8 ANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 Research and Development Topics in Analytical Chemistry The following are summaries of seven of the papers presented at a Meeting of the Analytical Division held on July 16th-I7th, 1990, in ICI C and P Ltd., Runcorn, Cheshire. Thermospray Interface for High-performance Liquid Chromatography-Diffuse Reflectance Fourier Transform Infrared Analysis Alan M. Robertson, Lindsay Wylie and David Littlejohn Department of Pure and Applied Chemistry, Thomas Graham Building, University of Strathclyde, Cathedral Street, Glasgow GI IXL R. John Watling Philips Scientific, York Street, Cambridge CB I 2PX Christopher J. Dowle ICI Wilton Materials Research Centre, Wilton, Middlesbrough TS6 8JE Over the past decade, the use of combined techniques for solving complex analytical problems has become increasingly popular. This is particularly true in the interfacing of chromato- graphy with molecular spectroscopy.Techniques such as gas chromatography-mass spectrometry, gas chromatography- Fourier transform infrared spectrometry and high-perfor- mance liquid chromatography-mass spectrometry have become relatively commonplace. Unfortunately, owing to the presence of the condensed mobile phase in liquid chromato- graphy, no successful interface has been developed for high-performance liquid chromatography-Fourier transform infrared (HPLC-FTIR) spectrometry. The use of infrared detection in HPLC has many advantages. The high scan speed and sensitivity of FTIR spectrometers have made it possible to record the infrared spectra of individual components separated by HPLC.Further, most organic solutes exhibit an infrared absorption spectrum, and so the desirable characteristic of “universal” detection is more of a possibility than with ultraviolet (UV) absorption or refractive index measurements. The infrared detector can also provide chemically specific information by monitoring particular absorption bands, which may simplify the separation condi- tions required for some HPLC analyses. Various procedures have been proposed to interface HPLC with FTIR.14 Much of this work has involved the use of flow cells,1-3 but many problems have been encountered due to the absorption of infrared radiation in important wavelength regions by HPLC solvents. More favourable results can be obtained if solvent elimination is applied prior to FTIR detection.4-6 A thermospray can be used to evaporate the solvent and deposit a concentrated spot of solute on to a substrate, which may then be positioned into the diffuse reflectance (DRIFT) accessory of an FTIR spectrometer.A thermospray device has been constructed for operation in normal and reversed-phase HPLC. The column effluent is de-solvated and deposited on to a moving ribbon which continually transfers the solutes into the DRIFT detector. Various aspects of the interface design have been investigated and optimised. The interface has been applied to the analysis of plastic additives, saccharides, aliphatic carboxylic acids and amino acids. Computer programs have been developed to Glycine Methionine Phenylalaninc Tyros i ne 0 20 40 60 80 Filenumber 0 5 10 15 20 Umin Fig.1 HPLC-FTIR trace for the analysis of amino acids. Functional group: amine deformation. Wavenumber window = (1660-1550 cm-1). All four amino acids detected vv 0 20 Tyrosine I 40 60 80 Filenumber I I 1 I I 1 0 5 10 15 20 Umin Fig. 2 HPLC-FTIR trace for the analysis of amino acids. Functional group: aromatic C-C. Wavenumber window = (1460-1440 cm-l). Specific detection of aromatic amino acidsANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 9 provide FTIR chromatograms based on data from a number of infrared spectral windows corresponding to different func- tional groups. Experimental Instrumentation The instrument used for HPLC analysis was a PU 4100 liquid chromatograph in conjunction with a PU 41 10 UV-visible detector.The FTIR instrument was a PU 9800 series FTIR spectrometer. Both instruments were provided by Philips Scientific, Cambridge, UK. Method Effluent from the HPLC system was transferred into the thermospray system via HPLC stainless-steel tubing, coupled to a 30 cm length of capillary tubing (125 pm i.d., 1.5 mm 0.d.). The capillary tubing was inserted into a 10 cm length of copper tubing (1.5 mm i.d., 4 mm 0.d.). Thermocoax heating wire was brazed on to the outside of this copper tubing. The wire was connected to a 60 V, 3 A variable power supply. The capillary tubing protruded about 1.5 cm from the heating assembly. The temperature of the emerging HPLC effluent was monitored using a ‘K’ type thermocouple [Fluke (GB), Watford, Hert- fordshire] attached to the microbore tubing, midway between the heated region and capillary tip.Effluent from the thermospray (flow-rate normally 1 ml min-1) was sprayed on to a moving stainless-steel ribbon positioned 1 cm below the thermospray, and at right angles to it. The ribbon was 20 pm thick and 12 mm wide, and moved at 1 cm min-1. DRIFT spectra of deposited solutes were collected continuously as the moving ribbon passed through the infrared beam. The FTIR spectrometer was coupled to a Dell system 200 computer enabling acquired data to be stored and processed using commercial and specially developed software. Results and Discussion Fig. 1 shows the FTIR chromatogram obtained for the reversed-phase HPLC separation of four amino acids. The 0 20 40 60 80 Fi lenurn ber I I I I I I 0 5 10 15 20 Urnin Fig.3 HPLC-FTIR trace for the analysis of amino acids. Functional group: unknown. Wavenumber window = (940-870 cm-1). Specific detection of glycine wavenumber window chosen is characteristic for amino acids and all four amino acids have been “universally” detected. Of the amino acids analysed, only two are aromatic, and these may be specifically detected by monitoring the wavenumber window 1460-1440 cm-1 (Fig. 2). Similarly, Fig. 3 indicates that the wavenumber 940-870 cm-1 is specific for the detection of glycine among the amino acids chosen. The analysis of amino acids was carried out using a 100% aqueous mobile phase and a 250 x 4.6 mm i.d. ODS 2 column. As a result, the thermospray was operated at a temperature of 280 “C and the flow-rate reduced to 0.5 ml min- l .Fig. 4 shows the spectral overlay of the HPLC-DRIFT interface spectrum and the standard DRIFT spectrum for tyrosine. Comparison of the two spectra indicates that identification has been achieved for this particular analyte and that no sample degradation has taken place via passage through the thermospray, even at an operating temperature of 280 “C. 105.0 100 - - 90 80 f 70 60 0, .- 5 C 50 c 4000 3000 2000 1 500 1000 Wavenum berlcm - 1 Fig. 4 Spectral overlay of HPLC-DRIFT interface spectrum (upper), and standard DRIFT spectrum (lower) for tyrosine. Upper spectrum baseline corrected and multiplied by a factor of two Similar results were obtained for the analysis of phenolic antioxidants, and also for non-UV absorbing species such as saccharides and aliphatic carboxylic acids.No sample derivati- zation procedures were required for the analysis of aliphatic carboxylic acids and the results obtained indicated potential improvements over differential refractive index detection. Conclusion The development of a HPLC-FTIR interface has provided both ‘universal’ and chemically specific information on non- volatile complex samples which cannot be handled by gas chromatogrgphy-FTIR. The interface has been applied to a variety of sample types with little or no sample degradation being observed. Spectral identification of HPLC separated components may be achieved by comparison of interface spectra with standard spectra, and the interface can be used with most normal and reversed-phase HPLC solvent composi- tions.References 1 2 3 4 Yeung, E. S., in Chemical Analysis, ed. Winefordner, J. D., Wiley, New York, 1986, ch. 3. Fujimoto, C., Vematsu, G., and Jinno, K., Chromatographia, 1985, 20, 112. Hellgeth, J. W., and Taylor, L. T., Anal. Chern., 1987,59,295. Kalasinsky, V. F., Whitehead, K. G., Kenton, R. C., Smith, J. A. S., and Kalasinsky, K. S., J. Chromatogr. Sci., 1987, 25, 273. Griffiths, P. R., in Analytical Applications of Spectroscopy, eds. Creaser, C. S., and Davies, A. M. C., Royal Society of Chemistry, London, 1988. Gagel, J. J., and Biemann, K., Mikrochim. Acta, Part 11, 1988, 185. 