ISSN:0003-2654    

DOI:10.1039/AN98712BP025

出版商:RSC    

年代:1987

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98712FX029

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALAO 1 12(8) 1093-1 196 (1 987)The AnalystAugust 19871093109711071113111711211127113111351139114311491155115911651169117111731177117911831185118911911193The Analytical Journal of The Royal Society of ChemistryCONTENTSGas Chromatographic Determination of Carbonates and Sulphides in the Corrosion Products of Metals-GianricoDetermination of Methylmercury in Fish by Gas Chromatography Direct Current Plasma Atomic EmissionAcid Dissolution of Soils and Rocks for the Determination of Boron by Inductively Coupled Plasma Atomic EmissionRapid Determination of Chromium in Bovine Liver Using an Atomic Absorption Spectrometer with a Modified CarbonUse of Fourier Transform Infrared Spectroscopy for Quantitative Analysis: A Comparative Study of Different DetectionSpectrophotometric and Analogue Derivative Spectrophotometric Determination of Trace Amounts of Iron UsingSpectrophotometric Determination of Iron in Boiler and Well Waters by Flow Injection Analysis Using 2-Nitroso-5-Extraction and Spectrophotometric Determination of Tungsten with Thiocyanate and Amides-Neera Mishra, S.K.Spectrophotometric Microdetermination of Tin: Use of a Sensitiser and Kalman Filtering t o Improve the SensitivityExtraction - Spectrophotometric Determination of Sulphur DioxideP. Selvapathy, T. V. Ramakrishna, N. Balasubra-Glass Transition Temperatures of Poly(Viny1 Chloride) and Polyacrylate Materials and Calcium Ion-selective ElectrodeDetermination of Camazepam and Bromazepam in Human Serum by Adsorptive Stripping Voltammetry-LucasEnzyme Sensor for the Determination of Lactate and Lactate Dehydrogenase Activity-Dietmar Weigelt, FlorianSingle Fibre Optic Fluorescence pH Probe-Ming-Ren S.Fuh, Lloyd W. Burgess, Tomas Hirschfeld, Gary D. Christian,Automation of Simple Instrumentation for Langmuir - Blodgett Technology-R. Stephen Brown, Ulrich J. KrullCastello, Anna Maria Beccaria, Gildo PoggiSpectrometry-Kenneth W. Panaro, Dor;ald Erickson, Ira S. KrullSpectrometry-Bernard A. Zarcinas, Brian CartwrightRod Atomiser-John W. Steiner, David C. Moy, Harvey L. KramerMethods-Peter S. Belton, Alfred M. Saffa, Reginald H. WilsonSulphonated 5-(3,4-Dihydroxyphenyl)-l0,15,20-triphenylporphine-Hajime Ishii, Katsunori Kohata(N-propyl-N-sulphopropy1amino)phenol-Noriko Ohno, Tadao SakaiSinha, K.S. Patel, R. K. Mishraand Selectivity-Yi-Ming Liu, Ru-Qin Yumanian, R. PitchaiProperties-G. J. Moody, B. Saad, J. D. R. ThomasHernandez, Antonio Zapardiel, Jose Antonio Perez Lopez, Esperanza BermejoSchubert, Frieder SchellerFrancis WangREPORT OF THE ANALYTICAL METHODS COMMllTEEHydroxyproline in PorkSHORT PAPERSRapid Application of Samples to Thin-layer Chromatographic Plates-J. Stephen CridlandRoutine Spectrophotometric Determination of Citric Acid in Milk Powders-Harvey E. Indyk, Andreas KurmannStudies of the P-Type Binuclear Chelates of the Lanthanides with plodochlorophosphonazo-Jinzhang Gao, RuyaoSeparation of Uranium from Neodymium in a Mixture of Their Oxides-Dawood M. MohammedExtraction - Spectrophotometric Determination of Hydrazine with 2-Hydroxy-I-Naphthaldehyde-J. Mafies, P.Microcomputer-aided Control of a Titrator Applied t o Precipitation Titrations-Shuko Fujieda, Nobue NishiSpectrophotometric Determination of Malathion with Methylene Blue-Kalluru Seshaiah, Pratapa MowliChen, Jingou Hou, Guangbi BaiCampillos, G. Font, H. Martre, P. PrognonCOMMUNICATIONAn Optical Potassium Ion Sensor-John F. Alder, David C. Ashworth, Ramaier Narayanaswamy, Richard E. Moss, Ian 0.SutherlandBOOK REVIEWSTypeset and printed by Black Bear Press Limited, Cambridge, Enqlan

ISSN:0003-2654    

DOI:10.1039/AN98712BX031

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, AUGUST 1987, VOL. 112 1093 Gas Chromatographic Determination of Carbonates and Sulphides in the Corrosion Products of Metals Gianrico Castello* Istituto di Chimica Industriale, Universita di Genova, Corso Europa 30, 16132 Genova, Italy Anna Maria Beccaria and Gildo Poggi Istituto per la Corrosione Marina dei Metalli, Consiglio Nazionale delle Ricerche, Via della Mercanzia 4, 16123 Genova, Italy A method has been developed for the gas chromatographic (GC) determination of sulphides in the corrosion products formed on the surface of metals immersed in sea water. The samples are treated at room temperature with 30% glycine solution, which dissolves the compounds of bivalent metals, but leaves unchanged other corrosion products and the metallic matrix. The evolved gases (carbon dioxide and hydrogen sulphide) are stripped with a flow of purified inert gas and trapped at liquid nitrogen temperature in the modified loop of a gas sampling valve.They are then thermally desorbed and injected into a gas chromatograph equipped with a Porapak Q column and a thermal conductivity detector. The method permits the simultaneous determination of COz and H2S and the measurement of the amounts of carbonates, oxycarbonates and sulphides in the corrosion layers. The detection limit is about 1 pg of H2S (with a precision of k15%). The GC method is therefore about ten times more sensitive than the spectrophotometric method using methylene blue. The performance of different methods was evaluated by analysis of the corrosion products of copper and iron specimens subjected to corrosion in surface and deep (200 m) sea water.Keywords: Metal corrosion products; sulphides; carbonates; gas chromatography; sea water The analysis of the oxidation layer formed on the surface of metals immersed in different corrosive solutions yields results that, correlated with other parameters (e.g. , electrode im- pedance, polarisation resistance, equilibrium potential) obtained by electrochemical measurements, permits the corrosion mechanisms and the corrosion resistance to be investigated. Some electrochemical methods are sometimes difficult or impossible to use in the natural environment, e.g. , in sea water, mainly owing to long exposure times. In this instance, chemical analysis is the main means of establishing the corrosion mechanism and, by permitting the characterisa- tion of various layers of different thickness formed in sequence, to trace back the changes in the parameters of the corrosive environment such as pH and dissolved oxygen (DO).For example, a first sulphide layer, formed under non-oxidising conditions at pH < 7, may be followed by a layer containing basic carbonates, formed at pH > 8 and with a high DO content (>8 p.p.m.),lJ showing that the metal was immersed in a non-oxidising environment owing to decom- position of algae in autumn and winter, followed by a high DO concentration due to algae proliferation in spring and summer.3 The analysis of such heterogeneous oxidation layers can be carried out by surface analysis methods such as Auger spectroscopy or ESCA, which permit the analysis of small areas, but sometimes require the use of reference standards which are difficult to obtain.Chemical analysis permits the determination of the ions in the corrosion layer, averaged over greater areas and, correlated with X-ray diffractometry, adds useful information to data obtained by electron spectroscopy. Chemical analysis methods require the selective dissolution of the different compounds of the oxidation layer with a suitable series of solvents (methanol, glycine) that leave the metallic matrix unchanged.4-6 The solubilised ions are then determined by atomic absorption spectrometry (AAS) (cations) or ion chromatography (anions), but some problems are encountered in the determination of carbonates and * To whom correspondence should be addressed. sulphides, which are converted into gaseous C 0 2 and H2S during the treatment with glycine.These gases can be determined by stripping them from the reaction vessel with a flow of inert gas and measuring their concentrations in a separate manifold. By following classical methods, H2S is removed from the stripping gas flow using zinc acetate at pH 5, in order to avoid its interference in the titration of C02 with sodium methoxide,7 and determined separately by the methylene blue method.8.9 Unfortunately, an appreciable amount of C 0 2 is absorbed by the zinc sulphide formed in the zinc acetate trap, or is coprecipitated with the zinc sulphide as zinc hydroxycarbonate, and is therefore lost in the subsequent titration. In a previous paperlo we described the gas chromatographic (GC) determination of carbonates in the corrosion products by analysis of the C02 evolved, in the absence of interferent sulphides.In this paper the simultaneous determination of H2S and C02 is described, and the GC method is compared with the spectrophotometric method.3.9 Experimental Equipment For the analysis of evolved C02 and H2S a Model 2800 gas chromatograph (Varian, Palo Alto, CA, USA) equipped with columns filled with 80-100-mesh Porapak Q (3 m X 4 mm i.d.) and with a thermal conductivity detector (TCD) was used. The analytical column was connected directly to a six-port gas- sampling valve (Varian, nut-type, stainless steel) and to a trap cooled with liquid nitrogen, in order to freeze and concentrate the C02 and H2S evolved.The trap was made of PTFE tubing (2 m X 2.4 mm i.d.) filled with nylon wire in order to increase the scrubbing surface, the heat exchange and the trapping efficiency. All tubing and connecting pieces were made of PTFE or nylon in order to avoid losses of H2S due to chemical reaction or adsorption on metal surfaces. Fig. 1 is a schematic diagram of the extraction and trapping apparatus. The determination of selectively dissolved cations was carried out by using a Uvidec-5 spectrophotometer (Jasco,1094 ANALYST, AUGUST 1987, VOL. 112 Tokyo, Japan), an atomic absorption spectrometer (Varian AA 6 ) with air - acetylene burner, an anodic stripping differential-pulse polarograph (Multipolarograph 472/WR; Amel, Milan, Italy) and a CGR Cristalloblock 50 X-ray diffractometer (Compagnie Generale de Radiologie, Lyon, France) using Cu K a radiation.The anions were measured by using an ion chromatograph (Dionex, Sunnyvale, CA, USA) with an HPIC AS 2 column, an HPIC AG 2 pre-column, 3 mM Na2C03 - 2 mM NaOH as the eluent and 0.0125 M H2S04 as the suppressing agent. Reagents Zinc acetate solution, 20%. N,N-Dimethyl-p-phenylenediammonium chloride solution, Iron(ZZI) chloride solution, 0.05 g ml-1. Glycine solution, saturated at 25 "C. 0.25% in 6 M HCl. General Procedure The corroded specimens, with average surface areas ranging from 2 to 20 cm2, were removed from the corrosive solution, rinsed with doubly distilled water and immediately subjected to selective dissolution in a manifold as described elsewhere ,4 which permits operation in an inert gas atmosphere in order to avoid further oxidation of the specimen, The procedure of the attack is shown in Fig.2. n G F ASC MS S L Fig. 1. Schematic diagram of the extraction and analysis apparatus. G, Inert stripping gas inlet; NV, needle valve; ASC, MS, ascarite and molecular sieve traps; S, reaction vessel; M, rubber membrane for injection of reagents; V1, V2, on - off valves; L, refrigerated samplin loop; GSV, gas-sampling valve in (A) sample collection and (Bf analysis position; F, flow meter; GC, gas chromatograph with thermal conductivity detector Corrosion products v 1st attack (methanol 65 "C, 30 min). Dissolution of CuCI2, ZnC12, NaCI, Cu and I Zn sulphates Dissolution of bivalent metal compounds: Cu and Zn oxides, oxycarbonates, oxysulphates, oxychlorides, sulphides I I Evolved C02 and H2S 1 Solutions A and B were analysed in order to measure the different ions by AAS (Na+, Mg2+, Cu2+), anodic stripping differential-pulse polarography , ASVPP, (Zn2+, Cu2+) or ion chromatography (Cl- , S042-) by following previously des- cribed methods.6 The C02 and H2S originating from the dissolution of the carbonates due to the addition of glycine solution to the corroded specimen in the manifold S, and stripped by a nitrogen or helium flow (Fig. l), were frozen in trap L, cooled with liquid nitrogen.A flow time of 20 min with a stripping gas flow-rate of 20 cm3 min-1 was suitable for nearly complete extraction of the evolved gases. The inlet and outlet of the trap were then closed by means of valves V1 and V2 and the trap was heated at room temperature for about 10 min.The gas-sampling valve was switched from position A to position B in order to deliver the evolved gases into the GC column, held at 70 "C. The TCD was operated with a filament current of 155 mA, using a flow-rate of 20 cm3 min-1 of purified helium as the carrier gas. A typical chromatogram is shown in Fig. 3. Quantitative calibration of the detector response was carried out by using an exponential dilution flask and injecting known amounts of C02 and H2S. The calibration was made within the range 5000-1 p.p.m. with an accuracy of _+2% by measuring the peak heights. Results and Discussion The results obtained by measuring C02 and H2S with three procedures (I, 11, 111), shown in Fig.4, were compared. In methods I and 11, H2S is measured by the methylene blue spectrophotometric method8.9 after absorption in a trap filled with zinc acetate solution while the C02 is titrated with sodium methoxide7 (detection limit about 40 pg with +50% precision and &lo% accuracy) or by GC1O (detection limit 0.5 pg with +_ 5 Yo precision and accuracy ) . Table 1 shows the results of the calibration of the spectrophotometric method carried out either with sodium sulphide solution or after absorption of known amounts of H2 in zinc acetate solution. The detection limit of this method is about 10 pg of H2S (twice the "blank" value of log Zdr), the precision is 22.5% and the accuracy is 210Y0. The method is not sensitive enough to permit the determination of sulphides F- 0 1 2 3 4 5 6 Ti me/m in Fig.2. of the corrosion products Schematic diagram of the procedure used for the dissolution Fig. 3. Chromatogram of the separation of H2S and C 0 2 . Column, 80-100-mesh Porapak Q (3 m X 4 mm i.d.); column temperature, 70 "C; helium flow-rate, 20 cm3 min-1ANALYST, AUGUST 1987, VOL. 112 in the corrosion products of some metals or their alloys (nickel and cupro-nickel), which have a thickness of a few nano- metres, the corresponding amount of S2- therefore being about five times less than the detection limit of the methylene blue method, even if the surface area of the sample is taken as 20 cm2. Further, during the adsorption of H2S by means of the zinc acetate solution, appreciable amounts of C02 are lost. Table 2 shows the C02 recovery as a function of the type of treatment followed in method I1 in Fig. 4 and of the sample composition.The direct GC of gases evolved from the heated trap permits the simultaneous determination of C02 and H2S without appreciable loss and with good sensitivity (about 1 pg with an average precision of +15%). Table 3 (column A) shows the results of the calibration obtained by direct injection of H2S into the GC column. Concentration in the cooled trap was almost completed within 20 min after the injection of the glycine solution; the procedure for heating the trap was critical. On increasing the temperature and the heating time, the amount of H2S lost (probably by reaction with the metallic surfaces of the gas-sampling valve) increased, as shown in Fig.5. At room temperature the H2S recovery was almost complete (>92%) even with long heating times, whereas at 50°C a large loss was observed. Table 3 also shows the effect of the dead volume of the gas-sampling valve, which was negligible (column B, showing the results of H2S injections through the trap), and of the dead volume of the on - off valves V1 and V2 (column C, showing the results of the complete procedure). The reproducible systematic error of about - 8% can be reduced by using microvalves of low dead volume Evolved C 0 2 and H2S in inert gas stream I 1 Q H2S absorbed r - l with Zn acetate in pyridine I coz absorbed Q Q with Zn acetate C 0 2 and H2S Cold trapping of I co2 GC analysis of COz and H2S Fig. 4. determination of H2S and C02 Schematic diagram of the different methods used for the Table 1.Spectrophotometric determination of H2S (methylene blue method). h = 670 nm; cell path length, 2 cm; final volume of blue methylene solution, 25 ml. Detection limit, 10 pg; correlation coefficient of equation log ZJZ = 0.0157Q + 0.446, 0.992; n = 5 Q4.g H2S Log IJZ 0.0 (blank) 0.052 f 0.0076 5.3 0.122 k 0.010 13.3 0.263 _+ 0.0058 26.6 0.456 f 0.0051 39.8 0.676 f 0.0060 1095 (Model H V , Hamilton, Reno, NE, USA) and shorter connecting lines. The precision of the determination of H2S by the complete freezing - heating method was f 10%. Analysis of Corrosion Products By following the described method, the corrosion products formed on Armco iron and copper specimens, immersed offshore for 6 months at different depths (1 and 200 m) with different amounts of DO (8 p.p.m.at the surface, 2 p.p.m. at 200 m) and of S2- (0.0 p.p.m. at the surface, about 0.2 p.p.m. at 200 m in winter), were analysed. The results (Table 4) showed good precision. It was impossible, however, to calculate the accuracy, firstly because no comparison stan- dards were available and secondly because the solubilities of the various compounds depend not only on their structure but also on their physical properties (grain size of the compound, thickness of the oxide film, etc.). Reasonable agreement between the results of X-ray and chemical analyses was observed (see Table 5) for the main components of the corrosion products (oxides, carbonates, oxychlorides, etc.). The minor constituents (sulphides) were only detected and measured by selective dissolution and GC, which gives no information on the structure of these com- pounds, but contributes to the knowledge of the composition of the corrosion layer obtained by using other analytical methods.100 0 r. - - A ” ” ” 80 8 U- z 60 a 0 2 40 Y 20 0 10 20 30 40 Heating time/rn in Fig. 5. Effect of desorption temperature on the recovery efficiency of trapped H2S. (A) 25°C; (B) 50°C. Stripping gas flow-rate, 20 cm3 min-1 Table 3. GC determination of H2S. n = 3 Amount of H2S Peak height* Pg A B C 5 7.6 16 k 3 10 15.2 41 _+3 3 8 k 4 25 37.9 104 f 5 104 k 5 50 75.9 220f10 2 1 9 f 8 203 k 9 75 113.8 316 f 12 100 151.8 423 k 23 427 f 10 389 k 37 * A, Direct injection into the column; B, injection through the trap without freezing; C, complete procedure with freezing (see text).Table 2. Effect of absorption of H2S in zinc acetate solution on the efficiency of recovery of C 0 2 . rz = 5 Sample Treatment COz recovery,% C02 (100 pl) . . . . . . . . . . None 100 CO2(100pl) . . . . . . . . Absorptioninzincacetate 92 k 1.3 C02 (100 p1) + H2S (100 pl) . . . . Absorption in zinc acetate 79 f 2.9 56 k 4.8 C02 (100 pl) + H2S (200 pl) . . . . Absorption in zinc acetate1096 ANALYST, AUGUST 1987, VOL. 112 Table 4. Chemical analysis of divalent metal compounds in the corrosion layer formed on Cu and Fe exposed in sea water for 6 months. n = 3 Ion/pg cm-2 Metal Depth/m Cu2+ Ca2+ Mg2+ Na+ Fez+ S042- c1- 52- C032- Copper . . 1 20055 10k0.5 3 k 0 . 2 2 0 f 2 - 3 f 0.5 35 k 2 - 25 rf: 2 200 280+8 14k0.3 5 k 0 .2 2 5 f 2 - 14+ 1 60 + 3 7 + 1 32f.2 Iron . . . . 1 - 8 f 0 . 7 4 + 1 4 1 k 2 5 7 5 f 5 - 139.5 + 11 - 32 + 2 200 - 7k0.4 5 + 1 7 3 f 5 1 8 3 f 4 - 153 f 5 - 40 f. 3 Table 5. Comparison of X-ray and chemical analyses Compounds identified by X-ray Metal DeptWm diffractometry compounds Stoicheiometrically calculated Copper . . . . 1 CuO, Cu20, CaC03, Cu(0H) (CO,), NaCl CuO, Cu20, CaCO,, MgS04, CU,(OH),(C~~)~ CuO, Cu20, CaCO,, Cu(OH)Cl, Cu(OH)C03 CuO, Cu20, CaCO,, MgS04, CU,(OH),(CO~)~, Iron . . . . 1 Fe(OH)*, NaCl, CaC0, Fe(OH),, NaCl, CaCO,, MgCO,, Fe,(OH),Cl, 200 CuS, Cu,(OH),Cl, 200 Fe(OH)*, NaCl, CaC0, FeS, Fe(OH)2, NaC1, CaCO,, Fe,(OH),Cl,, MgC03 Conclusions The GC method described permits the simultaneous determi- nation of H2S and C02 without previous absorption of the H2S in zinc acetate solution, a procedure that results in a loss of COZ.The minimum measurable amounts are 2 pg of H2S and 0.5 pg of C02 when the gases are evolved from the corrosion products of a 20 cm2 corroded surface. This permits the determination of sulphides, carbonates and oxycarbonates in thin layers ( 4 0 nm) of corrosion products, which cannot be measured with other methods, and adds useful information to the analytical results obtained by electron spectroscopic methods. References 1. Mor, E. D., and Beccaria, A. M., “4th Congress on Marine Corrosion and Fouling, Juan les Pins (Antibes), France, 1976,” Centre de Recherches et d’Etudes OcCanographiques, Bou- logne, 1977, p. 373. 2. 3. 4. 5. 6. 7. 8. 9. 10. Beccaria, A. M., Mor, E. D., Bruno, G., and Poggi, G., Werkst. Korros., 1982, 33, 416. Horne, R. A., “Marine Chemistry,” Wiley-Interscience, New York, 1969, p. 190. Mor, E. D., and Beccaria, A. M., “Proceedings of the International Congress on Corrosion and Fouling, Washington, DC, 1972,” Northwest University Press, Evanston, IL, 1973, Beccaria, A. M., and Poggi, G., Anal. Lett., 1985, 18, 2259. Beccaria, A. M., and Poggi, G., Anal. Lett., 1986, 19, 1205. Patchorniki, A., and Shalitin, Y., Anal. Chem., 1961,33,1887. Johnson, C. M., and Nishita, H., Anal. Chem., 1952,24,736. Sands, A. E., Grafins, M. A., Wainwright, H. W., and Wilson, M. W., US Bur. Mines Rep. Invest., No. 4547, 1949. Beccaria, A. M., Castello, G., and Poggi, G., J. Chromatogr., 1987,395, 641. p. 477. Paper A 7/40 Received February 9th, 1987 Accepted March l l t h , 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201093

