ISSN:0003-2654    

DOI:10.1039/AN99116BP011

出版商:RSC    

年代:1991

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN99116FX013

出版商:RSC    

年代:1991

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN99116BX015

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, APRIL 1991, VOL. 116 333 Application of Total-reflection X-ray Fluorescence Spectrometry to Elemental Determinations in Water, Soil and Sewage Sludge Samples Sarfraz Mukhtar and Stephen J. Haswell" School of Chemistry, Thames Polytechnic, Wellington Street, London SE 18 6PF, UK Andrew 1. Ellis Link Analytical, Halifax Road, High Wycombe, Buckinghamshire HP12 3SE, UK David T. Hawke Southern Science Ltd., Hillbarn Lane, Worthing, West Sussex BN 14 9QQ, UK Total-reflection X-ray fluorescence (TXRF) spectrometry represents a relatively new instrumental analytical technique for the determination of trace elements in liquid samples. In this study, a range of sample types common t o the water industry have been analysed for their elemental composition by TXRF and inductively coupled plasma (ICP) spectrometry.The TXRF method was found t o offer, in general, lower limits of detection than are possible with ICP spectrometry, 4-5 orders of magnitude range in calibration, a sample preparation precision of <3% and an instrumental precision of <I %. Results obtained by TXRF compared favourably with those acquired by ICP and with the reference values that were available. Preparation of the digested and aqueous-based samples for TXRF analysis offered a very simple internal standard method for calibration and required only very small volumes (=I0 PI). Keywords: Total-reflection X-ray fluorescence; inductively coupled plasma; soil; water; sewage sludge X-ray fluorescence (XRF) spectrometry has, over the past two decades, become an established non-destructive technique for multi-element determinations, applicable to a wide range of matrix types.In trace element analysis, the major disadvan- tage of conventional XRF has been the poor elemental sensitivity, which is mainly a consequence of high background noise levels, resulting from instrumental geometries and sample matrix effects.2 Total-reflection X-ray fluorescence (TXRF) spectrometry is a relatively new multi-element technique with a potential to achieve trace element determina- tions for a variety of sample types.3 The problem in detecting elements at the nanogram or sub-ppb level is basically one of being able to obtain a signal that can be clearly distinguished from the background. The detection limit is typically specified as the signal that is equivalent to three times the standard deviation of the background counts for a given unit of time.4 In XRF, the background is essentially caused by interactions of radiation with matter resulting from an intense flux of elastic and Compton-scattered photons.The background, especially in the low-energy region (<3 keV), is due mainly to Compton scatter of high-energy Bremsstrahlung photons from the detector crystal itself.5 In addition, the background might be elevated owing to impurities on the specimen support contri- buting to Compton scatter in the higher-energy region (17-20 KeV).6 The Auger effect does not contribute to an increased background, as the emitted electrons, of different but lower energy, are absorbed either in the Be foil of the detector entrance windows or in the air path of the spectrometer.6 A reduction in the spectral background can be effectively achieved by X-ray total reflection at the surface of a smooth reflector material such as quartz.If a collimated X-ray beam impinges onto the surface of a plane, smooth and polished reflector at an angle less than the critical angle, then total reflection occurs. In this instance the angle of incidence is equal to the angle of reflection, and the intensities of the incident and totally reflected beams should be equal (Fig. 1). The principles of TXRF were first reported by Yoneda and Horiuchi7 and further developed by Aiginger and Wobrau- schek.8 In TXRF, the exciting primary X-ray beam impinges at angles of incidence in the region of 2-5 minutes of arc below the critical angle on the specimen prepared as a thin film on an optically flat support. In practice, the primary radiation does not (effectively) enter the surface of the support, but skims the surface and irradiates any sample placed on the support surface.The scattered radiation from the sample support is virtually eliminated, thereby drastically reducing the back- ground noise. A further advantage of the TXRF geometry is that the solid-state energy-dispersive detector can be accom- modated very close to the sample (0.3 mm), which allows a large solid angle of fluorescent X-ray collection, thus enhanc- ing signal sensitivity and enabling the analysis to be carried out in air at atmospheric pressure. Since Yoneda and Horiuchi7 first reported the use of TXRF, various versions have been developed, in particular the use of a cut-off filter to reduce the background from X-raw / n Condensed matter * Present address: School of Chemistry, University of Hull, Hull HU6 7RX, UK.Fig. 1 Total reflection phenomena: ( a ) 8 > Ocrit; (6) 8 = ecrit; and ( c ) 8 < ecrit334 ANALYST, APRIL 1991, VOL. 116 high-energy photons (>20 keV).9-14 Recently, an X-ray generator with a fine-focus tube and multiple-reflection optics has been developed by Seifert and coupled with an energy- dispersive spectrometer fitted with an Si(Li) detector and a multi-channel analyser supplied by Link Analytical. Results obtained with such an instrument known as the EXTRA I1 are described in this paper. These results include an evaluation of the instrumentai performance and the suitability of instrument for multi-elemental quantification of trace ments in a wide range of liquid-based sample matrices.the ele- n X-ray tube Reflection unit 0 B F C D A / Experimental Instrumentation TXRF Element determinations were carried out using an EXTRA I1 TXRF spectrometer fitted with a multiple-total-reflection unit. A schematic diagram of the instrument is shown in Fig. 2. The EXTRA 11 is equipped with a Seifert (Hamburg, Germany) Model ID 3000 high-voltage generator fitted with the appropriate control and regulation hardware. The output voltage can be varied from 1 to 60 kV in steps of 1 kV, and output currents from 1 to 80 mA, in steps of 1 mA, can be selected. The total-reflection chamber has two reflector units, one optimized for Mo Ka and the other for W white spectrum radiation.The X-ray tubes were two fine-focused Model SFGO-K's (with Mo and W as anode materials), each with a maximum output of 2000 W and a line profile of 8 mm width and 0.4 mm depth. The X-ray detector and multi-channel analyser, supplied by Link Analytical (now Oxford Analytical, Abington, UK), consisted of an AN 10/55 analyser system. The Si(Li) detector crystals used had an active area of 80 mm2 with a resolution of S155 eV at 5.9 keV. The detector was fitted with a pulsed optical feedback pre-amplifier and pulse processor with dead-time correction, pulse pile-up rejection, and a 100 MHz ADC data-handling system consisting of a 512 kbyte CPU (20 MHz clock), and two floppy and one hard disk running software for auto-acquisition, data processing and instrumen- tal controls.ICP spectrometer A Jobin-Yvon JY 38 PI ICP (EDT Research) sequential- emission spectrometer, fitted with a Hook and Tucker auto-sampler, was used. The specifications of the spec- trometer were as follows. Generator. Durr JY22OOW; frequency, 55.5-56.6 MHz. Argon flow-rates. These were 18, 0.4, 0.6 and 0 dm3 min-* for plasma, coating, carrier and auxiliary, respectively. Torch. Jobin-Yvon three-piece demountable type; obser- vation