5 610 ANALYTICAL PROCEEDINGS. JANUARY 1991, VOL 28 Determination of Nitrite and Nitrate Ions by Electron Paramagnetic Resonance Spectrometry Eric P.K. Tsang, D. Thorburn Burns and Brian D. Flockhart School of Chemistry, The Queen‘s University of Belfast, Belfast BT9 5AG Alkali nitrites and nitrates are used as essential additives in a variety of foods to prevent bacterial spoilage and food poisoning. Nevertheless, concern exists over the presence of nitrate in the diet, not because of any intrinsic behaviour of the ion itself, but rather because of its ability to form nitrite by microbiological or other biochemical processes. Consequently, whenever nitrate is ingested in food, there is always the possibility that some of it will be reduced to nitrite. Under the acidic conditions in the stomach the nitrite ion can react with dietary components to form products that are often associated with carcinogenesis. 1 It is therefore highly desirable that the amounts of nitrite and nitrate in foodstuffs should be moni- tored closely.Various methods are available for this purpose .2-4 This paper describes the application of electron paramagnetic resonance (EPR) spectrometry to the determination of these ions. Principle of the Method The early work of Michaelis et aZ.5 on one-electron oxidation- reduction processes revealed many classes of organic com- pound exhibiting this behaviour. The compounds studied included phenothiazine and its derivatives. Later, Sawicki et aZ.6 showed that nitrite can oxidize such compounds to form coloured products for spectrophotometric determinations and suggested that free radicals were present. The present method is based on this principle, viz., Nitrite + reagent + stable free radical The formation of the free radical is autocatalytic6 and hence nitrite can be determined at trace levels by EPR spectrometry.Results A variety of compounds that can be oxidized by the nitrite ion were studied but only the most effective reagents are reported herein. Nitrate was determined after prior reduction to nitrite by spongy cadmium metal. With mixtures of nitrite and nitrate, the nitrite was first removed by volatilization as nitrosyl chloride7 and the nitrate then determined. The nitrite plus nitrate content was obtained by reduction followed by determi- nation of the total nitrite present. Method A. Phenothiazine A 0.1 ml volume of the test solution, delivered from an Agla micrometer syringe, was diluted to 10 ml with a solution of 0.05% phenothiazine in glacial acetic acid.The measurements were taken 50 min after mixing. The calibration graph was linear in the range 0-1.5 pprn of nitrite. The precision at 0.3 ppm was 1.2% (ten measurements). The detection limit at three times the standard deviation was 0.012 ppm of nitrite. Nitrate and other strong oxidizing agents interfere but reducing agents such as sulphite can be tolerated. Method B. N,N,N’,N’-Tetramethyl-p-phenylenediamine A 0.1 ml volume of the test solution, delivered from an Agla micrometer syringe, was diluted to 10 ml with a 0.05% solution of the reagent in glacial acetic acid. One millilitre of sulphamic acid was added to stabilize the signal which then remained unchanged for at least 2 h. Measurements were taken 50 min after mixing.The calibration graph was linear from 0 to 1.3 ppm; the precision at 0.3 ppm was 1.5% (ten measurements). The detection limit at three times the standard deviation was 0.025 ppm of nitrite. Nitrate and strong oxidizing agents interfere but reducing agents can be tolerated. Applications The methods were applied to the determination of the nitrite and nitrate contents in pre-packed cooked ham and in soft spreading cheese. The results were compared with those obtained by International Standards Organization (IS0)g and Association of Official Analytical Chemists (AOAC)9 stan- dard methods for these samples. Cooked Ham The samples were prepared for analysis by the I S 0 procedure.8 The results are shown in Table 1. Table 1 Nitrite and nitrate content of cooked ham Method A B I S 0 Nitrite 50.2 49.8 50.5 (PPm) Nitrate 65.2 64.9 65.6 (PPm) Soft Cheese The samples were prepared for analysis by the AOAC procedure.9 The results are shown in Table 2.Table 2 Nitrite and nitrate content of soft cheese Nitrate Nitrite A 18.5 62.4 B 18.2 61.6 AOAC 18.8 62.6 Method (PPm) (PPm) Conclusions The EPR methods outlined are simple and sensitive and extraction is not required. Further, the methods can tolerate large amounts of reducing agents which should permit their application to pollution studies, such as the determination of nitrogen dioxide in the presence of sulphur dioxide. An additional advantage is that they are applicable to coloured solutions. It is envisaged that residual interferences can be dealt with by ion chromatography.References Glidewell, C., Chem. Br.. 1990, 26, 137. Williams, W. J . , Handbook of Anion Determination. Butter- worths, London, 1979. Snell, F. D.. Photometric und Fluorometric Methods of Anulysis -Nun Metals, Wiley. New York, 1981. Marczenko, Z., Separation and Spectrophotometric Determina- tion of Elements. Ellis Horwood, Chichester, 1986. Michaelis, L., Granick, S . . and Schubert, M. P., J. Am. Chem. Soc., 1941. 63. 351. Sawicki, E., Stanley, T. W., Pfaff, J., and Johnson, H . , Anal. Chem.. 1963, 35, 2183.ANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 11 7 Velghc, N., and Claeys, A., Analyst, 1983, 108, 1018. 8 Meat and Meat Products, Determination of Nitrite Content (reference method), International Standards Organization, IS0 2918-1975(E), Sections 7.1-2, 8.3.1-5 and 8.4.1-4.9 Official Methods of Analysis of the Association of Official Analytical Chemists, ed. Williams, S . , Association of Official Analytical Chemists, Arlington, VA, 14th edn., 1984, Nitrate and Nitrite in Cheese, Section 16.278-16.283. Analysis of Fatty Acid Methyl Esters by Using Supercritical Fluid Chromatography with Mass Evaporative Light Scattering Detection Simon Cocks and Roger M. Smith Department of Chemistry, Loughborough University of Technology, Loughborough, Leicestershire LEI1 3TU The analysis of fatty acids and lipids is of great interest to the food, agricultural and pharmaceutical industries. 