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, AUGUST 1987, VOL. 112 1097 Determination of Methylmercury in Fish by Gas Chromatography - Direct Current Plasma Atomic Emission Spectrometry* Kenneth W. Panaro and Donald Erickson US Food and Drug Administration, Boston District Office, 585 Commercial Street, Boston, MA 02109, USA and Ira S. Krullt Department of Chemistry, The Barnett Institute, Northeastern University, 360 Huntington Avenue, Boston, MA 021 15, USA Gas chromatography (GC) has been interfaced very simply and inexpensively with a direct current plasma (DCP) atomic emission spectrometer in order to perform highly specific and selective determinations of methylmercury (MeHg) in fish samples. A simple, isothermal, low-cost GC was constructed which could be dedicated to the DCP, allowing routine qualitative and quantitative determinations of organomercury species in complex food matrices.Optimisation of the GC - DCP interface was accomplished, followed by a determination of the detection limits, the linearity of the calibration graph and comparison of the results with those obtained by GC - electron-capture detection (ECD) and total mercury by cold-vapour atomic absorption spectrometry. In most instances, qualitative and quantitative results for incurred and spiked levels did not agree for the GC - DCP and GC - ECD approaches. An additional extraction procedure has also been developed for MeHg from fish samples involving extraction with an organic solvent, concentration and injection on to the GC column. Depending on the particular organic solvent employed, artifact formation of MeHg can occur as a result of the extraction - GC conditions. Methods to avoid an artifact situation are suggested, and the possible implications of this for the currently accepted AOAC method involving GC - ECD.Keywords: Methylmercury determination; gas chromatography; electron-capture detection; direct current plasma; fish analysis The most commonly used method for organomercury speci- ation is currently gas chromatography (GC) followed by electron-capture detection (ECD). Methylmercury (MeHg) is a misnomer, a non-existent species, that has come to denote a chromatographic peak derived from a methylmercury(I1) anion specieshalt originally injected. It has been used in the analytical literature for many years, and we follow custom by using this trivial name in place of the more unwieldy alternative.There is some evidence to suggest that dimethyl- mercury is formed on injection of organomercurials on to the GC column, but only when metal surfaces are present.1-3 All of the glass column GC evidence has suggested that the peak obtained is the originally injected methylmercury salt. Studies utilising GC - mass spectrometry have indicated that, at least when standard methylmercury(I1) chloride is injected, the resultant peak using the GC conditions employed under Experimental is methylmercury(I1) chloride and not dimethyl- mercury.4.5 Total mercury levels are usually determined by cold-vapour generation followed by atomic absorption spectrometry (AAS). However, the poor selectivity of ECD for organo- mercury species and the often time-consuming and elaborate clean-up procedures necessary prior to the injection of many samples, have indicated the need for a simple, more specific approach for the determination of organomercury species.2.6Y7 Specificity in gas chromatographic determinations of mercury has been improved using AAS or microwave induced plasma (MIP) atomic emission detectors.3>&10 Although GC - MIP approaches have been demonstrated for MeHg in fish samples, such methods continue to have serious drawbacks, especially an intolerance to large sample volumes injected, plasma instability, poor reproducibility, poor precision and the inability to operate unattended for long periods of time." * Contribution No.307 from the Barnett Institute at Northeastern t To whom correspondence should addressed.University, Boston, MA, USA. There have been no reports of the application of GC - DCP for the determination of MeHg in fish samples, although this approach has been used for numerous other metal species." The earliest and certainly the most work in GC - DCP for organometal speciation has been reported by Uden et aZ.12 High-performance liquid chromatography (HPLC) has also been utilised for MeHg speciation, including its interfacing with electrochemical detection,13 AAS1417 and inductively coupled plasma (ICP) atomic emission spectrometry.18 However, depending on the particular sample, preparation and work-up can be extensive, as in GC, and accurate determination may prove as (or more) difficult. In view of the extreme volatility of MeHg and related organomercury species, GC still appears to be the preferred method of determination. HPLC can serve as a useful and practical alternative or for corroboration and confirmatory purposes when the initial analysis has been caried out by GC.Element specificity is possible with GC - atomic absorption or GC - atomic emission spectrometry (GC - AAS, GC - DCP, GC - ICP, etc.). We use the term element specificity, as opposed to element selectivity, as the GC - DCP approach used here is specific for mercury-containing species and their emission characteristics. The DCP has now been simply, quickly and inexpensively interfaced with GC to form a dedicated GC - DCP system for methylmercury in fish. A low-cost, isothermal packed-column gas chromatograph has been constructed which can resolve MeHg from the solvent front and ethylmercury (EtHg) in less than 5 min.The interfacing involves a quartz jet tube to convey the GC effluent directly into the DCP plume. This interface allows for prolonged use without replacement. Experimental Apparatus and Operating Conditions Construction of the gas chromatograph for GC - DCP A schematic diagram of the GC - DCP instrumentation is1098 ANALYST, AUGUST 1987, VOL. 112 given in Fig. 1. The outside surface of a 3 ft 8 in length of copper tubing was covered with furnace cement. The covered tube was baked overnight in an oven at 200°C. After the furnace cement had cured, 22 gauge nichrome wire was wound over the cement. The wire was wound with a 4 in space between each turn, except at the injection port end of the tube, where the wire was wound tighter, & in space per turn, for 2 in of tubing.The nichrome wire was triple twisted at each end of the wound tubing for electrical lead connections. The triple-twisted nichrome wire had mechanically fastened elec- trical connectors for connection to a variac for column temperature control. Additional furnace cement was applied to cover all the wound nichrome wire and then baked once more (overnight) at 200 "C. The cured tubing was installed in a 2.5 in 0.d. X 0.5 in i.d. fibre-glass insulation tube with heat-resistant wire insulation covering the triple-twisted nichrome wiring. The entire wired assembly was routed through the 1-in fibre-glass insulation. The injection port was one leg of a 4 in stainless-steel Swagelok tee. This tee had two septa sandwiched between washers and secured by a Swagelok end-cap, which had a hole drilled in the middle of the cap's plug end for the insertion of a syringe.The second leg of this Swagelok tee was connected to the GC column. The perpendicular leg of the tee was the input for the argon carrier gas. This input ran along the outside of the furnace cement, but inside the fibre-glass insulation, which presumably brought the gas temperature up to that of the column temperature. The outlet of the GC column had a 4 in stainless-steel Swagelok 90" elbow connection with a f in 0.d. X 1 mm i.d. quartz tube for transfer of the column effluent to the DCP plume region. The top of the flat-tipped quartz tube was ground into a cone to project the tube higher into the plume.The quartz tube was 6 in long and was wrapped with heating tape to maintain the temperature at 190"C, with the column maintained at 155 "C. Direct current plasma The DCP was a Spectraspan Model IIIb DC argon plasma spectrometer (Spectrametrics/Beckman, Andover, MA), operated in the active diagnostic mode (repeat dial = 0). Other operating conditions included a sleeve pressure of 50 lb in-2; zero nebuliser pressure; nebuliser, spray chamber, sample tube and peristaltic pump removed; a gain of 30; PMT voltage setting of 8 or 9 (900 or 950 V); input slit settings (vertical x horizontal) of 200 x 200 or 300 X 200; and an emission line of 253.652 nm. Atomic absorption spectrometer A Perkin-Elmer (Norwalk, CT) Model 403 atomic absorption spectrometer, with digital read-out, was used in the cold- Cathode Plume Column inlet wire column Fig. 1.Schematic diagram of the GC - DCP instrumentation vapour mode for total mercury determinations. A Westing- house mercury hollow-cathode lamp was operated at 253.6 nm at 15 A rating. GC - ECD operating conditions A Perkin-Elmer Model Sigma 2000 gas chromatograph was used with a 63Ni electron-capture detector. A 6 ft x 2 mm i.d. all-glass column packed with 5% DEGS-PS on 100-120 mesh Supelcoport was operated isothermally at 150 "C, with an injection port temperature of 210"C, a cell temperature of 350 "C and a nitrogen flow-rate through the GC column of 30 ml min-1. An additional 30 ml min-1 nitrogen flow was added as the cell make-up gas.Data were collected on a Perkin- Elmer Model LCI-100 laboratory computing integrator. Prior to analysis, the column was conditioned with one 20 pl injection of a 1000 p.p.m. solution of mercury(I1) chloride, according to the AOAC, 25.150(b).19 This is the recommen- ded procedure to passivate the column packing and to improve the chromatographic peak shape for MeHg. GC - DCP operating conditions The laboratory-made gas chromatograph was used for all the GC - DCP determinations. A 4 ft x 2 mm i.d. all-glass column packed with 5% DEGS-PS on 100-120 mesh Supelcoport was operated isothermally at 155 "C, with an injection port temperature of 175°C and an argon flow-rate through the column of 100 ml min-1. The DCP quartz jet inlet tube was maintained at 190 "C with external heating tape.Data were collected on a Perkin-Elmer Model LCI-100 laboratory computing integrator. Prior to analysis, the column was conditioned with one 20 pl injection of a 1000 p.p.m. solution of mercury( IT) chloride, according to the AOAC, 25.150(b). 19 Methods and Procedures for Sample Preparation Total mercury determination The established sample preparation method according to the AOAC, 25.131-132, 25.134-13519 was used for cold vapour atomic absorption spectrometry. GC - ECD sample preparation The work-up procedure used to extract and determine MeHg in swordfish was that of the AOAC, 25.146-25.152,19 modi- fied by Hight and co-w0rkers.6~7 This extraction procedure used toluene instead of benzene as the extracting solvent. It also used a 1.0 g composite sample of fish and 2.5 ml of 1 + 1 hydrochloric acid.The extract of 2-20 ml volumes of toluene was not evaporated, but was instead taken to the final volume of 50 ml for GC - ECD determinations. GC - DCP sample preparation The same method of Hight and co-workers6.7 was used to extract the MeHg from the swordfish. However, the sample preparation now involved concentrating the 50 ml of solution used in the GC - ECD determination step to 5 ml using a Kuderna - Danish (K - D) apparatus. This was performed in a warm water-bath (40-50OC) under a stream of nitrogen. The concentrated solution was directly injected into the GC - DCP. The evaporation step was one of the problems associated with the final method as it required approximately 90 min to perform.A second extraction procedure was developed using 20 ml of an extraction solution of 50 + 50 V/Vof diethyl ether - light petroleum to separate the MeHg from the sample matrix. The extraction solution was evaporated to 5 ml using a K - D apparatus in a warm water-bath (40-50 "C) under a stream of nitrogen. The time needed for this evaporation was less than 10 min. This extraction method was not without problems, in that the diethyl ether portion caused a ghost peak at the sameANALYST, AUGUST 1987, VOL. 112 1099 GC retention time as authentic MeHg when a solvent blank was injected in either GC - DCP or GC - ECD. Even though the results from these GC - ECD determinations compared favourably with the established toluene extraction GC - ECD method, it is recommended that diethyl ether and acetone should not be used.Other solvents may also lead to artifact formation of MeHg under certain GC conditions (see Results and Discussion). A third extraction procedure was developed using light petroleum as the extraction solvent. The MeHg was com- pletely extracted from swordfish when GC - ECD was used as the determination step (50 ml final volume). However, when the solution was concentrated to 5 ml final volume and determined by GC - DCP, the results were substantially lower than those obtained by GC - ECD. GC - DCP requires this 50 ml to 5 ml concentration step because of the differences in detection limits between ECD and DCP. We believe the differences were due to accidental losses of the MeHg during the evaporation.However, the loss was only seen with light petroleum. A fourth extraction procedure was attempted; this employed the first portion of the AOAC method19 with the modifications of Hight and GO-workers.697 However, instead of using toluene as the extracting solvent, 50 + 50 V/V light petroleum - toluene was used. The results obtained using this method were comparable to the results obtained using the standard AOAC method.19 No advantages were noted when using GC - ECD as the determination step and, when using GC - DCP, the only advantage noted was the reduction in the time required for the evaporation step. A fifth extraction procedure was also developed for use by the GC - DCP detection system only, now called the “rapid” method. The major differences in this procedure from that of the modified AOAC method19 were: a 50 + 50 V/Vmixture of light petroleum - toluene was used as the extraction solvent; one extraction of the fat from the product with acetone and one with toluene was used; no solvent clean-up of the 1 + 1 HC1 was needed; and evaporation of the final solution to 5 ml used a K - D apparatus in a warm water-bath (40-50 “C) under a stream of nitrogen.This rapid method took considerably less time per sample than the standard work-up procedure (2.5 vs. 5 h) from the initial sample extraction to final data acquisition. Recovery studies with MeHg were performed by directly analysing the fish samples and then spiking another portion of the original fish composite at a known level of MeHg and repeating the same determination.All analyses were carried out in duplicate, and each duplicate was injected at least in triplicate with averages k standard deviations being reported. The same fish sample extracts (standard work-up) used for GC - ECD determinations were concentrated from 50 to 5 ml and directly injected on to the GC - DCP. Separate recovery studies have shown that there was no measurable loss or formation of MeHg in the concentration step. Fish samples All of the fish examined were swordfish, except for NBS- RM50, which was tunafish containing a specified amount of Hg from the US National Bureau of Standards (NBS). Reagents and Solvents Mercury standard solutions were prepared fresh daily and were stored in a dark cool place when not in use. The GC packing material, 5% DEGS-PS on 100-120 mesh Supelco- port, was obtained from Supelco (Bellefonte, PA).Organic solvents, such as toluene, acetone, light petroleum and diethyl ether, were all purchased as distilled-in-glass grade from Burdick and Jackson (Muskegon, MI). Nitrogen (99.995% pure) and argon were obtained from Associated Gas Products (Everett, MA). The mercury standard for AAS was received as Bethlehem Instrument Mercury, triple distilled (Bethle- hem Apparatus, Hellertown, PA). The methylmercury(I1) chloride and ethylmercury(I1) chloride standards for GC were purchased from K and K Laboratories (Plainview, NY). American Chemical Society (ACS) certified grade concen- trated hydrochloric acid, mercury(I1) chloride, tin( 11) chloride, sodium chloride, hydroxylamine sulphate and vanadium pentoxide were purchased from Fisher Scientific (Boston, MA).ACS certified grade sodium sulphate was obtained from Mallinckrodt (Paris, KY). Concentrated nitric and sulphuric acids were obtained as Baker Instra-Analyzed grade (J. T. Baker Chemical, Phillipsburg, NJ). The NBS- RM50 mercury in tunafish reference material was obtained from the National Bureau of Standards (US Department of Commerce, Gaithersburg, MD). Results and Discussion GC - DCP Limits of Detection, Calibration Graph and Linearity of Calibration Graph A calibration graph was constructed which was linear from 0.2 to 20 pg per 8 pl injection, with a correlation coefficient of 0.9999. It is possible that linearity would extend beyond this range, but for the purposes of these determinations in fish, this linear dynamic range was adequate.The limit of detection was determined by injecting that concentration of MeHg (0.3 ng per 8 pl) which would produce a signal to noise ratio of about 3 : 1. Statistically, for 10 replicate injections, the analyte produced an average signal of 7.6 mm with a relative standard deviation (RSD) of 5.8%. At the same time, the background noise level was 2.5 mm (RSD 12.8%). GC - DCP Determinations of Methylmercury and Ethyl- mercury in Standards Although not displayed, standards of methylmercuryl(I1) chloride and ethylmercury( 11) chloride can be base-line resolved within a total of 5-6 min under the GC - DCP conditions indicated in Fig. 2. The peak shape and symmetry were acceptable. There was no serious solvent response on the DCP and the background noise level was low at the concentrations and attenuation employed.Fig. 2 is indicative of the GC - DCP chromatograms obtained with a swordfish sample, first analysed by GC - ECD using toluene as the extraction solvent. Four injections are shown: an authentic standard of MeHg, duplicate injections of the fish extract and a second injection of MeHg standard. Once again, peak shape was good, there was no interference from the solvent front and the total chromatographic time was less than 3 min. E l Fig. 2. GC - DCP chromatogram of actual swordfish sample extract using standard (toluene) extraction method, GC - DCP conditions as under Experimental. Retention times in minutes1100 ANALYST, AUGUST 1987, VOL. 112 A rapid sample work-up method was desired to improve the sample throughput and turn-around time.This procedure was developed and optimised. Fig. 3 illustrates a GC - DCP comparison of the MeHg content of the same swordfish sample by the standard method of extraction and the proposed rapid method. There were no obvious differences in the chromatograms for either sample work-up method. There was, however, a significant reduction in the time required (5 vs. 2.5 h) for one fish sample analysis from work-up to final GC - DCP chromatogram. Determination of MeHg in Swordfish and NBS Tunafish Samples In order to validate the GC - DCP method, a comparison was made between the GC - ECD and GC - DCP methods for incurred and spiked MeHg , using single blind methodology. Tables 1 and 2 summarise these results, including percentage recoveries of the spiked MeHg.The samples were analysed according to the AOAC method 25.146-25.152,19 modified by z 8 3 i!! cv c 0 Q C 0 w .- .- $ LL 0 Q Fig. 3. using (1 2.70 2.70 GC - DCP chromatograms of actual swordfish sample extract standard and (b) “rapid” methods of MeHg extraction and work-up. Retention times in minutes - Hight and co-workers.6>7 A poor statistical correlation was found between the two sets of data (Table 3) with one type of Student, l-tailed, t-test correlation term (t-value) indicated. Table 3 suggests real differences between the GC - ECD and GC - DCP levels of MeHg found in most fish samples, with the GC - ECD results consistently higher. Thirteen separate samples were compared by both methods, and in the majority of instances (9 out of 13), a significant lack of coincidence or correlation was evident from the t-values.In every single instance, the MeHg levels by GC - ECD were higher than by GC - DCP. Although the differences were minor in some instances (553RBM, 553KWP, 687RBM), in others, there were striking differences (725KWP, 725JRN, 841JRN, 841KWP). Even for the NBS-RM5O sample of tunafish, the GC - ECD level was statistically higher than that obtained by GC - DCP and the stated value (0.93 k 0.1 p.p.m.). The GC - DCP values have always been equal to or slightly lower than the stated Hg level in NBS-RM5O. The NBS has suggested that RM50 contains Hg only as MeHg, but this is not a standard reference material. The statistical approach used through these studies was a comparison of the means of two samples at the 95% ( P = 0.05) confidence level, for n - 2 degrees of freedom.20 Using the proposed rapid method of sample work-up and extraction, the same samples of fish were re-analysed by GC - DCP (Table 4).These results were in good agreement with those in Table 2, using the standard work-up, and the percentage recoveries were within acceptable limits. A Student t-test analysis of the two sets of data showed good correlation (Table 5), suggesting that the rapid method can provide the same qualitative and quantitative accuracy and precision as the standard work-up method. Determination of Total Mercury in Swordfish and NBS-RMSO by Cold-Vapour AAS. Comparison with GC - DCP and GC - ECD for MeHg In order to demonstrate the total mercury levels present in the fish samples, separate determinations were performed using cold-vapour generation atomic absorption spectrometry.Each sample was digested in triplicate. The cold-vapour AAS results are given in Table 6, with the corresponding standard work-up GC - DCP data. The data were compared by a Student t-test, with t-values at the 95% confidence level indicated. Eleven fish samples were compared, and of these, ~ _ _ _ _ _ Table 1. Determination of methylmercury in fish by standard work-up GC - ECD. Extraction procedure: AOAC 25.146-25.152,1Y modified by Hight and co-workers.6.7 GC - ECD conditions used: 6 ft x 2 mm i.d. glass column packed with 5% DEGS-PS on 10&120 mesh Supelcoport operated at 150°C, injection at 210°C, 30 ml min-1 N, flow-rate in GC plus 30 ml min-1 N2 make-up gas for ECD cell.Only two injections made, results quoted as average k S.D. Sample type Swordfish (n = 4) . . Tunafish (n = 3) Sample 458JRN 497JRN 553RBM 553KWP 687RBM 687JRN 720KWP 720JRN 725KWP 725 JRN 841 JRN 841KWP (0.93 2 0.1) NBS-RM5O Results k s.d., p.p.m. 2.10 2 0.01 1.07 k 0.02 1.05 k 0.05 1.10 * 0.01 1.15 f 0.02 1.19 f 0.05 2.21 f 0.06 2.12 2 0.03 2.51 2 0.04 2.64 k 0.02 1.99 f 0.02 2.08 k 0.02 0.97 f 0.01 Hg added/ !