height, just above the torch outer tube. Nebulizer. Glass (Meinhard), with the sample pumped at 1.0 ml min-* with use of a peristaltic pump. Monochromator. Czerny-Turner 1000M: focal length, 1 m; aperture, j76.8; master holographic grating, 2400 grooves mm-1; resolution, 0.01 nm at 200 and 400 nm; and wavelength step size, 0.002 nm.The monochromator was purged with nitrogen at a flow-rate of 18 dm3 min-1 when measurements were made below 200 nm (i.e., for determination of Hg and As). Duplicate peak scans were performed for samples and triplicate peak scans for standards. A peak area calculation mode was used for all elements except Pb, Ca and K, when peak height measurement was used. A background correction (one side of the peak) was used for Pb, and a two-sided correction was used for As and Hg. Re-calibration was performed after analysis of five samples. The choice of wavelengths used was as follows: Cd, 228.80; Cr, 267.72; Cu, 224.70; Ni, 231.60; Pb, 220.35; Zn, 213.86; As, 193.69; Hg, 194.16; K, 766.49; Ca, 317.93; Fe, 259.94; and Se, 196.03 nm.Fig. 2 Schematic diagram of the apparatus. A, Sample; B, sample support; C, diaphragm; D, reflector 1; E, reflector 2; F, primary reflection zone: G, X-ray tube anode; and H, slit-type collimator Digestion Apparatus A CEM MDS-81D microwave digestion system and a Tecam Dri-Block DB-4S electrically heated aluminium block were used. Reagents and Gases All reagents were of AnalaR or SpectrosoL grades (Fisons or BDH), and the water used throughout was distilled and doubly de-ionized. All the standards were prepared from 1000 mg dm-3 SpectrosoL solutions and were freshly made up in de-ionized water. For the inductively coupled plasma (ICP) spectrometry work, dilute solutions of metal standards were prepared by mixing 5.0 k 0.02 ml of a standard solution (in 10% v/v nitric acid) of Hg", Cd, Cu, Cr"', Pb, Ni or Zn (10.0 mg dm-3) with 50 k 1 ml of nitric acid (sp.gr. 1.41) and diluting to 500 ml in a calibrated flask. A standard solution of As"' or SelV, 10.0 mg dm-3, in 10% v/v hydrochloric acid, was similarly prepared using hydrochloric acid (sp. gr. 1.19). A mixed alkali-alkaline earth metal standard containing 200 mg dm-3 of Ca and 20.0 mg dm-3 of K was prepared by mixing 100.0 k 0.1 ml and 10.00 _+ 0.02 ml of stock Ca and K solutions, respectively, with 50 k 1 ml of nitric acid and diluting to 500 ml in a calibrated flask. Blank solutions of 10% v/v nitric and hydrochloric acids were prepared with omission of the stock metal solutions. The 10% acid concentrations were found to be suitable for the analysis of samples in both 12% nitric acid and 10% aqua regia [HCI-HN03 (3 + l)] matrices.For the analysis of liquid samples stabilized in 1% nitric acid, all of the above standard solutions and blanks were duplicated by substituting 1% for 10% acid matrix. The following Certified Reference Materials (CRMs) were used. Community Bureau of Reference (BCR) (Belgium) CRMs 143 Sewage sludge-soil; 144 Sewage sludge-domestic; and 142 Soil-light sandy. BOC 'Vac Spec' grade argon and nitrogen were used. Digestion Procedure Sludges (block digestion) A pre-weighed fraction of dried sludge (0.500 k 0.002 g) was transferred into a 50 ml borosilicate tube. De-ionized water (1.0 k 0.1 ml) was added to wet the sample, then nitric acid (6.0 k 0.1 ml) was added carefully and any initial reaction was allowed to subside.The tube was then placed in an electric heating block fitted with a polytetrafluoroethylene (PTFE) stopper and heated gently under reflux for 15 min. After cooling, the sample was transferred quantitatively through an acid-rinsed filter into a 50 ml calibrated flask, and then made up to the mark with de-ionized water. The final acid strength was 12% in nitric acid. Metal standards and blanks (1 .O ml of H20 + 6 ml of HN03 only) were subjected to the same procedure.ANALYST, APRIL 1991. VOL. 116 335 Soils, sediments (micro wave oven digestion) A pre-weighed fraction of the dried sample (0.500 & 0.002 g) was transferred into a Teflon PFA digestion vessel (acid- cleaned and dried). De-ionized water (1.0 & 0.1 ml) was added to wet the sample, then 3.75 k 0.02 ml of hydrochloric acid and 1.25 t 0.02 ml of nitric acid were.carefully added and the initial reaction was allowed to subside.The pressure-relief valve and cap were fitted and tightened to the correct torque. The following three-stage digestion programme was oper- ated. Stage 1: 10 rnin at 20% power; 2: 10 rnin at 70% power; and 3: 25 min at 50% power. The digestion vessel was removed, allowed to cool, then manually vented. The contents were transferred quantita- tively through an acid-rinsed filter into a 50 ml calibrated flask, then made up to the mark with de-ionized water. Final acid strength: 10% aqua regia. Preparation of Samples for Analysis by TXRF Reflector plate preparation. All the sample solutions were transferred by pipette (Gilson microman; 25 p1 adjustable pipette) directly (10 pl) onto highly polished quartz reflector plates (30 mm diameter) previously analysed to establish the plate background.The quartz plate was treated prior to sample addition with 5 1-11 of a pure silicone solution (Serva Feenbiochemical) to prevent excessive spreading of the sample solution. In the analysis for As, Hg and Se it was necessary to add a further 10 1-11 of 1% ammonium pyrrolidin-l-yldithioformate (ammonium pyrrolidinedithiocarbamate) to the sample on the quartz plate to minimize losses of the volatile analytes during drying. Aqueous samples. A 10 p1 aliquot of the aqueous sample was transferred by pipette onto a quartz plate and mixed on the plate with a 10 1-11 aliquot of an internal standard (30 pprn of Co).The plate was dried under an infrared lamp for 10 min prior to analysis. The Mo X-ray tube excitation conditions used were 50 kV and 10-50 mA, with a data acquisition time for each sample of 1000 s. Acid-digested samples. A 10 p1 aliquot of the digested sample was transferred by pipette onto the quartz plate and mixed with 10 1-11 of an internal standard (30 ppm of Co). The plate was then dried under an infrared lamp for 10 min prior to analysis. Results and Discussion Optimization of X-ray Excitation Radiation As the EXTRA I1 is supplied with fixed reflector optics, the only variables requiring attention from the analyst are tube voltage, tube current and sample presentation on the quartz reflector.An investigation of the spectral distribution in terms of energy profiles €or the determination of most elements of interest, identified 50 kV and 38 mA as the most suitable tube setting. Tube conditions are investigated more fully and relevant results are discussed under Detection Limits. The results shown in Table 1 confirm that the best compromise tube conditions are in the range 40-50 kV. Changes in the tube Table 1 Variation of MDL (pg dm-3) as a function of tube voltage and current in ambient atmosphere for aqueous standards Tubeconditions S Ca V Mn Co Zn Se Sr 59mA;20kV 260 63 39 21 17 12 11 24 59mA;25kV 233 54 28 16 12 8 8 1 59mA;30kV 121 31 16 10 8 5 4 5 28mA;40kV 118 29 17 9 7 5 4 4 20mA;50kV 135 30 16 10 7 5 4 4 12mA;SOkV 136 32 17 10 7 5 4 4 current were found to be necessary in order to minimize scattered radiation entering the detector.For a pure standard solution, low scatter was observed and a typical current setting of 38 mA was suitable. However, the presence of dissolved solids on drying caused increased scatter and produced a corresponding increase in the dead time of the counting statistics, and thus it became necessary to reduce the excitation current. For example, with the acid-digested samples used, a tube current of 10 mA would produce a typical dead time of 40%. The