical calcu1ations.g From these studies it follows as confirmed by our NMR data that in A, the most heavily ortho-substituted ring disposed cis with respect to the imine-substituted ring makes the largest angle with the nodal plane of the C=N bond.In compound B a conformational consequence of the existence of a hydrogen bridge and the tautomerism observed by UV is the co-planarity of the substituted ring with the imine double bond (Scheme 1). This explains the chemical shift differences of 2’-H (6’-H) in A and B at 7.72 and 7.16-7.03 p.p.m. respectively. The chemical shifts of C-4’ in A and B (131.5 and 129.2 p.p.m. respectively) reflect the twisting angle 8 of this ring with the imine bond and are free from steric compression.They may be estimated by the following approximationlo 4.74 cos28 = 6~ - 690” p.p.m. where 6y is the measured 4’-carbon atom chemical shift of A and B 690”is the 4’-carbon atom in a totally non-conjugated derivative where the imine double bond is reduced to a C-N single bond (6r = 127.9 p.p.m. for the dihydro derivatives of A and B), and 4.74 is an expression of the gain in p.p.m. of maximum overlap and is derived from various model compounds.11 The dihedral angles were found to be 57” for A and 28” for B. The 2-amino-5-chlorobenzophenone C was obtained commercially (Fluka) and desmethyldiazepam D was synthesised by the procedure of Sternbach et al. l2 Equipment lH and l3C NMR spectra were recorded on a Bruker WP 200 SY instrument with tetramethylsilane as an internal standard.Chemical shifts are reported in p.p.m. The following ab-breviations are used in describing the spectra s = singlet d = doublet t = triplet q = quadruplet m = multiplet br = unresolved broad signal. The IR spectra of KBr pellets were recorded using a Perkin-Elmer 399 spectrophotometer. UV -visible absorption spectral data were obtained using a Varian Superscan 3 spectrometer. A cell compartment was used between thermoplates throughout the kinetic study and the temperatures were maintained at 10 k 0.1 or 45 k 0.1 “C by a circulating water-bath. When UV scanning was necessary an automatic repetitive scanning device was used. The HPLC system consisted of a Rheodyne Model 7125 injection valve a Model 114 M solvent delivery module a 165 variable-wavelength detector and a 450 data sys-tem/controller all from Beckman Instruments.All the reagents were of analytical-reagent grade. The chromato-graphic solvents used for the separation were of HPLC grade. The mobile phase was a ternary mixture of phosphoric acid (0.05 M; pH 3.5 adjusted with ammonia solution - acetonitrile - methanol (55 + 30 + 15) and the flow-rate was 2.5 ml min-1. The separation was performed at ambient temperature using a 5-pm Spherisorb ODs-2 column (200 X 4.6 mm i.d.). The detector was set at 254 nm with a sensitivity of 0.05 a.u.f.s. The concentrations of A B C and D were determined by an external calibration with authentic reference compounds. The pK determination and pH measurements were carried out using a Tacussel T.S.70 N1 pH meter. Stock Solutions and Buffer Stock solutions (0.5 mM) of compounds A and B were prepared in methanol. All the reaction rate studies were performed in aqueous methanolic buffer (8 + 2) solutions. The buffer systems used were sulphuric acid pH d 2 citric acid and phosphoric acid (0.1 M) pH 2-8. The ionic strength of each was adjusted to 0.5 M with sodium chloride ANALYST SEPTEMBER 1986 VOL. 111 1053 Results and Discussion Spectral Changes and pK Determination The UV spectral characteristics of A and B in water containing 20% (VIV) of methanol are reported in Table 1. The spectral changes versus pH graphs for both compounds showed two successive steps. The first change in the pH of 0-2 did not enable us to obtain accurate quantitative information about the pKa value for the equilibrium [H3A2+ F HzA+ + H+], owing to the fast ring-closure reaction (see below).A second change was observed in the pH range 2-7 corresponding to the equilibrium [H2A+ HA + H+]. The main bands exhibited by A (2 isomer) are listed in Table 1 and shown in Fig. 1. With increasing pH a decrease of two of the absorption bands is observed at 275 and 390 nm. The changes in the spectra show an isosbestic point at 262 nm, indicating that a simple acid - base equilibrium is shifted. The value Of PKa was determined from graphs of 10g[(AH2A+ -A)@ - AHA)] versus pH at 275 nm where AH2A+ and AHA are the respective absorbances of the protonated and neutral species and A the absorbance of their mixture at a fixed pH: pH = PKa -k 10g[(AH2A+ - A)/(A -AHA)] * 9 (1) In the pH range 2-7 the band at 352 nm increases with increasing pH.The changes in the spectra show three isosbestic points at 242,318 and 378 nm indicating that an acid - base equilibrium is shifted. The pKa value was determined from graphs of log[(A - AH~A+)/(AHA - A)] versus pH: PH = PKa + log[(A - AH~A+)/(AHA - A)] - - (2) The measured pKa was apparent due to the occurrence of the tautometric equilibrium HA HB suggested by previous work dealing with { [ (2-hydroxy-5-chloropheny1)phenyl-methylene]amino}acetamide.5 As shown in Scheme 2 the hydrogen bond initiates the tautomeric equilibrium which generates the quinonoid form HB. The over-all equilibrium is quantitatively described by the following expressions: As reported previously,l3J4 a critical assumption made here was that only the quinonoid species HB may account for the The pKa was found to be 4.20 k 0.05 at 10 "C.At this temperature the hydrolysis and the ring-closure reactions (see below) are not fast enough to interfere with the pKa determination. The same approach was used for B ( E isomer). The main bands exhibited by B are listed in Table 1 and shown in Fig. 2. H CI &,* A Z isomer HZA' R = CHZ-CO-NHZ Scheme 2 0 HB B E isomer 1.2 a 6 0.8 e u) 2 0.4 n 500 300 400 Wavelengthhm 500 600 Fig. 1. Typical spectral changes due to increasing pH for Z isomer (A 1054 ANALYST SEPTEMBER 1986 VOL. 111 absorption band at 352 nm while the HA species no longer exhibits noticeable absorption at this wavelength.In agree-ment with this reasoning when the azomethine bond and the p-chloroaniline ring are not co-planar in the 2 isomer A the UV - visible absorption spectrum no longer exhibits the absorption band at 352 nm (Table 1). Hence it can be deduced that this absorption band is linked to the quinonoid resonance. In order to be able to calculate K, K and K', UV - visible absorption spectra were recorded in various solvents. As shown in Table 2 the absorbance in the wavelength range 352-365 nm increases with decreasing polarity of the solvent. If it is assumed that the wavelength shift from 365 nm (in diethyl ether) to 352 nm (in water - methanol 8 + 2) is due to the solvent effect and if one assumes that the molar absorbances are not changed markedly by shifts from aprotic to protic solvents then the percentage of quinonoid Table 1.UV spectra of A B C and D [A,,,, E,,,. in buffered water -methanol (8 + 2) solution] Compound Species A,Zisomer . . H2A+ HA B,Eisomer . . H2A+ HA * HB C . . . . . . H2A+ HA D . . . . . . H*A+ HA hlnm 240 275 390-450 243 290(sh) 240 270(sh) 390-450 232 260 (sh) 352 258 236 258(sh) 385 236 282 360 230 250(sh) 310 E/I mol-1 cm- 1 x 10' 21.25 10.0 2.75 21 .o 3.5 18.6 10.6 27.5 2.75 7.75 4.25 13.2 23.6 12.0 4.8 30.0 13.5 4.0 35.0 17.0 2.0 200 species HB under our experimental conditions can be deduced from the experimental ratio found for A,,,.(Table 2) i.e., 4.2516.65 = 0.64. Hence at 10 "C, [HB] 0.64 K--- - - 1.78 [HA] 1-0.64 t -Equations (3) and (4) enable us to determine values of K and K';. Hence at 10 "C pK = 4.70 k 0.05 and pK' = 4.45 k 0.05. The pK values of 2 (A) and E (B) isomers were found to be 4.20 and 4.70 respectively. The higher pK value of the E isomer is believed to be a consequence of the electron-donating effect exerted by the 2-amino group when the azomethine bond and the p-chloroaniline group are co-planar with the long-pair electrons on N-1 engaged in the quinonoid delocalisation. The aniline nitrogen is believed to be less basic than the azomethine nitrogen for reasons similar to those for medaze-pam. 13.15 The observation that acidic solutions of A or alternatively B exhibit similar UV - visible absorption spectra implies that in acidic media 2 or E isomers give a single conjugated H2A+ immonium species (Table 1).Hence it can be emphasised that the pK measured in the case of the E isomer B (pK = 4.70) is apparent. In this instance the E to 2 isomeric trans-formation is assumed to be very rapid in acidic medium as indicated in Scheme 2. Table 2. Ultraviolet absorption spectra of B in various solvents ( c = 5 x 10-4 M 25°C) Solvent ~ / 1 mol-1 cm-1 hlnm x 103 Diethylether . . . . . . 365 Dichloromethane . . . . . . 365 Methanol . . . . . . . . 365 Water - methanol (2 + 8) . . 360 Water - methanol (5 + 5 ) . . 356 Water - methanol (8 + 2) . . 352 6.65 6.65 6.15 5.50 5.00 4.25 300 400 Wavelengthlnm 500 600 Fig.2. Typical spectral changes due to increasing pH for E isomer (B ANALYST SEPTEMBER 1986 VOL. 111 1055 Table 3. Experimental first-order rate constants found for isomerism ( k ) hydrolysis (k2) and intramolecular ring closure (k3) listed against PH Isomer PH 1.2 1.3 1.6 2.0 2.85 3.2 3.6 3.85 4.05 4.3 4.6 5.5 6.05 6.6 7.0 7.25 E (B) (A) 103 kllmin-l 103 k,/min - 1 103 kJmin-1 10°C 45 "C 10 "C 45 "C 10°C 45 "C 400 320 280 125 124 120 118 220 66 190 250 44 7.9 29 16 6.3 9.5 3.6 1.8 7.1 0.63 3.6 1.6 4.0 1.3 1.1 3.2 \ \ \ -I 0 a! 0 -2 -J -3 8 - 4 PH Fig. 3. Dependence of log kexp on H at 45 "C A isomerism transformation ( k l ) ; B hydrolysis ( k J ; 8 intramolecular ring-closure reaction (k3) In order to determine whether the isomeric transformation has occurred during the pKa determination a sample of the E isomer was taken at pH 2.0 and then adjusted to pH 7.0.The spectrum recorded at pH 7.0 was identical with that of a fresh sample of 2 isomer. From this result it can be deduced that the isomeric transformation occurs rapidly in acidic medium (see Kinetic Behaviour). Kinetic Behaviour Rate Constant Determination and Rate -pH Profile The reactions under consideration were the transformation of benzylidenes A and B to the benzophenone C on the one hand and to desmethyldiazepam D on the other. The kinetic studies of the hydrolysis of A and B were followed spectropho-tometrically in the pH range 1-8.The HPLC method was also used to confirm the formation of D resulting from an intramolecular ring closure. The processes involved are presented successively for pH 4-8 and pH 14. Slightly acidic and neutral media pH range 4-8 With increasing time the E isomer B is involved in spectral changes implying an apparent two-step reaction both steps being first order. Spectral analysis at first showed a decrease of the band at 352 nm followed by an increase at 385 nm. Time plots of the logarithm of absorbance A352 at any given time, and time infinity consisted of one linear section according to the integrated first-order expression h(Ar - A,) = -kit . . . . . . (5) The apparent first-order rate constants listed in Table 3 were calculated from least-squares analysis of the slopes.The value of kl was found to be 29 x 10-3 min-1 at pH 5.5 and 45 "C. As confirmed by HPLC the first step consists of the isomeric transformation k E isomer (B) -3 2 isomer (A) If the absorptivities of A C and D at 352 nm are nearly zero, then kl was measured according to Moreover the plot of log kl against pH is linear in the pH range 4-8 with a slope of -0.6 (Fig. 3). When extrapolated at pH values corresponding to an acidic medium this linear section enabled us to determine the theoretical value of k1; log kl was found to be 0 at pH 3.0 and 45 "C. This result implies that (a) the experimental determination of k is no longer possible in acidic media using the methods presented herein and (b) the conjugate acid H2A+ species exists predominantly as the Z isomer A in these media as the rate constant kl shows high values.The second step implies a parallel reaction mechanism i.e., either a hydrolysis of A yielding benzophenone C or an intramolecular ring-closure reaction yielding desmethyldiaze-pam D in agreement with Scheme 3. As shown in Table 1 C C + H2NCH2CONH2 kr *A*D slow,E Scheme 3 exhibits a well defined band at 385 nm. Moreover in the pH range 5-8 A and D exist as free base forms HA [pKa (A) = 4.20; pK (D) = 3.401 which exhibit only a negligible A385 ,,,,, value and do not disturb the experimental determination of the concentration - time profile for C. Conversely in these media HPLC provides a convenient method for quantitative determination of [D] From the experimental determination of [C] and [D] the value of [A] may be deduced according to the expressions [A]o = a [C] + [D]m (a - x)t = [C]r + [Dlr Using the isomer A as starting material time graphs of the logarithm of (a - x ) were linear according to the integrated first-order expression In (a - x ) = - (k2 + k3)t .. . . (7) The value of (k2 + k3) was found to be 22.3 x 10-3 min-1 at pH 5.5 and 45 "C and 2.4 x 10-3 min-1 at pH 7.0 and 45 "C. From the value of [C],/[D] equal to 2.5 at pH 5.5 and 1.2 at pH 7.0 respectively the individual values of k2 and k3 (see Table 3) may be deduced using the expressio 1056 ANALYST SEPTEMBER 1986 VOL. 111 0 - 1 a X 22 m - 2 J -3 - 4 I I I 2 4 6 PH Fig. 4. Dependence of log kexp on pH at 10 "C A hydrolysis (k2); B, intramolecular ring-closure reaction (k3) Table 4.Kinetic data found for hydrolysis (k2) and intramolecular ring closure (k3) in the pH range 1-8 pH > pKa (45 "C) d[CI v = - = k2 [A] = k[A(HA S HB)] [H,O] [H+]o.8 dt v = - - d[D1 - k3 [A] = k' [A(HA HB)] [H+]o-5 . . dt k2 = k3 = 4 x 10-4 min-1 at pH 7.7 pH < pKa (10 "C) v = - - d[C' - k2 [A] = k [A(H2A+)] [H20] . . . . dt 4D1 v = - = k3 [A] = k' [A(H2A+)] [H+] k2 = k3 = 0.126 min-1 at pH 2.0 . . . . dt . . . . (9) . . . (10) . . . . (14) . . . . (13) In the pH range 5-8 the log k2 - pH relationship was a straight line (Fig. 4) whose slope (-0.8 at 10 "C and 45 "C) roughly agreed with an H+-catalysed hydrolysis [expression (9), Table 41. Using the E isomer (B) as starting material from the experimental determination of the y,/a ratio where y, designates the maximum concentration of the intermediate A species (Fig.5 ) and a the initial concentration of starting material B the value of the ratio k12/kl may be deduced using the mathematical procedure for the successive first-order react ion 16 Using kI2 = k2 + k3 = 22.3 x 10-3 min-1 at pH 5.5 and 45 "C, we found kl = 30 x 10-3 min-1 which is in good agreement with the values obtained by spectral determinations at 352 nm or by HPLC using E isomer (B) as starting material. Acidic Media pH < 4.0 In acidic media (pH d lS) when A or B (H2A+ samples) were withdrawn at appropriate times and subjected to spectral analysis the broad absorption band at 390 nm (Table 1) decreased with time while a band appeared at 360 nm (E = 4 x lo3 1 mol-1 cm-1).The change in spectra recorded in sulphuric acid solutions showed two isosbestic points at 320 and 410 nm indicating that a simple reaction had taken place. Time graphs of the logarithm of A360 at any given time and time infinity consisted of one linear section according to the integrated first-order expression In (Am -A,) = - k3t . . . . (12) By plotting the value of log k3 at 10 "C as a function of pH a linear section was obtained whose slope was found to be roughly equal to -1.0 (Fig. 4) in agreement with an H+-catalysed phenomenon [expression (13) Table 41. In this pH range the final spectrum recorded was identical with that of desmethyldiazepam D. The complete formation of D was confirmed by HPLC.These results are in agreement with the occurrence of an intramolecular ring closure H+-catalysed process with NH3 as the leaving group (Scheme 4). In less acidic media in the pH range 1.5-3.0 the imine bond hydrolysis (k2) and the intramolecular ring-closure reaction (k3) constitute a competi-tive reaction mechanism. The UV - visible absorption spectra exhibited by C (E = 4.8 x lo3 1 mol-l cm-1 at 385 nm) and D, H2A+ species (E = 4.0 x lo3 1 mol-1 cm-1 at 365 nm) are so close to each other than the bands overlap hence the changes in spectra did not enable us to determine the concentration -time profile for C and D. Moreover HPLC was no longer usable for kinetic determinations owing to the high rate of the chemical reaction.However by extrapolation of the log k2 (10 "C) vs. pH and log k3 (10 "C) vs. pH plots (Fig. 4) it may be deduced that at pH 2.0 k2/k3 = [C]/[D] = 1.0. Over the range 3.0<pH<4.0 log k2 remains independent of pH according to expression (14) (Table 4). The intercept (Fig. 4) with the linear section in the range 4.0 < pH < 8.0 corresponds to the pK of the immonium imine acid - base equilibrium in exact agreement with the value obtained from the UV - visible absorption spectra. In this pH range the ring closure reaction would constitute a minor route with respect to H H D Scheme ANALYST SEPTEMBER 1986 VOL. 111 1057 1 oc 8 i 2” 50 .- c C a C 0 0 0 2 4 Time/h Fig. 5. HPLC concentration - time profiles for starting material B, intermediate A and products C and D; pH 5.5 45 “C Fig.6. water (2 ( b ) 24 h I I 4.8 6.4 19.5 Time/min HPLC profile detected at 254 nm of B hydro1 sis in MeOH - + 8) pH 7.4 at 37 “C (a) before initiation of t i e reaction and after the reaction the hydrolysis reaction. However the observation that D is present in significant amounts in neutral medium [kZ (45 OC)/ k3 (45 “C) = 1.0 at pH 7.7; Fig. 31 implies that the ring closure reaction yielding D is operable throughout the entire pH range 1-8 according to an H+-catalysed process. Finally the kinetic data are consistent with the integrated first-order expressions listed in Table 4. By inspection of equations (9) and (10) and equations (13) and (14) it can be deduced that the ratio [k2]l[k3] must decrease (a) at pH > pK, with increasing values of pH; and ( b ) at pH < pK, with decreasing values of pH.Note that product E with the diazepine ring opened was detected in HPLC (Fig. 6) but was not present in significant amounts under our experimental conditions as reported previously. 1 7 ~ 8 Conclusions Compared with the behaviour of 2-{ [2-hydroxy-5-chlorophenyl)phenylmethylene]amino}acetamide derivatives reported in previous papers,&6 the 2-amino group seems to exert a decisive influence. For 2-hydroxy derivatives in water -alcohol (8 + 2) buffered solutions the phenol quinonoid tautomerism equilibrium constitutes a rapid antecedent step in slightly acidic and neutral media in the over-all hydrolysis reaction step hence yielding 2-hydroxy-5-chloro-benzophenone .The E isomer (B) of 2-{ [2-amino-5-chlorophenyl)phenyl-methylenelamino) acetamide yields kinetic data that obey the kinetic law for successive first-order reactions. The first step consists of the isomeric transformation B to A (2 isomer). The second step implies a parallel reaction mechanism i.e. either a hydrolysis yielding 2-amino-5-chlorobenzophenone C or an intramolecular ring closure reaction yielding desmethyl-diazepam D. Only the second step occurs when the 2 isomer, A constitutes the starting material. The over-all process is summarised in Scheme 3. The parallel route yielding D constitutes a major reaction at pH 2.0 and remains important in neutral media. As far as their medical use is concerned the 2-amino derivatives are more stable towards hydrolysis than the corresponding 2-hydroxy derivatives reported previously,4J and partial cyclisation into the well known desmethyldiaze-pam could be of therapeutic benefit.19 We thank Dr.A. E. Wick for helpful guidance and discussion, B. Raizon for preparing the compounds and G. PCtry and P. Poirier for their excellent technical assistance. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. References Snodgrass R. S. in Iversen L. L. Editor “Biochemical Studies of CNS Receptors,” Plenum Press New York 1983, Kaplan J.-P. Raizon B. Desarmenien M. Feltz P., Headley M. Worms P. Lloyd K. and Bartholini G., J. Med. Chem. 1980 23 702. Wick A. E. Mompon B. and Rossey G. in Bartholini G., Editor “Epilepsy and GABA Receptor Agonists Basic and Therapeutic Research,” Raven Press New York 1985 pp.53-62. Kaplan J.-P. and Raizon B. Fr. Pat. 2 470 116,1979; Chem. Abstr. 1982,% 052022e. Lemetais P. Maupas B. Fleury M. B. and Mompon B., Analusis 1984 12,20. Maupas B. Fleury M. B. and Mompon B. Analusis 1984, 12 72. Maupas B. Letellier S. Fleury M. B. and Mompon B., Analusis 1984 12 122. Fleury M. B. Letellier S. Maupas B. Mompon B. Poirier, P. and Raizon B. Analusis 1985 13 170. Bultsma T. Meijer J. Pauli F. Ramaker J. and Nauta W., Eur. J. Med. Chem. Chim. Ther. 1977 12 427. Dhami K. S. and Stothers J. B. Tetrahedron Lett. 1964,12, 631. SolladiC-Cavallo A. and SolladiC G. Org. Magn. Reson., 1977 10 235. Sternbach L. H. Fryer R. J. Metlesics W. Reeder E., Sach G. Saucy G. and Stempel A. J. Org. Chem. 1962,27, 3788. Maupas B. and Fleury M. B. Analusis 1982 10 187. Bell S. C. Conklin G. L. and Childress S. J. J. Org. Chem., 1964 29,2368. Barrett J. Smyth W. F. and Davidson I. E. J. Pharm. Pharmacol. 1973 25 387. pp. 167-239 1058 ANALYST SEPTEMBER 1986 VOL. 111 16. Frost A . A. and Pearson R. G. Editors “Kinetics and Mechanism,” Second Edition Wiley New York 1961 p. 168. 19. Fujimoto M. Tsukinoki Y. Hirose K. Hirai K . and Okabayashi T. Chem. Pharm. Bull. 1980 28 1374. 17. 18. Han W. W. Yakatan G. and Maness D. D. J . Pharm. Sci., 1977 66 573. Konishi M. Hirai K. and Mori Y. J . Pharm. Sci. 1982,71, 1328. Paper A6132 Received February 4th I986 Accepted March I8th 198