% 2.0 2.0 2.0 2.0 1 .o 1 .o 2.0 2.0 2.0 2.0 2.0 2.0 Determination of methylmercury in spiked sample, p.p.m. 4.14 3.30 3.13 3.23 2.27 2.20 4.19 4.15 4.56 4.61 3.93 4.03 Recovery, % 98.7 105.8 104.4 106.5 112.0 101 .o 98.5 101.5 102.5 98.5 96.5 101.5 NBS-RMSO 0.98 k 0.02 (0.93 k 0.1)ANALYST, AUGUST 1987, VOL. 112 1101 ~~~~ Table 2.Determination of methylmercury in fish by standard work-up GC - DCP. Extraction procedure: AOAC 25.146-25.152,19 modified. GC - DCP conditions used: 4 ft x 2 mm glass column packed with 5% DEGS-PS on 100-120 mesh Supelcoport operated at 155 “C, injection at 175 “C, total Ar flow-rate 100 ml min-1, quartz jet at 190 “C; DCP wavelength of emission at 253.652 nm Sample type Swordfish (n = 3) . . . . Tunafish(n=6) . . . . Sample 458JRN 497JRN 553RBM 553KWP 687RBM 687JRN 720KWP 720JRN 725KWP 725JRN 841JRN 841KWP (0.93 k 0.1) NBS-RM5O Results f s.d., p.p.rn. 1.94 f 0.11 1.02 k 0.02 1.03 f 0.07 1.07 f 0.03 1.12 f 0.05 0.97 k 0.06 2.10 f 0.06 2.09 f. 0.06 2.26 f 0.041- 2.22 f 0.04 1.88 f 0.05 1.96 f 0.05 0.89 k 0.03 * n = 6 .t n = 7 . Hg added/ Pg 2.0 2.0 2.0 1 .o 1 .o 2.0 2.0 2.0 2.0 2.0 2.0 - Determination of methylmercury in spiked sample f s.d., p.p.m. 3.90 f 0.08 3.09 f 0.03 2.75 f 0.06* 2.13 k 0.01 2.02 k 0.06 4.09 f 0.06 3.83 f 0.03 4.64 f 0.13 4.10 f 0.13 3.91 k 0.06 3.83 f 0.12 - Recovery, YO 103.0 103.2 84.0 101 .o 105.0 99.5 87.0 119.0 94.0 101.5 93.5 - - Table 3. Statistical comparison of MeHg levels in fish by standard work-up GC - ECD and GC - DCP methods Sample type Swordfish . . . . Tunafish . . . . Sample 458JRN 497JRN 553RBM 553KWP 687RBM 687JRN 720KWP 720JRN 725KWP 725JRN 841JRN 841KWP (0.93 k 0.1) NBS-RM5O GC - ECD f s.d., p.p.m. (n = 4) 2.10 f 0.01 1.07 f 0.02 1.05 k 0.05 1.10 f 0.01 1.15 f 0.02 1.19 f 0.05 2.21 k 0.06 2.12 f 0.03 2.51 f 0.04 2.64 f 0.02 1.99 f 0.02 2.08 k 0.02 0.98 f 0.01 GC - DCP f s.d., p.p.m.(n = 6) 1.94 f 0.11 1.02 f 0.02 1.03 f 0.07 1.07 f 0.03 1.12 k 0.05 0.97 k 0.06 2.10 f 0.06 2.09 k 0.06 2.26 k 0.04-l 2.22 f 0.04 1.88 f 0.05 1.96 f 0.05 0.89 k 0.03 t-Value (2.31)* 2.84 3.87 0.49 1.90 1.12 6.04 2.84 0.91 9.97 19.19 4.12 4.50 6.97 * Student’s t-value according to method described in Miller and Miller.20 For n = 8 degrees of freedom, t = 2.31 at P = 0.05, 95% t n = 7 . confidence level. Table 4. Determination of methylmercury in fish by GC - DCP using modified “rapid” extraction procedure. Extraction procedure: 50 + 50 toluene - light petroleum, no acid clean-up with organic reagents. GC - DCP conditions as in Table 2 Determination of methylmercury in Results k s.d., Hg added/ spiked sample + s.d., Sample type Sample p.p.m.(n = 6) p.p.m. (n = 3) Recovery, YO Swordfish . . . . . . 841JRN 1.87 k 0.06 2.0 3.94 f 0.02 103.5 458JRN 1.93 f 0.04 2.0 4.03 5 0.03 105.2 497JRN 1.04 k 0.02 2.0 3.03 f 0.04 99.3 (0.93 f. 0.1) Tunafish . . . . . . NBS-RMSO 0.91 f 0.02 Table 5. Comparison of MeHg levels in fish by standard and “rapid” work-up method with GC - DCP. GC - DCP conditions as in Table 2 Sample type Swordfish . . Tunafish . . . . . Standard method, results f s.d., Sample p.p.m. (n = 6) . 458JRN 1.94 f 0.11 497JRN 1.02 k 0.02 841JRN 1.88 f 0.05 841KWP 1.96 k 0.05 . NBS-RMSO 0.89 f 0.03 (0.93 f 0.1) “Rapid” method, results + s.d., t-Value p.p,m. (n = 6) (2.23) * 1.93 f 0.04 0.21 1.04 k 0.02 1.73 1.87 k 0.06 0.31 1.87 k 0.06 2.82 0.91 f 0.02 1.35 * Student’s t-value according to method described by Miller and Miller.20 For n = 10 degrees of freedom, t = 2.23 at P = 0.05, 95% confidence level.1102 ANALYST, AUGUST 1987, VOL.112 Table 6. Determination of total mercury in swordfish and NBS-RM5O by cold-vapour AAS and comparison with MeHg determined by standard work-up GC - DCP procedure. A wavelength of 253.6 nm was used for cold-vapour AAS (hollow-cathode Hg lamp). Standard work-up and GC - DCP conditions as in Table 2 Sample type Sample Swordfish . . . . . . 458JRN 497JRN 687RBM 687JRN 720KWP 720JRN 725JRN 725KWP 841JRN 841KWP Tunafish . . . . . . NBS-RM5O (0.93 k 0.1) * Student’s t-value according to method as described by t n = 7 . f For n = 8 degrees of freedom, t = 2.31 at P = 0.05, 95% 95% confidence level.I Total Hg MeHg determined determined by AAS, by GC - DCP results k s.d., results _+ s.d., p.p.m. p.p.m. t-Value ( n = 3) (n = 6) (2.36) * 1.95 k 0.01 1.04 f 0.01 1.04 k 0.01 1.04 k 0.01 2.08 f 0.02 2.08 f 0.02 2.41 f 0.01 2.41 k 0.01 1.91 k 0.01 1.91 k 0.01 0.94 k 0.01 1.94 k 0.11 1.02 f 0.02 1.12 k 0.05 0.97 k 0.06 2.10 f 0.06 2.09 k 0.06 2.22 k 0.04 2.26 k 0.04-t 1.88 k 0.05 1.96 k 0.05 0.89 f 0.03 0.15 1.60 2.66 1.94 0.55 0.27 7.85 6.21$ 1 .oo 1.66 2.73 Miller and Miller.20 For n = 7 degrees of freedom, t = 2.36 at P = 0.05, confidence level. Table 7. Determination of total mercury in swordfish and NBS-RM5O by cold vapour AAS and comparison with MeHg by “rapid” work-up GC - DCP. AAS and GC - DCP conditions as in Table 2 Total Hg determined by AAS results k s.d., p.p.m.Sample type Sample (n = 3) Swordfish . . . . . . 458JRN 1.95 f 0.01 497JRN 1.04 k 0.01 841JRN 1.91 f 0.01 Tunafish . . . . . . NBS-RM5O 0.94 f 0.01 (0.93 f 0.1) MeHg determined results k s.d., by GC - DCP p.p.m. t-Value (n = 6) (2.36)* 1.93 +_ 0.04 0.83 1.04 k 0.02 0.00 1.87 f 0.06 1.11 0.91 f 0.02 2.39 * Student t-value according to method as described by Miller and Miller.20 For n = 7 degrees of freedom, t = 2.36 at P = 0.05, 95% confidence level. Table 8. Determination of total mercury in swordfish and NBS-RMSO by cold-vapour AAS and comparison with MeHg by standard work-up GC - ECD. Cold-vapour AAS and GC - DCP conditions as in Table 2 Total Hg MeHg determined by determined results f s.d., by AAS GC - ECD results f s.d., p.p.m.p.p.m. t-value Sample type Sample (n = 3) (n = 4) (2.57)* Swordfish . . . Tunafish . . . . . . . . . 458JRN 497JRN 687RBM 687JRN 720KWP 720JRN 725JRN 725KWP 841JRN 841KWP (0.93 f 0.1) NBS-RM5O 1.95 f 0.01 1.04 k 0.01 1.04 f 0.01 1.04 2 0.01 2.08 f 0.02 2.08 f 0.02 2.41 rt 0.01 2.41 f 0.01 1.91 f 0.01 1.91 f 0.01 0.94 f 0.01 2.10 +_ 0.01 1.07 k 0.02 1.15 f 0.02 1.19 k 0.05 2.21 f 0.06 2.12 k 0.03 2.64 f 0.02 2.51 f 0.04 1.99 2 0.02 2.08 k 0.02 0.97 f 0.01-t 19.6 2.35 8.61 5.00 3.53 1.98 18.00 4.14 6.26 5.66$ 13.3 * Student’s t-value according to method as described by Miller and Miller.20 For n = 7 degrees of freedom, t = 2.57 at P = 0.05, 95% t n = 3 . $ For n = 8 degrees of freedom, t = 2.36 at P = 0.05, 95% confidence level. confidence level.ANALYST, AUGUST 1987, VOL.112 1103 Fig. 4. GC - ECD chromatograms of four different organic solvents. Illustration of artifact formation of MeHg peaks. Retention times in minutes four samples did not show an overlap (statistical coincidence) of Hg levels (687RBM, 725JRN, 725KWP and NBS-RMSO). The other seven samples have Hg levels that are of the same population. There was no reason why there should have been any coincidence or correlation of the data, unless all the Hg present was MeHg. These results suggested that in most of the fish examined, all or most of the Hg present was MeHg. More importantly, in 10 or 11 instances, the total Hg levels determined by AAS are equal to or higher than MeHg levels by GC - DCP. In only 1 out of 11 instances (687RBM) is the GC - DCP MeHg level statistically higher than total Hg determined by AAS.It is plausible that total Hg levels can be higher than the amount of MeHg present, as it is known that Hg may exist in several forms or species in fish. However, the reverse cannot be possible. A similar comparison has been made for total Hg levels determined by AAS and the MeHg levels found by the rapid GC - DCP approach (Table 7). Although the number of samples here was lower, 3 out of 4 have t values that indicate equality of the total Hg and the MeHg levels by GC - DCP. The AAS results were always equal to or higher than the GC - DCP values. A third comparison was made for total Hg levels and the MeHg levels determined by standard work-up GC - ECD. Table 8 summarises the relevant data and Student t-values, using the same statistical treatment as above.Eleven fish samples were compared, but there was no degree of correla- tion. There were only 2 out of 11 samples that showed any agreement (497JRN and 720JRN), but here and in all the other instances, the GC - ECD MeHg levels are higher than total Hg determined by AAS. This is a striking result, but was expected in view of the general disagreement between MeHg levels determined by GC - ECD and GC - DCP (Table 3). It was clear from Table 3 that one set of data was incorrect, but only when each set has been compared with total Hg levels by the accepted method of determination (cold-vapour AAS), does this become evident. Table 8 suggests that the GC - ECD method can, at times, lead to MeHg levels that are higher than one would expect when compared with the total Hg levels.Demonstration of Possible Artifacts due to Extraction Solvent and Injection Port Temperature in the GC - ECD and GC - DCP Determinations What was perhaps most disturbing were the statistically meaningful differences between total Hg determined by cold-vapour AAS and MeHg determined by GC - ECD (Table 8). That the MeHg levels were consistently higher than the total Hg levels, in virtually every sample, suggested a positive artifact formation of MeHg via the standard work-up GC - ECD method. This seemed an inescapable conclusion. The fact that the MeHg levels by GC - DCP were equal to or less m D 3.57 - Fig. 5. GC - DCP chromatograms of four different organic solvents. Illustration of reduced artifact formation of MeHg peaks.Retention times in minutes than total Hg by cold-vapour AAS suggested that a consistent positive artifact formation was absent. A possible explanation for the positive artifact formation could be pyrolysis of certain organic solvents within the injection port in GC - ECD, leading to the formation of methyl (alkyl) radicals. This could have been followed by combination with mercury atoms from the mercury(I1) chloride coating the GC column support to form an artifact MeHg peak. The injection port temperature in the GC - ECD was about 3040°C higher than that used in GC - DCP. In order to demonstrate a possible accidental (artifact) formation of a peak attributed to MeHg in GC - ECD, two separate studies were performed. In one of these (results in Fig.4), four different solvents (acetone, diethyl ether, light petroleum and toluene) were separately injected on to the column and determined by GC - ECD. These were just solvent blanks; no sample extraction or work-up was involved. Fig. 5 illustrates a set of similar GC - DCP chromatograms for the same injections of solvent blanks, but first concentrating each solvent from 50 ml to 5 ml. In both of these figures, a separate chromatogram is given for an injection of standard MeHg. Although the absolute amounts (ng) injected were different (higher for DCP), the volumes injected (higher for DCP) partially compensated, so that relative sensitivities for standards injected were about the same (five-fold difference, ECD more sensitive). A number of significant points were evident.The GC - DCP chromatograms were simpler than those from GC - ECD. This might be expected in view of the DCP's greater element selectivity. Secondly, in two out of the four solvents studied, diethyl ether and acetone, GC - ECD showed the presence of a peak that could be attributed to MeHg. In GC - DCP, a small, broad peak appeared at the correct retention time for MeHg with diethyl ether, but of lower peak height and area than in GC - ECD. Diethyl ether has not been used in any of the sample work-ups for the GC - DCP results reported. It is possible that these same artifact peaks could be formed in the other solvent GC - DCP chromatograms, but the injection port temperature used in GC - DCP was lower by about 40 "C than in GC - ECD. Another study was carried out in which GC - ECD deter- mination of the solvent blanks was carried out at three different injection port temperatures.With diethyl ether, as the temperature increased, more artifact MeHg formation occurred. At lower temperatures, the GC - ECD artifact level approached that of GC - DCP for the same solvent. With light petroleum and toluene, there was little or no artifact peak at any injection port temperature studied (170-230 "C). For acetone, the solvent front became broader with increasing temperature and artificial appearance was always evident, but was obscured at higher temperatures by the large solvent front.ANALYST, AUGUST 1987, VOL. 112 2.49 2.76 Fig. 6. (a) GC - ECD (standard work-up) and (b) GC - DCP (standard = “rapid” work-up) chromatograms of NBS-RMSO tuna- fish.Illustration of improved MeHg specificity by GC - DCP method. Retention times in minutes It seemed that a higher injection port temperature in GC - ECD could lead to a greater pyrolysis of certain solvents (diethyl ether - acetone) than others (toluene - light pet- roleum), which then caused artifact formation of MeHg. It is well known that simple alkyl ethers and acetone readily pyrolyse on metal surfaces to release methyl radicals. The presence of a mercury salt on the GC packing, as recommen- ded by the AOAC procedure,l9 may have contributed to this formation. In the future, it might be prudent to change the inorganic salt used to coat GC columns tor organomercury determinations. Alternatively, it might be recommended to use a capillary GC column, as recently reported for the determination of MeHg by GC - ECD, even though no actual sample results were reported.21 In the absence of confirma- tory GC - MS data, it is conjecture that the artifact peaks in GC - ECD were MeHg.The one artifact peak for diethyl ether in GC - DCP, at the correct retention time for MeHg and shown to contain Hg, is probably MeHg. Solvent fronts in GC - ECD are considerably broader and more pronounced than those in GC - DCP (Figs. 4 and 6) thereby increasing the detection limits for MeHg possible using such solvents, e.g., acetone. In Fig. 4, a series of solvent peaks can be seen using GC - ECD, leading to a more complex chromatogram than with GC - DCP. It was not absolutely clear that any of the toluene peaks were due to MeHg.However, the simplicity of the GC - DCP chromatogram for an actual fish extract, in comparison with GC - ECD, is striking (Fig. 6) , leading to improved analyte identification. Explanation of Standard Work-up GC - ECD Results for the Determination of Methylmercury in Fish How does this all reflect itself in the original comparison of MeHg levels by GC - ECD and GC - DCP (Table 3) where we alluded to differences? In GC - ECD, toluene and acetone alone were both used for sample work-up and extraction, but toluene (Fig. 4) did not cause a significant MeHg artifact peak. However, any residual, unseparated acetone in the final solution injected could have led to artifact MeHg. This was also a function of the age of the GC column, the amount of mercury(I1) chloride coating remaining and injection port temperature.It was possible to generate a much larger artifact peak under certain conditions. It was also possible that sample and/or matrix components, injected along with incurred MeHg, may have led to artifact formation. It is our belief that the GC - ECD results can or may be artifically higher, and that this may be due to the sample components, the solvents used and/or the operating condi- tions. Such a positive artifact formation can be avoided in the future. GC - ECD has better limits of detection for MeHg in fish but at the expense of selectivity. Conclusions These results have demonstrated several unique capabilities and advantages that GC - DCP possesses for metal-containing compounds in fish and related samples.Accurate, precise and reproducible determinations for MeHg in fish have been demonstrated. A comparison of total Hg levels in the same fish samples has shown that, in general, they are higher than those determined by GC - DCP, but often lower than MeHg by GC - ECD. Additional sample preparation was needed for GC - DCP, especially the ability to pre-concentrate the fish extracts before injection. This was not necessary wth GC - ECD, but then there was a serious trade-off in analyte specificity. Some problems have appeared with the organic solvents used for the extraction step, in that artifact formation of MeHg occurred within the GC. It is possible that simple matrix components also contributed to an artifact formation. Solvent and sample blanks must always be run alongside actual samples in order to demonstrate the presence or absence of artifacts, standard practice in elemental analysis laboratories.Additional problems may arise in the GC - ECD method, wherein co-eluting peaks from the fish sample (interferents) may respond on the ECD as if they were MeHg. The DCP, being Hg-specific, avoids such interferences, and thus can provide more accurate determinations of MeHg. The advantages of element-selective detection for MeHg or other metal species, either by GC or HPLC interfacing, are obvious. Finally, a very inexpensive, dedicated, isothermal gas chromatograph has been constructed for dedicated DCP interfacing, which could then be used routinely and continu- ously for the determination of organomercury species.It is hoped that this GC - DCP approach will find utility and application in other trace element analysis and speciation laboratories and applications, perhaps with modification of the metal salt used for column passivation. This work was performed at the Boston District Office of the US Food and Drug Administration (FDA), Boston, MA. We are grateful to the FDA for the opportunity to perform this work and to report the results. Acknowledgement is made to colleagues within the FDA who prepared blind, spiked fish samples for method validation purposes, including J. Noonan and R. Midwood. W. S. Adams and L. Gershman provided encouragement, time and guidance during these studies. Certain students at Northeastern University, especially C. M. Selavka, read drafts of the manuscript and provided helpful and constructive comments and suggestions. References 1. 2. 3. Fishbein, L., Chromatogr. Rev., 1970, 13, 83. Rodriguez-Vazques, J. A., Talanta, 1978,25,299. Bzezinska, A., Van Loon, J., Williams, D., Oguma, K., Fuwa, K., and Haraguchi, I. H., Spectrochim. Acta, Part B , 1983,38, 1339. 4.. Johansson, B., Ryhage, R., and Westoo, G., Acta Chem. Scand., 1970, 24, 2349. 5. Capar, S. G., personal communication. 6. Hight, S. C., and Capar, S. G., J. Assoc. Off. Anal. Chem., 1983, 66, 1121.ANALYST, AUGUST 1987, VOL. 112 1105 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. Hight, S. C., and Corcoran, M., J. Assoc. 08 Anal. Chem., 17. 1987, 70, 24. 18. Bye, R., and Paus, P. E., Anal. Chim. Acta, 1979, 107, 169. 19. Sklarew, D. S., Olsen, K. B., and Evans, J. C., Paper presented at the 189th ACS National Meeting, Miami Beach, Krull, I. S. and Jordan, S. W., Am. Lab. (Fairfield, CT), Uden, P. C., Barnes, R. M., and DiSanzo, F. P., Anal. Chem., 1978, 50, 852. MacCrehan, W. A., and Durst, R. A., Anal. Chem., 1978,50, 2108. Holak, W., Analyst, 1982, 107, 1457. Holak, W., J. Liq. Chromatogr., 1985, 8, 563. Holak, W., Paper presented at Innovative Techniques for the Analysis of Iodine and Methylmercury, 10th Annual Spring Training Workshop, AOAC, Dallas, TX, April 10, 1985. Talmi, Y., Anal. Chim. Acta, 1975, 74, 107. FL, April 28-May 3, 1965, Anal. 61. October 1980,21. 21. 20. Holak, W., J. Assoc. Off, Anal. Chem., 1983, 66, 1203. Krull, I. S., Bushee, D. S., Schleicher, R. G., and Smith, S. B., Jr., Analyst, 1986, 111, 345. “Official Methods of Analysis of the Association of Official Analytical Chemists,’’ Fourteenth Edition, Association of Official Analytical Chemists, Washington, DC, 1984. Miller, J. C., and Miller, J. N., “Statistics for Analytical Chemistry,” Wiley, New York, 1984, pp. 53-55. Cappon, C. J., and Toribara, T. Y., LCIGC Mag., 1986, 4, 1010. Paper A7113 Received January 19th, 1987 Accepted March 17th, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201097