ideal dead time for the system was found to be in the region of 30 to 50%; however, at present, there is no automated facility for self-adjusting the current output to match changes in the scatter from samples. Work is currently being undertaken to enable this aspect of instrumen- tal optimization to become fully automated.As indicated above, the dissolved solids content of the sample will affect the degree of scatter observed by the detector. To date, samples containing 2% dissolved solids have been analysed and indeed fine powders15 and biological tissues16 have been analysed directly. The distribution of the sample on the quartz surface does not appear to affect the quantification or the precision of the analysis, provided that the sample is spotted within the envelope of the excitation beam profile (8 mm width). A small PTFE jig enables spots to be easily placed within the zone of excitation. Calibration Multi-element standards covering five orders of magnitude, ranging from ppb to pprn concentrations, were prepared for Se, V and Zn.Each standard was spotted (10 p1) onto a quartz reflector plate, and K line intensities were measured relative to Co K lines. The three elements were chosen to represent a range of emission energies in the Mo X-ray excitation range. All three elements were found to yield good linear relation- ships, according to the following equations, where y is the normalized intensity count and n the concentration. Se: y = 1 . 0 2 ~ - 0.0311; V: y = 1 . 0 2 2 ~ - 0.192; Zn: y = 1.00~ + 0.0267. As the main thrust of this work was to determine trace element levels, no further examination of higher concentra- tions was carried out. It seems probable, however, that higher concentrations could be determined by the TXRF technique.Detection Limits The investigation into detection limits examined not only aqueous samples, but also acid-digested sample matrices. The elements selected for this section of the work were chosen to represent low (S) to high (Sr) range X-ray emission energies typically obtained with use of Mo X-ray excitation (2-20 keV). Detection limits with use of aqueous standards A multi-element standard was prepared containing 10 ppm each of Ca, Mn, S, Se, Sr, V and Zn, with Co acting as the internal standard at 10 ppm. A 10 p1 aliquot of this standard solution was transferred by pipette onto a quartz plate and air-dried. The signals were then collected for a 1000 s period for both the standards and a blank. The minimum detection limit (MDL) was calculated from the following equation: MDL = - I where c is the concentration of standard, Z is the total peak counts for the standard, ZB is the total background peak count, and c is time.The MDLs for various elements in an ambient atmosphere over a range of tube voltages and currents are shown in Table 1. The same procedure as above was then followed, except that a helium flush was introduced into the cavity between the336 ANALYST, APRIL 1991, VOL. 116 detector and sample plate, and the MDLs for the same range of elements were determined as a function of tube voltage and current (Table 2). From the data in Tables 1 and 2, it is clear that the MDL can be improved for most elements with the introduction of a helium flush. This effect was particularly pronounced in the lower energy emission range, as seen for S and Ca.As a general trend it was found that elements in the higher energy range (Sr and Se) afforded improved detection limits as the tube voltage was increased, with a corresponding reduction in tube current. This characteristic is thought to be associated with a reduction of the base line in this region of the spectrum, which normally results from the target of the Mo X-ray tube when high currents are used. However, a reduction in tube current will also lead to a loss of emission intensity, hence the apparent effect will be limited to the current ranges indicated; further reduction in tube current will subsequently lead to a loss in fluorescence yield. In the element range V-Zn, MDLs are seen to improve when a tube voltage of 30-40 kV and a current of 28-59 mA are used.For lighter elements at the lower end of the energy range, lower X-ray tube voltages and high currents appear to give the best MDL. As the technique is designed to be a simultaneous multi- element method of analysis, a suitable compromise of intermediate tube voltage and current would seem to be appropriate. However, for specific regions in the spectrum, optimum conditions of sensitivity could be better achieved by selecting the appropriate X-ray tube conditions. The MDL values obtained by TXRF were considered to be acceptable for a technique required for trace element analysis and comparable to other widely used trace element techniques. Detection limits in acid media and digested material The MDLs were determined for various elements in an acid matrix and for acid-digested reference materials.The internal standard (3 ppm of Co) was mixed with 10 pl of the acid-digested material, and the MDLs were calculated by using eqn. (1) and are summarized in Table 3. From the results obtained, various conclusions can be drawn. The most general trends are that the MDLs for all the elements examined are similar for both the acidic and aqueous standards and that MDLs improve for higher 2 elements, which corresponds to an increase in X-ray emission energies ~~~ ~ ~ Table 2 Variation of MDL (pg dm-3) as a function of tube voltage and current in a helium atmosphere Tubeconditions S Ca V Mn Co Zn Se Sr 59mA;20kV 230 68 39 21 16 13 11 23 59mA;25kV 140 36 22 13 9 7 6 7 59mA;30kV 103 32 17 10 7 5 4 4 28 mA; 40 kV 9 6 2 8 1 6 9 6 5 4 4 20mA;50kV 104 28 17 9 2 5 3 3 12mA;60kV 117 31 17 9 7 5 4 3 when using K lines.However, when the MDLs were calcu- lated for a reference material present in an acid-digested matrix it was obvious that a deterioration of the values obtained had occurred. This decrease in detection limits is approximately twice that for nitric acid-based matrices, except for some of the light elements in saline water, but the detection limit can be seen to improve by approximately five times for aqua regia matrices. The apparent determinator in detections limits observed for digested samples is thought to be a function of dissolved solids and scatter from the sample. As a general conclusion, it would appear that nitric acid used at strengths of up to 12% v/v represents the most suitable acid matrix for sample digestion.Precision of Sampling Methodology Ten quartz reflector plates, previously analysed for back- ground, were spotted with a solution of mixed standards, and, after drying, each plate was analysed ten times. The typical relative standard deviation (RSD) values of 2-3% for most elements were found to be similar to those for other manual injection techniques such as electrothermal atomic absorption spectrometry. The precision for Se was, however, considered to be unacceptably high (6.7% RSD) and this was found to be a feature of the drying step when samples were dried under an infrared lamp. During this stage the surface temperature of the quartz plate can reach 170 "C and this will be sufficient for significant amounts of volatile analytes, such as Se, Hg and As, to be lost.Air- or vacuum-drying has been found to reduce this effect, and a precision of 2-3% for the most volatile elements can be achieved in this way. The position of the 10 yl sample spot on the surface appeared to have little effect on the precision obtained, provided that the entire spot was con- tained within the 8 mm profile width of the excitation beam: this is not difficult to achieve practically. Instrumental Precision The evaluation of the instrumental precision