1 In the past, the need for the quantification of specific components as well as characterization in quality control has relied mainly on the established chromatographic techniques of gas chromato- graphy (GC) , high-performance liquid chromatography (HPLC)2 and thin-layer chromatography (TLC).For many samples HPLC has come to the forefront of these techniques as the conditions of separation lend themselves to thermally labile high relative molecular mass compounds.3 However, because of the relatively low separation efficiency isomeric components can be difficult to resolve and the absence of a chromophore means that detection of the lipids is a problem. The recent development of practical supercritical fluid chromatography (SFC) should provide a technique that is ideally suited to the separation of fatty acid methyl esters (FAMEs) and separations have been reported using both packed and capillary columns.4 The enhanced diffusivity and solubility of supercritical fluids with respect to liquids and gases potentially offer a fast, efficient separation.The low critical temperature and pressure of carbon dioxide, the most widely used mobile phase, are ideal for the analysis of non-volatile and thermally labile components, and the ability t o adjust the density of the mobile phase with temperature and pressure provides a wide range of possible conditions to effect a separation. Column eluent Light Nebulisation make-up gas- Nebuliser I Heated drift tube Scattered light detected source- md - I' t Exhaust Schematic diagram of a mass evaporative detector Fig. 1 Fatty acids possess a poor chromophore, hence ultraviolet detection, the most common non-destructive detection tech- nique, is of limited value.Although flame ionization detection is universal it is incompatible with modified mobile phases. Infrared spectroscopy is a possible method of detection; however, strong absorbances from carbon dioxide and modifi- ers limit the technique for on-line work. These difficulties led to the present study of the application of the evaporative light-scattering detector (LSD) to the SFC of fatty acids. It is a universal detector, the response of which should be indepen- dent of mobile phase composition. It has been successfully used for many years in liquid chromatography,5-7 and its advantages and disadvantages are well documented.8-10 Preliminary reports have appeared of its application in packed column SFC.11-13 The operation of the detector can be divided into three stages (Fig. 1).lo First, sample introduction: addition of make-up gas and nebulization. Second, evaporation: move- ment of the mist down the heated drift tube with evaporation of the volatile mobile phase components. Third, detection: scattering of the incident light by the remaining solute particles and detection by a photomultiplier tube. Discussion A range of fatty acids were examined in this preliminary study. They varied both in the degree of unsaturation and chain 1 0 8 6 4 2 Time/mi n Injection 20 pl 0 1 I I I 1 1 0 8 6 4 2 0 Ti me/mi n Injection 5 pI Fig. 2 Separation of CI8 fatty acid methyl esters on a Spherisorb column with supercritical fluid carbon dioxide 4 ml min-1 (density 0.871 g ml-1).Nitrogen make-up gas 60 lb in-2: (a), 20 p1 injection of analytes with small amount of glass wool in drift tube; (b), 5 pl injection with additional glass wool in drift tube showing reduced noise and increased sensitivity12 ANALYTICAL PROCEEDINGS. JANUARY 1991, VOL 28 length from CI2:o to CZ4: 1. All of the samples were made up in hexane to concentrations in the region of 1 x 10-6 g pl-1, and then analysed on a silica column (Spherisorb SSW), using an unmodified carbon dioxide mobile phase at selected densities in the region of 0.800 g ml-1. Pressure regulation within the SFC system was maintained using a heated Rheodyne 7037 pressure relief valve. The fluid conditions were extended into the nebulizer and detector of a Varex mass evaporative detector by means of a crimped tube.If the pressure was released between the regulator and detector the analytes condensed out in the connecting tubing, resulting in a small or negligible signal. It has been found to be essential to obtain good nebuliza- tion14 of the sample and the droplet size is controlled by the adjustment of the drift tube temperature and nebulizer gas flow-rate.15 The amount of light scattered is partially a function of the size of the nebulized particles as well as the wavelength of the incident light beam and the angle of collection. In contrast to the situation with HPLC, it was found that the addition of nitrogen as a make-up gas before nebulization was not necessary with SFC and could be omitted. Even at low flow-rates the large volumes of gas produced from the expansion of the carbon dioxide mobile phase gave good nebulization. Fig.3 24 20 15 12 6 0 Tirne/min Injection 10 KI Separation of C I ~ , Clh and CI8 fatty acid methyl esters on Spherisorb’column, eluent carbon dioxide (density 0.841 -g ml-I) 2.5 ml min-1 Initially, the detection of the FAMEs proved problematical owing to the presence of a large amount of base line noise. The noise was present only when carbon dioxide was passing into the detector,l6 and not when the system was being purged with nitrogen from the nebulizer head, thus suggesting that the noise was a function of the carbon dioxide. Removal of the column did not alleviate the problem. It appeared that as the carbon dioxide passed out of the interface into the detector small particles of solid carbon dioxide were formed during the adiabatic expansion, which then passed quickly down the drift tube and through the light source.Increasing the drift tube temperature up to 100°C did not eliminate the problem and worsened the detection limits as a result of partial evaporation of the more volatile analytes. However, it was found that the problem could be removed by plugging the base of the drift tube with a small amount of silanised glass wool. This appeared to act as a heat trap or heat exchanger, causing both turbulent flow and melting of the carbon dioxide particles. The base line noise was significantly reduced and the signal to noise ratio was further improved by additional plugging (Fig. 2). The addition of the glass wool had no apparent effect on the detection limits or peak efficiencies of the samples, which were easily detected at concentrations in the region of 1 x 10-7 g pl-1.The separation of the FAMEs followed a normal-phase type retention mechanism and was primarily dependent on the degree of unsaturation. The chain length played only a small part in the retention processes as C14:o was eluted slightly before Cl6:0 (Fig. 3). This was also true for the elution positions of Cl8:o and cz0:o esters. Separation of the isomeric oleic and elaidic acid methyl esters, c18:1, could readily be achieved with near base line resolution. In order to try to reduce the over-all retention time of the separation 1% methanol was used to modify the mobile phase, but the much shorter retention times brought about co-elution of the isomeric ClXz1 fatty acids.Conclusion The evaporative light scattering detector can be used with an SFC system if slight modifications are made to the drift tube. The results also show that SFC can be used successfully for the separation of both unsaturated and saturated FAMEs. A near baseline separation of the Clx:l isomers was achieved on a silica column. The authors thank the Trustees of the Analytical Trust Fund for the award of an SAC Studentship to S.C. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 References Hammond. E. W., J . Chromatogr.. 1981, 203, 397. Christie, W. W., Conner, K., and Nobel. R. C., J . Chromat- ogr., 1984, 298, 513. Shukla. V. K. S . , J . Disp. Sci. Tech., 1989, 10, 581. White, C. M.. J. High Resolut. Chromatogr.Chromatogr. Commun., 1985. 8,293. Stolyhwo, A., Colin, H., and Guiochon. G . , Anal. Chem., 1985,57, 1342. Robinson, J. L., Tsimidou, M., and Macrae, R., J . Chromat- ogr., 1985, 324, 35. Stolyhwo, A., Martin, M., and Guiochon, G., J . Liq. Chromat- ogr., 1987, 10, 1237. Stolyhwo, A., Colin. H., and Guiochon, G., J . Chromatogr.. 