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, AUGUST 1987, VOL. 112 1107 Acid Dissolution of Soils and Rocks for the Determination of Boron by Inductively Coupled Plasma Atomic Emission Spectrometry Bernhard A. Zarcinas and Brian Cafiwright CSIRO, Division of Soils, Private Bag No. 2, Glen Osmond, South Australia 5064, Australia The boron concentration in rocks, soils and standard reference materials was determined using hydrofluoric acid - aqua regia dissolution followed by inductively coupled plasma atomic emission spectrometry (ICP-AES) using the B I 249.773-nm line, corrected for spectral interference by iron. An excess of fluoride was complexed with aluminium to release boron from the stable fluoroborate ion and to protect the borosilicate and quartz components of the instrument. Boron was not lost by volatilisation during volume reduction.Soil and rock boron values determined using the recommended dissolution procedures were comparable to those obtained using the accepted sodium carbonate fusion procedure and by d.c. arc emission spectrophotometry, and those for standard reference materials showed good agreement and precision with the literature values. Keywords: Boron determination; h ydrofluoric acid - aqua regia dissolution; inductive1 y coupled plasma atomic emission spectrometry The concentration of boron in geological samples ranges from 10 pg g-1 in igneous rocks to more than 1000 pg g-1 in metamorphic rocks. Soil boron1 concentrations average about 30 pg g-1. Sea water contains 4.5 pg ml-1 of boron. The ability of clay minerals to complex boron has been used as an indicator of paleosalinity2 and to discriminate between modern marine and fresh water argillaceous sediments and between ancient shales.3 Boron is an important trace element in plant nutrition but is unusual because of the narrowness of the range of concentration between deficiency and toxicity4 in plants. Berger and Truogs indicated that less than 5% of total soil boron is usually available for crop use.A number of procedures have been developed for the determination of boron. These have been reviewed by Gladney et aZ.6 and include a variety of spectrophotometric and spectrometric techniques and also ion-selective electrode and mass spectrometric procedures, all of which require sample dissolution. However , flame insensitivity, serious interferences and tedious manipulations render many of these methods insufficiently sensitive and too time consuming for some geological samples.Owens et aZ.7 proposed the sodium carbonate fusion procedure of Gupta8 for the dissolution of geological materials prior to analysis by inductively coupled plasma atomic emission spectrometry (ICP-AES), but re- ported that the procedure is unsuitable for routine analysis with an unacceptably high detection limit. Walshg described a method for the determination of boron in rocks by ICP-AES on the aqueous leach of a potassium carbonate fusion melt after the bulk of the potassium had been precipitated using perchloric acid. However, Hall and Pelchatlo reported salt build-up at the top of the torch and clogging of the nebuliser when using the potassium carbonate procedure.Also, plati- num crucibles were attacked by the flux and, when nickel crucibles were used, flux creep over the walls occurred. A simple, rapid and accurate procedure with sufficient sensitivity was required for the determination of boron in silicate rocks and soils. The hydrofluoric acid - aqua regia dissolution method proposed in this paper achieved dissolu- tion of rocks and soils with a good detection limit, without loss of boron by volatilisation during volume reduction, and is suitable for routine chemical analysis. Good sensitivity for boron, relative freedom from interferences and a high rate of sample throughput have made ICP-AES the method of choice for the analysis of many major and trace elements in geological materials.Experimental Equipment PTFE beakers. Capacity 100 ml, 60 X 60 mm, Kartell brand. Pressure decomposition bombs. 70-ml capacity PTFE-lined bomb manufactured by Uniseal Decomposition Vessels, Haifa, Israel. The bomb was sealed using a closure tool set. Nickel crucibles. Capacity 100 ml, 64 X 60 mm, supplied by Arthur H. Thomas, Philadelphia, PA, USA. ICP-AES. A Labtest V-25 direct-reading vacuum spec- trometer was used, equipped with 26 analytical channels in the first order, including B (249.773 nm) and Fe (259.939 nm). Digest solutions were introduced into the plasma using a modified Babington pneumatic nebuliser (Type GMK) .I1 A Gilson Minipuls 2 peristaltic pump with a Tygon red - red (1.14 mm i.d.) pump tube was used for delivery of solution to the nebuliser .A stabilisation time of 30 s was followed by three 20-s integrations, the result being calculated as the arithmetic mean of these three readings. Argon flow-rates to the plasma torch were coolant 20 1 min-1, nebulisation pressure 280 kPa and UV optics purge 7 1 min-1. The viewing height was 15 mm above the load coil with the forward power set at 1.40 kW, producing <2 W of reflected power. The inductively coupled argon plasma source, spectrometer , scanning monochroma- tor, data aquisition system, wavelength profiles of potentially interfering analytes and on-line interference correction pro- cedure have been described elsewhere.12 Reagents All water was doubly distilled. Reagent bottles, volumetric ware and beakers were soaked overnight in 2 M hydrochloric acid, rinsed with water and dried at 60 "C.Aqua regia. Freshly prepared by mixing 180 ml of nitric acid (Ajax, Univar grade, 70% mlm) and 820 ml of hydrochloric acid (BDH Chemicals, AnalaR grade, 31-32% mlm). Hydrofluoric acid, 50% mlm. Ajax, Univar grade. Sodium carbonate. Merck, GR grade. Aluminium chloride solution, 20% mlV. Dissolve 200 g of aluminium chloride (Merck, extra pure) in 1 1 of water. Boron stock standard solution, 1000 yg g-1. Dissolve 5.719 g of boric acid (Merck, GR grade) in 1 1 of water. Dilute aliquots of this solution to produce the calibration standards.1108 ANALYST, AUGUST 1987, VOL. 112 Iron stock standard solution, 1000 pg g-1. Dissolve 7.234 g of iron(II1) nitrate (Merck, GR grade) in 1 1 of 1% mlV nitric acid. Dilute aliquots of this solution to produce the calibration standards.Sample Dissolution Owens et al.7 proposed using the sodium carbonate fusion procedure for the dissolution of geological materials with determination of boron by ICP-AES. However, this very salty digest solution caused severe calibration drift in the instru- ment, making the procedure impractical for routine analysis. The boron present in the sodium carbonate of 5-15 pg g-l (also reported by Troll and Saurerl3) resulted in an unaccept- ably high solution detection limit. Hydrofluoric acid, together with various inorganic acids, has been used for the dissolution of geological materialsl4-18 prior to analysis for many elements except boron, as residual hydrofluoric acid is complexed with boric acid on completion of the dissolution.Spiers et al. 19 evaporated hydrofluoric and hydrochloric acids to dryness for the removal of excess hydrofluoric acid. They reported unreliably high boron analyses but claimed that these were not due to hydrofluoric acid attack on the borosilicate glass nebuliser, spray chamber or quartz torch of their ICP-AES system. Sample presentation to the nebuliser must be as a liquid. A simple, rapid method for the determination of boron in geological materials required a dissolution technique that would produce a solution low in dissolved salts, to minimise instrumental drift, while realising that some boron-containing minerals, such as tourmaline, are not completely decomposed by hydrofluoric acid.20 On completion of dissolution, residual fluoride was complexed with aluminium (present as alumin- ium chloride) to release boron from the stable fluoroborate ion21 and to protect the borosilicate and quartz components of the instrument.Procedures Open beaker Weigh 1.000 g of rock or soil (ground to <200 pm) into a PTFE beaker. Add 5 ml of aqua regia and 5 ml of hydrofluoric acid. Place the beaker on a hot-plate at 140 "C (measured with a contact thermometer) and evaporate to a minimum volume (1-2 ml) without allowing the digest solution to dry, as loss of boron will result. Repeat with a further 5 ml of each acid and volume reduction. Add 10 ml of aluminium chloride solution and warm the solution for approximately 5 min. Allow to cool and dilute with water to 50 ml in a polyethylene calibrated flask.Sodium Carbonate fusion Weigh 1.000 g of rock or soil (ground to <200 pm) into a nickel crucible. Mix with 6.0 g of sodium carbonate and heat in a muffle furnace at 1000 "C for 1 h. Remove the mixture from the furnace, cool and cautiously add 5-ml aliquots of 6 M hydrochloric acid (with a covering slip in place to prevent loss) until effervescence ceases. Warm and break up the fusion cake to assist dissolution. Filter into a 100-ml polyethylene cali- brated flask using a Whatman No. 2 filter-paper. Wash the silicic acid gel thoroughly with water and then dilute to volume with water. Pressure decomposition bomb Weigh 1.000 g of rock or soil (ground to <200 pm) into the bomb and add 5 ml of aqua regia and 5 ml of hydrofluoric acid. Close the bomb using the closure tools provided and heat in an oven at 170 "C for 3 h.Cool, wash the digest solution into a PTFE beaker with water and continue as described for the open beaker procedure. Two reagent blanks were carried through with each of the decomposition procedures. Calibration Boron and iron calibration solutions were prepared separately to match the matrix of the digests, viz., in 4% mlV aluminium chloride and 6% mlV sodium carbonate solutions. Wallace22 has shown that of the several spectral lines available for boron determination the 249.773-nm line is the most sensitive. A comparative study using the 249.678-nm spectral line con- firmed that both analytical lines were relatively free from spectral interference. Only iron produced a significant inter- ference (Fig.1). On-line spectral interference correction was calibrated as 1.28 pg ml-1 of boron per 1000 yg ml-1 of iron for both the hydrofluoric acid and the fusion methods. Small changes in the profile setting of the spectrometer are known to affect the inter-element interference correction factors, especially if they are large. The interference correc- tion factor was therefore determined daily after the spec- trometer was profiled. On-peak correction has an uncertainty equivalent to approximately 10% of the total background interference effect.23.24 Hence, samples containing 10% of iron (National Bureau of Standards, Standard Reference Material 1633a, coal fly ash, Table 2) would produce an interference of 2.6 pg 8-1, implying a detection limit due to the interference correction of 0.26 pg g-1.The "adaptation effect" described by Maessen et al.25 was investigated for the matrices described above. Net line intensity as a function of time when 1 pg ml-1 of boron was aspirated is illustrated in Fig. 2. The results show that it is necessary to aspirate for a minimum period of 30 s to allow the analytical signal to stabilise before integrations are com- menced. B 249.773 Fe 249.653 I I Wavelengthlnm Fig. 1. Wavelength scans in the vicinity of the B 1249.773-nm line. B, 1 pg ml-1; Fe, 100 pg ml-* 220 > In C a c. .- c - 200 0 0 0 0 0 0 180 L L 10 20 30 40 Timels Fig. 2. Intensity of the B 1249.773-nm line as a function of aspiration time. (0) Aluminium chloride; (0) sodium carbonateANALYST, AUGUST 1987, VOL. 112 1109 Sensitivity Defining and determining the analytical sensitivity and detection limit of a procedure have been described by the ICP Detection Limits Committee.26 The 20 detection limit thus obtained for the hydrofluoric acid procedure was 0.005 pg ml-1 and that for the fusion method was 0.010 pg ml-1.When the errors introduced by the correction for spectral interferences are added to these 20 values, for a sample containing 10% of iron, the new detection limits become 0.265 and 0.275 pg ml-1, respectively. The higher detection limit for the fusion method was attributed to plasma instability induced by (a) the high salt concentration in the digest, (b) probable light scattering, (c) poorer aerosol injection into the plasma due to fouling of the torch injector tube and (d) decreased nebulisation efficiency due to a higher solution viscosity.For the inter-element interference determined above, the nomograms described by Church23 indicate that for a sample containing 10% of iron, boron determination is possible only if the sample contains more than 7.5 pg g-1. Memory Effects Memory effects are known to be a problem with boron when solutions containing dissolved lithium metaborate fusions27 are aspirated. A 1-ml volume of 20 pg ml-1 of boron (as boric acid), equivalent to 0.1% of boron in a sample, was analysed using the open beaker procedure to assess memory effects. The recovery of 20.2 k 0.4 pg g-1 of boron (n = 10) indicates no memory effects at the maximum concentration of boron generally found in rocks and soils. Boric acid, generally recommended to complex residual hydrofluoric acid, has been reported not to prevent attack on conventional glass nebulisers.28 No boron could be detected at the least quantitatively determinable amount, viz., 0.25 pg g-1, when a digest blank solution was aspirated continu- ously for 15 min.Results Recovery of Boron There have been few reports of the dissolution of geological materials using hydrofluoric acid for the determination of boron since Chapman et al.,29 using hydrofluoric and per- chloric acids, claimed a total loss of boron due to volatilisa- tion. Pritchard and Lee30 added mannitol to their hydrofluoric acid mixture to prevent the loss of boron during volatilisation of silica. The low results indicated that their procedure was unsuccessful.This was probably due to (a) evaporation of the digest solution to dryness, (b) the use of perchloric acid and (c) non-determination of some of the boron as the complexed stable fluoroborate ion. Volumes of 1 ml of 6 and 10 pg ml-1 boron (as boric acid) were analysed using the open beaker procedure to check the quantitative recovery of boron in an essentially matrix-free digest. The results were compared with those obtained using Table 1. Recovery of boron in matrix-free and synthetic soil digests Boron concentration*/pg g-1 Procedure Matrix-free “Soil” Hydrofluoric acid - aqua regia . . . . 6.1 k 0.3 5.8 k 0.3 10.2 5 0.3 Aquaregia . . . . . . . . . . 5 . 9 f 0 . 2 6.1k0.1 9.8 k 0.2 Sodium carbonate fusion . . . . . . 5.4 _+ 0.6 5.6 k 0.7 9.7 k 0.9 10.1 k 0.5 10.1 k 0.3 9.5 k 0.8 * Means for three replicates with standard deviations.only aqua regia, to assess any losses due to volatilisation, and the classical sodium carbonate fusion procedure (Table 1). The reproducibility of the determinations varied from 3% for the aqua regia procedure to 11% for the fusions. A small amount of boron was possibly lost using the fusion method. However, considering the experimental error of this very salty matrix due to the inherent difficulties outlined above and experienced by others,7 the loss was small. The slightly poorer precision of the hydrofluoric acid procedure (compared with the aqua regia procedure) was possibly due to the same plasma and nebuliser instabilities associated with the fusion method. Overall, there was good agreement between the procedures, indicating a negligible loss of boron using hydrofluoric acid.Evaporation of the digest to dryness for the removal of excess of hydrofluoric acid prior to elemental analysis is often recommended.15.18 When the hydrofluoric acid and the aqua regia systems were evaporated to dryness at 140 “C, the loss of boron varied from 5 to 25%. Previous experience with plant digestion using nitric acid - perchloric acids at 225 “C resulted in a 20% loss of boron on evaporation to 1 ml. Consequently, a hydrofluoric acid - perchloric acid mixture was not considered for sample dissolution. Matrix Effects The reasonable matrix matching of standards with samples for elemental determination by ICP-AES is desirable31 but usually not possible. A synthetic “soil” consisting of 0.45 g of analytical-reagent grade calcium carbonate , 0.45 g of silica and 0.10 g of goethite with 1 ml of either 6 or 10 pg ml-1 boron (as boric acid) was dissolved using the open beaker procedure. The results were compared with those obtained using only aqua regia, to assess any losses due to volatilisation, and the sodium carbonate fusion procedure (Table 1).As the standards and samples both contained the major matrix components, viz., aluminium chloride for the hydro- fluoric acid procedure and sodium carbonate for the fusion procedure, the effective matrix in the synthetic sample consisted of calcium, any silicon in solution and approximately 0.13% of iron. Analysis of variance using the MINITAB statistical soft- ware package32 showed that the results were not significantly different between the matrix and matrix-free digest solutions after spectral interference due to iron had been taken into account.Comparison of Dissolution Procedures The accuracy of the proposed open beaker and bomb procedures for sample dissolution and boron retention was confirmed by the determination of boron in a range of standard reference materials (Table 2). The over-all mean ratio of determined to literature values was 1.02 with a standard deviation of 0.04 for concentrations greater than 2 Samples of soils commonly found in south-eastern Aus- tralia35 and of Tindelpina Shale (consisting mainly of chlorite, muscovite, quartz, feldspar and calcite)36 from the Mount Lofty Ranges were chosen to cover a range of boron concentrations.Tables 3 and 4 report the results obtained using the recommended open beaker procedure with those using bomb dissolution and sodium carbonate fusion. A 1-ml volume of 50 pg ml-1 boron (as boric acid) was added to the soil and rock samples prior to dissolution as a check on the recovery of boron in a “real” sample matrix. For soil samples, analysis of variance of boron values showed no significant differences between the three dissolu- tion methods, except for sample No. 3082, for which the result obtained by the open beaker procedure was 15% lower. This g-l.1110 ANALYST, AUGUST 1987, VOL. 112 low result is not attributed to loss of boron by volatilisation but rather to incomplete dissolution of hydrofluoric acid-resistant minerals, as no loss occurred during volume reduction of the bomb digest.Complete dissolution of rock was not achieved using the open beaker method. Many minerals that are completely decomposed using the bomb procedure are not dissolved in open beakers.28 Analysis of variance of the boron values showed significant differences between the open beaker dissolution method and the other two methods (P < 0.001). There was no significant difference between the boron values of the sodium carbonate fusion and the bomb dissolution procedures. Quantitative recovery of boron in the standard reference materials and in the spiked soil and rock samples confirmed that (a) no losses occurred due to volatilisation, (b) matrix effects due to soil and rock concomitants were minimal and (c) compensation for the spectral interference due to iron was adequate.Triplicate analyses of the spiked and unspiked soil and rock samples showed that a precision of about 6Yo was achieved using the hydrofluoric acid method and about 10% for the sodium carbonate procedure. These are similar to the values obtained by Thompson and Walsh37 and Owens et al.,7 respectively. Sample Pre-treatment Fine grinding of the soil using a Siebtechnik mill and the destruction of soil organic matter by heating at 550 "C for 3 h in a muffle furnace had no significant effect on the recovery of boron. Comparison between analytical techniques An independent assessment of the concentration of boron in the soil and rock samples was obtained using d.c. arc emission Table 2. Boron concentrations (pg g- 1 ) in certified reference materials Sample Procedure* Fly ash- NBS1633a .. . . 1 2 CCRMP-SY2 . . . . 1 2 NBS278 . . . . . . 1 2 NBS688 . . . . . . 1 2 NBS91 . . . . . . 1 2 Silcates- * Procedure: 1 = open beaker; 2 = bomb dissolution. t Means for three replicates with standard deviations. $ Certificate of analysis. Analysis Literature values? value Reference 41.7 f 1.5 39.7 33 39.3 k 1.7 89.3 f 2.2 87 34 87.6 k 2.8 26.0 f 1.0 25 Cert . $ 27.7 f 1.5 1.6 f 0.2 1 33 1.5 + 0.2 296 f 4 302 33 295 f 5 Table 3. Comparison of boron values in soil obtained by the proposed and other procedures Sample Boron added/ Boron found$/ No.35 Soil type35 Procedure* Pg l3-I Pg g-l 588 Calcareous sand 1 - 18+ 1 50.0 70 f 1 2 - 19+2 50.0 72 f 4 3 - 19+ 1 50.0 71 f 1 574 Calcareous clay 3082 Red - brown earth 6061 Grey clay 1 - 84 4 3 50.0 128 5 3 2 - 83 f 6 50.0 130 f 7 3 - 88 f 4 50.0 134 + 4 - 1 2 3 1 50.0 50.0 50.0 50.0 - - - 244 k 10 300 4 15 282 k 20 351 f 21 286 -C 15 344 k 16 231 f 13 290 f 16 2 - 234 + 21 50.0 300 k 20 3 - 238 f 12 50.0 295 + 13 Boron initially present/pg g-1 1 8 f 1 20 k 1 19+2 22 _+ 4 1 9 f 1 21 2 1 84 f 3 78 f 3 83 f 6 80 k 7 88 f 4 84 f 4 244 f 10 2504 15 282 k 20 301 f 21 286 k 15 294 4 16 231 f 13 240 f 16 234 k 21 250 f 20 238 _+ 12 245 k 13 * Procedure: 1 = open beaker; 2 = sodium carbonate fusion; 3 = bomb dissolution.t Means for three replicates with standard deviations.ANALYST, AUGUST 1987, VOL. 112 1111 Table 4. Comparison of boron concentrations in rocks obtained by the proposed and other procedures 24 56A 46 Boron added/ Boron found?/ Ccg g-’ Ccg g-‘ Rock N0.36 Procedure* 76 1 - 352 1 50.0 84 k 1 2 - 42 f 3 50.0 94 f.4 3 - 46 f 2 50.0 97 f 2 1 - 100 k 4 50.0 145 f 4 2 - 130 f 10 50.0 183 f 11 3 - 145 k 6 50.0 208 f 7 1 - 128 f 7 50.0 195 k 9 2 - 175 f 13 50.0 230 f 12 3 - 182 4 10 50.0 239 f 11 1 - 213 k 10 50.0 261 f 11 2 - 271 k 20 50.0 326 f 21 3 - 267 f 13 50.0 333 k 12 * Procedure: 1 = open beaker; 2 = sodium carbonate fusion; 3 = bomb dissolution. t Means for three replicates with standard deviations. Boron initially presentlpg g-1 35 k 1 34 k 1 42 f 3 44 f 4 46 f. 2 47 k 2 100 f 4 95 f 4 130 f 10 133 f 11 145 k 6 158 f 7 128 f 7 145 f 9 175 f 13 180 f 12 182 ? 10 189 ? 11 213 ? 10 212 f 11 271 k 20 276 f 21 267 k 13 273 f 12 spectrophotometry .Thompson38 has shown that simple linear regression gives accurate estimates of bias between methods if (a) at least ten samples are used, (b) samples cover the concentration range from zero upwards fairly uniformly and (c) results from the method with the smaller variance are used as the independent (abscissa) variable. Boron values in soil and rock determined by d.c. arc spectrophotometry and standard reference materials determined by ICP-AES after dissolution using the open beaker procedure were used as the dependent (y-axis) variables while the boron values in soil and rock determined by ICP-AES after dissolution using the proposed procedures and the literature values reported for the standard reference materials were used as the independent (x-axis) variables to satisfy the conditions for generating the line of best fit using the MINITAB statistical software package,32 shown in Fig.3. The regression gave a slope of 1.01 400 I A I I I I 0 100 200 300 400 Literature boron values for reference materials and soil and rock by ICP-AES/pg 9-1 Fig. 3. Boron determined in the standard reference materials by the proposed dissolution procedure and in soil and rock by d.c. arc emission spectrophotometry as a function of the boron values reported in the literature for the standard reference materials and in soil and rock determined by the proposed procedures k 0.02, which is not significantly different from unity, and an intercept of -2.10 k 5.05, which is not signficantly different from zero, confirming the complete dissolution of soil (using an open beaker), rock (using a bomb) and the standard reference materials (using an open beaker and a bomb) whilst retaining boron during volume reduction.Conclusions The simple and rapid dissolution of silicate materials for the determination of boron was achieved using hydrofluoric acid and aqua regia without a loss of boron by volatilisation during volume reduction. Complexing residual fluoride with alumi- nium released boron from the stable fluoroborate ion and allowed determination by ICP-AES whilst protecting the borosilicate and quartz components of the spectrometer. The low salt concentration in the digest solution (compared with the fusion procedure) resulted in minimal calibration drift. Consequently, frequent re-standardisations of the instrument during routine analysis were not necessary. The increase in detection limit due to the inter-element interference produced by iron, with concentrations ranging up to lo%, did not degrade the quantifiable limit to the extent suggested by Thompson and Walsh,24 as evidenced by the recovery of boron in the synthetic “soil” and standard reference materials with a precision equal to or better than that suggested by Church? This is probably due to (a) daily monitoring of the spectral inter-element interference after profiling the spectrometer and (b) the use of a modified Babington-type nebuliser (Type GMK) as suggested by Mills.27 The boron concentrations in the soils, obtained by the open beaker method, and in rocks, obtained using the bomb procedure, were comparable to those obtained by accepted sodium carbonate fusion (with determination by ICP-AES) and d.c.arc emission spectrophotometry. Determinations of boron in the standard reference materials showed good agreement and precision with the published literature values.1112 Mills27 has also shown that, for the standard reference materials investigated, the recovery of boron with adequate precision is achievable using an open beaker procedure. However, dissolution of all silicates for the determination of boron using a single procedure must be treated with caution, as the comparison between methods for the determination of boron in soil, rock and standard reference materials shows. We express our thanks to J. Eames and N. Morgan of CSIRO, Division of Mineral Physics and Mineralogy, for the d.c.arc emission spectrophotometric analyses, J. C. Mills of the Broken Hill Prop. Co. Ltd. for helpful discussions and V. Gostin of the University of Adelaide for the rock samples. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. References Jackson, M. L., in Bear, F. E., Editor, “Chemistry of the Soil,” Second Edition, Reinhold, New York, 1964, p. 134. Dewis, F. J., Levinson, A. A., and Bayliss, P., Geochim. Cosmochim. Acta, 1972,36, 1359. Potter, P. E., Shimp, N. F., and Witters, J., Geochim. Cosmochim. Acta, 1963, 27, 669. Reisenauer, H. M., Walsh, L. M., and Hoeft, R. G., in Walsh, L. M., and Beaton, J. D., Editors, “Soil Testing and Plant Analysis,” Revised Edition, Soil Science Society of America, Madison, WI, 1973, p.173. Berger, K. C., and Truog, E., Soil Sci. SOC. Am. Proc., 1945, 10, 113. Gladney, E. S., Jurney, E. T., and Curtis, D. B., Anal. Chem., 1976,48,2139. Owens, J. W., Gladney, E. S., and Knab, D., Anal. Chim. Acta, 1982, 135, 169. Gupta, U. C., SoilSci. SOC. Am. Proc., 1966,30, 655. Walsh, J . N., Analyst, 1985, 110, 959. Hall, G. E. M., and Pelchat, J.-C., Analyst, 1986, 111, 1255. McKinnon, P. J., Giess, K. C., and Knight, T. V., in Barnes, R. M., Editor, “Developments in Atomic Plasma Spectro- chemical Analysis,” Heydon, Philadelphia, 1981, p. 287. Zarcinas, B. A., and Cartwright, B., CSZRO Aust. Div. Soils Tech. Pap., No. 45, 1983. Troll, G., and Sauerer, A., Analyst, 1985, 110, 283. Nadkarni, R. A., Anal. Chem., 1980,52, 929. McLaren, J. W., Berman, S. S., Boyko, V.J., and Russell, D. S., Anal. Chem., 1981, 53, 1802. Lechler, P. J., Roy, W. R., and Leininger, R. K., Soil Sci., 1980, 130,238. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. ANALYST, AUGUST 1987, VOL. 112 Uchida, H., Uchida, T., and Iida, C., Anal. Chim. Acta, 1979, 108, 87. Uchida, H., Uchida, T., and Iida, C., Anal. Chim. Acta, 1980, 116,433. Spiers, G. A., Dudas, M. J., and Hodgins, L. W., Commun. Soil Sci. Plant Anal., 1983, 14, 629. Barredo, L. B., and Diez, L. P., Talanta, 1976,23,859. Siemer, D. D., Anal. Chem., 1982,54, 1321. Wallace, G. F., At. Spectrosc., 1981, 2, 61. Church, S. E., Geostand. Newsl., 1981, 5 , 133. Thnompson, M., and Walsh, J. N., “A Handbook of Induc- tively Coupled Plasma Spectrometry,” Blackie, Glasgow ,1983, p. 126. Maessen, F. J. M. J., Balke, J., and de Boer, J. L. M., Spectrochim. Acta, Part B, 1982,37, 517. ICP Detection Limits Committee, ZCP Znf. Newsl., 1979, 5 , 295. Mills, J. C., Anal. Chim. Acta, 1986, 183, 231. Thompson, M., and Walsh, J. N., “A Handbook of Inductively Coupled Plasma Spectrometry,” Blackie, Glasgow , 1983, p. 134. Chapman, F. W., Marvin, G. G., and Tyree, S. Y., Anal. Chem., 1949,21,700. Pritchard, M. W., and Lee, J., Anal. Chim. Acta, 1984, 157, 313. Thompson, M., and Walsh, J. N., “A Handbook of Inductively Coupled Plasma Spectrometry,” Blackie, Glasgow, 1983, p. 130. Ryan, T. A., Joiner, B. L., and Ryan, B. F., “MINITAB Student Handbook,” Duxbury, Boston, 1976. Gladney, E. S., Burns, C. E., Perrin, D. R., Roelandts, I., and Gills, T. E., Nut. Bur. Stand. U.S. Spec. Publ., No. 260, 1984. Abbey, S., and Gladney, E. S., Geostand. Newsl., 1986,10,3. Stace, H. C. T., Hubble, G. D., Brewer, R., Northcote, K. H., Sleeman, J. R., Mulcahy, M. J., and Hallsworth, E. G., “A Handbook of Australian Soils,” Rellim, Glenside, 1968. Sumartojo, J., PhD Thesis, University of Adelaide, 1974. Thompson, M., and Walsh, J. N., “A Handbook of Inductively Coupled Plasma Spectrometry,” Blackie, Glasgow, 1983, p. 145. Thompson, M., Analyst, 1982, 107, 1169. Paper A61294 Received August 21st, 1986 Accepted March 1 Oth, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201107