was carried out in two ways. The first way was to introduce (via the auto- sampler) the sample plate ten times into the instrument, while the second way was to take ten consecutive measurements, with the plate remaining in the analytical position.From the RSD values obtained for both experiments it was found that an instrumental precision of <1% RSD could be obtained, and this was considered to be acceptable as an instrumental procedure and compared favourably with similar techniques. Accuracy Various reference materials were examined in order to assess the accuracy of the results obtained with TXRF. In addition, Table 3 MDLs for a range of elements determined in acid medium and for various matrix types (all values are in yg dm-3) Sample 1. Low concentration (1% HN03) 2. High concentration (12% HN03) 3. First sludge (12% HN03) 4. Second sludge (12% HN03) 5. Composite potable water (1% HN03) 6. Composite river water (1y0 HN03) 7. Saline water (1 % HN03) 8. Digested sludge (12% HN03) 9.Digested sludge (12% HN03) 10. High concentration (10'3'0 aqua regia) 11. Certified soil (10% aqua regia) 12. River sediment (10% aqua regia) 13. River sediment (10% aqua regia) 14. Composite soil (10% aqua regia) Ca 260 130 160 110 72 69 84 110 100 160 120 230 280 280 K Fe Cu Zn Ni Cr Pb Se 80 45 1 21 13 13 5.8 2.9 41 23 4 11 6.4 6.4 2.9 1.4 97 27 4.8 13 7.7 7.6 3.5 3.5 70 19 2.4 9.0 5.5 5.4 2.5 - 43 12 2.2 5.6 - 3.5 1.6 - 1.5 - 42 12 2.0 5.5 - - 1.9 - 70 20 3.5 9.1 5.3 5.5 2.5 1.2 62 17 3.1 8.0 4.9 4.7 2.2 1.1 140 25 2.0 12 7.3 7.5 5.9 1.5 110 19 3.0 8.7 5.5 5.5 4.4 1.1 200 35 5.7 17 10 10 8.2 8.2 250 43 6.9 20 13 13 10 - 250 43 6.9 20 12 - 10 - - 510 140 - 65 - Hg As 3.9 3.4 2.0 1.7 - 1.8 - 1.7 - 2.1 - 2.9 - 2.5 - - - - 5.2 2.3 3.8 1.7 7.2 3.2 9.2 4.0 - 3.8ANALYST, APRIL 1991, VOL.116 337 Table 4 Comparison of TXRF and ICP results for a variety of sample types Sample Synthetic standards (yg dm-3)- 1 . 1% HN03 2. 10% aqua regia 3. 12%HN03 Reference solids ( yg dm-3)- 4. BCR CRM 143 (12% HNO3) 5 . BCR CRM 144 (12% HNO3) 6. BCR CRM 142 (1 0% aqua regiu) Environmental solids (yg dm-3)- 7. Digested sludge (12% HN03) 8. Digested river sediment (10% aqua regiu) 9. Digested river sediment (10% aqua regiu) 10. Composite soil (10% aqua regiu) 11. Digested sludge (12% HN03) ICP TXRF ICP TXRF ICP TXRF ICP REF TXRF ICP REF TXRF ICP REF TXRF ICP TXRF ICP TXRF ICP TXRF ICP TXRF ICP TXRF Environmental liquid samples (yg dmP3)- 12. Composite potable water sample (1% HN03) ICP TXRF 13. Composite river sample (1% HN03) ICP TXRF 14.Saline water (1% HN03) ICP TXRF * ND = not detected. Ca 200 260 200 270 200 274 588 625 361 410 408 - 300 360 - 240 280 560 570 1290 1340 48 50 300 360 59 75 81 100 330 280 K 20 23 20 22 20 25 50 - - 14.1 14.0 - 19 - - 19 20 15 18 32 31 36 22 23 25 32 36 43 52 460 163 Fe Cu Zn Cr Ni Pb Se As Hg 5 5 5 5 5 5 1 1 5 5 5 5 4.7 5 5 1 1 5 50 20 20 20 20 20 5 5 10 4.6 12 50 20 20 19.8 20 22 5 50 20 20 20 20 20 5 5 10 50 20 20 20 20 20 5 5.4 9.0 190 2.3 12 1.5 0.79 12.2 ND* 0.11 ND 212 3.3 13 1.6 0.90 13.7 ND 0.21 ND 440 6.8 30.6 4.31 9.30 4.4 ND 0.012 ND 420 6.3 27.7 4.50 8.6 3.8 ND ND ND 214 2.36 13.01 2.08 0.93 13.12 0.006 0.16 - 460 6.94 30.55 4.94 9.47 4.79 0.023 0.09 - 136 0.27 0.80 0.31 0.26 0.25 ND 0.20 ND 160 0.27 0.81 0.26 0.28 0.20 ND 0.10 ND 190 0.25 0.80 0.44 0.28 0.30 0.008 0.17 - 58 5.22 7.36 1.70 1.40 4.09 - 0.09 74 5.21 7.20 1.50 1.20 4.20 0.02 ND 320 0.27 1.0 0.16 0.17 0.60 ND 0.12 340 0.30 1.1 0.21 0.21 0.36 ND 0.27 230 0.24 0.80 0.24 0.19 0.28 0.12 0.14 260 0.26 0.94 0.29 0.24 0.28 ND 0.18 340 0.38 1.34 0.54 0.30 0.48 ND 0.14 330 0.40 1.34 - 0.31 0.49 ND 0.20 89 68 9.8 1.4 2.0 5.3 0.18 0.16 94 70 9.7 1.6 2.0 5.6 0.03 ND 0.12 0.02 ND ND ND ND ND ND 0.28 0.02 ND ND ND 0.03 ND ND 0.31 ND 0.03 ND ND ND ND ND 0.44 0.02 0.05 ND ND 0.04 ND ND 0.01 ND ND ND ND ND ND ND 0.73 ND ND ND ND ND ND ND the elemental composition of each sample was also deter- mined by ICP, and the results are summarized in Table 4.From the data set obtained there can be seen to be, in general, good agreement between the ICP and TXRF values and close agreement with available reference values.There appear to be one or two spurious values, which tend to be associated with the lower 2 elements ( e . g . , Ca and K). In addition, Pb appeared to give a low value in sample 8 for the TXRF, with Pb values, for both the TXRF and ICP methods, in sample 6 being lower than the reference value. This latter effect could be due to the poor recovery from the extraction medium as results by both methods show reasonable agree- ment. Poor recovery could also account for the variation in the Cr values in sample 6. Within the data set, one or two disagreements between the two measurement techniques are apparent, e.g., Se in sample 9, As in sample 8, and Fe in samples 12 and 14. There appeared to be no apparent reason for the variations observed and these can only be attributed to random experimental error.In general, the extraction pro- cedures used for the soil and sewage sludge samples proved to be satisfactory, as did the instrumental method of analysis. Conclusions The aim of this work was to evaluate the analytical perfor- mance of TXRF as a suitable multi-element method for a range of liquid-based matrix types. The results indicate that TXRF is a suitable method for multi-element determinations for the range of liquid-based matrices studied. The perfor- mance characteristics of the technique have proved to be satisfactory, offering a simple and rapid method for sample preparation when performing TXRF analysis. In addition, the338 ANALYST, APRIL 1991, VOL. 116 small samples required for analysis were not decomposed and remained stable on the quartz reflector plate surface. The technique affords a reliable and simple method of calibration, and satisfactory results (within normal experimental error), in comparison with those of the ICP method and with reference values, have been obtained. References 1 Bacon, J. R., Ellis, A. T., and Williams, J. G., J. Appl. At. Spectrom., 1989,4, 199. 2 Berth, E. P., Principles and Practice of X-ray Spectrometric Analysis, Plenum Press, New York, 2nd edn., 1980. 3 Prange, A., Spectrochim. Acta, Part B, 1989,44,437. 4 Wobrauschek, P., and Aiginger, H., Spectrochirn. Acta, Part B, 1980,35,607. 5 Wobrauschek, P., and Aiginger, H., Adv. X-Ray Anal., 1986, 28, 1. 6 Wobrauschek, P., and Aiginger, H., Anal. Chem., 1975, 47, 852. 7 Yoneda, Y., and Horiuchi, T., Rev. Sci. Instrum., 1971, (July), 422. 8 9 10 11 12 13 14 15 16 Aiginger, H., and Wobrauschek, P., Nucl. Instrum. Methods, 1974, 114, 157. Knoth, J., and Schwenke, H., Fresenius Z. Anal. Chem., 1978, 291,200. Knoth, J., and Schwenke, H., Fresenius Z. Anal. Chem., 1980, 301,7. Knoth, J., and Schwenke, H., Fresenius 2. Anal. Chem., 1979, 294, 273. Schwenke, H., and Knoth, J., Trace Element Analytical Chemistry in Medicine and Biology, Walter de Gruyter, Berlin, New York, 1980, p. 307. Wobrauschek, P., and Kregsamer, P., Spectrochim. Acta, Part B, 1989, 44,453. Knoth, J . , Schwenke, H., and Weisbrod, U., Spectrochim. Acta, Part B, 1989,44, 477. Bohlen, A. V., Eller, R., Klockenkaemper, R., and Tolg, G., Anal. Chem., 1987,59, 255. Bohlen, A. V., Klockenkaemper, R., Tolg, G., and Wiecken, B., Fresenius Z. Anal. Chem., 1988, 331, 454. Paper 01049480 Received November 5th, 1990 Accepted November 16th, 1990

ISSN:0003-2654    

DOI:10.1039/AN9911600333