1983, 265, 1. Stolyhwo, A., Colin, H., Martin, M., and Guiochon, G., J . Chromatogr., 1984, 288,253. Macrae, R., Znt. Anal., 1987, 1, 14, 15, 21 and 24. Carraud, P., Thiebaut. D., Caude, M., Rosset. R., Laffosse. M., and Dreux, M., J. Chromatogr. Sci., 1987, 25, 395. Gorner, T., and Perrut, M., Liq. Chromatogr. Gas Chromat- ogr., 1989,7. 502. Nomura. A., Yamada. J . . Tsunoda, K . , Sakaki, K.. and Yokochi, T., Anal. Chem., 1989, 61, 2076.Mourey, T. H.. and Oppenheimer, L. E., Anal. Chem., 1984, 56, 2472. Charlesworth, J . M., Anal. Chem., 1978, 50, 1414. Nizery, D., Thiebaut, D., Caude, M., Rosset, R., Laffosse, M.. and Dreux, M.. J. Chromatogr., 1989. 467, 49.ANALYTICAL PROCEEDINGS. JANUARY 1991, VOL 28 13 Potassium-selective Electrode Based on a Dioxacalixarene: An Example of a New Series of lonophores Aodhamar Cadogan and Dermot Diamond* School of Chemical Sciences, Dublin City University, Dublin 9, Ireland Suzanne Cremin and Anthony M. McKervey Department of Chemistry, University College Cork, Cork, Ireland Stephen J. Harris Loctite (Ireland Ltd.), Whitestown Industrial Estate, Tallaght, County Dublin, Ireland The ability of calix[4]arenes to act as sodium sensors in ion-selective electrodes (1SEs) has been well established14 and successfully applied to the determination of sodium in human plasma samples.s.6 In contrast, poly(viny1 chloride) (PVC) membranes containing the larger hexameric calix[6]arenes have been used to produce caesium-selective electrodes .7 Hence, the ability of the calixarenes to act as host-guest receptors for alkali metal cations is clearly a size-related phenomenon.We now report on the properties of one of a new series of ionophores which possess polar cavities intermediate in dimensions between the tetramers and hexamers mentioned above. These ionophores differ from the normal tetrameric calixarenes in that they contain an additional ether-like oxygen in two of the four methylene bridges (Fig. 1). In the absence of any simple term to describe these substances, we propose the term “oxacalixarene” as a general trivial name, with the prefix “di-,” “tri-,” etc., depending on the number of modified methylene bridges.Bearing in mind the intermediate dimen- sions of this ligand, it was anticipated that it could be used to produce potassium-selective electrodes. Experimental Synthesis The ionophore [ 7,13,2 1,27-tetra-tert-butyl-29,30,3 1,32- tetrahydroxy-2,3,16,17-tetrahomo-3,17-dioxacalix[4]arene, Fig. 1, structure (lB)] was prepared by the procedure of Dhawan and Gutsche.8 A 1.17 g (0.006 mol) amount of tert-butyl bromoacetate, 0.65 g (0.0047 mol) of anhydrous potassium carbonate and 10 ml of AnalaR acetone were added to 0.5 g (0.0007 mol) of the parent dioxacalixarene [Fig.1, structure (lA)] and the mixture was refluxed under nitrogen with stirring for 120 h. After this time, all the volatiles were removed to give a sticky solid. This was taken up in 20 ml of dichloromethane and the extract washed twice with water. After drying with magnesium sulphate and removal of the solvent, 0.66 g (80%) of the product 1B was obtained as a pale yellow solid. Recrystallisation from tert-butanol gave 0.5 g of the colourless crystalline product 1B (m.p. 186.8- 187.8 “C). ISE Construction and Performance Materials The membrane components, PVC, 2-nitrophenyl octyl ether (2-NPOE) and AnalaR grade tetrahydrofuran (THF) were obtained from Fluka. AnalaR grade chlorides of sodium, potassium, lithium, caesium, magnesium and calcium were supplied by Riedel-de-Haen and dissolved in distilled, de- ionized water.Electrode fabrication and measurements The general procedure for the preparation of the polymeric membrane was as follows. The ligand (2.5 mg), plasticizer [2-NPOE (250 mg)] and PVC (125 mg) were mixed and * To whom correspondence should be addressed. dissolved in THF. The resulting solution was poured into a glass mould and left to set as described previously.9 Clear, flexible PVC membranes containing the ligand dissolved in the plasticizer were obtained on evaporation of the THF. Discs of suitable dimensions (diameter, 10 mm; thickness, about 0.1 mm) were cut from this master membrane and clipped into the tip of a blank Russell oxygen electrode body (reference number 98-7809). The internal reference electrolyte was 1 X 10-1 rnol dm-3 KCI.Fabricated electrodes were conditioned for at least 2 h prior to use. Potentiometric measurements were made relative to a saturated calomel electrode using a Philips PM9421 digital pWmillivoltmeter. The selectivity coefficients (kflq‘) were determined by the separate solution method using 1 x 10-1 mol dm-3 solutions of the salts of the cations involved (all chlorides). But But But But (lB) Fig. 1 the tetrabutyl ester [structure (lB)] Structure of tetrahomodioxacalix[4]arene [structure (lA)] and Results and Discussion Sensitivity and Limit of Detection Electrodes based on ligand (1) displayed near-Nernstian responses to potassium ions (Fig. 2) with detection limits in the range 1 x 10-3-1 x 10-4 rnol dm-3 KCl. In these preliminary investigations, however, a saturated calomel reference elec- trode was used, and it is very likely that leakage of KCI from the reference electrode salt-bridge affected the ISE signal in the less concentrated potassium standards (i.e., below 1 x 10-3 mol dm-3).Replacement of the KCI bridge with a different electrolyte should therefore extend the limit of detection reported above. Selectivity The selectivity data obtained with this ISE are summarised in Table 1. Reasonably good selectivity over sodium is evident, although not in the range of similar PVC electrodes incorporat-14 > y: E 80 c L' 60 40 20 ANALYTICAL PROCEEDINGS. JANUARY 1991, VOL 28 ' 140 120 100 I I 01 I I I I I -6 -5 -4 -3 - 2 - 1 Log aK Fig. 2 Calibration graph (e.m.f. versus log aK) for the ionophore incorporated in a plasticized PVC membrane.Average slope over the range 1 x 10-'-1 x mol dm-3 KCl is 53.7 mV per decade change in potassium activity ing the natural antibiotic valinomycin. However, these results demonstrate that the dioxa derivatives of the tetrameric calixarenes are potassium-selective, and can be used to produce functioning potassium-selective electrodes. Given the numerous options for structural modification of both calix- arenes and oxacalixarenes, and the attendant variable selectiv- ity, there is a clear possibility of producing a wide range of new Table 1 Selectivity coefficients (log k c ' ) for electrodes based on ligand 1. Values obtained by the separate solution method in 1 x 10-1 mol dm-3 solutions of each metal ion (all chlorides) Log k,qi"', i = K+ Na+ Li+ Cs+ Rb+ Ca2+ Mg2+ NH4+ -1.4 -2.4 -0.6 -0.1 -2.0 -2.1 -1.0 and more efficient ionophores for ISEs targeted at various cations.References Diamond, D., Anal. Chem. Symp. Ser.. 1986, 25, 155. Diamond, D.. Svehla. G.. Seward. E., and McKervey, M. A., Anal. Chim. Acta, 1988, 204, 223. Cadogan, A., Diamond, D.. Smyth. M. R.. Deasy. M., McKervey. M. A., and Harris, S. J., Analyst, 1989. 114. 