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, AUGUST 1987, VOL. 112 1113 Rapid Determination of Chromium in Bovine Liver Using an Atomic Absorption Spectrometer with a Modified Carbon Rod Atomiser John W. Steiner, David C. Moy and Harvey L. Kramer Department of Primary Industries, Animal Research Institute, Biochemistry Branch, 665 Fairfield Road, Yeerongpilly, Brisbane, Queensland 4 105, Australia Chromium in bovine liver was determined rapidly by atomic absorption spectrometry with a modified carbon rod atomiser. Pre-treatment of samples was minimised by the sequential introduction of mixtures of nitrogen, hydrogen, oxygen and methane into the atomiser. An optical device facilitated increased precision of injection of samples. The effectiveness of the method was confirmed by the determination of chromium in a certified sample.Keywords: Chromium determination; matrix modification; in situ gaseous pre-treatment; modified carbon rod atomiser; electrothermal atomisation atomic absorption spectrometry Chromium is an important trace element for mammalian nutrition, being essential for insulin action and maintenance of normal glucose metabolism. 172 Despite the recognition of the biological importance of chromium, its determination is still difficult, especially when older electrothermal atomisation atomic absorption spectrometers are used. Major difficulties of older electrothermal atomisation atomic absorption spec- trometry (ETA-AAS) instruments are inherent in the back- ground correction system using a deuterium lamp3.4 and in the lack of a variable ramp rate facility for the drying and ashing cycles.5 The determination of chromium in biological materials with older instruments depends largely on the pre-treatment of samples using the tedious and time-consum- ing chelation and solvent extraction procedure.6 We describe here an alternative approach using in situ treatment of the acid-digested sample within the atomiser.This utilises a carbon rod atomiser modified as described elsewhere.7 Use of the modified atomiser involves synchron- ised introduction of different gases into the atomiser during the drying, ashing and atomising cycle. This facilitates more uniform drying of the sample, which in turn improves the ashing efficiency and provides highly effective atomisation of the sample. Other benefits are an increase in the lifetime of the graphite tube and better precision.Experimental Apparatus A Varian Techtron AA-175 spectrophotometer equipped with a modified carbon rod atomiser (CRA-90) was used. The modifications consisted of the use of an optical device and vertical and transverse gas jets. The optical system facilitated the viewing of sample injection and the gas jets allowed the introduction of gases into the atomiser.7 Reagents Gases High-purity nitrogen, industrial dry hydrogen and medical- grade oxygen were obtained from Commonwealth Industrial Gases (CIG), Brisbane. Methane (99.9% pure) was obtained from Matheson. Standard solutions The concentrations of the working standards were 0, 0.01, 0.02 and 0.03 mg 1-1 of chromium. These were prepared daily from concentrated stock solutions (BDH Chemicals).Standard reference material Freeze-dried, powdered bovine liver standard reference material (SRM 1577) was obtained from the US National Bureau of Standards. The certified concentration for chro- mium was 0.088 k 0.012 mg kg-1 (95%/95% statistical tolerance limits). Procedure Glassware cleaning New borosilicate glassware was initially de-contaminated by soaking in 10% nitric acid (re-distilled before use). It was subsequently maintained in a clean condition using Decon-90 non-ionic detergent (Decon Laboratories). Prior to use, the glassware was rinsed thoroughly with doubly glass-distilled water. Preparation of SUM 1577 and test samples The SRM 1577 and test samples of liver tissue were prepared by weighing suitable amounts (approximately 1.0 g dry or 3.0 g fresh) into glass-stoppered borosilicate digestion tubes of 50-ml capacity.Concentrated doubly distilled, chromium-free nitric acid ( 5 ml) was added and the tubes were lightly stoppered before being placed in a shaking water-bath at 80 "C for 4 h. The partially digested samples were diluted to 25 ml before analysis. The samples used for recovery studies were fortified with a fraction of the analyte prior to digestion. Instrument optimisation and operation The optimisation and operation of the instrument have been described previously,7 as has the specific order in which absorbance readings were taken to evaluate the method for the determination of chromium in bovine tissue. The instrumental conditions and optimum flow-rates of the gases used for the continuous re-coating of the atomiser and for the in situ treatment of the sample are reported in Tables 1 and 2.Table 1. Instrumental conditions Parameter Value Volume of injection/pl . . . . . . . . . . 4 Drying ternperature/'C . . . . . . . . . . 95 Ashing temperature/'C . . . . . . . . . . 60 Atomising temperature/'C . . . . . . . . . . 2200 Drying time/s . . . . . . . . . . . . . . 50 Ashing time/s 30 Hold time/s . . . . . . . . . . . . . . 3.5 Ramp ratePC s-1 800 . . . . . . . . . . . . . . . . . . . . . . . . . . Residence time of transverse gas jet/s . . . . . . 121114 ANALYST, AUGUST 1987, VOL. 112 ~~ ~ ~ Table 2. Optimum flow-rates [mean k s.d. (n = determination of chromium Gas introduction system Gas Sheath . . . . . . . . . .. . Nitrogen Hydrogen Methane Vertical gas jet . . . . . . . . . . Methane Transverse gas jets . . . . . . . . Hydrogen Oxygen 6)] for the Flow-rate/ ml min- * 6719 f 51 119 k 2.1 100 f 1.2 9.1 & 0.8 216 k 1.2 177 k 1.3 Results and Discussion Evaluation of Method for Determination of Chromium The proposed method for the determination of chromium was evaluated by regular monitoring of accuracy, precision, repeatability and sensitivity. The effectiveness of using different gases and synchronising their use with the overall operation of the instrument is demonstrated in Table 3. From a sequence of 63 measurements, two intervals of 9 readings (periods 1 and 2) provide the determined chromium concen- tration. In conventional operation (Table 3, row 1) good precision but poor accuracy was obtained.However, a significant (P < 0.05) improvement in both was achieved (Table 3, row 8) when the appropriate gas mixture, in conjunction with the optical device, was used. Although the concentration of chromium in SRM 1577 is certified as 0.088 _+ 0.012 mg kg-1, only 0.081 k 0.004 mg kg-1 was found. Attempts were made to reduce the difference between the certified and experimentally determined concentrations of chromium in the SRM. This was done by changing the mixing ratios and flow-rates of nitrogen, methane and hydrogen while keeping the flow-rate of oxygen constant. No significant improvement was achieved until the flow-rate of oxygen-was elevated from 177 to 220 ml min-1 (while keeping the flow-rates of the other gases constant).This provided a value of 0.083 k 0.014 mg kg-1, but the increase in the oxygen flow-rate resulted in a subsequent reduction of the determined chromium concentration to 0.077 +_ 0.010 mg kg-1 after 40 firings. The determined value of 0.081 2 0.004 mg kg-1 lies within the range obtained by other workers314 for SRM 1577 using atomic absorption spectrometry, namely 0.084 f 0.001 mg kg-1 to 0.060 k 0.006 mg kg-1. The accuracy of our work is further supported by neutron activation analysis studies,g which provided a mean value of 0.0806 mg kg-1 for a range of 0.0772-0.0840 mg kg-1. Precision and accuracy tests were performed with the modified procedure on four separate days with four SRM- 1577 sub-samples (Table 4). The chromium content found in these samples was 0.081 2 0.004 mg kg-1 (mean f 95% statistical tolerance limits). Further assessments of precision, accuracy and recovery were performed at regular intervals (approximately every 250 analyses) using the analytical procedures described in this paper.During the course of over 3000 analyses, no variations were observed. Table 5 compares the results obtained with and without the modified procedure. The absolute error and the relative standard deviation (r.s.d.) are significantly (P< 0.05) lower using the modified procedure. The modified procedure also resulted in an improvement in repeatability and sensitivity (slope bl). Although the deuterium background correction system was continuously used during the entire investigation, the as- sociated difficulties reported by other workers3JO were not experienced, probably because the maximum background absorbance never exceeded 0.025 unit.There are recommen- dations for using a high and variable ashing temperature to minimise interference.10 This was not necessary, as the presence of the mixture of gases (nitrogen, methane, hydrogen and oxygen) in the atomiser during ashing effec- tively eliminated all interferences. Conclusion The introduction of different gases into the atomiser effected a direct pre-treatment of partially digested bovine liver samples. A careful synchronisation with the over-all instrumental Table 4. Recovery of chromium added to bovine liver samples. Four sub-samples, four analyses per sub-sample Amount of Cr Recovery, 0.3 94.3 4 3.2 0.6 94.6 & 1.3 0.9 92.0 k 1.6 added1 pg Yo f sad.Table 5. Comparison of performance using the modified and unmodified procedures Modification Analytical parameter With Without Absolute error*/mg kg-1 . . . . . . -0.007 -0.021 R.s.d.,% . . . . . . . . . . . . 4.1 19.8 Slope (b,) . . , . . . . . . . . . 6.909 6.220 measured and the certified values for SRM 1577. * The absolute error is defined9 as the difference between the Table 3. Experimental conditions and results of the repeated determination of chromium in SRM 1577 certified to contain 0.088 2 0.012 mg kg-l of chromium Gas introduction system Sheath N2 . . . . N,+H, . . N2 + H2 + CH4 N2 + H2 + CH4 N2 + H2 + CH4 N2 + H2 + CH4 N2 + H2 + CH4 N2 + H2 + CH4 Vertical jets - . . . . - . . . . - . . . . . . . . CH4 . . . . CH, . . . .CH4 . . . . CH4 . . . . CH4 Transverse jets - - - - H2 0 2 H2 + 0 2 H2 + 0 2 Optical - device Determined chromium concentration/mg kg-1* ~~~ ~ ~ ~ Period I t 0.067 2 0.014 0.061 k 0.012 0.068 2 0.018 0.062 f 0.016 0.068 -t 0.007 0.083 k 0.014 0.078 -t 0.007 0.080 k 0.003 Period 2f 0.065 f 0.007 0.110 f 0.037 0.057 k 0.021 0.063 f 0.020 0.064 f 0.009 0.077 f 0.010 0.077 t 0.008 0.081 t- 0.004 * Mean k 95'/0/95% statistical tolerance limits. t Firing interval 16-24 inclusive (based on n = 9). $ Firing interval 52-60 inclusive (based on n = 9).ANALYST, AUGUST 1987, VOL. 112 1115 operation resulted in a significant improvement in the 5. analytical data, shortened the time required for analysis and reduced the deterioration rate of the atomiser. We thank Mr. F. Duncalfe for the statistical analysis of the data. 6. 7. 8. 9. 1. 2. 3. 4. References 10. Schwarz, K., and Mertz, W., Arch. Biochem. Biophys., 1959, 85, 292. Mertz, W., Physiol. Rev., 1963,49, 163. Routh, M . W., Anal. Chem., 1980, 52, 182. Kayne, F. J . , Komar, G., Laboda, H., and Vanderline, R. E., Clin. Chem., 1978, 24, 2151. Matsusaki, K., Yoshino, T., and Yannamoto, Y. Anal. Chim. Acta, 1981, 124, 163. Chao, S. S., and Pickett, E. E., Anal. Chem., 1980, 52, 335. Steiner, J. W., and Kramer, H. L., Analyst, 1983, 108, 1051. Versieck, J., Hoste, J., De Rudder, J., Barbier, F., and Vanballenberghe, L., Anal. Lett., 1979, 12, 555. Christian, G. D., “Analytical Chemistry,” Xerox College Publishing, Waltham, MA, 1971, p. 456. Guthrie, B. E., Wolf, W. R., and Veillan, C., Anal. Chem., 1978, 40, 1900. Paper A61306 Received September 2nd, 1986 Accepted March 23rd, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201113