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, APRIL 1991, VOL. 116 339 Interlaboratory Determination of Copper, Chromium and Arsenic in Timber Treated With Preservative Bernard S. W. Dawson, Glenda F. Parker, Faye J. Cowan and Sung 0. Hong Forest Research institute, Private Bag 3020, Rotorua, New Zealand Precision analyses of timber treated with preservative are an important part of quality control in the New Zealand timber preservation industry. Determinations of Cu, Cr and As remain important criteria in this industry. Repeatability and reproducibility values for two interlaboratory trials are calculated. The repeatability for the Rotorua interlaboratory trial ranged from 0.0109 to 0.0497% for Cu, 0.0190 to 0.600% for Cr and 0.0121 to 0.0346% for As, while reproducibility values ranged from 0.0121 to 0.0663% for Cu, 0.0389 to 0.1951 % for Cr and 0.0342 to 0.1298% for As.The Queensland interlaboratory trial also produced repeatability and reproducibility values similar to those for the Rotorua trial. Significant differences between laboratories were found for the determination of Cu, Cr and As. Overall, the data from both interlaboratory trials were similar. Increased awareness of quality assurance programmes is seen as essential to improve the performance of laboratories involved in the timber preservation industry. Keywords: Copper, chromium and arsenic determination; wood; interlaboratory trial; precision; quality assurance Copper, chromium and arsenic (CCA) have excellent fungi- cidal (Cu) and insecticidal (As) properties in the treatment of wood.' The primary role of Crvl is in the fixation of Cu and As in the wood.These properties have been responsible for CCA preservatives being the most used worldwide for several decades. The importance of CCA preservatives in the New Zealand timber preservation industry is shown by the New Zealand Forest Service statistics.2 Out of a total of 1.4 X 106 m3 of timber treated, over 1 x 106 m3 were treated with CCA preservatives. The industry, therefore, places a high reliance on accurate determinations of the CCA components. The maintenance of national or industry standards of timber preservation is monitored by quality management pro- grammes involving treatment plants, the industry, auditors and analytical laboratories. Analytical techniques that have been employed to deter- mine CCA in the preservation industry include atomic absorption spectrometry (AAS) ,3-5 iodimetry,6 spectropho- tometry,7 X-ray fluorescence spectrometry,8>9 inductively coupled plasma atomic emission spectrometry10 and neutron activation analysis. 1 1 There are also several national standard methods available.12-17 The micro-distribution of CCA in various wood tissues has also been studied using energy dispersive X-ray analysis. 18 Industrial reliance on determinations of CCA has economic ramifications. To support the timber preservation industry, it was decided to conduct an interlaboratory trial on the determination of CCA, in various laboratories, primarily to establish precision data and variability of analytical methods. The data analysis was carried out following two methods in order to compare their usefulness for application to analysing interlaboratory trials.Data from a second interlaboratory study on CCA in wood were also analysed. Experimental Design of the Interlaboratory Programme for the Determina- tion of CCA in Treated Timber The design of the interlaboratory trial followed recommenda- tions contained in the International Standard I S 0 5725.19 Fourteen New Zealand, Australian and Fijian laboratories involved with the timber preservation industry were invited to participate. A written method was supplied and laboratories were asked to follow this method closely if it was similar to their normal laboratory method. If their normal method was dissimilar ( e . g . , methods based on different physical prin- ciples), they should use their own method and provide a brief description of it.Eighteen samples of Pinus radiata sapwood treated with CCA compounds [0.054.48% of Cu, 0.094.84% of Cr and 0.06-0.60% of As (all d m ) ] were analysed. These concentra- tions span the range of preservative retentions expected in timber when treated with CCA in accordance with current New Zealand specifications.20 The samples represented six concentrations to be determined in triplicate under repeatable conditions.19 (Raw analytical data are available on request.) Preparation of Treated Timber Samples A solution containing AnalaR grade CuS04-5H20, Na2Cr207.2H20 and As203 was prepared. Six aliquots of this solution were adjusted to a constant volume (75 ml) before addition to 150.0 g of milled (Wiley mill), oven-dried (105 "C) Pinus radiata sapwood.Each mass of timber was then mixed, frozen and freeze-dried, before being ground in a large stainless-steel ring mill to produce a timber flour and finally riffled ten times through a 12-channel riffle to ensure the homogeneity of each sample (see below). Sample lots of 3-6 g were distributed in plastic screw-cap jars identified by a laboratory/sample code. The laboratories were informed of the broad concentration ranges of the samples. Analytical Methods Used by Participating Laboratories Eleven laboratories responded to the invitation to participate in the interlaboratory trial. Three of these laboratories requested extra sample sets and performed the analyses by two methods. A brief description of the methods used by each laboratory for the determination of CCA is given in Table 1.The method used by laboratory C was the method supplied. Six laboratories followed this method and two others, F and L, had seemingly minor deviations. The techniques employed included digestion followed by titration, digestion or extrac- tion, followed by AAS and X-ray fluorescence spectrometry. Results and Discussion Method Validation For many years the supplied method of analysis ( e . g . , laboratory C , Table 1) has been used satisfactorily in our laboratory. The method consists essentially of two parts:340 ANALYST, APRIL 1991, VOL. 116 digestion of the sample and determination of the CCA analytes. Although several standard methods exist for each part of the analysis, and these standards constitute a method validation in themselves, there are several points that should be considered in the context of the present interlaboratory trial.Two of these are the limit of detection and 95% confidence intervals. Limit of detection Fifty analyses of a homogenized bulk of treated wood (means of 0.095% of Cu, 0.166% of Cr and 0.108% of As) gave relative standard deviations (RSDs) of 1.8, 2.9 and 3.2% for Cu, Cr and As, respectively. The RSD is an estimate of variability attributable to sample heterogeneity and the use of the method of analysis at the operational level. For single subsequent analyses involving use of these precision data, the limits of detection are 0.008% for Cu, 0.023% for Cr and 0.017% for As.21 Confidence limits Based on the standard deviation of the 50 analyses the 95% confidence interval can be calculated.21 The values are 20.0034% of Cu, +0.0096% of Cr and +0.0070% of As for a single analysis (at the specified concentrations). Sample contamination during preparation A check was made on possible contamination of samples from the iron-alloy ring mill (86% Fe, 12% Cr, 2% Si/Mn and Cu <0.05%; classified as a D3 mill by the American Iron and Steel Institute) by determining Fe in the wood after grinding.Successive 50 g lots of Wiley milled Pinus radiata were ring-milled from 2 to 8 min. There was no trend in the Fe content of the milled wood (n = 8, mean = 34 pg g-1, standard deviation = 5 vg g-1). When the wood was not ground, the Fe content was 14 pg g-1. Therefore, maximal Cr contamination (based on the mill’s composition) would be less than 5 pg g-1 (with Cu and As being much less).These levels are negligible compared with CCA concentrations in wood treated with preservative. Table 1 Brief description of the methods of analysis