ISSN:0003-2654    

DOI:10.1039/AN9861101051

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST SEPTEMBER 1986 VOL. 111 1059 Sampling and Gas Chromatographic Analysis of Volatile Sulphur Compounds and Gases at Sub-v.p.m. Levels in the Presence of Ozone Philip G. Slater and Leigh Harling-Bowen Southern Water Authority Lewes Road Falmer Brighton East Sussex BN I 9PY UK A method is presented for the sampling and analysis of sulphur gases (hydrogen sulphide and methanethiol) and volatile sulphur compounds (higher thiols and organic sulphides) in both ambient air samples and air containing ozone. The method has been developed for the quantitative analysis of hydrogen sulphide and for the semi-quantitative and qualitative analysis of other sulphur compounds. Field sampling and sample storage make use of simple cryogenic techniques and the analysis is carried out using a combination of thermal desorpt ion and cryogenic re-tra ppi ng with sul p h u r-specific gas chromatography.Sample volumes of 100 ml and collection times of 1 min per sample are typically used for hydrogen sulphide in the range 0.1-1.0 volume parts per million (v.p.m.) with a limit of detection estimated to be better than 70 volume parts per billion (v.p.b.). The technique has been used in connection with odour control and measurement at a sewage treatment works and in the assessment of odour control plant effectiveness. Keywords Sulphur gas analysis; gas chromatography; thermal desorption trapping and pre-concentration; cryogenic techniques Sulphur gases such as hydrogen sulphide (H2S) and methane-thiol (MeSH) together with volatile sulphur compounds for example simple thiols and organic sulphides and poly-sulphides are often implicated in nuisance odours at sewage treatment and disposal works.Levels of hydrogen sulphide in associated buildings pump houses wet wells etc. can be high enough to necessitate ventilation to reduce the level of gas to a safe concentration which may further aggravate the local odour problem. Various methods have been employed to reduce the ambient levels of sulphur compounds with varying degrees of success for example wet or dry ozonolysis and chemical scrubbing of airborne compounds or the dosing of sewage mains with oxidants or disinfectants such as hydrogen peroxide or sodium hypochlorite. Frequently the assessment of the effectiveness of control measures tends to be subjective as a result of the insensitivity of measurement techniques and often no attempt is made to relate plant performance to known odour threshold levels.l t 2 At a local sewage pumping station odour complaints and high levels of hydrogen sulphide led to the present investi-gation being deployed to assess the optimum operating conditions for control plant using wet ozonolysis to remove the sulphur and other odorous compounds. Hydrogen sulphide was considered to be the most important odorous compound, whereas other sulphur compounds were suspected of being present intermittently. The availability of gas chromatographic equipment and suitable specific detectors provided the basis of a method, although it was uncertain as to whether the conventional gas sampling methods were either sensitive enough or suitable for the reactive compounds involved.The distance of the site from the laboratory meant that samples would have to be transported and stored easily; the presence of ozone in the samples taken from the plant meant that losses of sulphur compounds during sampling and storage could be consider-able if steps were not taken to minimise oxidation reactions. Finally the complexity and special requirements of the gas chromatography equipment precluded the use of the instrumentation on site which might have been the ideal solution. A number of papers>-' have described methods for similar analytical requirements. Although several of the methods meet the requirements for the sensitivity and range required for sulphur compounds most were considered unsuitable for reasons of difficulty in sampling large sample volume requirements or excessively long sampling times and it was felt that none were likely to succeed in the presence of ozone.The extreme volatility of hydrogen sulphide and methane-thiol (boiling-points -60.7 and 6.2 "C respectively,lO indi-cated that a cryogenic technique offered the best chance of success and would have the added advantage of slowing post-sampling reactions. Methanethiol in particular was likely to decompose if heated to temperatures much in excess of 100 "C,3 which precluded the use of techniques involving high-temperature desorption or manipulation. Our investigation indicated that methanethiol and its decomposition product dimethyl disulphide were both present at the treatment works.System inertness was an important factor in the study and the determinands did not come into contact with materials other than glass fluorinated polymers ceramics or silicone rubbers. A solid adsorbent, Tenax GC (2,6-diphenylphenylene oxide) was chosen as a trapping material as it has been shown to be suitable as a column packing in the gas chromatography of sulphur compounds,5 and liquid argon (boiling-point - 186 "C) was chosen as the cryogenic medium in order to allow the use of nitrogen as a carrier gas in the chromatograph. Finally a small portable gas sampling pump primarily used for personal exposure monitoring was used for field sampling to give the option of a wide range of sample volumes with the minimum of equipment.Experimental Equipment Gas chromatograph A Pye Unicam GCV with a flame photometric detector (FPD) sulphur filter and temperature programming facility was used in the investigation. The carrier gas was oxygen-free nitrogen gas and zero grade hydrogen and compressed air were used for flame gases. All the gases were purified by passing through a Type 13X molecular sieve. Analytical column A pre-packed glass column (2 m x 2 mm i.d.) filled with 60-80-mesh Carbopack B with a 1.5% XE-60 and 1.0% phosphoric acid stationary phase was used (Supelco) 1060 ANALYST SEPTEMBER 1986 VOL. 111 Sample injection and flow switching valves Inert low-pressure valves (Rheodyne Model 5020) with a 100-pl loop were used for the syringe injections. Thermal desorber A GN Concentrator (GN Instrumentation) fitted with PTFE transfer lines a glass-lined stainless-steel cold-trap loop and additional flow and metering valves was used.Computing integrator A Hewlett-Packard 3380A or Trilab 2000 (Trivector Systems) was employed. Gas calibration equipment Calibration gas and pure air were mixed at a ratio of 1 100 using suitable flow meters (Platon Flowbits) and further diluted serially with flow meters to give an over-all dilution ratio of between 1 1000 and 1 10 000 at a final flow-rate of 5 1 min-1. The mixing of gases at each dilution stage took place in 300 x 30 mm Perspex tubes containing glass Raschig rings, and the final mix was sampled from a 250-ml round-bottomed flask into which the gas flowed.Static gas standards were prepared in a 300 X 300 X 300 mm Perspex cabinet equipped with a circulating fan a septum port for syringe or adsorbent tube sampling and a compressed air inlet for flushing. The approximate concentration of hydrogen sulphide used for dynamic gas calibration was monitored using a hydrogen sulphide monitor (Neotronics Model HS 102) and potentio-metric recorder (Servoscribe Model IS). Digital thermometer A Comark Model 5000 thermometer equipped with Type K thermocouples was used. De war flasks Various sizes of vacuum flasks for handling liquid argon and sample storage (Thermos and Day-Impex) were used. Adsorbent tubes Pyrex glass tubes 6 mm o.d. 4 mm i.d. and 89 mm long with their ends cut square packed with approximately 100 mg of 60-80-mesh Tenax GC and plugged at each end with glass-wool were used.The packing should be as central in the tube as possible leaving the ends empty although precise amounts and positioning of the packing is not critical. Tubes should be conditioned for at least 1 h six at a time, on a manifold of stainless-steel Swagelok union Tees at a temperature of 250 "C while purging with nitrogen at approxi-mately 40 ml min-1 per tube. Once conditioned the tubes were capped with plastic caps and stored in the dark at room temperature before use. Tubes prepared in this way showed no response for any of the compounds studied after storage for periods of several weeks, although for critical applications it would be advisable to recondition prior to use. Sampling blocks Blocks 60 x 60 x 60 mm were cut from suitable polystyrene foam.A high-density foam is the most suitable. A hole was cut using a cork borer in one face of the block of 30 mm diameter and 45 mm deep thus providing a container for the liquid argon. A second hole was cut completely through the block, approximately 15 mm below the top of the first hole and at right-angles to the hole. This hole should be small enough to provide a tight fit when a sampling tube is pushed through. The final configuration allows a sampling tube to have its central portion immersed in liquid argon with its ends protruding from the block for attachment of tubing. Sampling pump A sampling pump Model CS1 (Casella) with a flow-rate adjustable between 20 and 200 ml min-1 was used.Gas-tight syringes Syringes of 1 and 20 ml capacity (J. Young S.G.E.) were used. Reagents and Standards Hydrogen sulphide 99.6%. BDH Chemicals. Methanethiol 99.5%. BDH Chemicals. Ethanethiol 99.5%. BDH Chemicals. Dimethyl sulphide 98%. Aldrich Chemicals. Dimethyl sulphide 99%. Aldrich Chemicals. Calibration gas. 1000 v.p.m. hydrogen sulphide in nitrogen Ozone. Model Trailigaz Lab0 ozone generator. Ozone measuring tubes. Draeger Safety. (BOC Special Gases). Caution-Hydrogen sulphide and ozone are highly toxic. Thiols and organic sulphides are toxic and flammable. Suitable precautions should be taken at all stages when handling these materials. Chromatography and Thermal Desorption The gas chromatograph was set up according to the manufac-turer's instructions and the instrument optimised for sulphur analysis.The conventional septum injector was removed and the carrier gas supply taken to the inlet of the thermal desorber. The desorber has two gas outlets that are selected depending on the mode of use; one simply diverts the carrier gas back to the chromatograph where it flows via a switching valve and sample injection valve in series directly to the inlet of the analytical column. In this mode the gas chromatograph is used conventionally and gas samples are introduced by way of the injection valve and a gas-tight syringe. The other mode diverts the carrier gas through the desorption oven and cold trap and then back to the chromato-graph; flow switching is controlled automatically by the desorber and by the operator.Flows of gas are controlled at all times by metering valves and flow meters. The gas chromatographic conditions used were as follows: detector temperature 150 "C; injector temperature 60 "C; oven temperature 50 "C; temperature programme 2 min at 50 "C isothermal ramp at 15 "C min-1 to a final temperature of 110 "C held for 9 min; carrier gas flow-rate 40 ml min-1; flame gases hydrogen at 20 ml min-1 air at 30 ml min-1 air purge approximately 100 ml min- 1; amplifier attenuation, 32 X 102; output to integrator 1 V f.s.d.; integrator attenuation x 1024. The thermal desorber was modified by replacing the stainless-steel cold trap with a trap of the same dimensions made of glass-lined stainless steel in order to minimise any adsorption of reactive gases.In some experiments a length of 1.6 mm 0.d. PTFE tubing was passed through the trap if a smaller trapping volume was required. In practice either arrangement worked well. No packing was used in the trap as cryogenic re-trapping proved to be quantitative. All transfer pipework in contact with the samples was either 1.6 or 3.2 mm 0.d. PTFE tubing and where it was necessary to use stainless-steel couplings they were lined with PTFE or silicone-ru b ber tubing. An additional ball valve and fine metering valve were fitted in the sample transfer line and a switch was added to override the automatic flow switching sequence of the desorber. The latter allowed the temperature programmed runs to be completed without disturbing the gas flows.Finally a septum port was added to allow gas injections during optimisation of desorber conditions ANALYST SEPTEMBER 1986 VOL. 111 1061 The desorption and re-trapping conditions used for all the sulphur compound studies were as follows desorption tem-perature 110 "C; desorption time 3 min; cold trap (trap-ping) -186 "C; cold trap (volatilisation) 110 "C. A previous study11 indicated the need for conditioning the chromatographic system during sulphur gas analyses. The whole system was conditioned prior to a set of analyses by injecting 1 ml of 1000 v.p.m. H2S in nitrogen through the septum port on to the desorber. This conditioned all parts of the system in contact with the desorbed sample and was necessary only if the system was not used for more than 2-3 h.The column was conditioned immediately before each analysis by injecting 250 vl of 1000 v.p.m. H2S through the sample injection valve. Calibration Dynamic H2S gas standards A concentration of 1000 v.p.m. H2S in nitrogen was serially diluted with air to give standards in the range 0.1-1.0 v.p.m. H2S. Each mixture was monitored with the hydrogen sulphide monitor to indicate when the mixture was fully equilibrated; with a final flow-rate of 5 1 min-1 going into the sampling flask, a stable mixture is obtained in 4 min and where the large chamber was used at least 20 min were required for equilibration. Static gas standards Static gas standards of hydrogen sulphide and the other sulphides and thiols were prepared by injecting suitable volumes of gases and vapours into the large chamber through the septum port.A knowledge of the saturated vapour pressure of the pure volatile liquids at room temperatures allows semi-quantitative standards to be made up in the range 0.1-0.5 v.p.m. by serially diluting the saturated vapour using gas syringes and septum vials. When injected into the chamber the various compounds were allowed to mix for several minutes using the built-in fan. After mixing samples were taken immediately as described in the following sections and in order to minimise losses due to adsorption etc. no more than 5 min elapsed between mixing and sampling. After sampling the chamber was thoroughly flushed with air before preparing new stan-dards. Standards made up in this way were used only to provide a qualitative analysis of compounds other than H2S.Calibration procedure After the preparation of standards by one of the above procedures samples were taken for gas chromatographic analysis; the sampling technique was identical for either type of standard and with minor variations identical with the field sampling method. A freshly conditioned tube was placed in a polystyrene block. A short length of silicone-rubber tubing approximately 20 cm was attached to each end of the absorbent tube one piece of tubing being used to sample the calibration mixture and the other attached to the inlet port of the sampling pump. Liquid argon was poured into the chamber of the block to completely cover the exposed portion of the adsorbent tube. The tube and block were allowed to chill through for about 1 min topping up the argon as necessary.Care should be taken in handling the block at this stage particularly with regard to leaks of argon past the tube. Blocks should be tested before use and any that leak significantly rejected. With careful use blocks can be used many times before leaks occur. Tests with a micro-probe thermocouple indicated that a considerable temperature gradient existed both along the adsorbent bed and across it. Temperatures at the centre of the packing dropped to approximately - 10 "C whereas temper-atures of -166 "C were recorded at the inner wall of the tube directly in contact with the argon. These temperature grad-ients were considered beneficial however as it was likely that a more even spread of adsorbed and condensed materials occurred and the tube was less prone to blockages due to icing when ambient levels of water vapour were high.A sample was collected by switching the pump on for the required length of time. For the purpose of this study the sampling period was set at 1 min with a previously calibrated flow-rate of 100 ml min-1 which gave adequate sensitivity for 0.1 v.p.m. H2S. The sensitivity of the method may be changed as required by altering the sampling times or flow-rates within certain limitations. It was found that at least 1000 ml of ambient air of 35% relative humidity could be aspirated through the adsorbent tube before the flow-rate was reduced owing to ice formation but flow-rates should not exceed 100 ml min-1 otherwise trapping will not be quantitative.During aspiration the actual flow-rate was checked by connecting a bubble flow-meter to the outlet of the sampling Pump-When sampling was complete the pump was switched off, the silicone-rubber tubing removed argon returned to a Dewar vessel and the tube inserted immediately with forceps into the desorber oven. Approximately 10 s before the tube was inserted into the oven the automatic cycle of the desorber was started which allowed the cold trap Dewar to rise into position and thoroughly cool the glass-lined trap. At this stage a ball valve was closed and no flow of carrier gas took place through the adsorbent tube. When the tube was correctly positioned the ball valve was opened and a metering valve used to restrict the flow of gas to 4 ml min-1 during the desorption and re-trapping.The completeness of re-trapping at this stage was checked by injecting an aliquot of hydrogen sulphide into the system before the cold trap and observing any breakthrough of H2S on the chromatograph. Under the conditions described no breakthrough was recorded. The aliquot of hydrogen sulphide used to condition the system was injected through the sample injection valve immediately after placing the adsorbent tube in the oven and this peak appeared approximately 1 min before the first analytical peak. At the end of the pre-set desorption period the cold trap Dewar lowers automatically and the trap was heated quickly to 110 "C by resistive heating. At this point the full carrier gas flow was restored by opening the metering valve fully the integrator and temperature programming being started simul-taneously.For hydrogen sulphide and methanethiol the whole process was completed in less than 7 min with retention times of 0.6 and 1.5 min respectively whereas dimethyl disulphide eluted in approximately 13 min. On the column described all five sulphur compounds investigated were completely resolved; using dynamic gas standards a plot of log (peak area) against log (concentration) yields a linear calibration over the range investigated for H2S.11 Typical instrument responses for a series of hydrogen sulphide standards are given in Table 1. ~ ~~ Table 1. Hydrogen sulphide dynamic gas standards. Limit of detection based on the total standard deviation at 0.1 v.p.m.= 0.07 v.p.m. H2S Nominal hydrogen Mean peak Total sulphide concentration area standard deviation, v.p.m. counts x 106 counts x 106 1.00 0.75 0.50 0.40 0.30 0.20 0.10 2.58 0.24 (3)* 2.22 0.06 (2) 1.33 0.08 (2) 0.94 0.08 (2) 0.66 0.06 (2) 0.04 (6) 0.37 1.59 0.12 (4) * Degrees of freedom in parentheses 1062 ANALYST SEPTEMBER 1986 VOL. 111 Ozone Removal Investigation The presence of ozone in some of the field samples was expected to interfere with both sample storage and analysis. The reaction between ozone and hydrogen sulphide in the absence of water is slow at ambient temperatures but the technique employed in this investigation included long-term sample storage and thermal desorption both of which could lead to losses of sulphur compounds through oxidation.It was therefore decided to seek some means of selectively removing ozone prior to trapping the determinands on the Tenax adsorption tube. The laboratory ozoniser was set up using the ozone production graphs supplied by the manufacturer and the ozone gas was introduced into the dynamic gas calibration system in such a way that a final mixture containing pure air, hydrogen sulphide and ozone could be manipulated with flow meters to give any combination of gas concentrations in the ranges 0-10 v.p.m. of hydrogen sulphide and 0-50 v.p.m. of ozone. During tests with 10 v.p.m. of ozone and 1 v.p.m. of hydrogen sulphide it was noted that even with liquid argon cooled adsorption tubes some darkening of the Tenax material took place at the inlet side which indicated that reaction between ozone and Tenax was taking place and that the adsorption - desorption characteristics of Tenax so affected would be altered.This was confirmed by placing a Draeger ozone-indicating tube in the sampling line immediately after the Tenax tube and aspirating up to 20 v.p.m. of ozone through the Tenax; under normal sampling conditions no ozone was detected it having been adsorbed or reacted with the Tenax. Tests to find an effective and selective ozone adsorbent were conducted by placing the adsorbent under trial in an empty glass adsorbent tube with glass tubes containing strips of moistened starch - iodide indicator paper before and after the test material thus giving a semi-quantitative and convenient means of checking the efficiency of ozone re-moval.It was observed during tests that where a particular type of silicone-rubber tubing was used to connect the adsorbent and indicating tubes together ozone appeared to be significantly removed. Further tests showed that the 20 cm length of silicone-rubber tubing used to collect samples as described under Calibration was effective in removing ozone up to 10 v.p.m. when used under the sampling conditions of flow-rate, etc. Tests were carried out on a number of other tubing materials of similar dimensions. Synthetic and natural rubber tubing was not effective; nor was poly(viny1 chloride) (Tygon) tubing or certain other types of silicone-rubber such as opaque types. The most effective material that could be easily identified and obtained commercially was a translucent silicone rubber tubing 6 mm o.d.4 mm i.d. batch number 50318 manufactured by Esco Rubber. The efficiency of the tubing depended on the amount of ozone that it had been exposed to and its capacity for ozone removal decreased with exposure to ozone presumably as active adsorption - reaction sites were used up. A fresh length of tubing would reprodu-cibly remove 10 v.p.m. of ozone before the Tenax tube but attempts to re-activate the tubing by heating or exposure to hydrogen sulphide gas were not successful. In order to increase the lifetime of the silicone-rubber tubing tests were carried out with short pre-columns of adsorbants. The columns were made from 80 mm lengths of 6 mm 0.d.thin-walled PTFE tubing packed with adsorbent and plugged at each end with glass-wool. Silicone-rubber tubing was dipped into liquid argon and when brittle could be filed with a coarse file into a granulated form that was then used to pack the tube; this material proved very effective in removing ozone and a similar material described as silicone rubber “crumb” was obtained from Esco Rubber. This material was used for all subsequent tests and a fresh PTFE tube containing the crumb was attached to the inlet end of the silicone-rubber sampling tube when sampling for hydrogen sulphide in the presence of ozone. When these disposable tubes were used it was never necessary to change the silicone-rubber tubing and it was estimated that the capacity of the pre-column was at least 100 ml of sample containing 10 v.p.m.of ozone. The results of a test in which samples of air containing 1 v.p.m. each of hydrogen sulphide and ozone were subjected to immediate analysis and storage for approxi-mately 24 h followed by analysis are given in Table 2. The results of this test indicated that the finely divided silicone-rubber material was the best adsorbent for ozone and had little effect on the level of hydrogen sulphide after storage. A further check on this material was made by collecting a nominal 1.0 v. p. m . hydrogen sulphide sample on a number of Tenax tubes after passing the sample through a pre-column. Samples were collected both in the absence of ozone and in the presence of 1.0 v.p.m. ozone and the results of analyses are shown in Table 3.Comparison of the means by Student’s t-test indicated no significant difference at the 95% confidence level. Field Sampling Field sampling was carried out in a manner identical with that described under Calibration. The only additional equipment required for field sampling was a storage Dewar vessel to transport adsorption tubes from the site to the laboratory and plastic end-caps to seal the ends of tubes after a sample had been taken. A suitable Dewar vessel should have internal dimensions of approximately 30 cm depth and 1 cm diameter. A container with an internal diameter of 100 mm and a depth of 100 mm was provided. This could float on the surface of the liquid argon and allowed capped tubes to lie horizontally on the bottom of the container the objective being to maintain the tubes in close contact with the liquid argon.Provided that the argon liquid level was maintained as high as practically possible the storage temperature of the tubes could be maintained below - 150 “C. An efficient Dewar vessel should allow the storage of samples for several days without frequent topping up of argon. During later field sampling liquid nitrogen (boiling-point - 196 “C) was used as the trapping and storage liquid being approximately one third of the cost of argon but argon was retained as the cryogenic liquid in the desorber to prevent condensation of the carrier gas. Table 2. Effect of 1 v.p.m. of ozone on the storage of Tenax adsorbent tubes containing hydrogen sulphide in the presence of various pre-column materials.Nominal hydrogen sulphide concentration 1 .O v.p.m. All results quoted are a mean of two or more analyses Control Storage (immediate (analysis after Pre-column material analysis) v.p.m. 24 h) v.p.m. H2S H2S lOcm long PTFE tubing . . . . 0.73 0.52 20cmlongsilicone tubing . . . . 1.02 0.76 Finely divided silicone-rubber pre-column . . . . . . . 1.06 1.01 Silicone-rubber crumb pre-column 1.12 0.88 Table 3. Results of the analysis of nominal 1.0 v.p.m. hydrogen sulphide in the presence of 1.0 v.p.m. ozone using a finely divided silicone-rubber crumb pre-column No ozone (immediate 1.0v.p.m. ozone analysis) (24 h storage), v.p.m. H2S v.p.m. H2S Mean H2S concentration v.p.m. 1.053 1.027 Numberofanalyses . . .. . . 6 6 Total standard deviation . . . . 0.031 0.075 98 Recovery % . . . . . . . . ANALYST SEPTEMBER 1986 VOL. 111 1063 A check on the efficiency of storage was carried out by preparing dynamic gas standards of 0.2 and 1.0 v.p.m. H2S, analysing one batch of tubes immediately and storing others for periods up to approximately 30 h in the storage Dewar vessel. The results of these tests are given in Tables 4 and 5. Once a sample had been taken on site the tube was immediately sealed with standard 6 mm plastic or rubber chromatographic column end-caps that had previously been heated to 100 "C to remove trace amounts of volatile sulphur compounds. To analyse the capped tubes once removed from the storage Dewar vessel a tube was placed in a polystyrene block similar to that used for sampling with slots cut in the top instead of holes bored through.This allowed a tube to be semi-immersed in liquid argon while the caps were removed. In practice removal was difficult as the caps shrink tightly on to the glass tube when cooled and are no longer flexible; however by rotating the tube carefully with the caps between fingers it is possible quickly and safely to warm the caps to a point at which they can be removed whilst keeping the Tenax section cooled. A later refinement was to shorten commer-cially available caps to half their original length (15 mm), which helped the removal process. As soon as the caps were removed the tubes were analysed immediately as described previously. Results and Discussion Hydrogen sulphide was the only gas that was analysed using dynamic gas standards; the advantage of dynamic standards is that it can normally be assumed that at the low levels a gaseous component will reach equilibrium with its surroundings and that losses due to physical processes will be minimised; the disadvantage is that a relatively large supply of gas com-ponents is required preferably in undiluted form.Gas standards can be bought in (although stability would be a problem with reactive gases at low concentrations) or produced by serial dilution of a higher concentration or by some other means such as a permeation oven. In this investigation a commercial gas standard of H2S was used but further dilutions of between 1000- and 10 000-fold were required to achieve levels between 1.0 and 0.1 v.p.m.H2S. Errors in serial dilution are cumulative and it was considered that the problem of the production of stable calibration standards was the single most important reason for non-rep-roducible results and for some of the low recoveries in storage tests. The flow meters and dynamic standards technique used were adequate for the purposes of the application but it is recommended that for more critical measurements due regard should be given to the provision of high-quality flow controllers pressure regulators and control of other factors such as temperature. A permeation oven would overcome some of these difficulties and almost certainly would be essential for the production of dynamic standards of the non-gaseous sulphur compounds.The results of the storage tests (Tables 4 and 5) show variability in the recoveries obtained at the 1.0 and 0.2 v.p.m. levels more so at the lower level. It was considered significant for the purposes of the investigation that there was no loss of sample over the storage period and that once the tubes were installed in the storage Dewar vessel at liquid argon or nitrogen temperatures it would be unlikely that changes in the levels of adsorbed compounds wpuld take place over storage periods longer than those observed. As previously noted the variability observed in this investigation was almost certainly due to the difficulties in producing accurate low levels of the compounds under investigation and at the 0.2 v.p.m. level the flow meters used were at the bottom of their usable range.At this level flows were difficult to reproduce and maintain and as the cali-bration standards and storage samples were produced by the same technique it was concluded that this factor alone would account for variations. It was not possible to investigate this aspect further and the mean recoveries obtained during storage were applied to site samples stored for similar periods. A second important factor was the manipulation of samples. Calibration adsorption tubes were always placed quickly into the desorption oven hence minimising losses of determi-nands whereas field samples were capped and were subject to varying amounts of manipulation. This was partly confirmed when tubes were sorted without caps at all and gave recoveries as good as tubes that had been capped.The uncapped tubes were removed quickly from the sampling block and placed on the bottom of the storage Dewar vessel inner container, whereas others were placed inside screw-capped test-tubes and immersed directly into liquid argon. It should be noted that liquid argon must not be allowed to enter the sample tubes as the rapid evaporation of the liquid during manipula-tion causes severe loss of trapped materials (Table 6). A possible explanation for the good recoveries is that any losses from the tube during storage were balanced by the ease with which tubes could be transferred from the Dewar vessel to the desorption oven without delay. Storage of tubes as described under Field Sampling, ensures that the temperature of the adsorbent is kept well below the melting-point of any of the compounds studied and that losses due to physical processes should be minimal.In practice it is difficult to maintain a constant low temperature in the tube storage vessel especially during the removal of tubes for analysis. A possible solution might be to store tubes directly in the liquid argon phase but it would be necessary to find a means of sealing tubes against the ingress of liquid and no practical solution to the problem was found that was convenient to use under site conditions. A brief investigation into the processes involved in the sampling procedure was carried out by using sampling tubes packed only with GC-grade silanised glass-wool and samples of calibration (H2S) mixtures collected and analysed by the Table 4.Mean concentration and recovery of hydrogen sulphide after storage nominal 1 v.p.m. Time/h . . . . . . . . . . 6 18 24 30 Mean,v.p.m. H2S . . . . . . 0.83 0.83 0.95 0.99 Total standard deviation . . . . 0.06 0.07 0.11 0.05 Recovery YO . . . . . . . . 83 83 95 99 Meanrecovery(a1lresults) . . . . 90% Numberofanalyses . . . . . . 4 4 4 4 Table 5. Mean concentration and recovery of hydrogen sulphide after storage nominal 0.2 v.p.m. Time/h . . . . . . . . . . 6 18 24 30 Mean,v.p.m.H2S . . . . . . 0.11 0.13 0.14 0.16 Total standard deviation . . . . 0.05 0.03 0.06 0.02 Recovery,% . . . . . . . . 55 65 70 80 Meanrecovery(allresu1ts) . . . . 68% Numberofanalyses . . . . . 4 4 4 4 Table 6. Comparison of various methods of Tenax tube storage, nominal 0.5 v.p.m.H2S. Three samples of each tube were analysed Mean peaks Total Degrees Concen-counts x standard of tration, Calibration standard 106 deviation freedom v.p.m. - 0.5v.p.m. . . . . . . 1.34 0.16 2 Uncapped tubes, Uncapped tubes total 3hstorage. . . . . . 1.64 0.14 2 0.60 immersion 3 h storage 1.38 0.23 4 0.5 1064 ANALYST SEPTEMBER 1986 VOL. 111 Table 7. Recovery of hydrogen sulphide gas standards from tubes packed with glass-woo1 compared with the standard procedure Standard Glass-wool procedure Drocedure: H2S concentration, v.p.m. 0.98 0.79 0.60 0.41 0.21 Peak area, counts x 106 3.13 3.21 2.64 2.65 2.06 2.18 1.21 1.32 0.56 0.56 Mean calibrated concentration, v.p.m.Recovery % 1.10 112 0.88 111 0.70 117 0.40 98 0.16 76 Tenax procedure. The results are shown in Table 7 compared with conventional Tenax tube samples. The recoveries and precision are acceptable although non-linearity is evident. It is clear that the main process involved in sample collection is condensation assuming that the glass-wool itself has no adsorptive properties and provides only an inert support for the condensed materials. It was considered however that the use of Tenax was preferable as this probably combined condensation and adsorption and no attempt was made to use glass-wool only in any site investigations. As previously noted in the section dealing with experimen-tal conditions there is a possibility of extending the sensitivity of the method considerably; there is no evidence that blank or instrument noise levels would be a problem for an increase in sensitivity of several orders of magnitude.The method is potentially useful in the study of other sulphur compounds of environmental interest such as sulphur dioxide carbonyl sulphide and carbon disulphide and by changing the chromat-ographic or detection system the method offers a convenient means of qualitative or quantitative analysis of non-sulphur compounds. This paper was prepared with the permission of the Divisional Manager Sussex Division Southern Water Authority. The authors thank Mr. A. Lloyd for useful advice and discussion. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. References “Compilation of Odour and Taste Threshold Values Data,” ASTM Data Series DS 48A American Society for Testing and Materials Philadelphia 1978. Valentin F. H. H. and North A. A. “Odour Control a Concise Guide,’’ Warren Spring Laboratory Stevenage 1980. Bailey J. C. and Viney N. J. “Analysis of Odours by Gas Chromatography and Allied Techniques,” LR 298 (AP), Warren Spring Laboratory Stevenage 1979. Stendler P. A. and Kijowski N. Anal. Chem. 1984 56, 1432. Walker D. S . Analyst 1978 103 397. de Souza T. L. C. and Bhatia S . P. Anal. Chem. 1976,48, 2235. Meek D. M. and Reid W. J. “Identification of Some Odourous Compounds in Air Samples Taken Near Sewage Treatment Plants,’’ LR 969 Water Research Centre Steve-nage 1979. Young P. J. Effluent Water Treat. J . 1984 189. Pankow J. F. Isabella L. M. and Asher W. E. Environ. Sci. Technol. 1984 18 310. Weast R. C. Editor “Handbook of Chemistry and Physics,” Fifty-sixth Edition CRC Press Cleveland OH 1975. Hawke D. J . Lloyd A. Martinson D. M. Slater P. G. and Excell C. Analyst 1985 110 269. Paper A6112 Received January 14th 1986 Accepted April 2nd 198