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, AUGUST 1987, VOL. 112 1117 Use of Fourier Transform Infrared Spectroscopy for Quantitative Analysis: A Comparative Study of Different Detection Methods Peter S. Belton, Alfred M. Saffa and Reginald H. Wilson AFRC Institute of Food Research, Norwich Laboratory, Colney Lane, Norwich NR4 7UA, UK The value of Fourier transform infrared spectroscopy in the mid-infrared region for the quantitative analysis of protein - starch mixtures is examined. Attenuated total reflectance, diffuse reflectance and photoacoustic detection are compared. It is concluded that photoacoustic detection is the best method of analysis for protein but that only attenuated total reflectance is of use in wet systems. Diffuse reflectance gave poorly resolved spectra, which severely limited its usefulness.Keywords: Infrared spectroscopy; photoacoustic; attenuated total reflectance; diffuse reflectance; detection Fourier transform infrared spectroscopy is now a widely used technique in analytical and research laboratories. However, the use of mid-infrared spectra for quantitative analysis has been severely limited. This is in contrast to near-infrared reflectance (NIR), which has been widely exploited in quality control applications for obtaining quantitative information. As the price of Fourier transform instruments continues to drop, however, quality control applications become a more attractive option. The mid-infrared region has much to offer the analyst compared with NIR. Specific bands may be assigned to specific chemical entities and statistical correlation methods are not always necessary, although they are not excluded and may be required in very complicated mixtures.Several methods of sample presentation are now available that require little or no prior preparation. In this paper, three of the most used methods [attenuated total reflectance (ATR), diffuse reflectance infrared Fourier transform (DRIFT) and photoacoustic detection (PAS)] are compared. The test sample chosen was a mixture of protein and starch, which typifies much of the material currently analysed in the food industry by NIR. The results obtained indicate that, given a careful choice of experimental conditions, useful quantitative data are available on a routine basis. Experimental Mixtures of protein and starch were prepared using either a wheat gluten from an industrial supplier or casein (Sigma) with potato starch (BDH Chemicals). The mixtures were made by adding the appropriate mass of protein to starch and mixing well.In most experiments gluten was used, and the term protein will apply to that material unless otherwise designated. The efficiency of mixing and sampling was demonstrated by the reproducibility of replicate samples (three of each) of protein - starch mixtures containing 17.79 and 28.90% of protein. The mean PAS intensities and standard deviations were 13.74 _+ 0.037 and 17.87 _+ 0.039, respectively. In order to test the potential for quantitative measurements in the presence of water, 1 g of water per gram of dry solid gluten - starch was added and, after thorough stirring, the wet mixtures were allowed to stand overnight before use.The consistency of the wet systems varied from a moist powder at high starch concentration to an elastic mass at high gluten content. All infrared spectra were obtained on a Digilab FTS60 FTIR spectrometer equipped with a TGS detector and operating at 8 cm-1 resolution. The mirror velocity was 0.32 cm s-1, 256 interferograms were co-added before Fourier transformation and triangular apodisation was employed. Acquisition of spectra followed the conventional pattern. A single-beam background was collected with each sampling attachment either empty or with the appropriate reference material. A single-beam spectrum of the sample was then obtained and ratioed to the background before conversion to the relevant quantitative units.ATR spectra are presented in conventional absorbance units. DRIFT spectra are presented in Kubelka-Munk (KM) units.'.* PAS spectra are presented in terms of the normalised PAS signal intensity. DRIFT spectra were obtained with a Spectra-Tech COL- LECTOR attachment using a 14 mm diameter sample cup. No attempt was made to block the specular component of reflected light, as a trial carried out using the specular blocker showed little effect other than a considerable reduction in the over-all signal, This observation has been reported by other workers.3 For the dry mixtures, ground KBr was used as a reference material. However, the wet systems generally had flat, shiny surfaces producing a high degree of specular reflection. In ideal circumstances a reference and a sample with similar surface reflectivity are used in order that a true diffuse reflectance spectrum can be measured.The only reference available with a similar degree of surface reflectance to the wet doughs was an alignment mirror. Unless this was used as the reference, serious spectral distortions resulted. However, even in this instance there was severe distortion in the region 1800-1500 cm-1 caused by water vapour. ATR spectra were obtained with a Spectra-Tech continu- ously variable angle ATR attachment with a KRS5 crystal (45", 50 x 20 x 3 mm parallelogram). The best spectra were obtained with the incident angle set at 45". Powders were spread as evenly as possible over one face of the crystal and on the mounting plate on the opposite side of the crystal.Thus, when the attachment was assembled both sides of the crystal were covered. As much care as possible was taken to ensure both even coverage and even application of pressure. Wet samples were treated in a similar fashion. PAS measurements were obtained with a Digilab PAS cell fitted with a KBr window (CaF2 for wet systems) and a 3 mm deep sample cup. All spectra were ratioed to finely powdered carbon black. Scanning electron micrographs of pure casein and gluten powders were obtained on a Philips SEM 501 instrument. Results and Discussion Electron microscopy of gluten - starch mixtures showed the starch to have a bimodal particle size distribution. However, the starch particles were still considerably smaller than the protein particles.There was little indication of significant gluten - starch interaction, except for a few small starch particles lodged in large pores in the gluten. These represen- ted a very small proportion of the total starch content. Typical spectra obtained using the three techniques are shown in Fig. 1. All spectra, with the exceptions noted below,1118 ANALYST, AUGUST 1987, VOL. 112 000 1500 9002000 1500 900 Wavenum bericm- Fig. 1. Representative spectra produced usin (a) PAS (vertical scale = 8 PAS units er division); ( b ) DRIFTS fvertical scale = 3.2 KM units per division5; and ( c ) ATR (vertical scale = 0.1 absorbance units per division) of a dry mixture (50% protein). Also shown ( d ) is the ATR spectrum of the same mixture after the addition of 1 g of water per gram of dry material (vertical scale = 0.75 absorbance unit per division) show well separated signals from the protein at 1650 cm-1 (amide I)1 and 1540 cm-1 (amide II)4 and from the starch in the region 1180-800 cm-1 (C-0 and C-C stretching).5 It is interesting to compare the general features of these spectra.The ATR spectrum is that most closely resembling a conventional transmission spectrum. The bands in the region 2000-800 cm-1 are very well resolved and show a high degree of structure. The PAS spectrum is simlar to the ATR in appearance although the bands are, on the whole, slightly broader and the intensities of the low- and high-frequency peaks are reversed. The broadening is to be expected because of the functional relationship between the PAS intensity and the amount of light absorbed.Both techniques produce spectra with very good signal to noise ratios over the entire region, although there is definite curvature of the base line at lower frequencies. The DRIFT spectrum shows very poor resolution. Most notable is the appearance of a broad, relatively featureless area in the region 1400-900 cm-1 with the most intense peak about at 1500 cm-1. However, recent results3 have suggested that in DRIFT the geometry of the sampling optics is very important in eliminating surface reflection effects and that the optical geometry for DRIFT must be optimised for this purpose. This may not apply with some commercial accessories. Dilution in KBr also reduces3 specular effects. It has been suggested3 that the presence of specular effects, caused by either optical geometry or dilution differences, is largely responsible for the variation in the spectral appearance of samples run using different DRIFT attachments and differences between DRIFT spectra and spectra obtained with other techniques.Hence the results reported here may be improved upon by using a different diffuse reflectance cell. In our spectra the protein region from 1700 to 1500 cm-1 is also poorly resolved, the amide I1 band is reduced to a shoulder and the amide I band is broadened. The band at 1750 cm-1, due to lipid, is also reduced to a shoulder. A typical ATR spectrum of the wetted material is shown in Fig. 1. The presence of a water band at 1650 cm-1 obscures the amide I band, but the amide I1 and starch bands are relatively unaffected.Neither DRIFT nor PAS yielded useful spectra from the wet materials. The DRIFT spectra had severely sloping base lines and strong absorptions due to water vapour in the region 1600-1500 cm-1 that obscured the amide I and I1 bands. A similar problem with wet materials arose with photoacoustic spectra. That these effects were due to water vapour rather than surface liquid water was demonstrated by the sharpness of the spectral lines. In PAS the effect of water vapour was exacerbated by the strong photoacoustic response of vapours. An additional complication with PAS is that the heating effect of the incident radiation led to the generation of extra water vapour within the confined volume of the closed cell. Ideally, for quantitative analysis, the amount of absorbing material is measured by determination of the area under the absorption peak.However, where there are no line-shape changes or shifts of the peak maxima, simple intensity measurements are sufficient. Whichever type of measurement is selected, it is necessary to choose a correct and consistent base line from which to make measurements. In some spectra this was not straightforward, as the peaks were on sloping or offset base lines. Three different approaches to this problem were tried. The method of final choice was that which provided a calibration graph with the minimum of scatter. These methods were as follows: (A) a value for the base line was estimated from the flat region around 2000 cm-1, and this value was subtracted from subsequent peak-height measure- ments; (B) an over-all base line as in method A was used and the areas of defined bands above the base line were determined; and (C) a separate base line was defined for each band and a straight line was drawn between defining points on either side of the band and the area measured.When these techniques were applied to both DRIFT and ATR the over-all measured intensities of the bands were found to vary in an inconsistent manner. This was ascribed to irreproducibility in sample loading. The problems of sample loading in DRIFT have been reported by others3 and ATR suffers from variations in both crystal coverage and applied pressure. Therefore, in order to overcome these problems an internal ratio method was applied. For ATR on dry materials, all three quantitative methods were applied and the best calibration graph was obtained using method C.The calibration graph of the ratio ( R ) of the area (amide I + 11) to the area (1180-870 cm-1) versus protein to starch ratio is shown in Fig. 2(a). Usable spectra from wetted systems were obtained only with ATR. However, a useful calibration graph could not be constructed. The problem was that the amide I band was swamped by the large water peak and in some instances the amide I1 was superimposed on a severely sloping base line and was considerably offset. Method A was the only method usable for an attempt at quantitative analysis and the plot of the ratio ( R ) of the absorbance of the amide I1 to the starch band at 1018 cm-1 is shown in Fig. 2(a). Clearly, this does not constitute a useable calibration graph.Distortion of the carbohydrate region in DRIFT meant that the amide I1 band was effectively lost whereas the amide I band was superimposed on a strongly sloping base line. Further, the large number of overlapping peaks in the starch region coupled with a strongly offset base line made the choice of a base line for method C very difficult. The large number of peaks and the dissimilarity to the ATR or PAS spectra made the choice of band or bands for area measurements problem- atic. A prominent band was seen, however, in some DRIFT spectra at 1166 cm-1. This peak varied in intensity with starch concentration. Despite the fact that the origin of this peak is unknown, measurements were made at that frequency. Method A was used for quantitative analysis and a calibration graph [Fig.2(a)] of the ratio (R) of the amide I intensity to that at 1166 cm-1 was plotted against protein to starch ratio. Dilution in KBr did have an effect on the appearance of the DRIFT spectrum. The starch region showed some improve- ment in resolution and there was a trend towards the type of spectrum produced by PAS or ATR. However, the improve-1119 *-’ . .- L 4 MI . r I Y a 1 rn 1 1 . 2 . 0 1 2 3 4 Protein ANALYST, AUGUST 1987, VOL. 112 I.. I Starch (bl A A A A A.A ,‘@ 0 0 1 2 3 1 - I g-’ M P Fig. 2. (a) Graphs of R (as defined in text) versus protein to starch ratio (rnlm) for dry powders by DRIFT (m, 0.5 unit per division) and ATR (0,0.125 unit per division). Also shown is the same plot for the wetted system by ATR ( A , 1 unit per division).(6) Calibration graphs for PAS presented as the reciprocal protein band area (amide I + I1 (A, 0.65 unit er division) or the reciprocal protein band (amide I] peak height (&, 0.04 unit per division) versus the reciprocal of the mass of protein in the mixture, 1/M, (g-l) ment in starch band resolution was not sufficiently marked to allow full quantitative measurement. Dilution only marginally reduced the component at 1500 cm-1. It may be that further dilution with KBr would result in further spectral improve- ment, but this would have led to very low signals and hence long acquisition times or the need for more sensitive detectors. It was concluded, therefore, that further dilution was not a practical option.The problems of irreproducible sample loading did not arise with PAS and consequently internal ratioing was not required. PAS has been overlooked as a quantitative method until fairly recently. A useful and fairly general approximation6.7 to the equation of Poulet et al.8 is where H is the normalised signal intensity resulting from absorption of incident light, A’ is a parameter relating the energy absorbed to the final signal intensity6 in the sample and reference, p is the thermal diffusion length7 and p is defined by 2.303eMp p Mr MT P = where Mr is the relative molecular mass of the absorbing species of molar absorptivity E, p is the density of the mixture and M, is the mass of the absorbing species in the mixture of total mass MT. Assuming that there is no significant variation Fig.3. Variation of peak height at 1003 cm-1 starch band with mass fraction of starch in KBr using PAS in the parameter MTlp, then a plot of 1/H vs. l/Mp should be linear .6 Strictly these equations are valid only for homogeneous materials. However, previous work6 has shown that they do apply fairly well to powdered systems. All three methods of calculating intensity were tried on the gluten - starch mixtures. Methods A and B produced linear plots as expected when the reciprocal of the amide I peak height or area of amide (I + 11) was plotted against the reciprocal of the mass of gluten in the sample [Fig. 2(b)J. Method A produced a plot of slope 0.033, intercept 0.005 and correlation coefficient 0.998. Method B produced a plot of slope 0.054, intercept 0.042 and correlation coefficient 0.995. However, with the starch bands both methods gave intensities independent of concentration and method C showed an over-all decrease in the starch band intensity with increasing starch concentration.The problem was exacerbated by difficulty in obtaining a consistent base line in this region, the shape of which changed with starch concentration. The behaviour of the starch bands in the PAS spectra is due to photoacoustic saturation effects. When the term pp is large the signal intensity becomes independent of chromophore con- centration.8 In Fig. 3, the effects of dilution of starch with KBr is shown; above 15% by mass of starch there is no variation in band intensity. The general form of the curve is that predicted by Poulet et aZ.8 An additional problem in the starch region of the spectrum may be due to phasing effects.Poulet et aZ.8 suggest that the relative phase of the photoacoustic signal is dependent on the term pP. Phase correction on most Fourier transform instru- ments is concerned with symmetrisation of the interferogram and is only strictly appropriate for transmission spectra. These phase effects may account for the reversal of the intensity of the protein and starch bands in the PAS spectrum. Further complications in spectral intensities can also arise because the interferometer causes a variation in modulation frequency with wavelength. Hence the thermal escape depth, p, varies across the spectrum. A further series of experiments was carried using dry mixtures of starch with casein instead of gluten.Electron microscopy showed that the casein particles were larger than the gluten particles and resistant to grinding. ATR of casein - starch mixtures was not very successful because the protein bands were much weaker that those produced by gluten. This is presumably due to the larger size of the casein particles, which provide a poorer optical contact with the crystal. On the other hand, the PAS spectra of casein - starch mixtures were of very good quality, comparable to those from gluten - starch, showing that PAS is less sensitive to sample morphology. The three techniques used in this study are those which have become more widely used since the advent of Fourier transform instruments. They are generally considered to be1120 ANALYST, AUGUST 1987, VOL.112 Conclusions Overall, the ATR method proved to have the widest application. This, combined with its relatively low cost, makes it a very valuable quantitative analytical technique. For dry powdered samples or samples with a low vapour pressure of water, despite its relatively higher cost at the moment, PAS should be the technique of first choice. PAS is free from any of the distorting phenomena seen with DRIFT and for most materials is a technique requiring a minimum of preparation. The equipment has developed to the stage where long acquisition times are no longer required; the spectra shown here each took about 10 min to obtain. The mid-infrared region shows great promise for analytical work in the future. There has recently been an upsurge in near-infrared methods based on reflectance techniques that are of sufficiently low cost that they are becoming widely used in quality control environments.These techniques are ham- pered by the need for a high degree of statistical manipulation as the bands in this region are broad and overlap strongly. In contrast , the mid-infrared bands are intrinsically narrower and better resolved. There is, therefore, no need for statistical manipulation of data. The main factor limiting the more widespread use of the method is the relatively high cost of FTIR equipment. Fortunately, the cost is now beginning to be reduced and equipment is commercially available that matches the cost of other conventional quality control techniques. If this down- ward trend continues, the mid-infrared region should become more widely used.methods involving a minimum of sample preparation. There are now very few samples that cannot be analysed by at least one of these methods.The difficulty lies in the identification of the best one for a particular problem. Recently, DRIFT has been the first choice for many problems and PAS has been considered only as a last resort. The results presented here, however, suggest that PAS offers a very real alternative to DRIFT as a method of first choice. Recent work6>7 has led to a better understanding of the problems associated with quantitative PAS and the potential of the technique is clearly reflected in the calibration graphs that it produced. For many samples, PAS is the only true “no-preparation,’ method, although it is certainly much better suited to powdered samples. Moreover, it is apparently unaffected by the reflectance phenomena and less sensitive to sample morphology.Both PAS and ATR produce spectra that are useful for spectral interpretation and qualitative analysis. The spectra produced are sufficiently similar to transmission spectra for comparisons to be made with reference spectra and real spectroscopic information can be extracted. This is not always so with DRIFT. The problem of optical saturation that may be experienced with some absorbers can be overcome by dilution in KBr, although this does require the user to spend more effort in sample preparation. PAS is not alone in this respect. The DRIFT spectra show that in order even to produce a spectrum that approaches the quality of ATR and PAS, dilution in KBr is a necessity. Another problem with DRIFT is the spectral distortion partly due to specular reflectance.The bands in DRIFT spectra are considerably broader than in ATR or PAS, which reduces the spectral discrimination to the extent that the amide I1 band becomes indistinguishable. DRIFT, like PAS, is not very useful for very wet systems because of the effects of water vapour (although some useful PAS spectra of moist systems such as bread have been obtained by us). With the optical geometry of the DRIFT attachment used in this work, some burning of the surface of powders at the focal point was observed. Another problem was that the DRIFT cell could not be reproducibly loaded so that, unlike the PAS cell , internal ratioing was needed.Irreproducibility in sample loading was also found with ATR, as both crystal coverage and applied pressure are variables that are poorly controlled. ATR was also very sensitive to the nature of the sample. The relatively larger , harder particles of casein provided spectra with weaker protein bands than those from gluten. Materials with large, hard particles may also cause problems of crystal damage. The technique seems best with a material consisting of an even distribution of small particles that give good optical contact. Although ATR was unable to generate a usable calibration graph owing to the presence of a water band at the amide I frequency, it may be possible to improve the calibration by using the technique of absorbance subtraction. If the water band could be subtracted from the spectrum then the distorting effect on the amide I1 could be eliminated and even the amide I may prove accessible. The authors thank Dr. Mary Parker and Mr. Roger Turner for the electron microscopy. A. M. S. thanks the British Council for a TCTD award. 1. 2. 3. 4. 5. 6. 7. 8. References Kubelka, P., and Munk, F., Z . Tech. Phys., 1931, 12, 593. Kubelka, P., J . Opt. SOC. A m . , 1948, 38, 448. Brimmer, P. J., Griffiths, P. R., and Harrick, N. J., Appl. Spectrosc., 1986, 40, 258. D’Esposito, L., and Koenig, J. L., in Ferraro, J. R., and Basile, L. J., Editors, “Fourier Transform Infrared Spectroscopy,” Volume 1, Academic Press, New York, 1978, Chapter 2, p. 88. Tipson, R. S., and Parker, F. S . , in Pigman, W., and Horton, D., Editors, “The Carbohydrates,” Second Edition, Volume lB, Academic Press, New York, 1980, Chapter 27, p. 1398. Belton, P. S . , and Tanner, S . F., Analyst, 1983, 108,591. Belton, P. S . , in Chan, H. W-S., Editor, “Biophysical Methods in Food Research,” Blackwell, Oxford, 1984, p. 123. Poulet, P., Chambron, J., and Unterreiner, R., J. Appl. Phys., 1976, 105, 1076. Paper A7118 Received January 22nd, 1987 Accepted March 16th, 1987