1551. Kimura. K., Miura, T.. Matsuo. M.. and Shono, T., Anal. Chem., 1990, 62, 1510. Telting Diaz, M., Diamond, D.. Smyth. M. R.. Seward, E., McKervey. M. A., and Svehla, G., Anal. Proc., 1989.26,29. Telting Diaz, M.. Regan, F.. Diamond. D.. and Smyth, M. R., J . Pharm. Biomed. Anal., 1990, 115, 1207. Cadogan, A.. Diamond, D.. Smyth, M. R., Svehla.G., McKervey, M. A., Seward, E., and Harris. S. J., Analyst, 1990, 115, 1207. Dhawan, B.. and Gutsche. C. D., 1. Org. Chem., 1983,48.1536. Moody, G. J., and Thomas, J. D. R., in Chemical Sensors. ed. Edmonds. T. E.. Blackie. Chapman and Hall. New York, 1988. ch. 3, p. 76. Spectrophotometric End-point for the Phase-titration Determination of the Adulteration of Petrol With Kerosine M. Shahru Bahari, W. J. Criddle" and J. D. R. Thomas School of Chemistry and Applied Chemistry, University of Wales College of Cardiff, P.O. Box 912, Cardiff CFI 3TB The increasing number of instances of petrol adulteration with kerosine in countries where controls are difficult to exercise has led to various studies of procedures for detecting the adultera- tion. As adulteration occurs mostly during product delivery between refineries and petrol stations, researchers have endeavoured to find procedures for detecting the adulteration at this stage, in order to prevent the adulterated product reaching the public.. A mobile laboratory with a gas chromatograph set up in a van by deAndre Bruening and Branco in Brazil has been reported1 but no other work has been cited for an 'in-the-field' method of analysis. Although gas chromatography is arguably the best procedure for quantitatively detecting kerosine or other contaminants in petrol, the purchase cost could be a considerable problem in developing countries in the campaign against adulteration syndicates. Hence, an ideal instrument for detection should be simple to operate and affordable by petrol station owners, thus preventing the adulterated product from reaching customers.A phase-titration method, first investigated by Suri and co-workers,2 has recently been modified,3 the modified pro- cedure showing considerable improvements in linearity over the adulteration range of &20% v/v kerosine. The approach was based on the differences between the solubilities of petrol and kerosine in a water-based mixed solvent system arising from the variations in the hydrocarbon content. The introduc- tion of a surfactant, namely sodium dodecyl sulphate (SDS), * To whom correspondence should be addressed. Fig. 1 Schematic diagram of the flow-through phase-titration system: A. titrant; B, titration vessel; C. magnetic stirrer; D, peristaltic pump; E, spectrophotometer cell; F.spectrophotometer detector; and G. UV light source into the titrating solution has served to increase the titration volume difference for small percentages of adulteration, thus giving greater sensitivity and accuracy. However, the visual method of identifying the end-point of a phase titration has always been subject to uncertainties and hence criticism, and in order to reduce errors in the visual end-point it is necessary for the end-point to be determined instrumentally.ANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 15 A spectrophotometric approach permits effective end-point detection in turbidimetric titrations4 and the same approach can also be applied to the clarification titrations depending on 1 .o 1 0.5 a C m 8 16 24 32 40 removal of turbidity.Hence, a spectrophotometer has been used in this study as a detector for a modified phase-titration procedure designed to reduce errors which can occur during the visual detection of the end-points. Experimental Reagents and sample preparations for the turbidimetric phase-titration procedure have been described previously.3 20.0 15.0 m E Y 10.0 n v) v) Lc 0 4 8 12 16 20 Volume of titranb'crns Fig. 2 Typical absorbance results for (a) the turbidimetric [propan-l- 01 (30.0 cm3), petrol (Premium grade, 20.0 cm3); titrant: 0.27 mol dm-3 SDS], and (b) the clarification [ethanol (20.0 cm3), petrol (Premium grade, 10.0 cm3) and water (5.0 cm3); titrant: ethanol] end-point titration v) 0 4 8 12 16 20 rc a - 5 0 > 20.0 15.0 10.0 0 4 8 12 16 20 Kerosine in petrol (% v/v) Fig.3 Titration results on mixtures of propan-1-01 (30.0 cm3) and petrol samples (20.0 cm3) at different concentrations of SDS. (a) Visual method; and ( b ) spectrophotometric method (h = 600 nm). [SDS]: 0, 0.13; W, 0.16; A, 0.19; 0, 0.23; 0, 0.27; and A , 0.30 mol dm-3 0 4 8 12 16 20 10.0 t 7 0 4 8 12 16 20 Kerosine in petrol (% v/v) Fig. 4 Effect of batch variations on titration results of mixtures of propan-1-01 (30.0 cm3) and petrol samples (20.0 cm3) at 24.0 k 0.5 "C. (a) Visual method; and ( b ) spectrophotometric method (A = 600 nm). 0, Batch 1; 0, batch 2; and A , batch 3 15.0 m 1: C E 2 10.0 c v- E $ 20.0 > 15.0 10.0 0 4 8 12 16 20 0 4 8 12 16 20 Kerosine in petrol (% v/v) Fig. 5 (a) Visual and ( b ) spectrophotometric clarification titration results on mixtures of petrol (Premium grade, 10.0 cm3), ethanol (20.0 cm3) and water (5.0 cm3) at 24.0 k 0.5 "C16 ANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 For the clarification titrations, mixtures of absolute ethanol (BDH, 20.0 cm3) petrol (Premium grade, 10.0 cm3) containing kerosine (0-20% v/v) and water (5.0 cm3) were titrated with the same absolute ethanol.Both these phase-titration pro- cedures were based on a flow-through system operated at 24.0 k 0.5"C. The flow-through system, shown schematically in Fig. 1, consisted of a peristaltic pump (Cole Palmer) circulating the titration mixture continuously from the titration vessel to the spectrophotometer (Shimadzu UV-120-02), fitted with a flow- through cell, and back to the titration vessel.The wavelength of the spectrophotometer was set at 600 nm. Table 1 Statistical data for the titration results. Titration results at 24.0 -t- 0.5 "C for typical petrol samples with added kerosine Volume of titrant/cm' Kerosine in petrol (% v/v) Visual Spectrophotometric 0 2 4 6 8 10 12 14 16 18 20 19.74 f 0.10 18.46 f 0.05 17.44 f 0.03 16.44 k 0.19 15.28 f 0.05 14.19 k 0.06 13.23 f 0.14 12.18 k 0.13 1 1.39 f 0.06 10.60 k 0.08 9.95 k 0.10 20.70 f 0.08 19.60 -I- 0.22 18.60 f 0.08 17.50 k 0.18 16.50 f 0.08 15.50 k 0.34 14.56 k 0.10 13.50 f 0.14 12.40 f 0.14 11.50+0.18 10.80 k 0.08 Regression analysis: Regressionequation y = 19.41 - 0 . 5 0 ~ y = 20.58 - 0 . 5 0 ~ coefficient ( r ) 0.997 0.999 Aliquots of the titrating solution were introduced into the titration mixture through a burette and vigorously stirred with a magnetic stirrer.The flow of the mixture was controlled at a maximum of 2.0 cm3 s-1 in order to prevent the formation of bubbles which could interfere with the absorbance reading. The turbidimetric end-point was chosen as the point causing an increase in the absorbance of the mixture from a constant value, while the clarification end-point was that corresponding to a levelling off of absorbance to constant values (Fig. 2). Results and Discussion Repetition of turbidimetric end-point titrations at various SDS concentrations gave similar titration behaviour (Fig. 3) to the results obtained from the visual detection of the end-points. However, the procedure gives significantly better linearity over a wider range of 0-20% v/v kerosine in petrol compared with only a O-l6% v/v range in the visual end-point study.3 Such improvements in the linearity range are clearly seen in Fig.4. The plot shown in Fig. 4(b) essentially eliminates the deviation from linearity for the 0.27 mol dm-3 added SDS system that is clearly apparent in Fig. 4(a). Another noticeable advantage of the modified procedure is the reduction of the titration volume difference in batch variations. The 2% v/v adulteration error arising from this batch variation of the visual method has been reduced to 1% v/v adulteration with the introduction of the spectrophoto- metric approach. Table 2 Statistical data for the titration results. Data for the titration results of Fig. 5 Volume of titrant/cm3 Kerosine in petrol (Yo v/v) Visual Spectrophotometric 0 10.50 k 0.57 10.78 f 0.38 2 11.10 t- 0.62 11.91 f 0.86 4 12.31 k 0.19 11.99 k 0.35 6 12.63 k 0.15 14.09 f 0.37 8 13.72 k 0.56 15.15 f 0.40 10 15.14 k 0.71 