ISSN:0003-2654    

DOI:10.1039/AN9861101059

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1986, VOL. 111 1065 Effects of Slow Heating Rates on Products of Polyethylene Pyrolysis Thomas P. Wampler and Eugene J. Levy Chemical Data Systems, 7000 Limestone Road, Oxford, FA 19363, USA Polymeric materials, although they are not volatile themselves, may be analysed by gas chromatography (GC) through the use of analytical pyrolysis. Because the fragmentation pattern produced by heating is reproducible, the pyrolysates provide much information about the original macromolecule of which they were a part. Analytical pyrolysis, coupled with GC, mass spectrometry (MS), GC - MS and Fourier transform infrared (FTIR) spectroscopy has provided a wealth of information concerning polymer microstructure, branching, defect structure and degradation mechanisms. Traditionally, the pyrolyses have been carried out at very fast heating rates (for example, heating to 800 "C for 5 s at a rate of 10 000 "C s-1).Recently, however, interest has been shown in how polymers behave when heated to pyrolysis temperatures at slow rates ("C min-1) for long periods of time. Many polymers show considerable shifts in the kinds and relative amounts of pyrolysates produced at these slower rates. The polyolefins are good examples of polymers for which the pyrolysate distribution is sensitive to both the final temperature and the rate of heating. Examples are given indicating these differences for polyethylene in a series of pyrolyses in which the temperature is varied and the heating rate held constant, and for pyrolyses at a variety of heating rates to the same final temperature.Keywords : Pyrolysis; gas chromatography; polyethylene; degradation Analytical pyrolysis has been widely used in the study of polymers such as the polyolefins. The analysis of polyolefin pyrolysates has done much to elucidate structural anomalies such as branching and defect structures,1-3 and also to shed light on questions of kinetics and degradation mechanisms. When the polyolefins are pyrolysed, they fragment to produce volatile hydrocarbons that may be separated and studied using gas chromatography. Polyethylene is essentially a high relative molecular mass wax that degrades thermally to produce smaller hydrocarbon fragments. Most of the volatile compounds produced are unbranched hydrocarbons from the basic polymer chain, but branched compounds are also produced, and are of great importance when studying the degree and type of branching found in the polymer structure.4 The unbranched hydrocarbons have double bonds at one or both ends of the molecule.This produces a characteristic pattern consisting of a series of triplet peaks when the pyrolysates are chromatographed. Each triplet is composed of the alkane, alkene and the diene having the same number of carbons. Typically, a pyrogram contains a series of triplets representing hydrocarbons of every chain length to about 30 carbons. Examples of this are shown in Fig. 1, in which the pyrolysates were chromatographed on a 50 m x 0.25 mm i.d. SE-54 fused-silica capillary column. The order of elution here is the alkadiene first, followed by the alkene, with the alkane last.The relative abundance of the alkane and the alkadiene in each triplet is sensitive to both the pyrolysis temperature and the rate of heating used to effect the pyrolysis. Experimental All pyrolyses were performed using a filament coil type pyrolyser (Pyroprobe 122, Chemical Data Systems). The polymer samples were held in a quartz tube that was inserted into the platinum coil. Because pyrolyses at slow rates were to be studied, the pyrolyser was interfaced to a sample concen- trator (Model 320, Chemical Data Systems), which collected the pyrolysates on a Tenax filled trap. In this way, polymer samples could be processed for minutes or hours at slow heating rates, and the pyrolysates all collected for a single GC analysis.When the pyrolysis was complete, the trap was pulse heated and backflushed with the GC carrier gas. The desorbed Fig 1. Gas chromatograms showing effect of pyrol sis temperature on 01 ethylene, Conditions: ramp, 20 "C ms-1. 6) Temperature IOOh)"~ for 20 s, GC programme 50 "C for 2 min, 4 "C min-l to 280 "C. ( b ) 900 "C for 20 s; and ( c ) 700 "C for 20 s pyrolysates were transferred to a gas chromatograph (Model 3700, Varian) equipped with a 50 m x 0.25 mm i.d. SE-54 capillary column (Quadrex). A 60 : 1 split was established at the injection port of the GC and the column was programmed from 50 to 280 "C at a rate of 4 "C min-1. A flame-ionisation detector was used, and area data were obtained from a recording integrator (3390-A, Hewlett-Packard).1066 ANALYST, SEPTEMBER 1986, VOL.111 r 1 Results and Discussion The temperature at which a material is pyrolysed has a significant, and often predictable, effect on the nature and amounts of compounds produced. Polyethylene, when pyro- lysed at temperatures ranging from 600 to 1000 "C, produces the series of triplets each time [Fig. l(a)-(c)] but with some interesting differences. The relative amounts of the saturated and unsaturated products at each carbon number change as the pyrolysis temperature is changed. Specifically, for each triplet, the amount of alkane decreases and the amount of diene increases as the pyrolysis temperature increases. A ratio of the amount of diene to the amount of alkane (DIA) at specific carbon chain lengths may be calculated, and it is interesting to see how this ratio is affected by the pyrolysis temperature.Fig. 2 shows a graph of the DIA ratio for compounds of 10-19 carbons plotted against the number of carbons in the chain for pyrolysis temperatures from 600 to 1000 "C. From this graph it can be seen that the diene to alkane ratio increases with temperature, and that the ratio is roughly the same for most diene - alkane pairs having between 13 and 19 carbons. There are some fluctuations, with the DIA ratio peaking at tetradecane and heptadecane, and dipping slightly at pentadecane and hexadecane, regardless of temperature. The temperature dependence of this ratio may be seen more easily by selecting a specific number of carbons and comparing the DIA ratio at this carbon number for various pyrolysis temperatures.For example, the ratio of heptadecadiene to heptadecane could be chosen, and plotted against pyrolysis temperature. Fig. 3 is such a graph, showing that the relationship between the DIA ratio and pyrolysis temperature is, in fact, linear in the range 600-1000 "C. The above examples were all pulse pyrolyses, using a rate of 20 "C ms-1 to achieve the final temperature. It was of interest to study the converse situation, ie., pyrolysis to the same final temperature each time, but at different rates of heating. Rates between 5 and 60 "C min-1 were chosen, and the pyrolysis products compared with chromatograms of pulse pyrolyses (20 "C ms-1) of the same material. In order to achieve this, some means of retaining the compounds produced during a slow pyrolysis must be used.For example, if pyrolysis is performed at 5 "C min-1, to a final temperature of 900 "C, the heating process will take 3 h. It was desired to collect all of the volatiles produced during this time period and to inject them on to the GC column for a single chromatogram. For this reason, the CDS Pyroprobe 123 was selected, as it consists of a pyrolyser capable of heating coil or ribbon probes at rates as low as 5 "C min-l, and a sample concentrator that collects the pyrolysis products as they are produced. The unit uses a heated interface (into which the probe is inserted for pyrolysis) that is constantly swept with a stream of helium. This sample stream carries the pyrolysates out of the interface and into a Tenax packed trap, where they are retained while the helium is vented.After the pyrolysis is complete, the trap is backflushed with the GC carrier gas as it is pulse heated to desorb the pyrolysates from the Tenax. The GC carrier flow then takes the pyrolysates through a heated transfer line into the injection port of the gas chromatograph. Instrument parameters are listed in Table 1. Fig. 4(a) shows the material collected from a pyrolysis of polyethylene at 60 "C min-1 to 900 "C. Comparing this with Fig. l(b) (pulse pyrolysis at 900 "C for 20 s) shows the significant effect on the DIA ratio. If the heating rate is decreased to 30 "C min-1, the DIA ratio is further decreased, and a rate of 5°C min-1 diminishes the DIA ratio still more, as can be seen in Fig. 4(b) and 4(c). This decrease in the DIA ratio as the rate of pyrolysis is decreased (or an increase in the ratio as the rate is increased) mimics what was observed for the DIA ratio for pulse pyrolyses at various temperatures.In fact, if the rate were not known, or the pyrograms were assumed to be the result of pulse pyrolysis at a very rapid rate, the _ _ 1 1 1 I , , I I 1 1 11 12 13 14 15 16 17 18 19 Number of carbons Fig. 2. Pyrolysis of olyethylene at 20 "C ms-'. Diene to alkane ratio versus number orcarbon atoms 600 700 800 900 1000 Temperature/"C Fig. 3. Pyrolysis of polyethylene at 20 "C ms-1. Diene to alkane ratio versus temperature (b) Fig. 4. Gas chromatograms showing effect of pyrolysis rate on polyethylene. (a) Ramp 60 "C min-1, temperature 900 "C, GC programme 50 "C for 2 min, 4 "C min-1 to 280 "C.( b ) Ramp 30 "C min-1, temperature 900 "C. (c) Ramp 5 "C min-1, temperature 900 "CANALYST, SEPTEMBER 1986, VOL. 111 1067 Table 1. Operating conditions of the sample concentrator and gas chromatograph 320 sample concentrator- Valveoven , . . , 275 "C Transferline . . . . 275 "C Thermaldesorber . . 250 "C Trap desorption . . 275 "C for 2 min Trap bake . . . . 290 "Cfor2min Sample carrier . . 30 ml min-1 Trap carrier . . . , 30 ml min-l GCcarrier . . . . 60mlmin-l Gas chromatograph- Column , . . , 50 m x 0.25 mm i.d. SE-54 Initial . . . . . . 50 "Cfor2 min Ramp . . . . . . 4 "Cmin-l Final . . . . . . 280 "C Detector . . . . FID Carrier . . . . . , Heli~m32Ibin-~ chromatogram would resemble pyrolysis at lower tempera- tures. A pyrolysis at 5 "C min-1 to 900 "C, for example, produces a DIA ratio that resembles the products of a pulse pyrolysis at 625 "C.Similarly, 60 "C min-1 to 900 "C produces a pattern with a DIA ratio that would be expected from a pulse pyrolysis at about 750 "C. System reproducibility was evaluated through replicate pyrolyses and calculation of the ratios of the areas of peaks for the alkadiene and alkane containing thirteen carbons. At 900 "C this peak area ratio averaged 1.467 k 0.074, which produces a relative standard deviation of 5.04%. Conclusions Both the pyrolysis temperature and the rate at which that temperature is achieved have significant effects on the formation of pyrolysates from a solid polymer. Pyrolyses at very slow rates ("C min-1 rather than "C ms-l) produce pyrograms that resemble those of pulse pyrolyses at lower end-point temperatures. It is possible that the slower rates permit one to see more of the primary pyrolysis products, which may be formed at lower temperatures then volatilised and removed from the pyrolysis chamber before secondary pyrolysis reactions ensue. Pyrolyses at elevated temperatures, or at a faster rate to the same final temperature, may enhance secondary pyrolysis reactions. For polyethylene, the general trend is for the production of more terminally unsaturated dialkenes at higher temperatures or faster rates, whereas the corresponding alkane is enhanced at lower temperatures and slower rates. References 1. 2. 3. 4. Seegar, M., and Barrall, E. M., J. Polym. Sci., Polym. Chem. Ed., 1975, 13, 1515. Ahlstrom, D. H., Liebman, S. A., and Abbas, K., 1. Polym. Sci., Polym. Chem. E d . , 1976, 14, 2479. Sugimura, Y., and Tsuge, S., Macromolecules, 1979, 12, 512. Levy, E. J., and Walker, J. Q., J . Chromatogr. Sci., 1984, 22, 49. Paper A51385 Received October 28th I985 Accepted April Ist, I986