ISSN:0003-2654    

DOI:10.1039/AN9871201117

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, AUGUST 1987, VOL. 112 1121 Spectrophotometric and Analogue Derivative Spectrophotometric Determination of Trace Amounts of Iron Using Sulphonated 5-(3,4-Di hydroxypheny1)-I 0,15,2O=triphenylporphine Hajime lshii and Katsunori Kohata Chemical Research Institute of Non-aqueous Solutions, Tohoku University, Katahira, Sendai, 980, Japan Sulphonated 5-(3,4-dihydroxyphenyl)-lO,I 5,20-triphenylporphine (SDHP), a water-soluble porphyrin with the ability to complex with metal ions not only inside but also outside the porphine ring, has been synthesised. The analytical applications of the porphyrin, its chromogenic properties and reactivity with metal ions have been investigated. It was found that SDHP forms complexes with metal ions classified as so-called hard acids, such as aluminium(lll), iron(ll1) and tungsten(Vl), and also with those classified as soft or border-line acids, such as cadmium(ll), copper(l1) and lead(ll), all these complexation reactions proceeding rapidly at room temperature.As an example, the iron(ll1) - SDHP system has been studied in detail spectrophotometrically. SDHP reacts with iron(ll1) to form a 1 : 1 (metal : ligand) complex with a Soret band at 430 nm at pH 4 and a 1 : 2 complex with a Soret band at 415 nm at pH 7. Spectrophotometric and analogue derivative spectrophoto- metric methods are proposed for the determination of iron at ng ml-1 levels utilising these complexations. The apparent molar absorptivities were 1.09 x lo5 and 5.37 x lo5 I mol-1 cm-1 for the methods utilising the 1 : 1 and 1 : 2 complex formations, respectively.Keywords: iron determination; sulphonated 5-(3,4-dih ydroxyphen yl)- 10,15,20-triphen ylporphine; spectrophotometry; analogue derivative spectrophotometry Water-soluble meso-substituted porphyrins such as 5,10,15,20- te trakis(4-~arboxyphenyl)porphine, 5,10,15,20- tetrakis(4-sulphophenyl)porphine [T(4-SP)P] ,2,3 5,10,15,20- tetrakis( l-methylpyridinium-3-yl)porphine4-6 and 5,10,15,20- tetrakis( l-methylpyridini~m-4-y1)porphine~~7J3 are very useful as highly sensitive colour reagents for metal ions because they possess Soret bands with extremely large molar absorptivities (1 X 105-6 x 105 1 mol-1 cm-1). However, in general, the complexation reaction of the porphyrins with metal ions in aqueous media is very slow at room temperature.Hence several attempts have been made to accelerate it, including heating,l-3 the addition of an auxiliary complexing agent such as pyridine or imidazole4 or a reducing agent such as hydroxylamine or ascorbic acid5 and the utilisation of the metal-substitution reaction of the cadmium, lead or mercury(I1) - porphyrin complex.6 These attempts were fairly effective for the acceleration of the complexation reaction of the porphyrin with metal ions such as cobalt(II), copper(II), manganese(II), palladium(I1) and zinc(II), but were less effective or ineffective for that with many other metal ions. Hence we began to synthesise a series of water-soluble porphyrins with functional groups outside the porphine ring in order to improve the reactivity of the porphyrin and extend its analytical use.Sulphonated 5-(3,4-dihydroxyphenyl)- 10,15,20-triphenylporphine (SDHP or H8L) was the first of these porphyrins to be synthesised. In this paper an outline of the synthesis of SDHP, its chromogenic properties and its application to the spectro- photometric determination of trace amounts of iron are described. Experimental Reagents All reagents used were of analytical-reagent grade unless stated otherwise. All solutions were prepared with distilled, de-ionised water. SDHP solution. A 3 x 105 M aqueous solution was prepared using SDHP synthesised as described later. Its concentration was determined by photometric titration with a standard copper( 11) solution. Standard iron(III) and copper(I1) solutions. The solutions (1 x 10-2 M) were prepared from ammonium iron(II1) sulphate and copper(I1) sulphate pentahydrate, respectively.Working solutions were prepared by appropriate dilution. Apparatus A Hitachi 139 spectrophotometer and a Hitachi 556 dual- wavelength spectrophotometer were used for absorbance measurements and measurement of the absorption spectrum, respectively, the latter being used as an ordinary double-beam spectrophotometer throughout the measurements. In order to obtain the derivative spectrum, a modified Hitachi 200-0576 derivative unit composed of two analogue differentiation circuits was connected between the latter spectrophotometer’s output and a Hitachi 057 X - Y recorder input. The details of this apparatus and the principle and characteristics of the analogue derivative spectrophotometry have been described previously.6>9 Proton magnetic resonance spectra were recorded on a JEOL JNM-PS-100 NMR spectrometer and a Bruker CXP- 300 pulse Fourier transform NMR spectrometer. The chem- ical shifts are given in 6 values (p.p.m.) from tetramethylsilane (TMS) in CDC13 and 5-(trimethylsily1)propionic acid sodium salt (TSP) in D20.Infrared spectra were obtained on a JASCO A-3 IR spectrometer. Mass spectra were obtained on a JEOL DX-300 instrument using the fast atom bombardment (FAB) method. Thin-layer chromatography (TLC) was carried out on 0.2-mm E. Merck 60F-254 pre-coated silica gel plates. For column chromatography, E. Merck silica gel 60 (70-230 mesh) was used. Recommended Procedures for the Determination of Iron Ordinary spectrophotometry in acidic media (Procedure A ) To an aliquot containing up to 4 pg of iron(II1) in a 25-ml calibrated flask, add 4 ml of 3 x 10-5 M SDHP solution and 2 ml of 1 M acetate buffer (pH 4.1) and dilute to the mark with water.Measure the absorbance of the resultant solution at 442 nm against a reagent blank prepared under the same conditions using 1-cm cells.1122 ANALYST, AUGUST 1987, VOL. 112 Ordinary spectrophotometry in neutral media (Procedure B ) To an aliquot containing up to 1.3 pg of iron(II1) in a 25-ml calibrated flask, add 3 ml of 3 x 10-5 M SDHP solution and 2 ml of 0.2 M imidazole buffer (pH 7.2) and dilute to the mark with water. Measure the absorbance at 415 nm in the same manner as Procedure A. Second-derivative spectrophotometry When the iron content of the coloured solution prepared by Procedure A is too low to give a measurable absorbance, record the second-derivative spectrum from 500 to 360 nm against a reagent blank using a combination of both first- and second-order differentiation circuits (No.6 of reference 9) and a scan speed of 150 nm min-1 and measure the second- derivative value (the vertical distance from a peak to a trough or that from the base line to a peak). Procedure for the Synthesis of SDHP SDHP was obtained in two steps, the synthesis of 5-(3,4- dihydroxyphenyl)-l0,15,20-triphenylporphine (DHP) and its sulphonation. Synthesis of DHP Benzaldehyde (10.6 g, 0.1 mol) and 3,4-dihydroxybenzal- dehyde (6.9 g, 0.05 mol) were dissolved in hot (130 "C) propionic acid (490 ml). The mixture was heated to reflux and a solution of pyrrole (10.1 g, 0.15 mol) in propionic acid (10 ml) was added as rapidly as possible, caution being taken to prevent excessive heating as the reaction is exothermic. The resulting black solution was refluxed for 1 h and then cooled to room temperature. The mixture was allowed to stand in a refrigerator for several days and then filtered with suction to give a black - purple tarry matter.This is a mixture of 5,10,15,20-tetraphenylporphine (TPP) and its mono-, di-, tri- and tetra-substituted derivatives, which have one to four 3,4-dihydroxyphenyl groups in their 5-, lo-, 15- and/or 20-positions [RF values in TLC: TPP, 1.0; mono-, 0.29 in benzene - diethyl ether (9 + 1); di-, 0.58; tri-, 0.44; tetra-substituted derivative, 0.10 in chloroform - ethyl acetate (1 + l)].This mixture was dissolved in a minimum amount of methanol - chloroform, loaded on to a silica gel column (40 X 5 cm i.d. in cyclohexane) and then eluted with benzene. The first band contained TPP, which was confirmed by the following characteristic properties: h,,,, (CHC13) 647, 591, 550, 514, 420 nm; Y,. (KBr) 1590, 1470, 1440, 1350, 1060, 960, 760, 690 cm-1. These are identical with those of TPP independently prepared. After the eluate had become very light red in colour, the column was eluted with benzene - diethyl ether (9 + 1). The conclusion that the second band consisted of the target porphyrin, DHP, was based on the following visible, IR, NMR and mass spectra: h,,,, (CHCl3) 652, 595, 552, 516, 422 nm; Y,,,, (KBr) 3300-3500(m), 2900(w), 1590(s), 1500(m), 1460(s), 1430(s), 1340(s), 1260(s), 1240(s), 1180(s), 1150(sh), 1100(m), 960(s), 930(m), 790(s), 720(s), 690(s) cm-1; 1H NMR (CDC13) 6 (p.p.m.) 7.6-7.9 (m, 10H, phenyl), 8.0-8.4 (m, 8H, phenyl), 8.8 (s, 8H, pyrrole); FAB - MS, m / z (relative intensity) 648 (19, M+ + 1), 647 (47, M+), 646 (100, M+ - 1).Fractions showing these characteris- tic properties were collected, evaporated and dried in vacuo to yield purple crystals (610 mg, yield 3.8%). Sulphonation DHP (0.3 g, 0.46 mmol) and concentrated sulphuric acid (25 ml) were thoroughly mixed in an agate mortar with a pestle. The paste was transfered into a beaker and heated on a steam-bath for 5 h, then cooled in an ice-bath. About 50 ml of iced water was added slowly to precipitate green solids, which were filtered with suction and washed well with acetone.This product was subjected to column chromatography on silica gel. Elution with propan-2-01 - ethyl acetate - water (2 + 3 + 2, V/V) gave only one band, fractions of which were collected, evaporated and dried in vacuo to yield purple crystals (120 mg, yield 28.2%) with the following properties: RF 0.24 [propan-2- 01 - ethyl acetate - water (2 + 3 + 2, V/V)]; Y, (KBr) 320O-3500( s), 1620(s) , 1590(s) , 1550( s), 1450(s), 1430(sh), 1400(sh), 1340(w), 1030-1200(s), 1030(s), lOOO(s), 970(m), 790(s), 730(s), 620(s) cm-1; 1H NMR (DzO) 6 (p.p.m.) 8.1-8.2 (m, lH), 8.3-8.5 (m, 7H), 8.6-8.9 (m, lOH), 8.9-9.1 (s, 4H). Neither elemental analysis nor the mass spectrum of the product obtained here gave reliable results because the product has a high molecular mass and contains sulphonic acid groups.Hence the positive identification of this product as the desired porphyrin, SDHP, was carried out based on the following TLC tests. Three spots were detected by a TLC test, with propan-2-01 - ethyl acetate - water (2 + 3 + 2,V/V) as a developing solvent, on a reaction mixture sulphonated for 2 h and their RF values almost agreed with those of the di-, tri- and tetra-sulphonated derivatives of TPP (0.65, 0.48 and 0.24, respectively) obtained under similar sulphonation conditions. As the incorporation of a small diphenolic moiety is con- sidered to have almost no influence on the RF value in a high molecular mass compound such as SDHP (molecular mass 966.99), the three spots detected are considered to correspond to di-, tri- and tetra-substituted derivatives with 2-4 sulphonic acid groups in DHP.In contrast, the spots which correspon- ded to di- and tri-substituted derivatives completely disap- peared and only one spot, which corresponds to the tetra substituted derivative, was present in a TLC test on the product obtained when sulphonated for 5 h. On the basis of these results, the product obtained by the above procedure is presumed to be SDHP with four sulphonic acid groups. Three of the sulphonic acid groups were concluded to be at the 4-position of each phenyl group based on the analogy of T(4-SP)P,3 but the binding position in the 3,4-dihydroxyphenyl group could not be deduced. The prob- able structure of the synthesised SDHP is illustrated below.S03H Q Results and Discussion Properties of SDHP SDHP is soluble in water and stable in acidic solution, but is gradually oxidised in alkaline solution. SDHP behaves as a deca-basic acid, although the number of dissociable hydrogens in the molecule is eight. Its protonANALYST, AUGUST 1987, VOL. 112 1123 dissociation equilibria may be expressed as follows: ki HiL@ - i) - (1) where ki is the dissociation constant. kl and kz correspond to the dissociation of two pyrrole hydrogens, k3 and k4 the dissociation of two hydroxyl groups, k5 and k6 the protonation of two aza nitrogens and k7, k8, k9 and klo the dissociation of four sulpho groups. H4L4-, H5L3- and H6L2- correspond to the so-called free base, mono- and diprotonated forms, respectively. It was attempted to determine these proton dissociation constants spectrophotometrically. However, the only results obtained were that values of pk5 and pkb are very close together (around 5.2) and that pk4 is 8.52 at 20 "C, at an ionic strength of 0.1 (KN03).Their values could not be accurately determined because no clear spectral change corresponding to each individual proton dissociation was observed (i. e., the spectral change was as if one-step dissocia- tion occurred). The values of k l , k2 and k3 were not obtained because SDHP was not stable enough in alkaline solution to permit their exact determination. Fig. 1 shows the absorption spectra of SDHP at various pH values. Spectra A and B correspond to H6L2- and H4L4-, and Hi - 1L(9 - 4- + H+ (i = 1, 2 - - -, 10) 0.6 1 : I - c 0 (D n 2 8 0.4 +- 1 W C (D e 2 0.2 a n U (D 0.04 - W C (D e 0.02 4 a 0 " 400 500 600 700 Wavelengthlnm Fig.1. Absorption spectra of SDHP at various pHs. SDHP, 1.9 X H: A, 1.3-3.5; B, 6.9-7.6; C, 9.1; D, l l . S l 2 . 1 ; and E, 13.0-13.9 M; reference, water. the Soret bands, which are useful for the determination of trace amounts of metals, are at 438 and 415 nm, respectively. At pH values greater than 10 the Soret band undergoes a slight blue shift. Reactivity of SDHP with Metal Ions The reaction of SDHP with metal ions was analogous to that of Tiron (1,2-dihydroxybenzene-3,5-disulphonic acid), which has the same functional groups as SDHP. It formed complexes not only with cadmium(II), cobalt(II), copper(II), lead(II), manganese(II), mercury(I1) and zinc(I1) but also, as expec- ted, with gallium(III), molybdenum(V1) and tungsten(V1) around pH 4, iron(III), vanadium(1V) and vanadium(V) above pH 4, aluminium(II1) and titanium(1V) in the pH range 6-7 and rare earth ions in the pH range 6.5-8, at room temperature.As the complexes with the first seven of these ions gave absorption spectra similar to those of their com- plexes with porphyrins having no complexing groups outside the porphine ring, e.g., T(4-SP)P, these complexations can be assumed to occur inside the porphine ring, i.e., the metal ion seems to coordinate to four (or two) nitrogens as usual. The absorption spectra of some of these complexes are shown in Fig. 2. On the other hand, the complexations of the other eight ions and the rare earth ions seem to occur outside the porphine ring, i.e., the metal ion seems to coordinate to two oxygens, because (a) these ions, which either scarcely react or do not react at all at room temperature with the porphyrin with no complexing groups outside the porphine ring, rapidly formed complexes with SDHP at room temperature and (b) the spectral change of SDHP on complexation with these ions differed from that when the metal ion coordinates to four nitrogens inside the porphine ring. As an example, the spectral changes in the complexations with gallium(II1) at pH 3.8 and aluminium(II1) at pH 6.1 are shown in Figs.3 and 4, respectively. In the former the Soret band shifts from 438 to 428 nm and the Q band from 653 to 680 nm on complexation.On the other hand, in the latter complex the Soret band at 415 nm does not shift and only the decrease in its absorbance is observed on complexation. As for the Q band, its shape changes from four characteristic peaks to three rounded humps with an increase in absorbance. These two kinds of spectral changes in the complexations differ considerably from those seen when the metal ion forms 0.8 -0 0 n 5 0.6 0 52 ,-\ HeL2- W I T ( X f ) z a C (D 0.4 n 0.2 --_/--- 0 400 500 600 Wavelengthlnm - -0 C 0.06 2 cl W C (D n 0.04 2 0.02 1 1 Fig. 2. H4L4-, 6.9; complexes, 9.0. Reference, water Absorption spectra of SDHP and its metal complexes. SDHP, 2.2 X M; complexes, 2.2 X M. pH: H6L2-, 2.4;1124 ANALYST, AUGUST 1987, VOL. 112 - c -0 0.08 p n a 2 0.4 +d E I r n a a C c m - 0.04 2 s1 f s1 n n a a 0.2 n n " 400 500 600 Wavelengthhm 700 Fig.3. Absorption spectra of Ga"' - SDHP system. SDHP, 1.8 X 10-6 M; pH, 3.8; reference, water. Arrows indicate spectral trends observed when changing the SDHP to Ga ratio from 1 : 0 to 1 : 6.4 0.6 1 V 400 500 600 700 Wavelengthhm Fig. 4. Absorption spectra of A P - SDHP system. SDHP, 1.5 x M; pH, 6.1; reference, water. Arrows indicate spectral trends observed when changing the SDHP to A1 ratio from 1 : 0 to 1 : 4.7 a complex by coordinating to four nitrogens inside the porphine ring and are characteristic of the complexation in which the metal ion coordinates with functional groups outside the porphine ring (two oxygens in this instance). In addition, this spectral behaviour scarcely depended on the kind of metal ions, i.e., similar spectral changes to those in Figs.3 and 4 were also observed in the complexation with iron(III), molybdenum(V1) and tungsten(V1) at pH 4 and with titanium(1V) at pH 7, respectively. Hence it is finally concluded that for the complexation between SDHP and metal ions, ions classified as so-called hard acids form complexes by coordinating to the two oxygens outside the porphine ring and those classified as soft or border-line acids form complexes by coordinating to four (or two) nitrogens inside the porphine ring. Thus, the reactivity of the porphyrin can be remarkably improved by synthesising SDHP, a water-soluble porphyrin which has complexing groups outside the porphine ring. This suggests that the analytical use of the porphyrin can be considerably extended.In the subsequent work fundamental conditions for the spectrophotometric and analogue derivative spectrophoto- metric determination of iron at p.p.b. levels were carried out as an example of the application of the complexation reaction. Study of the Iron(II1) - SDHP System Absorption spectra Iron(II1) reacts with SDHP to form a 1 : 1 (metal: ligand) complex in the pH range 3-5.5 and a 1 : 2 complex in the pH range 5.5-9, as mentioned later. Both complexation reactions 0.1 0 A 2 -0.1 m n 2 e a c g -0.2 a C m o 0.5 D 0.4 0.3 0.2 0.1 0 t A 0.02 0.00 -0.02 4 I 0.10 0.08 0.06 0.04 0.02 400 500 600 700 Wavelengthhm Fig. 5. Absorption spectra of Fe"' - SDHP system at pH 4.1. SDHP, 2.0.X 10-6 M. Reference: (a), reagent blank; and (b), water.Arrows indicate spectral trends observed when changing the SDHP to Fe ratio from 1 : 0 to 1 : 4.8 proceed rapidly at room temperature. The spectral changes in the complexations between iron(II1) and SDHP at pH 4.1 and 7.2, which are shown in Figs. 5 and 6, respectively, are almost the same as those shown already in Figs. 3 and 4. At pH 4.1 the Soret band shifts from 438 to 430 nm and the Q band from 652 to 676 nm on complexation. The absorption spectra of the complex measured against a reagent blank give maxima at 426 and 694 nm and minima at 442 and 650 nm, 442 nm being the most preferable for the iron determination at this pH as it gives the highest sensitivity. At pH 7.2 the Soret band at 415 nm does not shift and only the decrease in its absorbance is observed by the complexation.The Q band changes in shape from four characteristic peaks to three rounded humps with the increase in absorbance. Stability of the absorbance The stability of the absorbance of the system in weakly acidic and neutral media was studied. The colour developedANALYST, AUGUST 1987, VOL. 112 1125 0.02 0.01 0 - -u (0 n 9 2 s I n a a C 0.04 0.02 0 400 500 600 700 Wavelengthhm Fig. 6. Absorption spectra of FeI" - SDHP system at pH 7.2. SDHP, 2.0 X 10-6 M. Reference: (a), reagent blank; and ( b ) , water. Arrows indicate spectral trends observed when changing the SDHP to Fe ratio froml:Oto1:1.7 -0.4 -0.3 a C m 9 -0.2 s n a -0.1 A 2 3 4 5 6 7 8 9 PH 0 Fig. 7. Effect of pH. SDHP, 3.5 x 10-6 M. Fe: A, 95.9; and B and C, 38.4 ng ml-1.Wavelength: A, 442; and B and C, 415 nm. Reference, reagent blank; B, imidazole buffer used; C, borate or phosphate buffer used immediately at room temperature in both media. The absorbance measured against the reagent blank remained almost constant for at least 2 h, but after that time it tended to decrease gradually. The oxidation of SDHP in alkaline media could be considerably reduced by shutting out the light. Effect of pH Fig. 7 shows the effect of pH on the formation of the iron(II1) - SDHP complex. In the pH range 2.6-5.5 a complex with a Soret band at 430 nm is formed, the absorbance of the complex solution measured against a reagent blank at 442 nm being almost independent of pH at 3.64.2. Above pH 5.5 another complex with the Soret band at 415 nm is formed, a constant absorbance being obtained at pH 6.4-7.5 when an imidazole - nitric acid buffer was used.However, when phosphate or borate buffer was used, the absorbance of the complex solution changed each time it was measured, probably owing to hydrolysis (see line C); the use of imidazole buffer is recommended in this pH region. -0.05 o*oolr----; -0.10 (D 9 0, 2 -0.15 -0.20 -0.25 A 0 - B 0.15 Q, (D 0.10 2 + s 0.05 9 0 0 1 2 3 4 5 [Fe]/[SDHPl Fig. 8. Molar ratio plots. pH, 4.1; SDHP, 2.0 X 10-6 M; and reference, reagent blank. Wavelength: A, 426; and B, 442 nm 0.10 r I I I 0.05 a C m 9 0 v) 9 -0.2 -0.4 -0.6 ' I I 1 I 0 0.2 0.4 0.6 0.8 1.0 [Fel/([Fel + [SDHP]) Fig. 9. Job plots. pH, 7.2; [Fe] + [SDHP] = 3.5 X reference, reagent blank. Wavelength: A, 434; and B, 415 nm M; and Effect of SDHP concentration The effect of SDHP concentration was investigated at pH 4.1 and 7.2 by measuring the absorbance at 442 and 415 nm, respectively, of solutions containing a fixed amount (76.8 and 38.4 ng ml-1, respectively) of iron(II1) and various amounts of the reagent.The results revealed that a constant, maximum absorbance is obtained, provided that 1.5-4 and 2.9-4.5 molar excesses of SDHP are added at pH 4.1 and 7.2, respectively, and in both instances a further excess of the reagent tended gradually to decrease the absorbance. Composition of the complex The composition of the complex was studied at pH 4.1 and 7.2 by Job's method of continuous variations and the molar-ratio method. Both methods indicated that the ratio of metal ions to ligand molecules was 1 : 1 at pH 4.1 and 1 : 2 at pH 7.2.Some of the results obtained are shown in Figs. 8 and 9. Calibration graph, sensitivity and precision Linear calibration graphs through the origin were obtained using the recommended procedures. The equations of the lines obtained by a least-squares treatment were Fe (ng ml-1) = 514A in procedure A and Fe (ng ml-1) = 104A in Procedure B, where A is the absorbance. The optimum ranges for the determination of iron, the sensitivities for an absorbance of 0.001 and the molar absorptivities calculated from the above1126 ANALYST, AUGUST 1987, VOL. 112 1.0 1 bd Table 1. Tolerance limits for foreign ions in the determination of 76.8 ng ml-1 of iron at pH 4.1 0, - 0.8 0, m .- c. .- 0.6 ? U 0 $ 0.4 cn 0.2 0 5 10 15 20 Iron/ng ml-1 Fig.10. Calibration graph for iron in second-derivative spectropho- tometry. Fe: a, 1.9; b, 3.8; c, 11.5; d , 19.2 ngml-1. SDHP, 1.1 X M; circuits, all No. 6 of reference 9; scan speed, 150 nm min-l; slit width, 1 nm; recorder sensitivity, x i ; cells, 10 mm; and reference, reagent blank. A, Peak to trough values plotted; and B, base line to trough values plotted equations were 0.24.0 yg, 0.514 ng cm-2 and 1.09 X 105 1 mol-1 cm-1 in Procedure A and 0.1-1.3 pg, 0.104 ng cm-2 and 5.37 x l o 5 1 mol-1 cm-1 in Procedure B, respectively. Two series of ten standard solutions each containing 1.9 and 0.96 pg of iron were analysed by the recommended pro- cedures. The results gave relative standard deviations of 0.9 and 1.0% in procedures A and B, respectively.Effect of foreign ions In order to study the effect of various ions on the determina- tion of iron, a fixed amount of iron(II1) was taken with different amounts of foreign ions and the recommended procedures were followed. Tolerances for foreign ions in procedure A (at pH 4.1) are summarised in Table 1. Fairly large amounts of cadmium(II), cobalt(II), chromium(III), lead(II), magnesium(II), manganese(II), nickel(II), thorium(1V) and zinc(I1) can be tolerated, but the interfer- ence of numerous foreign ions restricts the applicability of the reagent, the most important being aluminium(III), cop- per(I1) , molybdenum(V1) , tungsten(V1) , vanadium(1V) and vanadium(V), which also react with SDHP to give complexes, and their Soret bands show maxima at wavelengths close to that of the iron(II1) complex.No effective, appropriate masking agent for these interfering ions could be found. Interferences in Procedure B (at pH 7.2), which are not shown here, were severe. The interferences by molybdenum(V1) and tungsten(V1) disappeared, but many transition metal ions interfered. Hence the determination at pH 7.2 is very sensitive, but less selective, prior separation being necessary for the application to real samples. Analogue Derivative Spectrophotometry to Improve Sensitivity Procedure A, utilising the 1 : 1 complex formation, was the most sensitive of the known spectrophotometric methods for iron, but less sensitive compared with Procedure B utilising the 1 : 2 complex formation. Further sensitisation was there- fore tried by introducing an analogue second-derivative spectrophotometric technique.6.9 In second-derivative spectrophotometry both the time constant of the analogue differentiation circuit and the scan Ion APT .. Cd" . . CO" . . Cr"' . . CrVI . . CUT' . . HgII . . MgII . . MnI1 . . MoV1 . . Tolerance limithg ml-1 Tolerance limithg ml-I 3% * 7.3 840t 8201. 8601- 90 55 930t 460 400 35 * Tolerable error. t Maximum tested. 5% * Ion 12.2 Ni" . . PblI . . Pd" . . ThIV . . 150 TiIV . . 65 VIV . . v v . . 770 Wvl . . 660 ZnII . . 58 3% * 5% * 8lOt 510 90 150 340 5607 90 110 12 20 7.5 12.5 9.4 15.7 630 speed of the spectrophotometer affect the sensitivity and selectivity, so both need to be optimised to give a well resolved large peak. Experimental results indicate that a combination of circuit No.6 of reference 9 (which has the largest time constant, 2.0 s, the time constant increasing with circuit number in our apparatus) and a scan speed of 150 nm min-l [or that of No. 5 (time constant: 0.96 s) and 300 nm min-11 give the best sensitivity and resolution (i.e., selectivity) for the iron determination. The calibration graph prepared under the recommended conditions by plotting the second-derivative value versus the iron concentration was linear and passed through the origin when either the peak to trough values or the base line to trough values were plotted (shown in Fig. 10). The equations for the lines were Fe (ng ml-1) = 21.70, and Fe(ng ml-1) = 32.30, where 0 is the second-derivative value converted into absorbance. It will be seen that iron can easily be determined down to 1.9 ng ml-1 in this manner. Conclusions SDHP has been synthesised and its properties and reactivity with metal ions investigated spectrophotometrically. It was found that analytical use of the porphyrin can be extended by introducing complexing groups into the substituent outside the porphine ring. Extremely sensitive spectrophotometric and analogue derivative spectrophotometric methods for the determination of iron have been proposed. 1. 2. 3. 4. 5. 6. 7. 8. 9. References Ishii, H., Koh, H., and Okuda, Y., Nippon Kagaku Kaishi, 1978, 686. Koh, H., Kawamura, K., and Ishii, H., Nippon Kagaku Kaishi, 1979, 591. Ishii, H., and Koh, H., Nippon Kagaku Kaishi, 1978, 390. Ishii, H., and Koh, H., Talanta, 1977, 24,417. Ishii, H., and Koh, H., Bunseki Kagaku, 1979,28, 473. Ishii, H., and Koh, H., Nippon Kagaku Kaishi, 1980, 203. Igarashi, S., Kobayashi, J., Yotsuyanagi, T., and Aomura, K., Nippon Kagaku Kaishi, 1979, 602. Igarashi, S., Yotsuyanagi, T., and Aomura, K., Nippon Kagaku Kaishi, 1981, 60. Ishii, H., and Satoh, K., Fresenius Z . Anal. Chem., 1982,312, 114. Paper A61304 Received September lst, 1986 Accepted February 23rd, I987