16.29 k 0.57 12 15.47 k 0.74 17.18 k 0.36 14 16.49 k 1.65 18.11 k 0.48 16 17.04 k 0.90 19.67 k 0.47 18 (Not recorded) 20.02 f 0.25 20 18.23 k 0.72 21.65 k 0.68 Regression analysis: Regression equation Correlation coefficient ( r ) 0.994 0.999 y = 10.53 + 0.41 x y = 10.78 + 0 .5 4 ~ A linear correlation is also observed for the clarification end-point titration (Fig. 5 ) . However, the relatively large standard deviation limits the sensitivity of this approach, and although the procedure can still be an effective monitor for field-type analysis, we would not recommend this procedure in preference to the turbidimetric end-point.A statistical analysis of the data quoted in this paper is shown in Tables 1 and 2. Conclusion The modified phase-titration method for determining the adulteration of petrol with kerosine has been made more sensitive, repeatable and precise by using a spectrophotometric end-point detector and the flow-through system described is a stage in the development of a miniaturized portable instrument designed to combat adulteration syndicates. The authors thank the Public Services Department of Malaysia, Kuala Lumpur, and the University of Technology of Malaysia for financial support to M.S.B. References 1 2 3 4 deAndre Bruening, I . M. R.. and Branco, V. A. C., Bof. Tech. Petrobras, 1980, 23, 117. Suri, S. K ., Ahluwalia. J . C., and Rogers, D. W., Tafanra, 1981, 28, 281. Bahari, M. S . , Criddle. W. J . , and Thomas, J. D. R.. Analyst, 1990, 115,417. Rogers, D. W . , and Ozsogomonyan, A., Talanta, 1964,ll. 652. Determination of Trace Amounts of Bacitracin by Differential-pulse Adsorptive Stripping Voltammetry at a Hanging Mercury Drop Electrode Josino C. Moreira, Royston D. Miller and Arnold G. Fogg Department of Chemistry, Lo ug h boro ug h University of Tech nolog y, Lo ug h boro ug h, L eicesters h ire LEI1 3TU Bacitracin consists of a group of polypeptide antibiotics Bacitracin has been added to animal feed for growth promo- containing several amino acid residues such as L-cysteine, tion, feed efficiency and disease control as bacitracin base or as D-ornithine , L-lysine, L-histidine, D-aspartic acid, D-glutamic a zinc or methylene disalicylate salt .* acid, L-isoleucine, L-leucine and D-phenylalanine.Bacitracin Bacitracin is usually assayed by microbiological methods A is the most potent and main bacitracin component. such as the plate agar diffusion technique or a turbidimetricANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 17 method.* The presence of bivalent metal ions such as zinc, cobalt, manganese, calcium or copper in the solution can interfere with the biological activity of bacitracin. Copper causes a negative bias in the microbiological assay and must be removed before the assay.3 Alternative methods are important in instances where the microbiological assay results are uncertain or when the interest is in biological transformation or degradation products.These methods include thin-layer chro- matography with ultraviolet detection, electrophoresis, high- performance liquid chromatography and pulse polarography . Bacitracin A exhibits a double wave at the dropping mercury electrode over the pH range 1-8. Its polarographic activity is probably due to reduction of the >C=N- group in the thiazoline ring.4 By using differential-pulse polarography, a concentration of 5 pg ml-1 (approximately 4 x 10-6 mol dm-3) can be detected. In this paper a voltammetric method capable of determining bacitracin A at levels as low as 30 ng ml-1 is described. Experimental Differential-pulse adsorptive stripping voltammetry was car- ried out by using a Metrohm 626 Polarecord with a 663 VA stand in conjunction with a multi-mode electrode in the hanging mercury drop electrode (HMDE) mode.The three- electrode system was completed by means of a glassy carbon auxiliary electrode and a silver-silver chloride reference electrode. -40 - 35 - 30 -25 5 -20 L 3 0 -15 -10 - 5 A D 1 0 I I I 1 I -0.10 -0.30 -0.50 -0.70 -0.90 -1.10 -1.30 PotentialN Fig. 1 Differential-pulse adsorptive stripping voltammograms of a 1 .O x lo-' M solution of bacitracin in acetate buffer, pH 4.5, at different accumulation times: A, no accumulation; B , 1 min; C, 2 min; and D, 3 min. Accumulation at -0.1 V All potentials given are relative to this silver-silver chloride electrode. A pulse amplitude of 50 mV was used with a scan rate of 10 mV s-' and a pulse interval of 1 s.pH measurements were made with a Corning combined pH-reference electrode using a Radiometer PHM 64 pH meter. Bacitracin was obtained from the Sigma Chemical Company and was used without further purification. A 0.001 mol dm-3 solution of bacitracin was prepared by dissolving 0.014 g of bacitracin in water containing a few drops of an acetate buffer solution (pH 5.0) in a 10 ml calibrated flask. This solution had to be freshly prepared. Bacitracin F solution was prepared by leaving the bacitracin A standard solution at 37°C and pH 7.5 for 2 weeks. 30 20 P ? 2 3 Y m a 10 0 0 2 4 6 8 10 Accumulation time/mi n Fig. 2 Influence of the accumulation time on the peak current of a 1 .0 x 10-7 M solution of bacitracin in acetate buffer, pH 4.5. Accumulation at -0.1 V Procedure The general procedure used to obtain differential-pulse adsorptive stripping voltammograms was as follows.A 20 ml aliquot of 0.1 mol dm-3 acetate buffer solution was placed in a voltammetric cell and the required amounts of standard bacitracin solution were added. The stirrer was switched on and the solution was purged with nitrogen gas for 6 min. Subsequently, a 15 s deoxygenation was made between adsorptive stripping cycles. After forming a new HMDE a 3 min accumulation was effected at -0.1 V whilst stirring the solution. At the end of the accumulation period the stirrer was switched off, and, after 20 s had elapsed to allow the solution to become quiescent, a negative potential scan was initiated between the accumulation potential and -1.2 V. Results and Discussion The differential-pulse adsorptive stripping voltammograms of bacitracin A at four different accumulation times are shown in Fig.1. The peaks at -1.05 and -1.15 V are due to bacitracin A. Bacitracin F, an inactive form of bacitracin formed from bacitracin A, gave a peak at -0.67 V and can be differentiated from the active form. A shift of 59.5 mV pH-l to more negative potentials was observed for the bacitracin A peak potentials with increasing pH from 1 to 7, showing consumption of hydrogen ions in the electrode reaction. The peak shape and the peak separation were not affected greatly by varying the pH from 1 to 7. The influence of the pH on the peak current is shown in Table 1. Acetate buffer, pH 4.5, was selected as the optimum because of the higher response obtained and the stability of bacitracin in this medium.1 The effect of accumulation time on the peak height for bacitracin in 0.1 mol dm-3 acetate buffer, pH 4.5, is shown in Fig. 2. The peak current increased rectilinearly with accumula-18 ANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 Table 1 Influence of pH on the height of the peak for bacitracin at -1.05 V. Bacitracin concentration = 1.0 x mol dm-'. Accumula- tion step: 2 min at -0.1 V PH i,lnA 1.0 4.5 3.0 3.5 4.5 5.9 7.0 5.2 tion time up to 5 min. After this time a deviation of the linearity was observed suggesting saturation of the electrode surface. A small peak was observed even with a 'no accumulation' scan showing that accumulation of the antibiotic occurs at the electrode surface during the scan. An accumulation time of 3 min was used generally in this work.The effect