ISSN:0003-2654    

DOI:10.1039/AN9861101065

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1986, VOL. 111 1069 Determination in Urine of Diisocyanate-derived Amines from Occupational Exposure by Gas Chromatography - Mass Fragmentography Christina Rosenberg and Heikki Savolainen Institute of Occupational Health, Department of Industrial Hygiene and Toxicology, Haartmaninkatu I , S F-00290 Helsinki, Fin Ian d Hexane-1,6-diamine was determined in the urine of car painters exposed to paint aerosols based on 1,6-bis(carbonylamino)hexane. The concentration of inhaled functional NCO groups during a 15-min exposure averaged 2.8 f 0.8 pmol m-3 (standard deviation, n = 5). The peak diamine concentration in urine of 63 F 33 nmol (mmol creatinine)-1 (standard deviation, n = 5) occurred 30 min after the end of exposure. The diamine was extracted from the urine with Sep-Pak silica gel cartridges and determined as the perfluoroamide derivative by capillary gas chromatography - mass fragmentography.The determination of diisocyanate- derived diamines in urine offers a selective and sensitive means of biological monitoring of occupational isocyanate exposure. Keywords : 1,6- Bis (ca rbon yla m in 0 ) h exane; h exane- 1,6-diamin e; isoc ya na tes; urinalysis; occupational exposure Organic diisocyanates are used in, e.g., foams, synthetic rubber, adhesives and paints. Articles made of polymers of diisocyanates (polyurethanes) combine outstanding material properties with economic processing costs for various applica- tions. The main isocyanates in industrial use are the aromatic diisocyanates, bis(carbony1amino)toluene (a 2,4 - 2,6- isomeric mixture) and 4,4'-bis(carbony1amino)diphenyl- methane.In addition to the monomeric isocyanates, various isocyanate-terminated pre-polymers derived from monomeric sub-units have been introduced. 1 These have higher relative molecular masses and, consequently, lower vapour pressures. One such polyisocyanate is the biuret structure of 1,6-bis(car- bony1amino)hexane. This oligomer is the major building block in two-component polyurethane spray paints used in car paints and many other coatings. The isocyanate component usually contains less than 1% of the volatile monomer.2 Conjugated diamines from the percutaneously absorbed monomeric aromatic isocyanate were excreted in the urine of rats.3 Likewise, the inhalation of vapour of monomeric 4,4'-bis(carbonylamino)diphenylmethane caused dose- related 4,4'-bis(amino)diphenylmethane excretion in urine.4 In this paper we describe a novel gas chromatographic - mass fragmentographic method for the determination of exogenous aliphatic diamines in urine.Extracts of acid hydrolysed samples were purified on Sep-Pak silica gel cartridges. The isolated diamine was converted into its perfluoro-substituted amide and determined by capillary gas chromatography - mass fragmentography . The peak hexane- 1,6-diamine concentration in urine occurred 30 min after the end of the exposure. Experimental Analysis of Isocyanates Isocyanates were sampled from air on reagent impregnated glass-fibre filters2.5 for 2-15 min at a flow-rate of 1 1 min-l. A 300-mg amount of N-[(4-nitrophenyl)methyl]propanamine hydrochloride (Regis Chemicals, Morton Grove, IL, USA) was dissolved in 25 ml of distilled water.Sodium hydroxide solution (15 ml, 1 M) was added, and the mixture was extracted twice with 25 ml of hexane (HPLC grade). A 10-fold dilution of the combined extracts with hexane dried over sodium sulphate gave a 2.6 x 10-3 M impregnation solution. A 20-ml volume of this solution was transferred into a beaker with 20 glass-fibre filters. The filters, without binders and 13 mm in diameter, were obtained from Millipore (Bedford, MA, USA). The beaker was protected against UV light and the solvent was evaporated with a gentle stream of nitrogen. Evaporation was continued until the filters no longer clung to the beaker. The dry filters were placed in 13 mm filter holders (Swinnex, Catalogue No.SX0001300, Millipore). The filters were stored in the dark at -20 "C and were stable for at least 4 weeks. Reference standards were prepared by adding known amounts of the respective isocyanate dissolved in dichloro- methane on to the filters. 1,6-Bis(carbonylamino)hexane was purchased from Fluka (Buchs, Switzerland) and the biuret oligomer from Bayer (Leverkusen, FRG). The standard and sample filters were placed in tubes containing 1 ml of the chromatographic solution, shaken for 15 s and then centri- fuged. A 100 pl volume of this solution was injected into a Kontron Model 6000 liquid chromatograph operated with a Uvicon 720 LC variable-wavelength detector at 275 nm. A Hypersil ODS 5 pm, 125 x 4.6 mm i.d. column was used for reversed-phase chromatography. The urea derivatives of the isocyanates were eluted isocratically at 1.5 ml min-1 with acetonitrile - water - triethylamine (64 + 35 + 1) at pH 3. Determination of Hexane- 1 ,ddiamine in Urine Urine aliquots ( 5 ml) were heated with 200 pl of concentrated sulphuric acid at 100 "C for 1 h to hydrolyse the acetylated diamine conJugates.3 The pH of the cooled hydrolysate was adjusted to 9.0-9.3 with saturated sodium hydroxide solution.The sample was loaded on to the Sep-Pak silica disposable column (Waters Associates, Milford, MA, USA). The cart- ridge was rinsed with 5 ml of water and the diamine was eluted with 5 mi of 0.05 M sulphuric acid. The first 2 ml of the eluate were discarded. The diamine was recovered from the last 3 ml by extraction with 1 ml of toluene after the addition of 1.5 g of sodium chloride and 6 ml of saturated sodium hydroxide solution.The organic phase was removed and 20 pl of heptafluorobutyric anhydride (HFBA) (Pierce Chemicals, Rockford, IL, USA) were added. The mixture was shaken for 30 s, and after 20 min the excess of derivatisation reagent was removed by extraction with 1 ml of 1 M dihydrogen phosphate1070 ANALYST, SEPTEMBER 1986, VOL. 111 buffer (pH 7). The toluene layer was removed and dried over sodium sulphate. The diamine was determined by capillary chromatography - mass fragmentography. A Hewlett-Packard 5890A capillary gas chromatograph was linked to a Hewlett-Packard 5970A quadrupole mass selective detector. Ionisation was achieved by electron impact (70 eV) and mass spectra were recorded using full (rnlz 15-550) or multiple ion monitoring (mlz 226 + 339). A fused-silica column, 25 m x 0.2 mm i.d., coated with 5% phenylmethylsilicone (cross-linked, Hewlett-Packard) was coupled directly to the ion source.A linear temperature gradient from 115 to 210 "C at 10 "C min-1 was used. The carrier gas was helium with a linear gas velocity of 27 cm min-1. The injector was operated in the splitless mode with an inlet temperature of 220 "C and splitless time of 1.5 min. The heptafluoroacyl derivative of hexane-l,6-diamine eluted at 169-170 "C. Aliquots of a standard solution of hexane-l,6-diamine in 0.05 M sulphuric acid were added to control urine samples and analysed together with the actual specimens.These standards were used for quantitation. The hexane-l,6-diamine results were corrected for the excretion of creatinine determined by the alkaline picric acid method. Exposure Conditions and Sampling The 1,6-bis(carbonylamino)hexane-based paint was nebulised by means of compressed air in an exposure chamber (9.7 m3) for approximately 30 s. The average concentration of the oligomer was 34% and that of the monomer 0.24% in the hardener.2 After nebulisation, air samples in the closed chamber were collected on reagent-impregnated glass-fibre filters in the breathing zone of five car painters exposed for 15 min.2 Urine voided 30 min to 8 h after the exposure was sampled and immediately deep-frozen and stored at -25 "C until analysis. The average isocyanate exposure during a work shift in the paint shop was determined for one car painter.The painter used a respirator with a compressed air supply. Air samples were collected both outside and inside the respirator. Urine voided after the work shift was analysed. Results and Discussion Sep-Pak cartridges of silica gel and CIS have been used previously for sample clean-up of biological fluids for quanti- tative analysis by gas or liquid chromatography.6.7 The aliphatic diamine was extracted from urine using a silica gel cartridge in this study. The procedure is a modification of that described by Brossat et aZ.6 The clean-up of urine samples was much more efficient with the silica gel cartridge than the acid back-extraction procedure used in our previous studies.3 The more effective purification of the sample is needed for the gas chromatography with the electron-capture detection.The silica gel step did not affect the over-all recovery, which was 82 k 7% (standard deviation, n = 9). The recovery of hexane- 1,6-diamine depended strongly on the alkalinity of the extraction mixture, i.e., 2 ml of saturated sodium hydroxide solution per 1 ml of 0.05 M sulphuric acid were needed to obtain an acceptable recovery. The same effect was shown in an earlier study for the aromatic toluenediamine.8 The total ion chromatogram of the heptafluoroamide of hexane-l,6-diamine is shown in Fig. l(b). The acylated derivative was synthesised using a standard procedure for symmetric diamides.gJ0 The molecular ion of rnlz 508 appeared with a weak relative abundance, 2.8%, in the electron-impact spectra [Fig.l(a)]. The base peak of mlz 226 is formed when the carbon chain is split leaving the fragment C3F7CONHCH2. The ions of mlz 69 and 169, corresponding to CF3 and C3F7, were obtained with high relative abundances (75 and 46%), but as they are not characteristic of the parent loor 226 339 (Peak identifier) 100 200 300 400 500 rnlz n a 0 , 1 5 7 9 1.1 Time/mi n Fig. 1. (a) Mass spectra and (b) total ion chromatogram of HFBA derivative of hexane-1,6-diamine (115 ng pl-1). The ions correlate to the following uresumed fragmentation: rnlz 508 (M+'); 339 M - C3F7); 3f1 *(M - C,F,CUO); 296 (M - C,F$ONH); 282 M - C3F7CONHCHZ); 226 (M - C3F7CONH(CH&; 169 (C,F,); 114 (M - 2C2F7CO); 69 (CF?). Conditions: column, fused silica, 25 m x 0:2 mm i b ., coated with'5% phenylmethylsilicone; carrier gas, helium, 27 cm min-1; injector, splitless, 1.5 min at 225 "C; and co.lumn linear temperature gradient, from 115 to 210 "C at 10 "C min-1 molecule, these fragments cannot be used for the peak identification. The residual ion of mlz 282, with a relative abundance of 43%, formed when m/z 226 is split off, could be used to obtain structural information. However, it is close to rnlz 281, the impurity peak of septum or silicone methyl coatings, and is therefore not recommended for the identification. Hence, the ion of rnlz 339, produced when C3F7 is split off, was chosen as the peak identifier. It appeared with a relative abundance of 32%. The quantification was effected by mass fragmentography of the base peak of rnlz 226.Peak areas were integrated and the limit of detection, defined as the minimum amount of diamine injected with more than 40 integrator counts, was 0.2 pmol. A 5-fold concentration of the urine sample was achieved by the purification on the silica gel cartridge resulting in a detection limit of 2 nmol (mmol creatinine)-1. The base line for the control showed only noise and the integration over the region spanned by the amine peak gave no more than 10 integrator counts. The calibration graph for the perfluoroacyl derivative was linear (y = 2 . 6 5 ~ - 26.82, where y is area counts and x the amine concentration in ng ml-1) over the range 20-900 pg with a 1-pl sample injected ( r = 0.997). The urea derivative for isocyanate determinations is well established and has been tested for several diisocyanates, including the aliphatic 1,6-bis(carbonylamino)hexane and its oligomer with the biuret structure.2,5 Hexane-l,6-diamine derived from the parent isocyanate was demonstrated in workers after aerosol exposure forANALYST, SEPTEMBER 1986, VOL.111 1071 Table 1. Concentration in urine of amine derived from diisocyanate after aerosol exposure for 15 min Concentration of hexane-l,6-diamine/ nmol (mmol creatinine)-' after exposure for: Exposure level/ Subject pmolNCOm-3* 30min 4h 6h 8 h 1 2.8 16 13 < 2 <2 2 4.0 110 37 9 <2 3 2.2 49 56 < 2 <2 4 3.4 48 25 < 2 <2 5 1.8 90 48 < 2 <2 * Derived from 1,6-bis(carbonylamino)hexane. 15 min (Table 1). The peak concentration of the amine occurred 30 min after the end of the exposure.The concen- tration of the inhaled functional NCO groups in the breathing zone, originating from 1,6-bis(carbonylamino)hexane, varied from 1.8 to 4.0 pmol m-3. The spray chamber in the car paint shop was approximately 85 m3 in size and had roof to floor exhaust ventilation with a nominal air exchange rate of 11.5 m3 min-1. The average isocyanate monomer concentration in the work-room air was 0.2 ymol m-3. No monomer was detected inside the air hood of the painter, and no hexane-l76-diarnine was observed in his urine sample. Conclusions Amine metabolites have been identified in the urine of painters exposed to 1,6-bis(carbonylamino)hexane. The gas chromatographic technique coupled with mass fragmen- tography allowed the detection of the isocyanate-derived amine in the 0.2 pmol y1-1 range. We thank Ms. Raija Vaaranrinta for technical assistance. References 1. Hardy, H. L., and Devine, J . M., Ann. Occup. Hyg., 1979,22, 421. 2. Rosenberg, C., and Tuomi, T . , Am. Znd. Hyg. Assoc. J . , 1984, 45, 117. 3. Rosenberg, C., and Savolainen, H., J. Chromatogr., 1985,323, 429. 4. Rosenberg, C., and Savolainen, H., J. Chromatogr., 1986,358, 385. 5. Tucker, S. P., and Arnold, J. E., Anal. Chem., 1982,54,1137. 6 . Brossat, B . , Straczek, J., Belleville, F., Nabet, P., and Metz, R., J . Chromatogr., 1983, 277, 87. 7. Mumtaz, M., Narasimhachari, N., and Friedel, R. O., Anal Biochem., 1982, 126, 365. 8 . Rosenberg, C., Analyst, 1984, 109, 859. 9. Drozd, J., J . Chromatogr., 1975, 113, 303. J . Chromatogr., 1984, 303, 89. 10. Skarping, G., Smith, B. E. F., and Dalene, M., Paper A51427 Received November 18th, 1985 Accepted March 25th, 1986