for each laboratory-Rotorua trial Laboratory Method A, B Cu, iodimetric titration with thiosulphate following digestion of sample with H202-H2S04 (3 + 1) Cr, back-titration of excess of ammonium iron(]]) sulphate with standard dichromate following digestion of sample with H20-H2S04 (3 + 1) following digestion of sample with As, titration with standard cerium(iv) nitrate H202-H2S04 (3 + 1) C, D, I, J, M, P Modification of American Wood Preservers Association Standard A7-75. * Digestion of sample with H202-H2S04 (3 + 1).No perchloric acid added to digested sample. Determination by AAS Portable XRF (Asoma LCA). Calibrated each day with manufacturer’s calibration disk As for Laboratory C, except different preparation of standards used for AAS Digestion with HN03-HC104 (5 + 1) on temperature-programmable digestion block. Analysis by AAS E F 0 H, N L British Standard 5666 : Part 3. Method lt As for Laboratory C except temperature- programmable digestion block used. Require H202-HN03 than expected to affect digestion * Reference 14. t Reference 13. These points indicate that the method supplied to the laboratories can produce precise results and will, therefore, be a suitable method with which to compare the performance of the methods employed by the various laboratories.Data From Queensland Interlaboratory Trial For comparative purposes, data generated by the Queensland Forest Service22 have been included in this paper. In this trial, duplicates of six samples were sent to nine laboratories worldwide, all of which analyse large numbers of treated timber samples annually. Sample treatment included grinding to pass through a 1 mm mesh. The sample concentration range was less than that for the Rotorua interlaboratory trial. Analyses by participating laboratories involved a range of methods; there was no recommended method of analysis. Statistical Analyses Statistical analyses, based on the I S 0 572519 and the Ana- lytical Methods Committee (AMC)23,24 procedures for calcu- lating the repeatability (for within-laboratory variation, denoted by subscript e) and reproducibility (for between- laboratory variation, denoted by subscript L) and for checking outliers, were carried out, together with conventional data exploratory analysis.As every single test result (yj,) is assumed to be the sum of a general mean ( u ) , laboratory variation (bj) and random error (ejJ occurring in every test, the basic model in the present study can be written as: y j , = u + bj + eij where i = 1 . . . p laboratories, and j = 1 . . . n replications in each laboratory (in this trial p = 14 and n = 3). An analysis of variance (ANOVA) was performed to estimate the interlaboratory performances and to determine variance components as given in Table 2. The repeatability value ( r ) and reproducibility value ( R ) were calculated19 from the variance components of the mean squares of the ANOVA as follows: r = 2.8 or where or = = repeatability standard deviation R = 2.8 OR where OR = vat + 02 = reproducibility standard deviation and MSL - 0: 0 2 = n There was a difference between the I S 0 5725 and the AMC model in the procedure for outlier checking.The I S 0 5725 model identifies outliers, by the Cochran and Dixon tests, and advises their removal before calculating r and R. The AMC model, on the other hand, tests for outliers only to allow correction of spurious values and to identify outlying labora- tories. A major concern of the AMC is the possibility of underesti- mating the variability encountered in routine analyses by rejecting outliers, thereby reducing the variability of results. They contended that this was effectively providing variability parameters for perfect analytical conditions.In this trial, the tests for outliers were carried out following both models. The outliers identified by the Cochran and Dixon tests were, Table 2 Analysis of variance (ANOVA) to estimate interlaboratory performance and to determine variance components Degrees of Mean Variance Components Source of variation freedom square of mean squares Between laboratories p- 1 MSL 0; + n 0; Within laboratories p ( n - 1) MS, 0:ANALYST, APRIL 1991, VOL. 116 341 however, not removed prior to data analysis, in order to produce realistic values of the precision parameters. The other statistical procedure was to apply a least significant difference (LSD) test following an ANOVA, when the result was significant.Interlaboratory Trial Results The data submitted by participating laboratories for CCA in timber in the Rotorua trial were tested for normal distribution and skewness. All data were found to be normally distributed with homogeneous variance and no skewing. Determination of Precision Parameters Although a standard test method was supplied to all labora- tories, the variety of methods used was expected. It is still of value, however, to determine the repeatability and reprodu- cibility values. The r and R values were calculated from raw data, following I S 0 5725.19 The mean, r and R values for CCA for the Rotorua trial are given in Table 3. The repeatability and reproducibility values are used to test the difference between two analytical results from the same laboratory and from different laboratories, respectively. The ranges in r values were 0.0104-0.0497% for Cu, 0.0190-0.0600% for Cr and 0.0121-0.0418% for As.As a check on these values, there are 252 possible differences between individual results in one laboratory (14 laboratories x 6 concentrations x 3 possible differences per cell) for Cu and Cr, and 248 for As (which has two cells with a member missing). There are 21 (Cu), 19 (Cr) and 16 (As) differences exceeding the r values [or 8.3% (Cu), 7.5% (Cr) and 6.5% (As) of differences], which is a result compatible with a 95% probability level. The R values range from 0.0121 to 0.0663% for Cu, 0.0389 to 0.1951% for Cr and 0.0342 to 0.1298% for As. The 95% critical difference (Cr Dg5) can be used to compare labora- tories performing analyses19 (Table 3).Of the 546 differences between the means of each laboratory (14 laboratory means have 91 possible differences per concentration and there are six concentrations), 29 for Cu, 19 for Cr and 49 for As exceed the critical difference values for the respective concentrations (or 5.3% for Cu, 3.5% for Cr and 9.0% for As). This again is consistent with a 95% probability level. The Y and R values in Table 3 can be used in comparisons within and between laboratories. It is probable that in a collaborative trial (where a method is specified) lower r and R values would result than in a co-operative trial (where there is no method specified). The raw data for the Queensland trial were also normally distributed with homogeneous variance and no skewness.The mean, r and R values for CCA are presented in Table 4. The ranges in r values were 0.0011-0.0159% for Cu, 0.0066 0.0252% for Cr and 0.00954.0265% for As. As a check on these r values, there are 54 possible differences between individual results in each laboratory (9 laboratories X 6 concentrations x 1 possible difference per cell) for each element. Therefore, there are one (Cu), two (Cr) and three (As) [or 1.9% (Cu), 3.7% (Cr) and 5.6% (As)] differences exceeding the corresponding r value, which is compatible with a 95% probability level for r. The R values span the ranges 0.0148-0.0454% (Cu), 0.0136-0.0512% (Cr) and 0.0267- 0.0876% (As). Using the 95% critical differences (Table 4) to compare laboratories, of the 216 possible differences between the means of each cell for each laboratory (9 laboratories have 36 possible differences per concentration), eight (Cu), nine (Cr) and eight (As) [or 3.7% (Cu), 4.2% (Cr) and 3.7% (As)] exceed the appropriate 