of the accumulation potential on the peak height of bacitracin is shown in Table 2. The highest peak was obtained by accumulating at -0.1 V and this accumulation potential was selected as ideal. The calibration graph obtained under the optimized condi- tions was linear. A rectilinear relationship between the peak current and the bacitracin concentration was observed from 2.5 x 10-8 to 3.5 x 10-7 mol dm-3. The slope obtained was 10.7 x lo7 nA mol-1 and the correlation coefficient was 0.996. At concentrations higher than 3.5 x 10-7 mol dm-3 the deviation of the linearity observed in the calibration graph suggests saturation of the electrode surface. No interference of surfactants at the sub-microgram per litre level was observed on the peak height of bacitracin.The peak, however, was suppressed by the presence of5 x 10-8 mol dm-3 albumin in the solution. Small decreases in the peak height of bacitracin were observed in the presence of zinc, copper(II), manganese, nickel and chromium(II1) ions. This is probably Table 2 Influence of the accumulation potential on the peak height for 1.0 x mol dm-3 of bacitracin in acetate buffer, pH = 4.5. Accumulation time: 3 min idnA at E x , JV -1.05 V -0.1 9.5 -0.3 7.0 -0.5 6.5 -0.7 6.0 associated with complex formation between the antibiotic and the metal ions."6 This kind of interference can be avoided by the addition of EDTA. References 1 Froyshov, 0.. Drugs Pharm. Sci., 1984,22, 694. 2 Ragheb, H. S., J .Assoc. Off. Anal. Chem., 1981, 64, 980. 3 Grynne, B . , Hoff, E., Silsand, T.. and Vaaje, K., Analyst. 1973, 98, 906. 4 Jacobsen, E., and Pederstad, J. H.. Anal. Chim. Acta, 1977,91, 121. 5 Garbutt, J. T., Morehouse, A. L., and Hanson, A. M., Agric. Food Chem., 1961, 9, 284. 6 Cornell, N. W., and Guiney, D. G., Jr., Biochem. Biophys. Res. Commun., 1970, 40, 530. Determination of Selenium in Copper Metal Using Flow Injection Hydride Generation Atomic Absorption Spectrometry With Continuous Flow Matrix Isolation Stephen G. Offley and Nichola J. Seare Department of Chemistry, University of Technology, Loughborough, Leicestershire L E 1 I 3TU Julian F. Tyson Department of Chemistry, University of Massachusetts, Amherst, MA 0 1003, USA Helen A. B. Kibble Development Department, Philips Scientific, Analytical Division, Cambridge CB 1 2PX Hydride generation atomic absorption spectrometry (HGAAS), although a highly sensitive technique for the determination of hydride-forming elements, has the disadvan- tage of being susceptible to interference from a large number of transition metals.' One of the most severe interferences is that of copper on the determination of selenium.l-4 Many attempts A R H ~ r 5 ,Abw Ar Fig.1 Schematic diagram of the FI manifold with continuous flow matrix isolation unit: P, peristaltic pump; S, sample; H, water; A, HCI; R, sodium tetrahydroborate(II1); W, waste; V1, switching valve; V2, sample injection valve; C , micro-column; and G, gas-liquid separator have been made to reduce this interference but with varying degrees of success.3-7 Recent developments involve the introduction of both continuous flow8 and flow injection (FI)9.10 methodology.Flow injection hydride generation was first reported by Astromg for the determination of bismuth. A significant reduction in interference effects was reported. Although on-line matrix isolation has been applied to flame AAS (FAAS) using FI, to date only one or two attempts have been made to apply such methodology to interference removal in HGAAS.11-12 This paper reports the determination of selenium by FI-HGAAS. Elimination of the interference from copper is achieved in a continuous flow matrix isolation manifold containing a micro-column of cation-exchange resin. This is coupled to the FI manifold via the injection valve.Experimental Reagents Analytical-reagent grade water, produced by a LiquiPurc RG System (reverse osmosis followed by ion exchange), was used for all solutions and as a carrier stream. A sodium tetrahydro-ANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 19 borate(m) solution (1% m/v in a 0.1% m/v NaOH solution) was prepared using sodium tetrahydroborate(rI1) pellets (Spec- trosol, BDH) and filtered through a Whatman 541 filter- paper. With refrigeration this solution was usable for up to 3 days. The hydrochloric acid (6 mol dm-3) reagent solution was of SpectrosoL grade (BDH). All selenium(1v) standard solutions were prepared by dilution of a standard solution of selenous acid (SpectrosoL, BDH) containing loo0 pg ml-1 of Se'V. For the interference investigation work copper(11) sulphate pentahydrate (AnalaR, BDH) was used to prepare interferent standards.Conditioning of the T-shaped silica atomization cell was carried out using a 5% v/v solution of hydrofluoric acid (AnalaR, BDH). High-purity argon was used as the purge gas (99.998% Ar, BOC). The digestion of certified copper metal standard reference materials was carried out using nitric acid (Aristar, BDH) and hydrochloric acid (Aristar, BDH). The cation-exchange resin used was Dowex 50W-X8 (Drymesh 100-200, hydrogen form, 8% cross linkage, Sigma). Two certified copper metal standard reference materials, National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 454 and Bundes- anstalt fur Materialforschung und -priifung (BAM) SRM 361 , were obtained from the Bureau of Analysed Samples (Middles- brough, Cleveland).Table 1 Optimized variables for HGAAS with continuous flow matrix isolation HGAAS- Reagent Concentration Flow-rate/ml min-1 H20 carrier - 6.0 HCI 6 mol dm-3 4.2 NaBH4 1 .O% m/v 3.2 Argon - 600 (Injection volume, 409 PI) Reagent Concentration Flow-rate/ml min-1 Continuous flow matrix isolation- Sample 61000pgml-1Cu 2.0 HCI (column regenerant) 1.2 rnol dm-3 2.0 Apparatus A Philips Scientific SP9 atomic absorption spectrometer equipped with a Philips data coded selenium hollow-cathode lamp, operated at 7.5 mA, was used for all determinations. A spectral bandpass of 1.0 nm was used with the 196.0 nm selenium spectral line. The signals were recorded on a Tekman TE 200 chart recorder (2-10 mV range), all measurements being expressed as peak-height absorbance.A 50 mm Philips Scientific universal burner was used to support the air-acety- lene flame heated T-shaped silica cell (air flow setting 28, acetylene flow setting 15). The FI hydride generation manifold shown in Fig. 1 was developed around the gas-liquid separator and hydride atomization systems of a Philips Scientific PU9360 continuous flow vapour system. Two peristaltic pumps were employed: a Gilson Minipuls 3 peristaltic pump used for the hydride generation manifold and a Gilson Minipuls 2 peristaltic pump for the matrix isolation unit. Control of flow-rates was achieved by application of different bore standard manifold tubing (Altec). All manifold tubing consisted of 0.8 mm i.d. PTFE tubing (Anachem).Manifold channels were connected by three-way connector T-pieces (Anachem) which aided reagent mixing. A microbore glass column (50 X 3.0 mm i.d., Anachem) fitted with porous 25 pm PTFE frits was incorpor- ated in the external sample loop of a rotary sample injection valve (Anachem). Sample injection was achieved using a Rheodyne Model 5020 fixed volume loop injector valve operated by an electrically activated universal valve switching unit (Anachem). Sample loops of various volumes were prepared by using PTFE tubing (0.5-1.5 mm i.d.) cut to appropriate lengths. Preliminary Experiments The manifold design allows the two manifolds to be optimized independently. The parameters optimized are shown in Table 1. The efficiency of the micro-column