ISSN:0003-2654    

DOI:10.1039/AN9861101069

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1986, VOL. 111 1073 Determination of Organic Sulphides by Enthalpimetry Using Chromyl Chloride Mieczyslaw Wronski and Awn. S. Abbas Department of Chemical Technology and Environmental Protection, University of Lodi, Nowotki 18, Poland Thermometric titration with chromyl chloride and direct injection enthalpimetry with the same reagent have been used for the determination of sulphides in hydrocarbon solvents; the second procedure was found to be superior. Chromyl chloride and sulphides react in a 1 : 1 ratio, producing a brown precipitate, the heat of reaction being 250 kJ mol-I. The limit of determination is 3 p.p.m. and the relative standard deviation 3.3% at 60 pap.m. of sulphide-sulphur. The method can be applied in the presence of unsaturated compounds and polycyclic aromatic hydrocarbons.Polycyclic aromatic hydrocarbons also react with chromyl chloride, but the reaction with sulphides is dominant. The method has been suggested for the determination of sulphide- sulphur in petroleum products. Keywords: Chrom yl chloride; enthalpimetry; sulphide-sulphur; petroleum products; thermometric titration In spite of a variety of methods for the determination of sulphides in petroleum products1-3 the analysis of real samples from petroleum refineries can present great difficulties. Recommended methods based on the oxidation of sulphides to sulphoxides give high results. Ultraviolet spectro- photometry of complexes with iodine is not particularly useful for the direct measurement of sulphides in petroleum samples.In our efforts to determine sulphides using an enthalpi- metric approach, we have examined several chemicals that react easily with sulphides, e.g., chlorine, nitrogen trichloride, chloramine T and chromyl chloride, and we have found chromyl chloride to be the most satisfactory. Chromyl chloride4 is readily formed by the reaction of hydrogen chloride with chromium(V1) oxide: Cr03 + 2HC1 = Cr02C12 + H20 In the presence of aliphatic hydrocarbons, a red solution is formed; It has been found that in the presence of sulphides, chromyl chloride gives slightly soluble complexes and the heat evolved can be evaluated by enthalpimetry. Experimental Apparatus The apparatus used for the enthalpimetric analysis was as described in the literature.5>6 The 60-ml reaction cell is placed in a 250-ml Dewar flask.In the lid of the reaction cell are fixed two thermistors, a stirrer and a PTFE 2-mm diameter tube to act as an inlet for the reagent and calibration heater. The enthalpimetric apparatus also includes a regulated, variable, low-voltage power supply, a digital voltmeter, a timer for the calibration heater, a temperature measuring bridge and recorder. The reagent stream is produced by means of a 338 B variable-speed dispenser, made in Poland. Chemicals and Reagents The samples of sulphide were purchased from Aldrich Chemie, Steinheim, FRG. Hexane and octane were of analytical-reagent grade. The solution of chromyl chloride in octane is prepared by shaking 5 g of chromium(V1) oxide, 10 ml of concentrated hydrochloric acid, 3 ml of concentrated sulphuric acid and 250 ml of octane.The clear red organic layer is separated and stored at 0 "C in a brown flask for not longer than 3 d. As the solution is decomposed by water to form chromic acid, its concentration (ca. 0.1 M) can be determined by titration with sodium thiosulphate in the presence of potassium iodide and sulphuric acid. The solution is then diluted with octane as required. Qualitative Analysis The solution of chromyl chloride in octane can be used for the detection of sulphides and other compounds by adding 1 ml of the reagent to 5 ml of sample and noting the change in colour. Sulphides and disulphides change the colour to brown and in more concentrated solution a brown precipitate is formed. The limit of detection is 5 p.p.m. for sulphide-sulphur and 50 p.p.m. for disulphide-sulphur.The monocyclic aromatic compounds change the colour to brown but no precipitate is formed. The limit of detection in this instance is only 2%. Polycyclic aromatic homo- and heterocyclic compounds such as anthracene, phenanthrene, pyrene and thianthrene change the colour to black and form black precipitates. The sample remains clear but the colour changes to yellow if compounds with active hydrogen are present. Quantitative Analysis Two procedures for the determination of sulphides have been examined, one based on a thermometric titration with chromyl chloride solution and the other on direct injection enthalpimetry of sulphides into an excess of chromyl chloride solution. Procedure 1. Thermometric titration A 40-ml hydrocarbon sample containing 20-150 pmol of sulphide and not more than 25% of monocyclic aromatic hydrocarbons is placed in the reaction cell and the lid is fixed.After stirring for a few minutes at constant temperature, the 0.64.8 M solution of chromyl chloride in octane is added with a constant flow-rate of 1-2 ml min-1 until the end-point can be determined on the graph indicated by the recorder. Procedure 2. Direct injection enthalpimetry A 0.02 M solution of chromyl chloride in octane is placed in the reaction cell and the lid fixed. After stirring for a few minutes at constant temperature the 0.5-2-ml samples are injected into the cell, followed by washing with 0.5 ml of hexane. Another 1 ml of hexane is added in order to determine the heat of dilution, which is subtracted from the result for the total volume of the sample.1074 ANALYST, SEPTEMBER 1986, VOL. 111 Results and Discussion Typical thermometric titration curves for sulphides are shown in Fig.1. All curves have a sharp beginning and end-point and are similar for different sulphides. In the blank sample the addition of chromyl chloride solution results in a heat evolution of ca. 26 kJ per mole of Cr02C12, which can be assumed to be the heat of dilution. When the reaction with the sulphides is complete, the excess of chromyl chloride evolves the heat of dilution, but the value is not constant and decreases with increasing amounts of sulphides. The results for the determination of the heats of reactions and molar ratios found for different sulphides are summarised in Table 1.From an analytical point of view, it is an advantage that the molar heats of reaction are essentially, within the limits of error, very similar and can be taken as 250 kJ mol-l. Dithiane behaves as a simple sulphide and the second sulphur atom is not indicated. The next point of analytical importance is that the molar ratio of chromyl chloride consumed to sulphide may be assumed to be unity. The consumption of chromyl chloride calculated from the time of reaction, flow-rate and concen- tration of reagent is not as useful and sensitive as the heat of reaction, but it may be involved in the analysis of complex mixtures. The mean relative standard deviation for the thermometric titration of sulphides at 60 p.p.m. of sulphide-sulphur can be accepted as 3.3%, compared with 6% calculated for consump- tion.The problem of selectivity is discussed on the basis of results f f Time -+ Fig. 1. Enthalpimetric titration curves of sulphides with 0.08 M chromyl chloride in hexane. Flow-rates: A, B, F, G, H and I, 2 ml min-l; C, D, E and J, 1.5 ml min-l. A, Dimethyl 135; B, diethyl 111; C, dibutyll37; D, di-sec-butyl 110; E, di-tert-butyl99; F, didodecyl80; G, diallyl 101; H, dibenzyl 112; I, 1,3-dithiane 106 pmol; J, blank shown in Fig. 2. Consider line A. In spite of a 140-fold excess of hexene with respect to diethyl sulphide, the end-point is still determinable and the result is in agreement with the expected value. The more active cyclohexene is tolerable only at excesses of less than 14-fold. Line C shows the titration of anthracene.It resembles the titration curves of sulphide. However, on the addition of 56 pmol of diethyl sulphide, a distinct end-point corresponding to the added sulphide can be observed. The result this time is too high and a correction should be introduced. The lines E and F demonstrate the titration of phenanthrene and phenanthrene with added sulphide. No shift of the end-point can be observed. Conse- quently, when dealing with an unknown sample, it may be recommended that the enthalpimetric titration is repeated with some sulphide added. Table 2 gives the heats of reaction determined by the injection of sulphides into the solution of chromyl chloride given in Procedure 2. The heat of reaction is independent of the type of sulphide and is equal to 300 kJ mol-1, with the exception of diphenyl sulphide. The increase in the heat of reaction from 250 to 300 kJ mol-1 may be explained by assuming that in the presence of a great excess of chromyl chloride solution, complexes of the composition R Z S ( C ~ O ~ C ~ ~ ) ~ are formed.Disulphides react with chromyl chloride, but the direct titration (Procedure 1) does not give a sharp end-point. The heat of reaction can be conveniently determined using Procedure 2. Table 3 and Fig. 3 demonstrate the combined ~~ Table 1. Molar heats of reaction and mole ratio as determined by the thermometric titration of sulphides with chromyl chloride solution, f standard deviation (n = 5) Heat of reaction/ Sulphide kJ mol -1 Dimethyl . . . . 243 f 8 Diethyl . . . .. . 266 f 8 Di-n-butyl . . . . 254 2 8 Di-sec-butyl . . . . 251 _+ 10 Di-tert-butyl . . . . 248 k 11 Di-n-dodecyl . . . . 240 _+ 8 Diallyl . . . . . . 258 f 9 247 f 8 Dibenzyl . , . . . . 1,3-Dithiane . . . . 254 f 9 Molar ratio, Cr02C12/R2S 0.97 f 0.07 1.02 f 0.05 0.99 k 0.06 1.05 f 0.08 1.05 f 0.05 0.99 f 0.07 1.01 f 0.05 1.02 f 0.07 1.04 k 0.08 Table 2. Molar heats of reaction determined by direct injection enthalpimetry of sulphide into an excess of chromyl chloride solution, k standard deviation ( n = 5) Amount taken/ Sulphide pmol Diethyl . . . . . . 22 Di-n-butyl . . . . 12 Di-n-butyl . . . . 24 Di-n-dodecyl . . . . 24 Diphenyl.. . . . . 120 1,3-Dithiane . . . . 28 Di-n-butyl disulphide . . . . 25 Heat of reaction/ kJ mol -1 298 f 10 312 k 17 302 If: 11 298 f 9 118 _+ 8 301 f 12 407 f 19 Table 3.Combined analysis of petroleum products Sulphur content, % m/m Heat evolved with chromyl chloride/J ml- 1 Chromyl chloride Excess Ti tration Excess KBr - KBr03 Titration Sample Sp. gr. (Procedure 1 ) (Procedure 2) Total* titration (Procedure 1) (Procedure 2) Kerosene1 . . . . 0.774 0.54 32.8 0.006 0.32 0.009 0.45 Kerosene11 . . . , 0.692 2.85 3.70 0.055 0.055 0.055 0.057 Kerosene111 . . 0.783 9.75 22.6 0.15 0.29 0.14 0.31 Diesel fuel . , . . 0.843 20.3 200 0.35 0.90 0.31 2.52 Crudeoil . . . . 0.853 115 242 1.86 - 1.73 3.03 * Determined by combustion.ANALYST, SEPTEMBER 1986, VOL. 111 1075 Time - Fig. 2. Enthalpimetric titration curves of sulphides with chromyl chloride in the presence of interfering substances. A, 56 pmol of diethyl sulphide plus 7850 pmol of hexene; B, 56 pmol of diethyl sulphide plus 800 pmol of cyclohexene; C, 22 pmol of anthracene; D, 56 pmol of diethyl sulphide plus 22 pmol of anthracene; E, 22 pmol of phenanthrene; and F, 56 pmol of diethyl sulphide plus 22 pmol of phenanthrene Time - Fig.3. Enthalpimetric titration curves of petroleum products with chromyl chloride. Total volume 40 ml diluted with hexane. A, 5 ml of kerosene I; B, 5 ml of kerosene 11; C, 1 ml of kerosene 111; D, 2 ml of kerosene 111; E, 1 ml of diesel fuel; F, 0.1 ml of crude oil; and G, 0.2 ml of crude oil results of the analysis of some petroleum products, including diesel fuel and crude oil. The samples of known sulphur content determined previously by combustion were titrated with a bromide - brornate solution as described for determina- tion of sulphides,’ and were analysed using Procedures 1 and 2.The sulphur content was calculated from the equations 3.2 Qi s1 =- 250 d 3.2 Q2 s2 =- 300 d where Ql and Q2 are the heats of reaction determined by Procedures 1 and 2 and d is the density. The sulphur content determined by thermometric titration contributes 8&100°/~ of the total sulphur. The difference can be assumed to be due to thiophene and elemental sulphur, which do not react with chromyl chloride. The consumption of bromine gives results that are evidently too high, with one exception. There must be other substances present that react with bromine, apart from sulphides. The highest sulphur content is found by using Procedure 2, indicating the presence of substances that react with chromyl chloride but not with bromine. The difference can be assumed to be due to polycyclic compounds. Consequently, it may be expected that the combined analysis of petroleum will provide some new important information. It may be added that chromyl chloride has been used for the separation of sulphur compounds from hydrocarbons followed by chromatographic examination.’ References 1. Karchmer, J. H., Editor, “The Analytical Chemistry of Sulfur and its Compounds, Part 2,” Wiley-Interscience, New York, 2. Ashworth, M. R. F., “The Determination of Sulphur- containing Groups,” Volume 3, Academic Press, New York, 3. Karaulova, E. N., “Chimia Sulfidov Nefty,” Nauka, Moscow, 1970. 4. Mellor, J. W., “Comprehensive Treatise on Inorganic and Theoretical Chemistry,” Volume XI, Longmans Green, Lon- don, New York and Toronto, 1948, pp. 390-399. 5. Vaughan, G. A., “Thermometric and Enthalpimetric Titri- metry,” Van Nostrand Reinhold, London, 1973. 6. Jordan, J., Grime, J. K., Miller, D. C., Cullis, H. M., and Lohr, D., Anal. Chem., l976,48,427A. 7. Wronsky, M., and Abbas, A. S . , J. Chrornatogr., submitted for publication. Paper A51333 Received September 19th, 1985 Accepted April 17th, 1986 1972, pp. 1-79. 1977, pp. 1-113.

ISSN:0003-2654    

DOI:10.1039/AN9861101073

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1986, VOL. 111 1077 Description of Air Pollution by Means of Pattern Recognition Part 2* Geed Jan H. Roelofs, Frans W. Pijperst and Gfred A. P. E. Jakobs Institute for Theoretical Physics, Faculty of Sciences, University of Nijmegen, Toernooiveld, 6525 ED Nijmegen, The Netherlands Based on meteorological observations and hourly measurements of chemical constituents at various locations in the city of Schiedam in the Rhine-mouth area near Rotterdam, The Netherlands, a learning machine has been constructed for the description and prediction of complaint situations for polluted atmospheres. It is demonstrated that a seven-parameter model may classify about 80% of the complaint and between 50 and 60% of the non-complaint situations correctly. Sometimes a prediction of a complaint situation up to 6 h in the future is possible, but different meteorological conditions may hamper the predictive ability of the learning machine.The conditional probability describing the burden on the population in the area runs parallel with the reaction of the human nervous system to an exposure of increasingly concentrated noxious smelling air in test panel experiments. It is shown that the seriousness of the burden influences the total number of complaints filed within a certain period of time. Keywords: Pattern recognition; environmental control; Ba yes classification; air pollution In a previous paper' a method was developed to describe an air pollution problem by means of pattern recognition. This was shown to yield only moderate success when the weather conditions were stable; unstable weather conditions gave no success at all.Using this method, a learning machine was constructed based on the stability of the weather, SO2 concentrations measured at three different locations in the area under investigation and concentrations of hydrocarbons (saturated and olefinic) and ozone. This learning machine was based on 116 hourly reports between 10.00 and 18.00 h in a period from April to June 1979. In this period, no missing features were encountered, the weather was stable and about equal numbers of complaint hours and non-complaint hours were registered. The measurements were made in the Rhine-mouth area near Rotterdam, The Netherlands (Fig. 1). In this learning machine a sharp separation between the two categories of patterns was seen.However, owing to the severe limitations mentioned, this learning machine could not be used as a predictor of burden on the inhabitants of the area. In order to improve the predictions, a search was made for better descriptors, yielding the following criteria: a complaint hour was denoted as valid after consideration of the previous 2 h. When in one of these two hours a complaint was reported, the complaint hour itself was included in the training set. In order to combine this category with an about equally populated category of non-complaint hours, for each com- plaint hour a non-complaint hour was selected 24 h before or after the complaint. According to this procedure daily or seasonal effects were eliminated. Apart from this, another restriction was posed on the non-complaint hour: there should not be any complaint in the period between 5 h before and 5 h after the selected non-complaint hour.The training set was tested by hours taken from measurements in 1981 and 1982. It was learned from experience that sometimes a complaint was communciated 1-2 h after the burden was noticed. The concentrations of chemical compounds in air exhibit autocor- relation functions with characteristic decay times between 2 and 4 h. Because of these two observations, it was felt that not only were the actual concentrations measured at a given * For Part 1, entitled "Failures and Successes with Pattern Recognition for Solving Problems in Analytical Chemistry," see reference 1. t To whom correspondence should be addressed. instance of importance, but also the feature values in the immediate past (some hours before).Another consideration that was taken into account was the logarithmic response of the human nervous system to stimuli from the outside, which means that an increase in burden caused by noxious smells should be rated on a logarithmic rather than on a linear scale. These considerations dictated the feature transformations as listed below (Practical Considerations). Thus a learning machine was constructed that could serve as a classifier with more success than that published earlier.1 Practical Considerations According to the criteria mentioned in the Introduction, for the year 1982 a total number of 499 complaint hours and 8261 non-complaint hours was found (see Appendix for statistical information).From this set a training set was selected consisting of 35 non-complaint patterns and 38. complaint patterns. The features describing the learning machine are SO2 concentrations measured at four different locations, wind direction and the time dependence of two SO2 concentrations. The SO2 concentrations were transformed according to where x represents the hour of observation. The resulting values were autoscaled. The direction of the wind was rated 1 when coming from the industrial area (between south and west) and 0 otherwise. This rating was applied for the hour of observation and the previous 4 h, which gives a rating from 0 up to and including 5. The evolution in time of SO2 concentrations at a particular location is given by dSO2ldt = 1n{[S02(x) + SO2 (x - 1)]/ These feature transformations were found empirically, but showed the statistically relevant advantage of a better skewness and kurtosis in comparison with the non- transformed features.Thus a better correspondence with a normal distribution function is found. It was also noticed that the weight of the features for the class separation increased (Table 1, variance weights). ( S 0 2 ) n e w = ln[S02(4 + S02b - 1)1 [so2(x - 2) + so2 ( x - 311) (1)1078 ANALYST, SEPTEMBER 1986, VOL. 111 Results and Discussion The pre-processed features, logarithmised (except the wind direction) and autoscaled, were weighted (Table 1) and combined according to the Karhunen - Lokve transformation. A set of orthonormal eigenvectors was obtained, from which the two with the highest eigenvalues were selected (Table 2).These two eigenvectors, in which the SO2 concentrations at various locations contribute highly, span a plane in which about 72% of the information is accounted for. On this plane the pattern space is projected, including the patterns of the test set (Fig. 2). It is seen that an area with only complaint hours (at the left) is separated from an area with only non-complaint hours by an area where both categories are found. This is an indication of the probability distribution that describes the situation. In such a situation, the non-linear classification according to the nearest neighbour method may give an indication of the quality of the learning machine. The results for such a classification for five nearest neighbours are given in Table 3.Table 3 is based on a Euclidian distance matrix in the seven-dimensional feature space spanned by the features listed in Table 2. It is seen that the non-complaint hours (77% correctly classified) are less predictable than the complaint hours (87% correctly classified). These classifications alter only a few per cent. when less nearest neighbours are consulted. The results of a test set, obtained in the same manner, are listed in Table 4. The classification is only to a minor extent less satisfactory than that of the learning machine itself. In order to show the ability of this learning machine for the prediction of complaint situations hidden in the near future, a test set of 25 successive hours (May 17th 1982, from 01.00 to 24.00 h) is projected on the application of the same Table 1.Comparison of variance weights for the two category separation of direct and of logarithmised features Variance weights? Feature (xi)* X LnU dSO,(lO)/dt - S02(10) . . . . 1.77 2.34 S02(11) . . . . 1.77 2.25 SOZ(12) . . . . 1.30 1.68 soz(13) . . . . 1.36 1.67 . . . . 1.39 dSO,(ll)/dt . . . . - 1.39 * In the notation SO&), y denotes the sampling location (see t Calculated for the two-category situation 1 and 2 according to geographical map). where 3ti denotes the expectation value for the square of feature j in category i, Ni the population of category i, Zj,i the expectation value of the feature j in category i and o ,,i2 its variance. Table 2. Karhunen-Lokve transformation of logarithmised features Eigenvector, YO Feature First Second Sulphur dioxide (10) .. . . 34 0 Sulphur dioxide (11) . . . . . . 26 32 Sulphur dioxide (12) . . . . . . 11 21 Sulphur dioxide (13) . . . . . . 10 36 Directionofwind . . . . . . 2 4 dS02( 10)ldt . . . . . . . . 11 4 dS02( 1l)ldt . . . . . . . . 6 3 Eigenvalue (= information) 58.3% + 13.4% = 71.7% mathematical process as used for the learning machine (Fig. 3). Here it is demonstrated that at the beginning of the observation period, the hours are located in the non-com- plaint area. It is shown by the drawn lines that gradually the successive hours move towards the region where complaint hours should be expected and indeed two complaint hours are 4 48 438 SO2 - measuring location * Measuring station DCMR a Industrial area City area Fig.1. automatic SOz sampling and measuring locations Map of the Rhine-mouth area west of Rotterdam showing the V 0 m 0 0 0 00 0 0 0 D X = 1 Karl Fig. 2. Karhunen-Lokve projection of logarithmised features: 0, non-complaint hours (training set); 0, complaint hours (trainin set); 0, non-complaint hours (test set); W, complaint hours (test se8ANALYST, SEPTEMBER 1986, VOL. 111 1079 found (at 13.00 and 14.00 h). Hereafter the trace moves out of the complaint area and the observation period ends with non-complaints in the non-complaint region. The nearest neighbour technique operates well here. In order to demonstrate the failure of the same learning machine, a test set of 24 successive hours from July 9th, 13.00 h, to July loth, 12.00 h, 1982 is applied in the same way as that of May 17th (Fig. 4).Here all hours linger in the same area, denoted by a moderate probability of complaints. Sometimes a complaint is actually made, but prediction is virtually impossible. Application of the nearest neighbour technique on these 24 hours does not yield a good classifi- cation here. The results so far show the need for a more probabilistic descriptor, which will be given in the next section. Bayes Statistics Probablistic Description of Complaint Situations From an observation of the Karhunen - Loeve projections of pattern vectors denoting complaint and non-complaint hours, it is seen that no sharp distinction can be made between a complaint area and a non-complaint area. There exists an in-between area with a non-zero probability for each of the two categories.The assumption can be made that on a line connecting the centres of gravity of the non-complaint area and the complaint area the probability for a complaint increases from about 0 to about 1 in a sigmoidal way. Such a relationship is also observed when a test panel is exposed to an increasing concentration of unpleasant smelling (noxious) air. The number of persons from the panel indicating that they could smell something bad increases 'gradually with increasing logarithm of the concentration of pollutant. In order to describe these phenomena, a Bayesian statistical approach is chosen.3 In accordance with such an approach, a probability is attributed to each position in the pattern space given the chance of a complaint situation.This probability function is constructed from knowledge derived from a training set with a statistical amount of complaint and non-complaint situations. The reliability of this function depends critically on the availability of a representative training set. For a two-category model, as encountered here, the probability for x (x being a member of one category out of n categories) is given by Pbl = P[xla(l)l*p[a(l)l + p[xla(2)1*p[a(2)1 + - * - +p[xla(n)]*p[a(n)] . . . . . . * (2) where a(i) are categories with i = 1,2,3, . . ., n. p[a(i)] equals the a priori probability for the ensemble a(i) andp[xla(i)], the conditional probability for encountering x when ensemble a(i) is given. In Bayes statistics the question is reversed: if an event x is encountered, then what is the probability that this event belongs to the ensemble a(i)?, or, find p[a(i)Ix].The prob- ability that event x belongs to ensemble a(i) is then compared to the probability that event x belongs to all other ensembles For a two-category model, this comparison results in a aCi) * decision function Table 4. Classification according to five nearest neighbours for a test set consisting of 13 non-complaint hours and 13 complaint hours (YO) Predicted category 0 0 0 0 0 0. 0 0 00 0 :- 0 0 0 0 0 0 0 0 0 0 0 0 0 Table 3. Classification according to five nearest neighbours for the training set consisting of 35 non-complaint hours and 38 complaint hours (Yo) Predicted category Input category Non-complaint Complaint Non-complaints .. . . 77 23 Complaints . . . . . . 13 87 Input category Non-complaint Complaint Non-complaints . . . . 69 31 Complaints . . . . . . 15 85 I I I X = 1 Karl Fig. 3. Evolution of a complaint situation in time for May 17th 1982, followed from 01.00 h until 24.00 h, projected on the training set of Fig. 2. 0, Hours of May 17th; 0, start (01.00 h); 0, end (24.00 h) 0 0. 0 . 0 0 I l o - 0 0 0 0 I 0 X = 1 Karl Fig. 4. Evolution of a complaint situation in time for July 9th and 10th 1982, followed from 13.00 h until 12.00 July 10th. 0, Hours of July; 0, start (9th, 13.00 h); 0, end (loth, 12.00 h)1080 ANALYST, SEPTEMBER 1986, VOL. 111 where ~(~1x1) represents the probability function for event x belonging to category 1, p(xlX2) the same function for category 2, p(1) equals the a priori probability of category 1 and p(2) that of category 2.L(1) represents the loss function for category 1, a number between 0 and 1 describing the penalty on a misclassification in the other category; the same applies to L(2). The classification of an unknown pattern x in category 1 occurs when s > l ; otherwise it is classified in category 2. This decision function is only valid if a one-dimensional probability function p ( x ) can be calculated from all features in the pattern space. The Bayes classifier for the ARTHUR computer package, however, does not allow this calculation. Instead, the prob- ability function is calculated for each individual feature and all these probability functions are combined afterwards. Thus, the probability for the occurrence of x in category k based on the knowledge of the value for feature j equals ~ ( k ) Uk)p(xjlXjd pj(xjklxj) = K - * (4) k = l ~ ( k ) L(k) p(xjlX,k) in which K represents the number of categories (here 2).The probability distribution function for feature j is estimated from the histogram of this feature, derived from the training set patterns. For the combination of the individual distribution functions per feature, the ARTHUR package gives two alterna- tives: J PtotadXkb) = C * ~ X P { z lnbj(X,klxj)I) * * ( 5 ) ; = l or, alternatively J j = 1 Ptotal(Xkb) = C(a) * z bj(qkIxj)]u . I where C and C(a) represent normalisation constants. Implementation of the Bayesian Classifier for our Problem The 25 class resolution histograms of all features except the wind direction had a Gaussian distribution function.The resolution of the distribution over 25 classes proved experimentally to be sufficiently accurate. The selection of a loss function encounters a basic difficulty; because no knowledge is available, each selection is a subjective one. In such a situation the maximum likelihood classification is chosen, i.e., an equal loss factor, valued 1, for all categories. Another assumption made is that all new, unknown patterns of the test set are distributed according to the same distribution function as those from the a priori known patterns of the training set. Selection of test sets from the available data Three test sets were selected: test set one consists of 20 complaint hours and 19 non-complaint hours in 1982; test set two consists of 18 complaint hours and 18 non-complaint hours in 1981; test set three consists of all complaint hours in 1982 as far as no missing feature values are met.It contains 332 pattern vectors, but not all of these complaint hours satisfy the conditions mentioned in the introduction. The training set consists of three categories, described as follows: category one with high values for NO,, low for SO2 and low wind speed, yielding complaints from the population; category two, having high values for SO2 and a wide distribution for the values of NO,, wind coming from the industrial area, yielding complaints from the population; category three, no complaints, features show widely diverging values. The features for the description of the training set are NO, N02, SO2 measured at four different locations, the wind velocity and the wind direction (0 or 1) summed over a period of 5 h.In order to optimise the predicting power of the learning machine, the following a priori probabilities (non-normalised) were chosen:p(l) = O.l,p(2) = 0.3 andp(3) = 1.0. With these probabilities and the histograms for all features, the individual feature probabilities were combined according to equation (5). In Table 5 the results of the classification of the combined classes 1 and 2 (denoted “complaint”) and those of class 3 (denoted “non-complaint”) are presented for the training set and the test sets one and two. From Table 5 it can be seen that the complaint hours are well classified (80-81 % correct), which differs only to a minor extent from that of the training set (84%).However, a test with all the 332 complaint hours of 1982 (without missing features) reveals that only 65% of all hours are classified correctly. This may be interpreted as decisions resulting from an intermediate area in the pattern space, where the prob- ability for classification as complaint hour is almost equal to that for non-complaint hours. This interpretation is illustrated in Fig. 5, where the probability that the patterns of test set three belong to category 2, calculated by the Bayes classifier according to equation ( 5 ) , is plotted. It is seen that this classifier behaves as a yesho decision function, because all intermediate probabilities have a population of practically zero. This observation is in conflict with the expectation of an intermediate area where intermediate probabilities should be denoted.In a search for a better representation of this area, the feature probability functions were combined according to equation (6), with application of an optimised value for a and p(i) ( i = 1, 2 and 3). The optimisation, following an iterative procedure, was aimed for a maximum amount of correctly classified training set patterns in all categories, and resulted in the parameter values given in Table 6. From a comparison of the classifica- tion results, presented in Tables 5 and 6, it is seen that the actual classification hardly shows any difference. Test set three, comprising all complaint hours from 1982, is here only 64% correctly classified.However, the probabilities for category 2 of these complaints (Fig. 6) behave differently. Here the intermediate region is well populated and therefore this Bayesian classifier is not a simple yesho one. For the decision function probability(x) for class 2 probability(x) for class 3 s(x) = operating on all complaint hours of 1982, Fig. 7 is found, which on integration yields Fig. 8. The shape of this function corresponds well with a graphical representation of the experiment, mentioned before, where observations were made of a test panel that was exposed to air h 0 0.2 0.4 0.6 0.8 1 .o P,,t( x, I x) Fig. 5. Bayesian classifier equation (5) Histogram of complaint hours of class 2 as a function of theANALYST, SEPTEMBER 1986, VOL. 111 7 7 c - c 1081 - - - I with an increasing concentration of a noxious smelling odour.Here the number of observations on bad odour was plotted as a function of the logarithm of the concentration. The correspondence of these two functions suggests a relationship between the probability function s(x) and the burden felt by the population in the area, resulting in an increasing number of complaints. This will be elaborated under Quantification of the Burden of Air Pollution. Quantification of the Burden of Air Pollution One of the possibilities offered by pattern recognition is the ability to estimate unmeasurable properties of an object by the observation of measurements. Here the measurements are concentrations of chemical constituents of air and meterolog- ical conditions, as a function of time and location. From this base one aims for a yardstick for the amount of burden that is felt by the inhabitants of the area, which cannot be “measured” directly.In a search for the relationship between the amount of burden and the value of the decision function s(x), the interval between the minimum and maximum values of s(x) encoun- tered in practice was subdivided into nine intervals. The spacing of these intervals was chosen such that about equal amounts of complaint hours are found in each of the intervals (Table 7). The measurements describing each of the individual inter- vals are fed as separate categories to the ARTHUR computer program. Histograms for the various features are constructed and the weights for category separations are calculated. From the inspection of the results of these calculations it is seen that the velocity and the direction of the wind are mainly responsible for elevated SOz concentrations, provided that there is high industrial activity, and for high NO and NO2 concentrations, resulting from automotive traffic.A high wind velocity produces a high mixing and dilution of these components, resulting in a limited burden on the population and a low value for the decision function s(x). The direction of the wind becomes more important with a decrease in wind velocity: a prolongated wind direction from the industrial area results in an enhanced concentration of, e.g., SO2 and thus an increased burden; s(x) also increases here. A further decrease in wind velocity is accompanied by enhanced concentrations of NO and NO2.In combination Table 5. Classification with a Bayesian classifier, according to equation (5). p(1) = O.l;p(2) = 0.3;p(3) = 1.0 Predicted category Set Input category Non-complaint Complaint Training set . . Non-complaint 69 31 Complaint 16 84 Complaint 20 80 Complaint 19 81 Test set one . . Non-complaint 53 47 Test set two . . Non-complaint 61 39 Table 6. Classification with a Bayesian classifier according to equation (6). a: = 0.2; p(1) = 3.27; p(2) = 3.29; p(3) = 3.44 Predicted category Set Input category Non-complaint Complaint Training set . . Non-complaint 73 27 Complaint 15 85 Complaint 20 80 Complaint 22 78 Test set one . . Non-complaint 53 47 Test set two . . Non-complaint 61 39 with low industrial activity, SO2 concentrations will remain low and hourly reports according to category 1 of the training set will be encountered.However, a low wind velocity generally means that the direction of the wind will be undefined. This type of weather, in combination with high industrial activity, means that the concentrations of SO2 increase at all locations. Here discrimi- nation based on the location of the measurement becomes less significant. In combination with the low mixing and dilution, the decision function becomes very high, together with the burden felt. Because of this, one may conclude that s ( x ) increases with an increase in SO2 concentration, which itself serves as a tracer for all kinds of noxious smelling substances produced by industrial activities. Thus s(x) is interpreted as a yardstick for the burden felt by the population. In order to elaborate on this point of view, a search is made for the correlation between s ( x ) and the number of filed complaints.For every hour of 1982, as far as no missing feature values were encountered, s(x) is calculated and classified in one of the classes of Table 7. It is seen in this table that complaints were filed in 332 hours out of the 5511 tested hours, yielding a 60 F 40 C nr Q .- - : 20 0 0 PtOt(X21 x) Fig. 6. Histogram of complaint hours of class 2 as a function of the Bayesian classifier equation (6) 60 n 0.90 1.05 1.20 ~tot(X,Ix)l~to,(X3lx) Fig. 7. probability ratio for class 2 over class 3 Histogram of all complaint hours of 1982 as a function of the 70 0 0.60 0.75 0.90 1.05 1.20 1.35 Integrated number of complaints for 1982 as a function of the ~tOt(X,lX)l~t,,(X3lX) Fig.8. probability ratio for class 2 over class 31082 ANALYST, SEPTEMBER 1986, VOL. 111 Table 7. Boundary values between successive s ( x ) classes, population of classes (hours and complaint hours), YO complaint hours and number of complaints per 100 h (i.e., burden on inhabitants of the area) for the year 1982 Average number of Boundary Total Number of Yo complaints Number of Class between number Complaint Complaint Per complaints i and i- 1 in i in i in i hour 100 h value, s(x) i of hours hours hours complaint Per - 1 2 0.849 3 0.918 4 0.965 5 0.999 6 1.015 7 1.031 8 1.061 9 1.087 Total (average): 1808 913 768 542 303 276 402 249 250 5511 35 37 36 40 31 34 38 36 45 477 1.9 4.1 4.7 7.4 10.2 12.3 9.5 14.5 18.0 (6.0) 1.11 1.08 1.42 1.14 1.67 1.53 1.82 1.46 1.67 (1.43) 2.1 4.4 6.7 8.3 17.3 18.8 17.4 21.3 31.6 (8.6) / total of 477 complaints.The distribution of these hours over the various classes is such that the number of complaints increases with a higher class number. The average number of complaints per complaint hour does not vary in a systematic fashion over the classes (Table 7). Column 7 in Table 7 is constructed by multiplication of the number of complaint hours per class with the average number of complaints per hour and divided by the total number of hours per class. Column 7 of Table 7 gives an indication of the burden felt by the inhabitants of the area, as a function of the class number, i.e., the value of the decision function.It is seen that in classes 1-3, the number of complaints per 100 h gradually increases. In classes 4 and 5 this number is almost doubled. Classes 5, 6 and 7 have almost the same value whereas in classes 8 and 9 a jump in the value of this number is again noticed. This table indicates that there is a relationship between the weight of the burden and the amount of complaints, although the number of hours with more than one complaint does not change significantly. The relative number of complaint hours does increase. It should be noted that for the compilation of Table 7, all hours with no missing feature values are taken into account. This means that hours between 02.00 and 08.00 h are also included. From the distribution of complaints over the day (see Appendix), it is known that in this period almost no complaints are made.When a correction is made for this observation, the number of complaints per 100 h should be increased by almost 30%. Conclusions The load of the burden of noxious air on the population has been characterised by concentrations of NO, NOz and SO2. These concentrations depend on wind direction and wind velocity. In Schiedam, a city in the Rhine-mouth area, a wind direction from the south west will give enhanced concen- trations at locations 10 and 11 (Fig. 1). A low wind velocity gives rise to a low mixing of air and a cumulation of all eventual noxious constituents. A very low wind speed results in a varying wind direction and a wider area over which noxious constituents may be spread. In feature space, there does not exist a sharp boundary between hours with complaints and hours with no complaints.Instead a probablistic description should be handled. Bayes statistics, based on past experience, offer the possibility of constructing such a probablistic description. A decision function can be made that is based on the ratio of the probability for complaints and the probability for non- complaints. The numerical result of this decision function can be regarded as a measure of the seriousness of burden on the population, which in turn is reflected by the complaints filed. According to this method, it may be possible in practice to estimate the number of complaints at a given time, once the necessary features are obtained: the burden, “measured” by this method, is an indication of this.When the time-dependent effects of the burden on the population have been investigated more thoroughly, this method may not only help to predict the burden but, more importantly, help to prevent it. Thanks are due to J. E. Evendijk, P. H. C. Eilers and P. J. W. M. Miiskens, D. C. M. R., for making the measured data on air constituents available to us; to R. Wolters for his critical reading of the original manuscript and his useful suggestions; and to B. R. Kowalski for making the computer program ARTHUR available to us. Mr. Tjeerd de Jong is also thanked for providing Fig. 1. APPENDIX Some Statistical Information on the Observations During 1982 Total number of hours: 8760. Hours with one or more complaints: 499. Total number of complaints: 770. The frequency distribution of complaints is indicated in Table 8 and histograms of the number of complaints are shown in Figs. 9 and 10.ANALYST, SEPTEMBER 1986, VOL. 111 Table 8. Frequency distribution of complaints in complaint hours 96 -I Number of complaints/h 1 2 3 4 5 6 7 8 9 10 > 10 Number of hours 381 73 19 6 8 2 1 5 0 1 3 60 15 0 I n 1 6 12 18 24 Hours of the day Fig. 9. Histogram of the number of complaints as a function of the hour of the day r 0 2 4 6 8 Month 1 10 2 Fig. 10. Histogram of the number of complaints as a function of the month of the year 1. 2. 3. 4. 5 . 6. 7. 8. References Pijpers, F. W., Analyst, 1984, 109,299. “Annual Report 1982,” Dienst Centraal Milieubeheer Rijnmond, Schiedam, 1982 (in Dutch). Varmuza, K. , “Pattern Recognition in Chemistry,” Springer Verlag, Berlin, Heidelberg and New York, 1980. Mendel, J. M., and Fu, K. S. , “Adaptive Learning and Pattern Recognition Principles,” Addison-Wesley, Reading, MA, 1979. Jurs, R. C . , and Isenhour, T. L., “Chemical Applications of Pattern Recognition,” Wiley, New York, 1975. Isenhour, T. L., Kowalski, B. R., and Jurs, R. C., CRC Crit. Rev. Anal. Chem., 1974, 4, July, 1. Kateman, G., and Pijpers, F. W., “Quality Control in Analytical Chemistry,” Wiley, New York, 1981. Duewer, D. L., Koskinen, J. R., and Kowalski, B. R., “ARTHUR,” Laboratory for Chemometrics, Department of Chemistry BG 10, University of Washington, Seattle, WA. Paper A6197 Received March 24th, I986 Accepted April I Oth, I986