95% critical difference.This is consistent with a 95% probability level for R. The values of r and R for both the Rotorua and Queensland trials are plotted in Fig. 1. The equations expressing r and R as a function of concentration are presented in Table 5 . Although the concentration range for the Queensland data is about half that for the Rotorua trial, the r and R values are very similar functions of concentration except for Cr and As, where the Queensland r values are not linear functions of concentration.The values of r and R are similar for both trials, suggesting that laboratories participating in both trials have the same extreme Table 3 Values for repeatability and reproducibility (g of metal per 100 g of wood) for Rotorua data Mean or. ros RSD(%) OR R95 c u 0.048 0.0039 0.0109 8.1 0.0043 0.0121 0.094 0.0037 0.0104 3.9 0.0053 0.0148 0.189 0.0097 0.0271 5.1 0.0110 0.0307 0.272 0.0078 0.0219 2.9 0.0144 0.0404 0.357 0.0097 0.0271 2.7 0.0198 0.0556 0.440 0.0177 0.0497 4.0 0.0237 0.0663 Cr As As? 0.087 0.0108 0.0302 12.4 0.0139 0.0389 0.177 0.0068 0.0190 3.8 0.0169 0.0472 0.349 0.0163 0.0457 4.7 0.0294 0.0822 0.507 0.0106 0.0297 2.1 0.0453 0.1267 0.666 0.0187 0.0523 2.8 0.0633 0.1773 0.824 0.0214 0.0600 2.6 0.0697 0.1951 0.056 0.0055 0.0155 9.8 0.01 50 0.0419 0.119 0.0043 0.0121 3.6 0.0122 0.0342 0.231 0.0082 0.0225 3.5 0.0167 0.0468 0.339 0.0118 0.0332 3.5 0.0333 0.0934 0.449 0.0149 0.0418 3.3 0.0357 0.1001 0.553 0.0124 0.0346 3.3 0.0464 0.1298 0.056 0.0037 0.0103 6.6 0.0 147 0.0413 0.116 0.0043 0.0122 3.7 0.0084 0.0236 0.228 0.0081 0.0226 3.6 0.0135 0.0379 0.0498 0.331 0.0121 0.0339 3.7 0.0178 0.0838 0.443 0.0148 0.0415 3.3 0.0299 0.542 0.0177 0.0328 3.3 0.0182 0.0509 Cr Do5* 0.0081 0.0121 0.0213 0.0362 0.0510 0.0524 0.0301 0.0446 0.0732 0.1244 0.1721 0.1888 0.0399 0.0327 0.0430 0.0894 0.0941 0.1267 - - - - - - * Cr Dgj (I Y1 - Y21) = T Excluding Laboratory 0 data.342 ANALYST, APRIL 1991, VOL.116 0.04 Table 4 Values of repeatability and reproducibility (g of metal per 100 g of wood) for Queensland data Mean o r r95 RSD (Yo) OR R95 c u 0.031 0.0024 0.0066 7.7 0.0053 0.0148 0.037 0.0004 0.001 1 1.1 0.0063 0.0176 0.090 0.0044 0.0124 4.9 0.0075 0.0209 0.0183 0.135 0.0034 0.0096 2.5 0.0065 0.175 0.0057 0.0159 3.3 0.0151 0.0424 0.223 0.0051 0.0142 2.3 0.0162 0.0454 - 0 00 0 0 I I I I Cr As 0.0118 7.5 0.0049 0.0136 0.056 0.0042 0.049 O.OO24 0.0066 4.9 0.0089 0.0251 0.152 0.0037 0.0102 2.4 0.0082 0.0230 0.224 0.0049 0.0138 2.2 0.0094 0.0263 0.258 0.0090 0.0252 3.5 0.0178 0.0498 0.379 0.0040 0.0111 1.1 0.0183 0.0512 0.062 0.0034 0.0095 5.5 0.0095 0.0267 0.028 0.0034 0.0096 12.1 0.0109 0.0304 0.183 0.0059 0.0166 3.2 0.0166 0.0465 0.272 0.0031 0.0087 1.1 0.0220 0.0617 0.296 0.0095 0.0265 3.2 0.0313 0.0876 0.250 0.0048 0.0133 1.9 0.0209 0.0584 Cr D95* 0.0140 0.0175 0.0189 0.0170 0.0409 0.0442 0.0108 0.0246 0.0218 0.0244 0.0465 0.0506 0.0259 0.0296 0.0450 0.0614 0.0856 0.0576 rL 2 * Cr Dg5 (IY, - Y21) = JR2-- (Reference 19).0.05 .- c 0.04 - .- % 0.03 CI 9) g 0.02 a 0.01 0 0 0 0 O Q O n~ I 1 I 0.04 0.03 0.02 0.05 0.04 0.03 0.02 0.01 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 ( d ) 0.07 0.06 0 0 - .g 0.05 0.04 0.03 .- 3 2 !2 0.02 K 0.01 0.16 o.20 I 0.08 O.I2 I 0 0 0 0 0 0 0 0 0 O D 0 0 0 0.1 0.2 0.3 0.4 0.5 0.6 (f-) 0.14 I 0 0.12 0.08 0.06 0 . 0.04 i;: 0 0.02 1 1 1 1 1 1 0 0.1 0.2 0.3 0.4 0.5 0 0.2 0.4 0.6 0.8 1.0 0 0.1 0.2 0.3 0.4 0.5 0.6 Concentration of Cu (%) Concentration of Cr (%) Concentration of As (%) Fig. 1 Plots of repeatability and reproducibility for Cu, Cr and As as a function of concentration for both the Rotorua and Queensland data (0, Rotorua data; and 0, Queensland data) Table 5 Repeatability and reproducibility, parameters from the Rotorua and Queensland trials, as a function of concentration Coefficient of Coefficient of Rotorua determination* tSlopet Queensland determination* tslopet Cu r = 0.08185 x concentration + 0.00510 81.2 A r = 0.05777 X concentration + 0.00332 66.7 A R = 0.14297 x concentration + 0.00329 99.5 B R = 0.15747 x concentration + 0.00842 79.4 A ns Cr r = 0.04499 x concentration + 0.01991 66.9 A R = 0.23014 x concentration + 0.01112 98.3 B R = 0.10418 x concentration + 0.01209 74.2 A ns As r = 0.05514 x concentration + 0.01056 81.3 A R = 0.19366 x concentration + 0.01798 91.9 B R = 0.18395 x concentration + 0.01845 85.0 B - - * Coefficient of determination (per cent.of variation due to regression; is the square of the correlation coefficient). t-Test for slope difference from zero: A = significant at the 5% level; B = significant at the 1% level; ns = not significant. value ranges and similar sizes of errors within laboratories and between laboratories. According to the AMC23 the variation between batches is usually of a comparable size to oe and oL (which can be calculated from Y and R values) and, therefore, the maximum error for any further analytical determination, as estimated by a, and oL, would be an underestimation of the true error. An advantage of using oe and oL values as opposed to Y and R values is that the source of the components of error in measurement can be directly seen.When ANOVA was applied to both the Rotorua and Queensland raw data (Rotorua data not including target values) for CCA, there were only three instances (all Cu)ANALYST, APRIL 1991, VOL. 116 343 where ANOVA was not significant at the 1% level. When considering, firstly, the Rotorua data for all concentrations combined, the LSD tests for CCA all show differences in the means of each laboratory (Table 6). The methods used for analysis (Table 1) can be divided into one group of six laboratories (C, D, I, J, M and P) and two groups of two laboratories (A, B and H, N), the remainder of the methods being used by only one laboratory. The method used by the largest group of laboratories (the supplied method) has not been applied with uniform results. As such, the data from this method have not been analysed separately.For the method used by Laboratories A and I, the results for Cr were significantly different, but not for Cu and As. The use of the BS 5666 Method13 (Laboratories H and N) achieved similar results for each element. From the above observations it appears that the BS 5666 method13 is possibly the most precise (a greater number of laboratories using this method would be necessary to confirm this). This method: (1) is a leaching method rather than a digestion method; (2) employs a temperature of 75 "C, which is less than that required for digestion; and (3) uses a leaching time of 30 min in contrast to many digestions where heating can be prolonged; a further point that may be relevant is the Table 6 Least significant difference test (5% level) for CCA on Rotorua raw data.Means with a bar in the same column are not significantly different Laboratory Mean cu B C D F 0 N M H J L I E P A Cr J M I 0 L C D F N H A B P E As 0 D F I N A J H L P M C E B 0.250 0.243 0.242 0.239 0.239 0.238 0.238 0.237 0.229 0.225 0.224 0.223 0.221 0.220 0.495 0.489 0.471 0.459 0.458 0.449 0.427 0.424 0.416 0.41 1 0.410 0.405 0.397 0.380 0.358 0.300 0.298 0.295 0.295 0.294 0.291 0.289 0.284 0.282 0.280 0.276 0.273 0.262 LSD = 0.010 LSD = 0.032 addition of sodium sulphate to the solutions of leachates and standards for atomic absorption analysis. These points could be responsible for ensuring more control over sample prepara- tion and analysis and hence improved precision. The greater variation of data obtained with the other methods offers an opportunity for improvement in precision.The LSD tests for CCA for the Queensland data, with all concentrations combined (Table 7), also suggest different performance by different methods. Determination of Statistical Outliers Stragglers and outliers