ISSN:0003-2654    

DOI:10.1039/AN9871201121

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

ANALYST, AUGUST 1987, VOL. 112 1127 Spectrophotometric Determination of Iron in Boiler and Well Waters by Flow Injection Analysis Using 2-Nitroso-5-( N-propyl-N=sulphopropylamino)phenol Noriko Ohno and Tadao Sakai* Department of Chemistry, Asahi University, 185 1 Hozumi, Hozumi-cho, Gifu 50 1-02, Japan 2-Nitroso-5-( N-propyl-N-sulphopropylamino)phenol (nitroso-PSAP) forms a water-soluble chelate with iron. The molar absorptivity of the complex is 4.3 x lo4 I mol-1 cm-1 at 753 nm. Iron (4-100 pg I-') is easily and selectively determined in a flow injection system because divalent transition metals (Ni, Co, Cu, Zn and Cd) at the 500 pg 1-1 level do not interfere. The method has an average recovery error of +1% and the sampling rate is 25 samples h-1. A simple procedure for the determination of iron in boiler and well waters is described.Keywords: Iron determination; flow injection analysis; spectrophotometry; water analysis Nitrosophenol and nitrosonaphthol derivatives have been synthesised as selective and sensitive photometric reagents for metals such as cobalt,' iron,2 palladium3 and nickel.4 These complexes are designed for use with solvent extraction and the sensitivity for some metals is excellent. Tsurubo and Sakai5 reported that iron(I1) forms a complex cation with 3-(2- pyridyl)-5,6-diphenyl-l,2,4-triazine (PDT) in aqueous solu- tions which can be extracted into 1,2-dichloroethane as an ion associate with tetrabromophenolphthalein ethyl ester (TBPE). The calibration graph was linear over the range 0-0.25 mg 1-1 of iron.However, in order to eliminate the solvent extraction process, water-soluble reagents such as 2-nitroso-5-(N-ethyl-N-sulphopropylamino)phenol (nitroso- ESAP) and 2-nitroso-5-(N-propyl-N-sulphopropylamino)- phenol (nitroso-PSAP) have been synthesised by Saito et al.6 These water-soluble reagents are preferable for the determi- nation of trace amounts of metals in a flow injection system. In an earlier paper,7 we reported the determination of iron with nitroso-ESAP and Wada et a1.8 reported the determination of iron with 2-(3,5-dibrorno-2-pyridylazo)-5-[N-ethyl-N-(3-~~1- phopropyl)amino]phenol (3,5-diBr - PAESPAP) using flow injection analysis. These reagents are very selective, but are not sensitive enough for the analysis of water samples. This paper reports the highly sensitive and selective determination of iron in boiler and well waters using nitroso- PSAP in a flow injection analysis system.Experimental Reagents All reagents used were of analytical-reagent grade. Standard iron(II) solution. Iron( 11) ammonium sulphate was dissolved in 0.1 M sulphuric acid to give a 1 g 1-1 stock solution of iron(II), which was standardised by complex- ometry. The stock solution was diluted to an appropriate concentration as required after adding 0.05 M sulphuric acid. 2-Nitroso-5-(N-propyl-N-sulphopropylamino)phenol (ni- troso-PSAP) solution, 2 x 10-4 M. Prepared by dissolving 0.0302 g of nitroso-PSAP (Dojindo Laboratories, Japan) in 500 ml of 5 x 10-3 M sulphuric acid. Sodium ascorbate solution, 1 X 10-3 M. Prepared by dissolving 0.099 g of sodium ascorbate in 500 ml of distilled water.Buffer solutions. Buffer solutions (pH 9-12) were made from equal volumes of 0.15 M potassium dihydrogenphosphate and 0.05 M sodium borate, the pH being adjusted with 1 M sodium hydroxide solution. * To whom correspondence should be addressed. Apparatus A Hitachi Model 556 double-beam spectrophotometer with 10 mm quartz cells and a Hitachi Model 057 X - Y recorder were used for absorption spectra. A Hitachi-Horiba Model F-711 pH meter with a glass electrode was used to measure the pH. General Procedure (Batch Method) Mix 3 ml of sample solution containing up to 5 mg 1-1 of iron, 2 ml of 1% sodium ascorbate, 1 ml of 4 x 10-3 M nitroso-PSAP solution and 5 mi of phosphate buffer solution (pH 8.0) in a 25-ml calibrated flask, dilute to the mark with distilled water and shake thoroughly.Measure the absorbance at 753 nm in a 10-mm cell. Flow Injection Procedure The manifold of the flow injection system used is shown in Fig. 1. Two double plunger micro-pumps (Sanuki Kogyo Model DM2U-1026, Japan) are used to pump the solutions. The reagent solution containing 5 x 10-3 M sulphuric acid and the carrier solution of sodium ascorbate are pumped at a flow-rate of 0.8 ml min-1 by pump A and the samples (120 pl) containing up to 100 pg 1-1 of iron are injected into the carrier stream by a six-way injection valve to which a volume control loop is attached. The sample and reagent are mixed in the 150-cm reaction coil. After adding the buffer solution (pH 10.5, 0.35 ml min-1) via pump B, the colour is developed in the 200-cm coil.The absorbance of the complex is monitored by a spectrophotornetric detector (Japan Spectroscopic, Model UVIDEC-100-VI) fitted with a micro flow cell (8 pl, 10 mm path length). The absorbance measured at 753 nm is recorded as peak-shaped signals. A Toa Electronics Model FBR-251A recorder is used. The PTFE tubing is of 0.5 mm i.d. except for the back-pressure coil which is 0.25 mm i.d. ( 5 m long). - RS-Pp, I CS-- Waste S U Fig. 1. Schematic diagram of flow system. CS, Carrier solution (1 x M nitroso-PSAP); BS, buffer solution, pH 10.5; PA and PB, pump flow-rates (PA 0.8 ml min-1, PB 0.35 ml min-1); R1 and R2, reaction coils (Rl, 0.5 mm i.d. X 150 cm, R2, 0.5 mm i.d. x 200 cm); I, sample injector; S, sample (120 PI); BPC, back-pressure coil (0.25 mm i.d. X 500 cm); D, spectrophotometric detector; Rec, recorder M sodium ascorbate solution); RS, reagent solution (2 x1128 ANALYST, AUGUST 1987, VOL.112 Results and Discussion Absorption Spectra and pH of Formation of Iron(I1) Chelate Compounds A 1.5-ml portion of a 10 mg 1-1 iron(I1) standard solution was placed in a 25-ml calibrated flask and the reagents were added. The mixture was diluted according to the general procedure described under Experimental. After standing for 5 min, the absorption spectra and pH on formation of the chelate were measured against a reagent blank. Iron(I1) reacted with nitroso-PSAP to form a chromogenic chelate compound, which had absorption maxima at 425 and 753 nm. The molar absorptivities of the iron chelate compound were 13400 1 mol-1 cm-1 at 425 nm and 43000 1 mol-1 cm-1 at 753 nm.However, other metals such as cobalt(II), copper(I1) and nickel reacted with nitroso-PSAP to form complexes which had absorption maxima at 390-450 nm. As a result, iron was determined sensitively and selectively at 753 nm. Fig. 2 shows the effect of pH on the formation of the iron complex by the batch method. Graph 1 was obtained when the reagents were added in the order iron, sodium ascorbate, buffer solution and nitroso-PSAP. The absorbance decreased considerably in the pH range 8-9, but at higher pH (9.7-10.3) maximum and constant absorbance was obtained after 1 h. The effect may be attributed to the dissociation constant of nitroso-PSAP (8.4) ,9 the hydrolysis product or hydroxo complex formed and the very slow reaction rate of complex formation in the above pH range.At 80 "C the maximum absorbance was quickly obtained even at pH 8-9. As a result, the order of reagent addition was seen to be important for the rapid formation of chelate compounds. Effect of Nitroso-PSAP Concentration in FIA System The effect of the nitroso-PSAP concentration on the colour development was studied for 50 and 100 pg 1-1 of iron(I1). The nitroso-PSAP concentration was varied from 1 X 10-5 to 3 x 10-4 M. The peak signals were maximum and constant at nitroso-PSAP concentrations above 1 x 10-4 M, as shown in Fig. 3. In this work, a 2 X M nitroso-PSAP solution was used. Effect of pH on Iron Complex Formation in FIA System The dependence on pH of the complex formation with nitroso-PSAP was investigated (Fig. 4).The peak signals were maximum and constant in the pH range 9-12. The optimum pH range was very wide and acceptable. Above pH 12, the peak signal decreased because the iron was hydrolysed. The pH of the buffer solution in the FIA system was adjusted to 10.5. \ /' \ \ I I I I 7 8 9 10 11 0.2 L I 6 PH Fig. 2. Effects of pH and order of reagent addition on formation of iron chelate compounds by the batch method. (1) Iron + sodium ascorbate + buffer + nitroso-PSAP, absorbance measured after 1 h; (2) iron + sodium ascorbate + nitroso-PSAP + buffer, absorbance measured after 5 min. Iron concentration, 0.6 mg I - l ; sodium ascorbate concentration, 0.08% ; nitroso-PSAP concentration, 1.6 X M; wavelength, 753 nm However, the complex was quickly formed without hydroly- sis when the reagents were added in the order iron, sodium ascorbate, nitroso-PSAP, buffer solution, in the pH range 6.4-10.3 (Fig.2). Accordingly, as can be seen in Fig. 1, the buffer solution was added after mixing the chromogenic reagent and samples. The iron(I1) - nitroso-PSAP chelate compound was stable for at least 3 h. Effect of the Reaction Coil Lengths The effect of the lengths of the reaction coils (R, and R2 in Fig. 1) was examined and the R1 coil length was varied in the range 50-400 cm. The reagent and iron(I1) react to form the coloured chelate compound in R1 (0.5 mm i.d.). The largest peak height was obtained in the range 50-150 cm, but the reproducibility was poor when the R1 coil length was less than 100 cm.A 150-cm coil is recommended. The same flow-rates were used for both the reagent and the sample channels (0.8 ml min-1). The effect of the R2 coil length was studied over the range 100-600 cm. The peak heights decreased with increasing R2 length because of dispersion. Accordingly, a 200-cm reaction coil length was used for R2. A flow-rate of 0.35 ml min-1 was recommended and 120 pl was found to be the most suitable sample volume. Effect of Reductant Iron(II1) does not form a complex with nitroso-PSAP, but it is able to form a chelate in the presence of a reductant. The effect of the concentration of sodium ascorbate as a reductant was examined. The reduction of iron(II1) was complete above 3 x 10-4 M sodium ascorbate. Therefore, a 1 X 10-3 M sodium ascorbate solution was chosen.8ot / 2o i 1 I 1 I I ) 0 1 2 3 Concentration of nitroso - PSAPI10-4 M Fig. 3. Effect of nitroso-PSAP concentration in FIA. (1) 50 pg 1-1 iron(I1); (2) 100 kg 1-1 iron(I1). Sodium ascorbate concentration, 1 x M; pH, 10.5; wavelength, 753 nm; sample volume, 120 p1 I I I I 1 8 9 10 11 12 13 PH Fig. 4. Effect of pH on iron complex formation in FIA. Iron(I1) concentration, 50 pg 1-1; nitroso-PSAP concentration, 2 X lop4 M; sodium ascorbate concentration, 1 x 10-3 M; wavelength, 753 nmANALYST, AUGUST 1987, VOL. 112 1129 Table 1. Effect of foreign ions on the determination of 50 pg 1-l of iron(I1) Amount added/ Recovery, Amount added/ Recovery, Foreign ion Added as mg 1-1 Yo Foreign ion Added as mg 1-1 % F- KF 50 99.6 PtIV H2PtC1, 10 101.2 C1- .. . . . . KC1 50 99.0 Tartaricacid . . . . 10 98.0 Br- KBr 50 99.0 PdII PdC12 5 100.0 C03,- Na2C03 50 100.1 SbI" SbC1, 5 100.2 CH3COO- CH,COONa 50 100.0 PbII . . . . . . Pb(N03), 2.5 99.6 C104- . . . . . . NaClO, 50 98.1 25 100.0* NH, Aqueousammonia 50 100.0 BaII BaCl, 2.5 99.2 Citricacid . . , . 50 98.0 SeIV . . . . . . SeCl, 1 100.0 Lil . . . . . . LiCl 25 100.1 25 100.0* SiIV Na,SiO, 25 100.0 CO'* COCI, 1 98.7 As"' NaAsO, 25 99.8 CuII CuCI, 0.5 100.6 SrII . . . . . . SrCl, 25 99.5 5 98.91- Hg' HgN03 25 99.0 Nil' NiS04 0.5 100.0 HgII Hg(CH3C00), 25 98.6 AIIIx AIC13 0.5 98.5 Cd'I . . . . . . Cd(N03), 16 100.9 2.5 98.9* Ca" . . . . . . CaC1, 10 100.4 . . . . . . Bi(NO& 0.005 100.0 Zn" . . . . . . ZnS0, 10 99.3 25 100.0* MgI' . . . . .. MgCl2 10 101.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I- KI 50 100.4 WVI Na2W04 5 99.0 NO3- NaN0, 50 98.0 CrlI1 CrCI, 5 100.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . * 0.04% Tiron was added to the buffer solution. t 4 x 10-4 M potassium cyanate or 2 X lo-, M bathocuproinedisulphonic acid was added to the sample solution. Table 2. Determination of total iron in boiler and well waters Iron found/mg 1-1 Sample Proposed method TPTZ method* Boiler water 1 . . 0.46 0.47 Boilerwater2 . . 0.51 0.50 Boilerwater3 . . 0.07 0.06 Well water 1 . . 0.03 0.04 Wellwater2 . . 0.01 - * Spectrophotometry using 2,4,6-tris(2-pyridyl)-s-triazine (TPTZ) chelate reagent.Effect of Flow-rate The flow-rate of pump A was varied between 0.5 and 2 ml min-1 with the flow-rate of pump B constant, and also with the flow-rate of pump B varied between 0.2 and 1 mi min-1. When the flow-rate of pump A increased, the peak height of the signal became larger because of the small dispersion of the complex. When the flow-rate of pump B increased with a constant flow-rate of pump A, the peak height became smaller owing to the dilution effect on the complex. When the flow-rate of pump A was 1.6 ml min-1 and the flow-rate of pump B was 0.35 ml min-1, that is, when the ratio (PA/PB) was 4.5, the peak height was maximum and constant. When PA/PB was >5, the effect of the buffer decreased, the colour development was incomplete and double peaks appeared.Calibration Graph Fig. 5 shows the recorder traces for a series of runs in duplicate. The calibration graph was linear for 4-100 yg 1-1 of iron and the dynamic range was wide. The detection limit was 1 pg 1-l(signa1 to noise ratio = 2). The relative standard deviation was below 1% for ten runs on a 50 yg 1-1 iron sample. The sampling rate was about 25 samples h-1 in the proposed manifold. Interferences For each interference, solutions were prepared which con- tained fixed concentrations of other ions and their interfer- 0.01 absorbance Fig. 5. Flow signals for iron(1I). Iron(I1) concentration (pg 1-1) : A, 0; B, 4; C, 10; D, 20; E, 30; F, 40; G, 50; H, 80; and I, 100 ences were investigated by the recommended flow injection analysis procedure. The tolerable amounts of various ions were taken as that amount which caused an error of +2% on each peak height and the tolerance limits for various metals ions are listed in Table 1.There was no interference from 50 mg 1-1 of C1-, I-, NO3-, S042-, CH3COO-, C104- and citric acid. Most metal ions did not interfere at levels from 2.5 to 25 mg 1-1 (50-500-fold for iron). CU" and NiII at 0.5 mg 1-1 and CoII at 1 mg 1-1 had no significant effect because the absorbance was measured at 753 nm, where the absorption spectra did not overlap each other. When a 0.04% Tiron concentration was added to the buffer solution, the interfer- ence of Bi"' was reduced and the tolerance limit raised considerably for AlI", Pb" and SeIV. Moreover, 5 mg 1-1 of Cur1 (100-fold for iron) could be tolerated by the addition of potassium cyanate or bathocuproinedisulphonic acid. Conse- quently, the method was shown to be selective and sensitive for the determination of iron.Determination of Total Iron in Boiler and Well Waters Total iron in three boiler- and two well-water samples was determined by the proposed flow injection method. The results obtained are summarised in Table 2. It is important to monitor the corrosion of the boiler so that the iron in boiler1130 ANALYST, AUGUST 1987, VOL. 112 and well waters is determined. The boiler water and/or well water was heated for 30 min at 80 "C after the addition of sulphuric acid to give a final concentration of 0.05 M. A 120-yl portion of the sample solution was injected into the flow injection system. The same samples were also concentrated to one-tenth of the original volume and determined by spectro- photometry using 2,4,6-tris(2-pyridyl)-s-triazine (TPTZ, Dojindo Laboratories, Japan). The results obtained by the proposed method were in good agreement with those obtained in TPTZ batch spectro- photometry. In conclusion, micro-amounts of iron in boiler and well waters can be determined spectrophotometrically using water- soluble nitroso-PSAP. The method was applied successfully to flow injection analysis. This method is very selective for the determination of micro-amounts of iron without interference from other metal ions. References 1. Toei, K., and Motomizu, S . , Bunseki Kagaku, 1973,22, 1079. 2 . Korenaga, T., Motomizu, S . , and Toei, IS., Anal. Chim. Acta, 1973,65, 335. 3. Toei, K., Motomizu, S., and Hamada, S . , Anal. Chim. A m , 1978, 101, 169. 4. Motomizu, S . , and Toei, K., Anal. Chim. Acta, 1978,97,335. 5. Tsurubo, S., and Sakai, T . , Analyst, 1984, 109, 1397. 6 . Saito, M., Horiguchi, D., and Kina, K., Bunseki Kagaku, 1981, 30, 635. 7. Sakai, T., and Ohno, N., Bunseki Kagaku, 1984, 33, 331. 8. Wada, H., Nakagawa, G . , and Ohshita, K., Anal. Chim. Acta, 1983, 153, 199. 9. Yoshida, I., and Ueno, K., Bunseki Kagaku, 1985,34, 77. Paper A 7/44 Received February 9th, 1987 Accepted March 16th, I987

ISSN:0003-2654    

DOI:10.1039/AN9871201127

出版商:RSC    

年代:1987

数据来源: RSC