ISSN:0003-2654    

DOI:10.1039/AN9861101077

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1986, VOL. 111 Optical Flow Cell Studies With lmmobilised of an Optical Fibre Sulphide Sensor fibre link Ramaier Narayanaswamy" and Fortunato Sevilla, 111 Department of Instrumentation and Analytical Science, Pump + Flow cell 4 - 1085 Reagents for the Development Reservoir UMIST, PO Box 88, Manchester M60 IQD, UK Several reagent systems, such as 2,6-dichlorophenolindophenol, phenanthroline complexes and dithio- fluorescein complexes, have been immobilised on polymeric solid supports and their suitability as interphases in optical fibre sensors for sulphide ions investigated. These studies were carried out in a flow cell arrangement. The reflectance of the reagent phase was measured using optical fibres and was found to be related to the concentration of the sulphide ions in the solution.The dynamic range for sulphide ion determination varied with the reagents. The reagent phases prepared from 2,6-dichlorophenolindophenol and from the complex formed by dithiofluorescein with o-hydroxymercuribenzoic acid could be regenerated and are considered suitable for the development of an optical fibre sensor for sulphide ions. Keywords: Sulphide determination; optical fibre sensor; irnmobilised reagent systems; flow cell The advantages of chemical sensors based on optical fibres have been widely recognised, and consequently the develop- ment of sensors of this type has been pursued with intense interest. These sensors are based on the change in the optical characteristics of a reagent phase incorporated on the tip of an optical fibre, through which the change is detected.The reagent phase usually consists of a chemically specific reagent immobilised on a polymeric solid support. A variety of optical fibre chemical sensors have already been reported. Optical sensors for pH,lJ moisture3 and ammonia4 have been devised, based on a change in the absorbance or reflectance of a reagent phase in the presence of the analyte. Fluorescence sensors for pH,5,6 some metal ions,7-9 halide ions ,lo oxygen ,I1 carbon dioxide ,I2 glucose13 and haloethanes14 have also been constructed using optical fibres. This paper presents preliminary studies on the development of an optical fibre sensor for sulphide ions. Several reagent phases have been immobilised on a solid support and their reaction with small amounts of sulphide ions was investigated.The suitability of the reagent phases in an optical sensor have been evaluated. Based on these results, an optical fibre probe for sensing sulphide ions is being explored.15 This further study will also demonstrate the applicability of the reagent systems in optosensing as detectors for sulphide ions in a flow system similar to that described by Rfiiitka, and Hansen.16 The development of an optical fibre sulphide ion sensor could provide a simple and rapid method for routine environ- mental measurements. Small amohts of sulphide ions in effluents and atmospheres produce effects toxic to living organisms, together with corrosion in metallic and concrete structures. Although its characteristic odour could be a means of detection of sulphide, this is not reliable because the nose is desensitised by high concentrations of sulphide ions.An optical fibre sensor provides an alternative to the electro- chemical sensors now in use as it is capable of being used for remote monitoring and is not susceptible to electromagnetic interference. Experimental Instrumentation The instrumentation (Fig. 1) used in this study has been adapted from a system previously described for pH measure- ment with an optical fibre.1 Optical radiation was supplied by a quartz - halogen lamp (12 V, 100 W) and modulated by an optical chopper (Bentham 218) set at a frequency of 360 Hz. I .c I I Quartz Photo- Lock-i n halogen * Optical Mono- .) multiplier lamp chopper chromator tube amplifier Chart recorder Fig.1. Schematic diagram of the instrumentation system used for reflectance measurements * To whom correspondence should be addressed,1086 ANALYST, SEPTEMBER 1986, VOL. 111 The reagent phase, contained in a flow cell, was irradiated through one branch of a bifurcated fibre system, constructed from a 16-polymer optical fibre bundle (Optronics). Light reflected from the reagent phase was collected and guided to a monochromator (ISA Instrument H-1061) by the other branch of the optical fibre system. Measurement of the reflected radiation was carried out with a photomultiplier (Hamamatsu R446), a current amplifier (Bentham 286) and a lock-in amplifier (Bentham 223). The reflectance spectra and the response graphs were recorded on a chart recorder (Servoscribe RESll).The flow cell (Fig, 2) was fabricated from a Perspex block and featured a bore (2.2 mm in diameter) perpendicular to the axis of a cylindrical compartment (6 mm long and 3 mm in diameter). A fine wire mesh was placed at one end of the compartment to block the reagent phase, which was loaded into the flow cell with the aid of a peristaltic pump (Watson-Marlow 21) and packed in such a manner that it completely surrounded the sensing end of the fibre. The reagent phase was continuously flooded with either water, a buffer or an analyte solution during reflectance measure- ments. Bubbles would disrupt the arrangement of the micro- spheres in the packing and alter the reflectance; consequently, care was exercised to exclude them from the flow. Reagents Several spectrophotometric reagents for sulphide ion determi- nation were used in the study.Solutions of phenolindo-2,6- dichlorophenol (2,6-dichlorophenolindophenol, DIP), tris( 1,lO-phenanthroline)iron(III) sulphate and bis(2,9- dimethyl-1 ,lo-phenanthroline)copper(II) sulphate were pre- pared from chemicals purchased from BDH Chemicals. Solutions of the complex of dithiofluorescein (Fluorochem) with silver nitrate (BDH Chemicals) and with o-hydroxymer- curibenzoic acid (Koch-Light) were prepared according to the method of Wronski.17918 Silica gel and some porous organic polymers, all obtained from BDH Chemicals, were investigated as possible supports for the immobilisation of the reagents. The organic materials used were Amberlite XAD-2, XAD-4 and XAD-7. Whereas XAD-2 and XAD-4 are cross-linked copolymers of styrene and divinylbenzene, XAD-7 is a cross-linked polymer of methyl met hacrylate. From reservoir Tubing with fitting I t To pump Fig.2. Cross-section of the flow cell assembly Immobilisation Procedure Immobilisation consisted in soaking the immobilising agent (0.5-1.0 g) in 25 ml of a solution of the reagent (0.05-1.0%) for 2-4 h. At the end of this period, the solid phase, which contained the adsorbed reagent, was thoroughly washed with water until the washings were colourless. The optimum conditions (i. e . , solution concentration, amount of adsorbent and immobilisation period) for each reagent were determined by spectrophotometric analysis of the liquid phase in the system. For the organic immobilising agents, the microporous polymer beads were washed successively with acid, water and methanol.The non-polar polymers, XAD-2 and XAD-4, had to be hydrated before they could be used in aqueous solutions. This was achieved by soaking the polymer first in methanol and then in water. Reflectance Measurements The reflectance spectra of each reagent phase were recorded before and after its reaction with sulphide ions. The wavelength of maximum divergence between the two spectra was identified and used in subsequent reflectance measure- ments. Results and Discussion 2,6-Dichlorophenolindophenol (DIP) Sulphide ions decolourise DIP, reducing it from the blue azoquinoid form into the leuco form. The reagent was best immobilised on Amberlite XAD-2 microspheres, which acquired a violet colour as a result of the adsorption of the dye.The colour was discharged in the presence of sulphide ions and recovered through oxidation with oxygen. The above reactions of the reagent in the immobilised state were observed to be faster than those in solution. The reflectance spectra recorded for the immobilised reagent before and after reaction with sulphide ions are illustrated in Fig. 3. The response obtained from reflectance measurements at 598 nm was a function of the concentration of the sulphide ions. The relative reflectance of the reagent phase increased linearly with the logarithm of the concentration of sulphide ions in the range 8-20 mmol l-1 (Fig. 4). The reaction rate was affected by the pH of the analyte and was greatest at pH 8. The response time varied between 2 and 25 min, depending on the sulphide ion concentration. Reflectance signals were there- fore recorded after 2 min of reaction. Wavelengthhm Figure 3.XAD-2 (A) before and (B) after reaction with sulphide Reflectance spectra obtained from DIP immobilised onANALYST, SEPTEMBER 1986, VOL. 111 1087 Log c Fig. 4. Semi-logarithmic graph of the reflectance of immobilised DIP with different concentrations of sulphide (mol 1-1) Phenanthroline Complexes Tris( 1 ,lo-phenanthroline)iron(III) is converted into a bright red complex ion through the reduction by the sulphide ion of the iron atom in the coordination sphere. This complex could be suitably held on the surface of Amberlite XAD-7 micro- spheres. The resulting immobilised reagent was sensitive to small amounts of sulphide ions and showed a 25% reduction of its reflectance at 550 nm in a solution containing 0.16 mmol l-1 of sulphide.However, it was noted to be unstable, and was transformed into its reduced form during storage, particularly in the presence of light. Similar observations have been reported in solution studies.19 Bis(2,9-dimethyl-l, l0-phenanthroline)copper(II) reacts with sulphide ions and produces a bright orange complex ion as a result of the reduction of the copper ion. This coordi- nation compound could be adsorbed on silica gel and on Amberlite XAD-7. A weak luminescence, which has been reported for the complex in non-aqueous solutions,20 was also detected in the reagent immobilised on silica gel. This emission interfered with the reflectance measurements.On the other hand, the copper complex held on XAD-7 showed a marked change in its reflectance spectra in the presence of sulphide ions. The response at 485 nm was rapid and reproducible. The reflectance signals read after 33 s of reaction showed a linear dependence on the logarithm of the concentration of the sulphide ion in the range 0.13-0.50 mmol 1-1. It was noted that the reflectance approached the same equilibrium value for all the concentrations of sulphide ion used in the study. The equilibrium was found to be insensitive to the pH of the system. Regeneration of the reagent phase through oxidation with oxygen gas or with hydrogen peroxide was attempted, but proved unsuccessful. This irreversibility is also observed in solutions of the reagent and has been attributed to the stabilisation of the copper(1) complex by the steric hindrance of the methyl substituents.21 Dithiofluorescein Complexes The coordination compounds formed by the reaction of dithiofluorescein (DTF) with silver nitrate and with o-hy- droxymercuribenzoic acid (HMB) were observed to react with sulphide ions, releasing the blue DTF dye.Immobilisation of these reagents from their aqueous solutions resulted in a solid phase that was not responsive to the presence of sulphide ions. A sulphide-sensitive reagent phase was produced through a preliminary immobilisation of DTF on a suitable matrix (XAD-2) and a subsequent reaction of the adsorbed dye with silver nitrate or HMB. A plausible explanation for the difference in the properties of the reagent phases resulting from a direct immobilisation and a stepwise formation is that there is a difference in the orientation of the adsorbed reagent 400 500 600 Wavelength/nrn Fig.5. XAD-2 (A) before and (B) after reaction with sulphide Reflectance spectra obtained from Ag - DTF immobilised on a m tn a - .- c - 2 - > v) C c .- c .- C 0 U -0 .- c .- E 4- a E a Fig. 6. on XAD-2 (A) before and (B) after reaction with sulphide Reflectance spectra obtained from HMB - DTF immobilised molecule on the surface of the supporting solid and in the degree of accessibility of the metal atom for reaction with the sulphide ion. A distinct change in the reflectance spectra of the reagent phases occurred on their reaction with sulphide ions, the greatest change occurring at 590 nm (Figs. 5 and 6).The response obtained at 590 nm varied linearly with the logarithm of the concentration of sulphide ions in the range 0.063-0.63 mmol l-1 for the silver complex and 0.31-2.5 mmol l-1 for the HMB complex (Fig. 7). The response was rapid, reaching a minimum after 20 s of reaction. A constant value for the reflectance of the reacted reagent was not observed because of the photo-instability of the resulting DTF.22 The reagent phases were regenerated by complexing the liberated DTF with a solution of silver nitrate or of HMB passing through the flow cell. The recovery of the original phase was possible for the immobilised HMB complex, but not for the immobilised silver complex. Silver sulphide formed1088 ANALYST, SEPTEMBER 1986, VOL.111 References -2.5 -3.5 -4.5 I I Log c Fig. 7. Semi-logarithmic graphs of the reflectance of (A) Ag - DTF and (B) HMB - DTF immobilised on XAD-2 with different concentrations of sulphide (mol 1-1) during the reaction of the immobilised silver complex with sulphide ions was found to be deposited on the surface of the polymer support and prevented the recovery of the colourless reagent phase. Conclusions The results of this study demonstrate the suitability of the immobilised reagents for measuring small amounts of sulphide ions. In general, the reactions are observed to be faster than those occurring in solution, and the responses are found to be reproducible. The reflectance of the reagent phases studied varies linearly with the logarithm of the concentration of sulphide over certain ranges. The immobilised DIP and DTF - HMB complexes are renewable and could therefore be used in optical fibre probes for sulphide ions.These reagent phases can also be used in flow-through detectors for sulphide ions in a manner similar to that described by RfiiiEka, and Hansen.16 The immobilised copper - phenanthroline complex and the silver - DIF complex cannot be regenerated, but would nevertheless be useful in disposable sensors and alarm systems based on optical fibres. The authors thank Elf UK Ltd. for their support of this work. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Kirkbright, G. F., Narayanaswamy, R., and Welti, N. A., Analyst, 1984, 109, 1025. Peterson, J. I., Goldstein, S. R., Fitzgerald, R.V., and Buckhold, D. K., Anal. Chem., 1980, 52, 864. Russell, A. P., and Fletcher, K. S . , Anal. Chim. Acta, 1985, 170, 209. Giuliani, J. F., Wohltjen, H., and Jarvis, N. L., Optics Lett., 1983, 8, 54. Saari, L. A., and Seitz, W. R., Anal. Chem., 1982, 54, 821. Zhujun, Z., and Seitz, W. R., Anal. Chim. Acta, 1984,160,47. Saari, L. A., and Seitz, W. R., Anal. Chem., 1983, 55, 667. Saari, L. A., and Seitz, W. R., Analyst, 1984, 109, 655. Zhujun, Z . , and Seitz, W. R., Anal. Chim. Acta, 1985, 171, 251. Urbano, E., Offenbacher, H., and Wolfbeis, 0. S., Anal. Chem., 1984, 56, 427. Peterson, J . I., Fitzgerald, R. V., and Buckhold, D. K., Anal. Chem., 1984, 56,62. Zhujun, Z., and Seitz, W. R., Anal. Chim. Acta, 1984, 160, 305. Schultz, J. S . , Mansouri, S . , and Goldstein, I. J., Diabetes Care, 1982,5, 245. Wolfbeis, 0. S., and Posch, H. E., Anal. Chem., 1985, 57, 2556. Narayanaswamy, R., and Sevilla, F., to be published. Rfiiitka, J., and Hansen, E. H., Anal. Chim. Acta, 1985,173, 3. Wronski, M., Chem. Anal. (Warsaw), 1960, 5 , 457. Wronski, M., Biochem. J., 1967, 104, 978. Novak, J., and Arend, H., Talanta, 1964, 11, 898. Dietrich-Buchecker, C. O., Marnot, P. A., Sauvage, J. P., Kirchoff, J. R., and MacMillin, D. R., J. Chem. SOC., Chem. Commun., 1983, 513. Arce, J. A., Spodine, E., and Zamudio, W., J. Inorg. Chem., 1976, 38, 2029. Wronski, M., Talanta, 1977, 24, 347. 21. 22. Paper A51460 Received December 23rd, I985 Accepted March 25th, 1986