to retain copper was assessed by pumping a lo00 pg ml-1 copper standard solution through the column and continuously monitoring the copper concentration of the column eluent by FAAS.Breakthrough of the column was adjudged to have occurred when the copper concentration of the column eluent exceeded 1.0 yg ml-1. Sample Preparation Weigh accurately approximately 0.5 g of copper metal sample into a 100 ml Pyrex beaker. Add 10 ml of 8 rnol dm-3 nitric acid and cover with a watch-glass. Heat to near dryness on a hotplate and after cooling add 10 ml of 6 rnol dm-3 hydrochloric acid. Cover again with a watch-glass and heat on a steam-bath for approximately 15 min to dissolve the sample residue then allow to cool. Transfer the cool digest solution into a calibrated flask (100 ml) and dilute to volume with analytical-reagent grade water.Prior to analysis dilute the digest solutions further to produce working samples containing <lo00 yg ml-1 of Cu. The NIST SRM 454 and BAM SRM 361 working sample solutions contained 75 and lo00 pg ml-1 of Cu, respectively. Column Packing and Conditioning Pack the micro-column with an aqueous slurry of Dowex 5OW-X8-200 resin under suction (250 mg of dry resin). Prior to use, pump a solution of hydrochloric acid (1.2 rnol dm-3) through the column for approximately 5 min (2 ml min-1) to condition the cation-exchange resin and achieve optimum efficiency for copper retention. l 2 1 I I I 0 1 .o 2.0 3.0 Flow-rate/m I m in - Fig. 2 the micro-column Retention of copper as a function of sample flow-rate through Conditioning of Silica Atomization T-cell To achieve optimum system sensitivity and precision new T-cells should be pre-treated as follows.Soak the T-cell in a solution of 5.0% v/v hydrofluoric acid for a period of 4 h in order to etch the surface of the silica cell and therefore aid hydride atomization.*3 At the start of each analysis inject a standard solution of 1000 ng ml-1 of Se*V two or three times prior to the analysis of standards and samples, respectively. Operation of FI Manifolds Operate the hydride generation and matrix isolation manifolds (Fig. 1) according to the optimized variables shown in Table 1. To obtain an SelV calibration, pump standard solutions requiring no matrix isolation through the injection valve, by-passing the micro-column. Activate the injection valve intermittently (load time 20 s, injection 10 s).Following the introduction of standards switch the micro- column into the sample line and pump analytical-reagent grade20 ANALYTICAL PROCEEDINGS, JANUARY 1991, VOL 28 water continuously through it to remove the hydrochloric acid. After washing, pump sample through the column in the same way. Pump continuously for 60 s to fill the void volume of the column and pump tubing and sample the column eluent as described earlier. After carrying out triplicate injections switch the micro-column back into the HCI regenerant line to regenerate the column and, by pumping analytical-reagent grade water, wash out the sample line. Following a short period of regeneration (30 s) switch the column back in-line and repeat the procedure for subsequent samples.In order to prevent any introduction of air on to the column, during transfer of the sample uptake tube from sample to analytical- reagent grade water the sample pump should be stopped. 0 1 .o 2.0 3.0 4.0 5.0 Fig. 3 Retention of copper on the micro-column as a function of sample solution pH Sample solution pH Results and Discussion Investigation into parameters affecting the performance of the hydride generation manifold resulted in the optimum variables shown in Table 1. The relationship between copper retention and sample flow-rate (lo00 pg ml-1 of Cu, pH 4.0) is shown in Fig. 2. The efficiency of the resin improved with reduced sample flow-rate because of the increased contact time between copper ions and the active sites of the resin. With the matrix isolation unit being independent of the hydride generation manifold, the sample flow-rate through the column could be kept low to keep column efficiency high without having to compromise the sensitivity of the hydride generation manifold as reported for the system of Riby et al.12 By pumping sample at low flow-rates, problems with resin compaction in the column and build up of high back-pressures were eliminated. Column compaction was further prevented by introducing the HC1 regenerant solution in the opposite direction to the sample flow. A sample flow-rate through the column of 2.0 ml min-1 was chosen after consideration of both column efficiency and sampling rate. At a flow-rate of 2.0 ml min-1 the 409 PI sample loop of the injection valve could be filled in 20 s.As shown in Fig. 3 the efficiency of the resin at 2.0 ml min-1 reached a maximum at about pH 2.0. Even at a pH of 1.0 (typical of a working sample solution containing 1000 pg ml-1 of Cu with no pH adjustment), the column capacity exceeded 5.0 mg of Cu permitting triplicate analyses of each sample before column regeneration was needed. Table 2 Analysis of copper metal standard reference materials SetV certified/ SeIV found*/ Yg g-' Yg g- NIST SRM 454 479 k 8 476.0 k 7.2 BAM SRM 361 36 k 0.6 37.1 k 0.7 * t95% Confidence interval. As shown in Table 2, the results for the analyses of the two standard reference materials, NIST SRM 454 and BAM SRM 361, agree with the certified values. Performance characteris- tics of the system for the determination of copper metal are summarized in Table 3. These figures give evidence for the benefits of the proposed manifold design in comparison with previous systems. 11712 The performance characteristics of the resin were retained after continued use for 3 months without any need for column repacking. Table 3 Performance characteristics of the system for copper metal determination Characterization concentration 1 .O ng ml- 1 SeIV Limit of detection* 2.1 ng ml-1 SeIV Relative standard deviation 1.5% (10ngml-l SeIV, n = 12) Sample throughput 17 h- * (triplicate analyses) * Reference 14. As the manifold components are similar to those of a totally automated system reported previously,ls automation of this system is possible. The same system could be used, with minor modifications, for the determination of other hydride-forming elements in the presence of a variety of interfering species such as Ni, Fe, Co and Ag. Financial support for S. G. 0. by the SERC and both provision of equipment and financial support from Philips Scientific (Cambridge) is gratefully acknowledged. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 References Smith, A. E., Analyst. 1975, 100, 300. Meyer, A., Hofer. C., Tolg, G., Raptis, S . . and Knapp. G., Fresenius Z. Anal. Chem., 1979, 296. 337. Welz. B., and Melcher, M., Analyst, 1984, 109, 569. Welz, B., and Schubert-Jacobs, M.. J . Anal. At. Spectrom., 1986, 1, 23. Peacock. C. J., and Singh. S. C., Analyst, 1981, 106. 931. Bedard, M., and Kerbyson, J. D., Can. J . Spectrosc.. 1976,21, 64. Hershey, J . W., and Keliher, P. N., Spectrochim. Acta, Part B, 1989. 44. 145. Narasaki, H . . and Ikeda, M.. Anal. Chem., 1984, 56, 2059. Astrom. 0.. Anal. Chem., 1982, 54, 190. Fang, Z . , Xu, S., Wang. X., and Zhang, S.. Anal. Chim. Acta, 1986,179,325. Ikeda. M.. Anal. Chim. Acra, 1985, 170. 217. Riby, P. G.. Haswell. S. J., and Grzeskowiak, R., J. Anal. At. Spectrom., 1989, 4, 181. Welz, B., and Melcher, M.. Analyst. 1983, 108. 213. Miller, J. C., and Miller. J. N., Statistics for Analytical Chemistry, Ellis Horwood, Chichester, 1984, p. 96. Bysouth, S. R., Tyson, J . F . , and Stockwell. P. B., J . Autom. Chem., 1989, 11. 36.