ISSN:0003-2654    

DOI:10.1039/AN9861101085

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1986, VOL. 111 1089 Collaborative Studies of Methods for the Detection of Residues of Monensin in Chicken Tissues Analytical Methods Committee* Royal Society of Chemistry, Burlington House, Piccadilly, London W1 V OBN, UK Some published methods for the detection of residues of monensin in chicken tissues were examined. In some instances the limits of detection claimed by the authors could not be achieved. A method was evaluated in which the monensin was extracted from the tissues with methanol, partitioned into carbon tetrachloride and then purified on a silica gel column. Monensin was determined in the eluate from this column by TLC. Alternative procedures for detecting the monensin on the TLC plate by bioautography or by formation of a fluorescent derivative are given.Keywords: Monensin residues; chicken tissues; collaborative studies; thin-la yer chromatography The Analytical Methods Committee has received and has approved for publication the following report from its Veterinary Residues in Fresh Meat Sub-committee. Report The constitution of the Sub-committee responsible for the preparation of this Report was: Dr. J. F. C. Tyler (Chairman, until April 1984), Dr. N. T. Crosby (Chairman, from May 1984), Mr. J. B. Aldred, Mr. A. Anderson, Mr. K. J. Briant (fromMay 1984), Mr. P. M. Brown, Mr. J. Ganley, Mr. H. L. Hatfield, Mr. A. Hobson-Frohock, Mr. A. F. Lott, Mr. A. F. Machin (until April 1984), Mr. M. P. Quick (from September 1984), Mr. R. Ryden, Dr. G. Shearer and Mr. G. M. Telling, with Mr. J. J. Wilson as Secretary. Introduction Monensin (Chemical Registry No.17090-79-8) is a polyether antibiotic produced by Streptomyces cinnamonensis. It is added in the form of the sodium salt to broiler chicken feeds as a coccidiostat at levels of 100-120 mg kg-1 until 3 d before slaughter. It is also used as a growth promoter in cattle'feeds. Monensin (C36 H6* OI1) has a relative molecular mass of 670.9 and the structure shown below. HO Similar techniques have been utilised for the determination of monensin in animal tissues. Donoho and Kline4 described a semi-quantitative thin-layer bioautographic method, which was later modified by Okada et a1.5 Tihova and Peneva6 also used TLC to separate monensin from co-extractives, but employed p-anisaldehyde (4-methoxybenzaldehyde) as a detection reagent in place of bioautography as the end determination step.Other medicinal additives such as lasalo- cid, narasin and salinomycin, which are very similar in chemical structure to monensin, may also be present and would not be separated by the above methods. However, whilst this work was in progress, Owles7 reported a qualitative test for the separation of these antibiotics in feeds and pre-mixes using a TLC system. The Sub-committee was charged with the task of develop- ing and evaluating a method for the detection and determi- nation of monensin residues in chicken tissue at a level of 0.1 mg kg-I or below. The Sub-committee began by examining the method of Okada et al.5, which uses a thin-layer bioautographic technique at the determinative stage.The authors claimed that recoveries of monensin from fortified tissues were 92.9% from fat, 86.0% from liver and 104.4% from muscle tissues. The detection limit achieved was 0.01 , CH30 OH CH(CH31COOH The compound itself is non-volatile and does not absorb in the UV - visible region. This restricts the number of chemical techniques that are available for its determination in animal feeding stuffs. The low levels likely to be present in tissues further exacerbate the problem. Kline et al. 1 proposed a microbiological assay for monensin in feeds using Bacillus subtilis, whereas Golab et a1.2 described a spectrophotometric procedure based on the reaction of monensin with vanillin (4-hydroxy-3-methoxybenzaldehyde). An HPLC method using a refractive index detector has been reported by Macy and L0h,3 but this is only applicable to pre-mixes.* Correspondence should be addressed to the Secretary, Analytical Methods Committee, Analytical Division, Royal Society of Chemistry, Burlington House, Piccadilly, London WIV OBN. mg kg-1 in fat and 0.0125 mg kg-1 in other tissues. No detect- able residues of monensin were found in the fat of animals 48 h or more after withdrawal from the feed, or in other tissues 24 h or more after withdrawal. In later work, the method of Tihova and Peneva6 was evaluated. This also used TLC to separate monensin from co-extractives followed by detection with p-anisaldehyde reagent to form a fluorescent derivative. This method has the advantage that it only uses equipment and experience normally found in chemical laboratories and does not require microbiological techniques.The authors claimed a better sensitivity (0.003 mg kg-1) than Okada et aZ.5 and a mean recovery of 97% from sheep and pig tissues. A simple solvent partition stage was recommended for the removal of co- extractives, in contrast to the column chromatography system used by Okada et al.51090 ANALYST, SEPTEMBER 1986, VOL. 111 Experimental Investigations were carried out using samples of chicken tissues kindly supplied by the Lilly Research Centre. These were obtained from animals fed on diets known to be free from monensin. The feeding of chickens with monensin at recommended levels in the diet would produce residual levels in tissues that are likely to be too low to be useful for the evaluation of an analytical method.On the other hand, the birds might not accept feeds containing significantly higher levels of monensin than those recommended. Hence, it was decided to work only on monensin-free tissues to which known amounts of monensin had been added in the laboratory. Table 1. Recovery of monensin residues from fortified tissue. Bioautographic method Monensin Estimated Laboratory Tissue added/mg kg-1 recovery, % Muscle Liver Muscle Muscle Muscle Liver Muscle Liver Muscle Liver 0.05 0.1 0.05 0.1 0.0125 0.1 0.0125 0.025 0.1 0.0125 0.025 0.05 0.1 0.2 0.05 0.1 0.1 0.1 0.017 0.033 0.05 0.066 0.05 67,67 67,53 94,76 89 72,72,65 40,80,70 o,o, 57,81,33,53 50,64,81,64 84,75,106 o,o,o, 070 74,73,63,73 82,94,68,62,77, 62,67,53,62 54,55,63,64,64 62,70,58 108,109,118,99,89 117,150,140 125,112 50,60 110 78 88,88,89,93 79 87,77,72,72 Evaluation of the Method of Okada et aZ.5 Preliminary studies of the bioautographic procedure estab- lished that levels of monensin down to 0.1 mg kg-1 could be detected without much difficulty.At lower levels, difficulties were encountered, such as poor recovery of monensin from the tissues, interference from fat and other co-extractives and, for liver, the presence of compounds producing zones of inhibition with the test organism. Some results are shown in Table 1, and these were considered to be satisfactory, although the limit of detection (0.0125 mg kg-1) claimed by the authors could not be achieved with every sample. The layer of agar applied should be as thin as possible (50-80 ml per plate, 230 mm square).A plot of dose against response on a log - linear graph was a straight line in the range 0.2-0.05 yg of monensin. Below this level, the response was found to be erratic. Some participants reported that the zones observed were elliptical rather than circular. Separation of Monensin from Other Ionophores Narasin, salinomycin and lasalocid are very similar in chemical structure to monensin. The method of Okada et al.5 failed to separate monensin from the other three compounds on the TLC plate. Two laboratories looked at modifications of the method involving (1) developing the TLC plate twice with ethyl acetate and (2) using water-saturated ethyl acetate as the development solvent. The RF values obtained are shown in Table 2. Both modified development systems clearly separate monensin from the other ionophores.Ethyl acetate saturated with water (system 2) gave a marginally better separation of monensin than the other ionophores, whereas ethyl acetate double development (system l ) , although more time consuming, was able to resolve all four ionophores. Similar findings were published by Owles,7 whose results are also included in Table 2 for comparison. However, the use of a fixed ratio of ethyl acetate to water (97 : 3 V/V) was thought to be preferable to the method recommended by Owles.7 Evaluation of the Method of Tihova and Peneva6 The authors claim that monensin reacts with p-anisaldehyde (4-methoxybenzaldehyde) in acidic solution to form a fluores- cent compound that is more sensitive than that formed in the Table 2.Separation of ionophores on a TLC plate R , values Developing solvent Ethyl acetate (twice) . . Water-saturated ethyl acetate(once) . . Ethyl acetate (twice) . . Water-saturated ethyl acetate(once) . . Water-saturated ethyl acetate(twice) . . 100 ml of ethyl acetate plus 3 ml of water7 . . Monensin Salinomycin Narasin Lasalocid . . 0.25 0.43 0.48 0.50 . . 0.33 0.53 0.59 0.61 . . 0.30 0.51 0.58 0.76 . . 0.31 0.55 0.59 0.61 . . 0.52 0.77 0.81 0.83 . . 0.38 0.51 0.60 0.71 R, values Ethyl acetate (twice) . . Water-saturated ethyl acetate(once) . . Ethyl acetate (twice) . . Water-saturated ethyl aeetate(once) . . Water-saturated ethyl acetate(twice) . . 100 ml of ethyl acetate plus 3 ml of water7 . . . . 1.0 1.71 1.92 2.0 . . 1 .o 1.61 1.79 1.85 * .1 .o 1.70 1.93 2.54 . . 1.0 1.77 1.90 1.97 . . 1.0 1.48 1.56 1.59 . . 1 .o 1.34 1.58 1.87 Laboratory 2 2 3 3 3 Owles7 2 2 3 3 3 Owles7ANALYST, SEPTEMBER 1986, VOL. 111 1091 Table 3. Recovery of monensin residues from fortified tissue. First collaborative trial Laboratory Tissue 1 Muscle Fat Liver 2 Muscle Fat Liver 3 Muscle Liver Fat Liver Fat 4 Muscle Muscle Liver * Not quantifiable. Monensin Estimated addedlmg kg-1 recovery, yo 1.0 0.6 0.2 0.1 0.05 1 .o 0.6 0.2 0.1 0.05 N.q.* 1 .o 0.6 0.2 0.1 0.05 1 .o 0.6 0.2 0.1 0.05 N.q.* 1 .o 0.6 0.2 0.1 0.05 1 .o 0.6 0.2 0.1 0.05 0.1 0.1 0.1 0.2 0.1 0.05 1 .o 0.6 0.2 0.2 0.1 0.05 1 .o 0.6 0.2 110-120 90-95 95-100 95-100 95-100 50-60 75-80 125-130 90-95 90-95 90-100 90-100 90- 100 90-100 90-100 90-100 90-100 90-100 90-100 90-100 90-100 90-100 90-100 90- 100 90-100 65-85 65-85 65-75 55 55 N.q.* 70-80,70-80 after clean-up 70-80,80-90 after clean-up 100 33-100 100-200 N.q.N.q. N.q. N.q. N.q. N.q. N.q. N.q. N.q. reaction with vanillin (4-hydroxy-3-methoxybenzaldehyde). However, they gave no directions for the formation of the colour apart from a general instruction to “warm the plate slightly for 2-3 min. ” Accordingly, the Sub-committee initially examined the colour formation on the TLC plate. Monensin reacts with the reagent to produce a complex that is visible in daylight and is also fluorescent under UV irradiation. At low temperatures the complex is yellow and the fluorescence is a maximum at 366 nm. At higher tempera- tures the complex is orange, although the wavelength of maximum fluorescence does not appear to change.Heating the plate above 80°C for 5 min destroys the fluorescent compound , increases the background fluorescence of the plate and, above 100 “C, causes charring. Prolonged exposure to UV irradiation may cause the fluorescent spots to fade. Some collaborators obtained satisfactory results by heating the plate for 5 min at 60 “C. Other workers preferred to allow the colour ~ Table 4. Recovery of monensin residues from fortified tissue. Second collaborative trial Laboratory Tissue Muscle Liver Fat Muscle Liver Fat Muscle Liver Fat Muscle Liver I Fat * Not quantifiable. Monensin Estimated addedlmg kg-1 recovery, % 0.2 0.1 0.05 0.2 0.1 0.05 0.2 0.1 0.05 0.2 0.1 0.05 0.2 0.1 0.05 0.2 0.1 0.05 0.05-0.2 0.2 0.1 0.05 0.2 0.1 0.05 0.2 0.1 0.05 0.2 0.1 0.05 100 100 100 60-65 60- 125 0 55-60 35-40 55-60 80 80 80 50 50 50 25 25 25 N.q.50 50 80 50 100 67-100 67 67-100 67-100 67 100 100 67 to develop at room temperature over a 10-20 min period.-The other ionophores also react with the reagent, but with lasalocid the reaction takes several hours at room tempera- ture. The method was then evaluated using a similar protocol to that designed for the bioautographic method. Preliminary work by three laboratories gave estimated recoveries in the range 80-100% and was sufficiently encouraging for a further study to be arranged using muscle, liver and fat tissues. The results obtained have been collected together in Table 3. In general, satisfactory recoveries were obtained from muscle tissue.Recoveries from liver tissue could not be determined unless an additional clean-up procedure was used. A second collaborative trial was therefore organised, in which partici- pants were asked to use the column clean-up procedure in the method of Okada et aZ.5 for fat and liver samples, in the hope that the spots on the TLC plate would be less distorted by the lipid material present in the extract applied to the plate. The results are shown in Table 4. It is clear that for levels of monensin of 0.1 mg kg-1 and below, the method is really only semi-quantitative. Finally, a third collaborative trial was then arranged in which participants were supplied with “blind” samples of chicken tissue, some of which had been fortified with monensin at a level of 0.1 mg kg-1.The column chromatographic clean-up procedure was used for all three tissues and either the bioautographic or the TLC - spectropho- tometric method at the end determination stage. The primary objective was to determine whether or not residues of monensin in chicken tissue could be reliably detected at the 0.1 mg kg-1 level. The results are shown in Table 5 . Apart from two results from laboratory 2, correct identifications were made. Laboratory 2 subsequently repeated their work1092 ANALYST, SEPTEMBER 1986, VOL. 111 Table 5. Detection of monensin residues in fortified chicken tissues. Third collaborative trial* Fortified . . Laboratory 2 3 4 6 * ?, Ambiguous result; I Negative in first test. $ Positive in first test. Muscle Liver A B C D A B . . . .. . . . N Y Y N Y N Method Chemical - Chemical - + + NS + Chemical NS + + Chemical - + + NS + NS + + -$ NS +t - Bioautograph - Bioautograph - + + NS ? - + - - - - + - - NS, sample not supplied; -, negative; +, positive; N, no; Y , yes. Fat A B N Y ? NS NS NS + NS + NS NS + - - and obtained satisfactory results as shown in the table. In general, recoveries were estimated at about 50%. One member examined the use of HPTLC and found that the limit of detection was 10 ng as opposed to 100 ng by conventional TLC. However, the extraction and clean-up stages may not be able to cope at such low levels, and overall there was little advantage to be gained by using HPTLC. Conclusions Both methods examined were found to be semi-quantitative for the detection of monensin residues at levels below 0.1 mg kg-1.The detection limits and recovery values claimed by the original authors could not be confirmed in this study. Neither method was thought to be significantly better than the other. Hence, both methods are given in the Appendix and are recommended by the Sub-committee for the detection of monensin residues in chicken tissue. The analyst can make a choice based on the facilities available in his or her laboratory. Alternatively, as both methods use the same extraction and clean-up procedures, it may be advisable to use both end-detection systems to confirm positive findings in chicken tissues. APPENDIX 1 Detection of Monensin Residues in Chicken Tissue. Bioautographic Method Principle Monensin is extracted with methanol and partitioned into carbon tetrachloride.After evaporation to dryness, the residue is purified on a silica gel column and examined by thin- layer chromatography using bioautography with agar innocu- lated with Bacillus subtilis. Reagents All reagents should be of analytical-reagent grade unless otherwise stated. Caution.-Attention is drawn to the toxic nature of some reagents. Suitable precautions should be taken. Agar medium. Containing 10 g of glucose, 2.5 g of dried yeast extract, 0.69 g of K2HP04, 0.45 g of KH2P04 and 15 g of agar powder in 1000 ml. Sterilise and check that the pH is about 6. Further details of the preparation and use of agar can be found in Donoho and Kline4 or in reference 8. Carbon tetrachloride. Chloroform. Remove any ethanol present as stabiliser by distillation, or by washing with water and drying over anhydrous sodium sulphate.Developing solvent for TLC. Ethyl acetate - water, 97 + 3 Ethyl acetate. Hexane. Methanol. Micro-organism. Bacillus subtilis ATCC 6633 (NCIB 8054). Monensin. Available from Eli Lilly, Indianapolis, from the International Laboratory for Biological Standards, Central Veterinary Laboratory, Weybridge, Surrey, or from Sigma, Poole, Dorset. (V/V>* Monensin standard solution, 10 ug ml-1 in methanol. Silica gel. Kiesegel60, Merck, or equivalent. Sodium sulp hate. Anhydrous. Apparatus sterile and disposable. Assay plates. 230 mm square (e.g., Nunc) or equivalent, Centrifuge. Capable of accepting 100-ml tubes. Chromatography tank for TLC. Glass columns. 250 x 11 mm fitted with a glass filter, Homogeniser .Rotary film evaporator. Separating funnel, 250 ml. Thin-layer plates. Kieselgel 60 without fluorescent indi- stopcock and reservoir. cator, pre-coated, 200 x 200 mm, 0.25 mm thick. Procedure Extraction Liver and muscle. Add 30 g of tissue to a 100-ml glass centrifuge tube containing 60 ml of methanol. Homogenise and then centrifuge for 10 min. Extract the supernatant three times, each with 30 ml of carbon tetrachloride. Evaporate the carbon tetrachloride solution to dryness in a rotary film evaporator at 40 "C. Fat. Add 30 g of sample to a 100-ml centrifuge tube containing 60 ml of methanol. Homogenise and then centri- fuge for 10 min. Evaporate 40 ml of the supernatant to dryness in vacuo in a rotary film evaporator at 40 "C. Clean-up Prepare a suitable column by adding approximately 1.5 g of silica gel to a column containing hexane, to give a layer 12 mm deep.Stir to eliminate air bubbles and pack uniformly by tamping. Then add a 20-mm layer of anhydrous sodium sulphate on top of the column. Transfer the extracted residue to the prepared column using several 5 ml portions of hexane. Allow the liquid to fall to the level of the top of the packing. Wash the column with 40 ml of chloroform and elute monensin from the column using 25 ml of chloroform -ANALYST, SEPTEMBER 1986, VOL. 111 1093 methanol (95 + 5, VIV). Evaporate the eluate to dryness in a rotary film evaporator and transfer the residue into a 10-ml test-tube using three portions of hexane (1-2 ml each). Remove the hexane by evaporation in a stream of air or nitrogen and take up the residue in 0.2 ml of methanol. Thin-layer chromatography Apply 20 pl of the prepared extract to a TLC plate 30 mm from the bottom edge.Apply, alongside the sample spot, equal volumes of a range of monensin standard solutions containing 0.05, 0.1 and 0.2 pg of monensin, respectively. Develop the chromatogram until the solvent front is 120 mm from the origin. Allow the solvent to evaporate from the plate in air. Melt the agar medium in boiling water, cool to about 60 "C and spray on to the surface of the plate. Prepare inoculated agar by cooling 100 ml of melted agar medium to about 60 "C and inoculate with 0.2 ml of B. subtilis spore suspension. Mix the inoculated agar gently and pour 50-80 ml evenly over the surface of the TLC plate contained in a suitable assay plate.Allow the plate to cool until the agar solidifies and then incubate at 37 "C for 16-18 h. Measure the longest and shortest diameters of each inhibition zone and, by reference to the standard spots, calculate the amount of monensin present in the extract. APPENDIX 2 Detection of Monensin Residues in Chicken Tissue. Chemical Method Principle Monensin is extracted with methanol and partitioned into carbon tetrachloride. After evaporation to dryness, the residue is purified on a silica gel column and examined by thin-layer chromatography. A fluorescent derivative is pre- pared with p-anisaldehyde. Reagents All reagents to be of analytical-reagent grade unless otherwise stated. Caution.-Attention is drawn to the toxic nature of some reagents.Suitable precautions should be taken. Acetic acid, glacial. Carbon tetrachloride. Chloroform. Remove any ethanol present as stabiliser by distillation, or by washing with water and drying over an hydrous sodium sulp ha t e . Developing solvent for TLC. Ethyl acetate - water, 97 + 3 Ethyl acetate. Hexane. Methanol. Monensin. Available from Eli Lilly , Indianapolis, from the International Laboratory for Biological Standards, Central Veterinary Laboratory, Weybridge, Surrey, or from Sigma, Poole, Dorset . (V/V). Monensin standard solution. 10 pg ml-1 in methanol. Silica gel. Kieselgel 60, Merck, or equivalent. Sodium sulphate, anhydrous. Sulphuric acid. Sp. gr. 1.84. Detection reagent. Dissolve 0.5 ml of p-anisaldehyde (4- methoxybenzaldehyde) in 8.5 ml of methanol.Add 0.5 ml of sulphuric acid and five drops of acetic acid and stir. Apparatus Centrifuge. Capable of accepting 100-rnl tubes. Chromatography tank. Glass columns. 250 x 11 mm i.d. fitted with a glass filter, Homogeniser . Rotary film evaporator. Separating funnel, 250 ml. Thin-layer plates. Kieselgel 60, without fluorescent indi- UV lamp. With a filter at 366 nm. stopcock and reservoir. cator, pre-coated, 200 x 200 mm, 0.25 mm thick. Procedure Extraction Liver and muscle. Add 30 g of tissue to a 100-ml glass centrifuge tube containing 60 ml of methanol. Homogenise and then centrifuge for 10 min. Extract the supernatant three times, each with 30 ml of carbon tetrachloride. Evaporate the carbon tetrachloride solution to dryness in a rotary film evaporator at 40 "C.Fat. Add 30 g of sample to a 100-ml centrifuge tube containing 60 ml of methanol. Homogenise and then centri- fuge for 10 min. Evaporate 40 ml of the supernatant to dryness in vacuo in a rotary film evaporator at 40 "C. Clean-up Prepare a suitable column by adding approximately 1.5 g of silica gel to a column containing hexane to give a layer 12 mm deep. Stir to eliminate air bubbles and pack uniformly by tamping, then add a 20 mm layer of anhydrous sodium sulphate on top of the column. Transfer the extracted residue into the prepared column using several 5-mi portions of hexane. Allow the liquid to fall to the level of the top of the packing, then wash the column with 40-ml of chloroform and elute monensin from the column using 25 ml of chloroform - methanol (95 + 5 , VIV).Evaporate the eluate to dryness in a rotary film evaporator and transfer the residue into a 10-ml test-tube using three portions of hexane (1-2 ml each). Remove the hexane by evaporation in a stream of air or nitrogen and take up the residue in 0.2 ml of methanol. Thin-layer chromatography Apply 20 p1 of the prepared extract to a TLC plate 30 mm from the bottom edge. Apply, alongside the sample spot, equal volumes of a range of monensin standard solutions containing 0.05, 0.1, 0.02 and 0.2 pg of monensin, respec- tively. Develop the chromatogram until the solvent front is 120 mm from the origin. Allow the plate to dry in air and spray with the detection reagent. Allow the plate to stand for 15 min at room temperature then examine it under UV light. Monensin appears as a yellow - orange spot with an RF value of approximately 0.3-0.4. Calculate the amount of monensin present by reference to the standard spots. 1. 2. 3. 4. 5. 6. 7. 8. References Kline, R. M., Stricker, R. E., Coffman, J. D., Bikin, H., and Rathmacher, R. P., J. Assoc. Off. Anal. Chem., 1970, 53,49. Golab, T., Barton, S. J., and Scroggs, R. E., J. Assoc. Off. Anal. Chem., 1973, 56, 171. Macy, T. D., and Loh, A., J. Assoc. Off. Anal. Chem., 1983, 66, 284. Donoho, A. L., and Kline, R. M., in "Proceedings of the Annual Conference on Antimicrobial Agents and Chemo- therapy, 1967," American Society for Microbiology, Bethesda, Okada, J., Higuchi, I . , and Kondo, S., J. Food Hyg. SOC., (Jpn.), 1980, 21, 177. Tihova, D., and Peneva, V., Vet.-Med. Nauki, 1982, 19, 52. Owles, P. J., Analyst, 1984, 109, 1331. The Medicines (Animal Feeding Stuffs) (Enforcement) Regu- lations 1985, SI 1985 No. 273, HM Stationery Office, London, Paper A6196 Received March 26th, 1986 MD, 1968, pp. 763-766. 1985, pp. 71-2.

ISSN:0003-2654    

DOI:10.1039/AN9861101089

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, SEPTEMBER 1986, VOL. 111 SHORT PAPERS 1095 Simple Fibre Optic pH Sensor for Use in Liquid Titrations Nira Benaim," Kenneth T. V. Grattant and Andrew W. Palmer Measurement and Instrumentation Centre, School of Electrical Engineering and Applied Physics, The City University, Northampton Square, London EC? V OHB, UK This paper describes the development of a fibre optic sensor incorporating an LED source and solid-state photodiode detection for the determination of the pH of solutions containing the indicator phenol red. The sensor can be used in, for example, liquid titration studies. This inexpensive and rapid response device is shown to have a reproducible characteristic output, with a low response change at elevated temperatures. Keywords: Optical pH sensor; optical fibre sensor; light-emitting diode source; phenol red The research field of fibre optic sensing is one that has seen significant growth in recent years and many fibre optic sensors for the measurement of important parameters in industrial situations have been described.' Hence, parameters such as temperature, pressure, fluid flow and liquid level are frequently measured by novel optical devices, but the sensing of chemical parameters is less frequently tackled by fibre optic means, and very few devices have been reported.This is surprising as some of the well known advantages of fibre optic transducers are particularly valuable for use in the environ- ment of a chemical process. Fibre optic sensors are non- electrical, and as no currents are flowing at the sensor head there is no risk of explosion from sparks from short circuits.In most instances, there is no danger of causing ignition by the light used to address the sensor being coupled into the flammable material owing to its comparatively low level of emission. Most transducers could be constructed entirely from passive insulating materials, if necessary, and therefore they can be used in a corrosive atmosphere or solution in which some conventional sensors would be attacked or would be bulky due to the addition of protective coatings or casings. Additionally, if the chemical process to be studied involves the use of high magnetic or electrical fields, the optical sensor is unaffected by their presence and the fidelity of the transmitted signal is preserved.In this paper, a simple, inexpensive fibre optic pH sensor is described for use in the sensing of liquid solutions to which an indicator dye has been added, for example, in acid - base titrations. The indicator dye changes colour as a result of a change in the optical absorption spectrum. This sensor determines the change quantitatively and hence the pH of the solution over a pre-determined range. Previous work in this area has concentrated on, for example, a narrow pH sensor for biological applications,2 a flow cell device with a wider pH measurement function3 and an immobilised indicator on a polymer base connected to a plastic fibre bundle. These devices use bulky, conventional lamp sources and optical filters. Peterson et a1.2 used a mechanical filter cycling wheel to determine a reference signal.The devices of Kirkbright et af. 3~ required extensive and expensive optical processing, using a grating monochromator and photomultiplier to detect the returned signal. A fluorescence monitoring pH sensor5 again used large, expensive optical components, such as a xenon arc lamp, monochromator and a commercial spectro- fluorimeter. The use of such hardware severely increased costs, thereby reducing the advantage of cheap fibre optical components. * Permanent address: Ministry of Defence, PO Box 2250, Haifa t To whom correspondence should be addressed. 31021, Israel. The device described here is designed with simplicity of components included as a key feature. A small, bright light-emitting diode is used as the source and detection is with a Si p - i - n diode.The use of these components means that this sensor is comparatively inexpensive, with a very rapid response, and can be packaged in such a way as to be portable and, if necessary, battery operated. Such a pH meter, with an interchangeable optical probe, could then be a direct substi- tute for the small, pocket-sized electrical pH meters available for use in titration studies. Experimental The device is illustrated schematically in Fig. 1. An ultra- bright green LED (RS components) with a peak wavelength of 565 nm (40 nm full width at half-maximum) and output intensity of 120 mcd is modifided by polishing the clear plastic lens flat to within ca. 1 mm of the emitting semiconductor chip. This is then mounted in a standard SMA fibre optic device housing (TO18 size, Radiall).The light emitted is launched into a fibre bundle (B,) consisting of two 600 pm diameter core fibres (PCS 600, Quartz et Slice) and one 200 pm diameter core fibre (PCS 200) held together in a modified SMA connector barrel. This 2 m long bundle is separated after the connector, with the narrow fibre (F) being led to a Si p - i - n diode (D1) (BPX 65) and the two 600 pm fibres being led to the input end of the sensor. The large size of the LED emitting surface (ca. 1.5 mm across) enables light to be coupled into each fibre. The sensor head is manufactured in this instance from a small cylinder of stainless steel with an internal length of ca. 1 cm. The end of the cylinder is closed with a threaded stainless-steel cap, which is polished to provide a high level of reflectivity.Both ends are demountable for cleaning and f7 Fig. 1. Schematic diagram of the sensor1096 ANALYST, SEPTEMBER 1986, VOL. 111 re-polishing, if necessary. As polished stainless steel is used, this can readily be cleaned after use and re-polishing should therefore be infrequent. Should this happen, re-calibration can easily be performed with a solution of pH 7 or 9, with reference to the calibration graph. At the input end of the cylinder, four further 600 pm diameter fibres, arranged around the two input fibres, collect light that has passed through the solution twice, having been reflected by the polished end-cap. This fibre bundle (Bz) is led to the input of a second Si p - i - n diode (DZ), where the light intensity level is determined.The device is used by placing the sensor head in the solution containing the indicator dye and the absorption characteristic of the indicator, averaged over the narrow emission band of the LED, is determined. The fibre bundles used were made from individual fibres, stripped of their plastic coating at the end and cemented with Epotek 353ND epoxy resin. The SMA connector barrels and the six fibre inputs of the sensor head were constructed in this way and a polished surface was obtained at the fibre interfaces by conventional means, in order to ensure good optical coupling characteristics. The resin - silica fibre combination is inert and is resistant to both strongly acidic and alkaline solutions. The sensor cylinder itself could be made from a plastic material, e.g., PTFE or nylon, when solutions in which the stainless steel is attacked are investigated. The electrical design of the device is described below.A simple configuration is employed, in which the light from the 200 prn diameter fibre is detected and used to monitor any change in the emissivity of the LED. Such ultra-bright diodes must be operated close to the maximum current (ca. 30 mA) and therefore a gradual change in the emission level may occur, so it is important to monitor this directly with a fibre optic connected to the main input fibre bundle (BJ. In normal use lengths of able above a few metres are not necessary for a laboratory instrument, and as a fibre bundle is used it is preferable to design the sensor with the required length of cable.However, this does not preclude an initial design with long bundles for use in remote sites. The output from each detector is amplified with conventional integrated circuit devices and the signal levels are displayed in this instance on digital voltmeters. The ratio of these signals reflects the pH of the solution. Indicator and Reagents The indicator dye used in this work was phenol red (phenolsul- phonphthalein). This was chosen because it can be used over a comparatively wide range of pH and has a strong absorption feature at the peak emission wavelength of the LED. It exists in two tautomeric forms, which results in the sizes of its two absorption peaks varying in response to the pH of the solution. The absorption spectrum in the range 350-600 nm is shown in Fig.2 for four sample solutions of pH 6.8-9.7, using a 1-cm absorption cell in a conventional spectrophotometer. This shows clearly the peak that is monitored in this device, at ca. 565 nm. The device was calibrated against a conventional electronic pH meter (Kent EIL 3055) referenced to standard buffer solutions, which were carefully made in the usual way. Solutions and chemicals were purchased from BDH Chemicals. Solutions were prepared containing ca. 0.5 ml of indicator and the calibration was performed in a solution, stirred continuously, which contained the fibre optic probe and the conventional electronic meter probe. Measurements were performed in ambient light, although the beaker containing the solution was wrapped in an opaque layer.The construction of the probe is such that the fibre bundle at the sensor head is shielded from direct light. For use in very bright environ- ments, the LED may be modulated electronically and discrimination against the unmodulated background light can be performed easily in the signal processing circuitry. Calibration The response of the device is shown in Fig. 3, with the ratio of the signals measured by the two photodiodes plotted as a function of the pH of the solution. Excellent reproducibility is shown by the coincidence of three successive experimental runs. The non-linear response is not a problem as this calibration may conveniently be stored in, and re-called from, a simple microprocessor, The resolution of the sensor is k0.05 pH unit, arising from the inaccuracy in the measurement of the absorption peak, in the approximately linear region of the device response.The stability of the reading is good, <0.5% variation over a period of 1 h for a constant solution. A major advantage of this sensor is that its response time to changes in the pH of the solution is very rapid (and diffusion limited only) owing to the optical method of measurement with detectors with a sub-microsecond response. This con- trasts with the conventional electronic meter, which requires several seconds to readjust and show the same change. The temperature coefficient may be expressed as a change of pH per degree and was determined as 0.013 pH unit K-1 over the temperature range 2440°C around pH 7.5. This compares closely with the results from other workers (0.017, phenol red2; 0.013, bromothymol blue3) for indicator dyes and is significantly better than that seen with a conventional glass electrode.0 ljbx?-l 600 500 400 Wavelength/nrn Fig. 2. pH. A, 9.7; B, 8.0; C , 7.2; and D, 6.8 Absorption spectrum of the indicator in solutions of different 100 90 a, (0 v) - = 80 Lc a, a m C a, c 70 a . .- c m a 60 pH value Fig. 3. Transmission through the sensor as a function of the pH of the indicator solution for three separate experimental runs. 0, First run; 0, second run; and X , third runANALYST, SEPTEMBER 1986, VOL. 111 1097 Conclusion and Future Work This study has demonstrated a simple, inexpensive fibre optic pH sensor for use in liquid solutions. The opt0 - electronic components used were an LED and Si p - i - n detectors, with simple signal processing and no optical filtering elements.A rapid response sensor results, which offers considerable benefits in terms of cost, safety, reliability and applicability to certain corrosive environments. Additionally, a source moni- toring reference channel, incorporated in the input fibre bundle, is included to correct for possible source fluctuations. Future developments lie in two directions. Work is proceed- ing towards the inclusion of a reference channel, the fibre optic of which is fully incorporated in the input and output optical connectors of the sensor itself. Alternatively, a balanced intensity approach to referencing using two separate fibre paths for the modulated and reference signal, with both signals transmitted through both channels, can be used.6 A second optical source at the same wavelength as the first is used and the output does not (theoretically) depend on source intensities, fibre and connector attenuations or detector sensitivities. Further, microprocessor techniques may be used to calculate the signal ratios involved directly.Even with the inclusion of such additional features, the cost of the sensor can still be kept comparatively low and yet a sophisticated and accurate device produced. Such a sensor has a temperature characteristic superior to that of a conventional electronic meter and, in contrast to some proposed fibre optic pH sensors, will not be affected by parameters such as humidity changes in the air or a small change in ambient temperature.7 The authors acknowledge technical assistance from Mr. R. A. Valsler in the construction of the device, and helpful suggestions from Mr. R. K. Selli regarding signal ampli- fication. 1. 2. 3. 4. 5. 6. 7. References Grattan, K. T. V., MeasurementJ. Int. Meas. Confed., 1984,2, 134. Peterson, J. I., Goldstein, S. R., Fitzgerald, R. V., and Buckhold, D. K., Anal. Chem., 1980, 52, 864. Kirkbright, G. F., Narayanaswamy, R., and Welti, N. A., Analyst, 1984, 109, 15. Kirkbright, G. F., Narayanaswamy, R., and Welti, N. A., Analyst, 1984, 109, 1025. Wolfbeis, 0. S., Fresenius 2. Anal. Chem., 1985, 320, 271. Culshaw, B., Foley, J . , and Giles, I. P., SPIE Proc., 1985,574, 117. Attridge, J. W., Leaver, K. D., and Cozens, J . R., “Proceedings of the Conference on Sensors and their Applications, Southampton, UK, September, 1985,” Institute of Physics, London, 1985, p. 103. Paper A6144 Received February 12th 1986 Accepted April 17th 1986

ISSN:0003-2654    

DOI:10.1039/AN9861101095

出版商:RSC    

年代:1986

数据来源: RSC