in both data sets were identified following recommendations of both I S 0 5725 and the AMC. When using the Cochran and Dixon tests (IS0 method), several stragglers and outliers were found (Table 8) (stragglers and outliers are data points significant at the 5 and 1% levels, respectively). Also included in Table 8 are points lying more than three standard deviations from the mean of each concentration. From the results of these three tests, it is clear that coincidence of straggler/outlier identification was un- usual.None of the stragglers/outliers could be tagged as spurious as could have occurred from transcription errors or an order of magnitude error. For ANOVA stragglers were retained, while outliers were discarded and were replaced with values generated from a missing value estimation. Statistical outliers found for the Queensland data were from Laboratory D (Cu), lowest concentration (0.040% of Cu), and Laboratory E (Cr), second lowest concentration (0.070% of Cr). In using Cochran's test, based on variance ratio, only the signficant concentration level is identified. The variance for each cell is then inspected and the cell with the highest variance is flagged as the outlier. The offending member of the cell is then found by inspection.This identification can sometimes be ambiguous and often Cochran's test needs to be re-applied. ~ ~~~ Table 7 Least significant difference test (5% level) for CCA in the Queensland samples. Means with a bar in the same column are not significantly different Laboratory Mean c u 1 G D A C F H E B Cr G C F H 1 A D E B 0.125 0.123 0.118 0.117 0.117 0.114 0.111 0.109 0.102 0.199 0.194 0.188 0.187 0.186 0.184 0.183 0.183 0.172 As G 0.218 1 0.187 0.186 0.185 0.183 0.181 0.173 0.167 0.158 LSD = 0.009 I LSD = 0.011 LSD = 0.013344 ANALYST, APRIL 1991, VOL. 116 Following the AMC approach,23J4 values of ai and aii statistics were calculated; ai values greater than 2.5 identify a laboratory with data significantly different to the grand mean. The aii value is a within-concentration cell statistic, indicating variability within a laboratory.Values of (xi and a!, greater than 2.5 suggest a systematic laboratory bias or a discrepant replication, respectively (Table 9). For As, the ai values were significant for three of the six concentrations for Laboratory 0; Laboratory 0 was therefore tagged as being an outlying laboratory, and the As data were re-analysed without the Laboratory 0 data. Twelve values of ai, for Cu and Cr were greater than 2.5 as were nine for As; without Laboratory 0 data for As, six values were greater than 2.5. When ai and aii for the Queensland data were calculated, one (xi was greater than 2.5, namely, 2.57 (Cu, Laboratory B; mean concentration 0.223% of Cu). The significant aii values were 2.99 (Cu, Laboratory D; 0.030 and 0.040% of Cu; lowest concentration cell) and 2.99 (Cr, Laboratory E; 0.060 and 0.070% of Cr; second lowest concentration cell).No ai and aii values for As were greater than 2.5. Again there was a lack of agreement between outliers that resulted from tests following either the I S 0 5725 or the AMC method. The AMC calculation was more sensitive in that for the Rotorua data it picked 33 points as outliers compared with 16 points for the I S 0 method. Comparison With Nominal Target Values Comparison of the laboratory analyses with the nominal target values was assessed in two ways. Firstly, the recovery of metal from the samples was considered. Secondly, with outliers identified by the Cochran and Dixon mean +30 tests removed, ANOVA was carried out.Least significant differ- ence tests were performed when ANOVA was significant. By using the mean values obtained from determinations of each concentration level, it was found that 95% (Cu), 100% (Cr) and 93% (As) [or 91% (As) if Laboratory 0 data were excluded] of the nominal target concentrations were recovered (Table 10). The LSD tests were employed to compare the means obtained by each laboratory for each element. In Table 11, the LSD test is for combined concentration data. Of the six laboratories, three were significantly above and three signifi- cantly below the nominal target values for Cr. For As, only one laboratory was above the nominal target value with eight below, while for Cu, no laboratories were above the nominal Table 8 Identification of stragglers and outliers in the Rotorua data [IS0 method (reference 19)] Mean c u 0.048 0.094 0.189 0.272 0.357 0.440 Cr 0.087 0.177 0.349 0.507 0.666 0.824 As 0.056 0.119 0.231 0.339 0.449 0.553 * ns = not significant. t Significant at 1% level.$ Significant at 5% level. Mean k3a (0.032; M)? ns ns ns ns (0.040,0.036; M)t ns ns ns ns ns ns ns (0.177; C)t (0.439,0.446; 0)t ns (0.772,0.713; 0)t (0.152; P)t Outlier M (0.032) - P (0.152) - P (0.312) M (0.383) M (0.040,0.036) - - - P (0.562) A (0.724) 0 (0.085) C (0.177) C (0.277); 0 (0.439,0.446) 0 (0.712,0.713) c (0.102) - Table 9 Values of ai and auii for CCA from the Rotorua data c u Cr As As (excluding Laboratory 0 data) Concentration level ai (Lab.) 1 2.62 (M) 2 3 - - - - - - - 4 - - 5 - - - 6 - aij (Lab. ; concentration) 2.72 (B; 0.043) 2.62 (M; 0.032) 2.75 (M; 0.101) 2.92 (A; 0.169) 2.92 (A; 0.215) 2.83 (P; 0.152) 2.87 (A; 0.267) 2.92 (B; 0.314) 2.57 (P; 0.312) 3.50 (P; 0.360) 4.17 (M; 0.502) 4.05 (M; 0.383) - - a,j (Lab.; ai (Lab.) concentration) 2.66 (M) 2.61 (M; 0.036) - 4.77 (M; 0.101) - 2.71 (M; 0.193) - 2.71 (M; 0.223) - 2.98 (I; 0.331) - 3.03 (P; 0.268) - 2.58 (J; 0.603) - 2.54 (A; 0.654) - 2.84 (A; 0.572) - 3.43 (P; 0.647) - 2.63 (A; 0.724) - 4.51 (A; 0.849) - - - - ajj (Lab.; ai (Lab.) concentration) 2.87 (B) 3.63 (0; 0.085) - 3.34 (0; 0.053) 2.80 (0) 3.51 (C; 0.102) - - 2.59 (C; 0.216) - 3.24 (C; 0.177) - 2.74 (I; 0.193) 3.23 (0) 2.52 (A; 0.381) - 3.76 (C; 0.335) aii (Lab.; ai (Lab.) concentration) 2.79 (B) - 3.48 (C; 0.102) - 2.62 (C) 2.63 (C; 0.216) - 3.29 (C; 0.177) - 2.78 (I; 0.193) - 3.68 (C; 0.335) - - 3.41 (0) 3.21 (B; 0.552) - 3.38 (B; 0.552) - -ANALYST, APRIL 1991, VOL.116 345 Table 10 Relative standard deviations, recovery of nominal target samples and 95% confidence limits (CL) for both trials (g of metal per 100 g of wood unless other units given) Rotorua Queensland c u Cr As Mean 0.048 0.094 0.189 0.272 0.357 0.440 0.087 0.177 0.349 0.507 0.666 0.824 0.056 0.119 0.231 0.339 0.449 0.553 As* 0.056 0.116 0.228 0.331 0.443 0.542 Nominal RSDR(%0) target 9.0 0.050 5.5 0.099 5.8 0.196 5.2 0.291 5.4 0.384 5.3 0.476 15.9 0.088 9.3 0.174 8.3 0.345 8.7 0.512 9.3 0.675 8.3 0.835 * Excluding Laboratory 0 data. 26.1 0.063 10.0 0.125 7.1 0.247 9.6 0.366 7.8 0.483 8.2 0.598 23.4 0.063 7.1 0.125 5.9 0.247 5.3 0.366 6.6 0.483 3.3 0.598 Recovery (%) (mean x 100hominal target) 96 96 98 94 93 93 95 * 2 102 102 101 99 99 99 89 96 94 93 93 93 93 * 2 89 93 92 90 92 91 91 k 1 l o o k 2 CL 0.046-0.049 0.093-0.096 0.185-0.192 0.267-0.276 0.35 1-0.363 0.433-0.448 0.082-0.091 0.172-0.182 0.340-0.358 0.494-0.52 1 0.6474686 0.803-0.845 0.052-0.061 0.1 15-0.123 0.226-0.236 0.329-0.349 0.438-0.460 0.539-0.567 0.052-0.060 0.114-0.119 0.224-0.233 0.326-0.337 0.434-0.453 0.536-0.548 Mean 0.031 0.037 0.090 0.135 0.175 0.223 0.056 0.049 0.152 0.224 0.258 0.379 0.062 0.028 0.183 0.272 0.296 0.250 RSDR(%o) 17.1 17.0 8.3 4.8 8.6 7.3 8.8 18.2 5.4 4.2 6.9 4.8 15.3 38.9 9.1 8.1 10.6 8.4 CL 0.028-0.033 0.034-O.040 0.086-0.093 0.132-0.139 0.167-0.182 0.2 15-0.23 1 0.053-0.058 0.045-0.053 0.148-0.156 0.2 194.228 0.2494266 0.370-0.388 0.057-0.067 0.023-0.034 0.175-0.191 0.261-0.283 0.281-0.312 0.240-0.260 target value and eight laboratories were below.This spread of results is consistent with the recoveries of metal obtained from the participating laboratories. Quality Assurance The ranges of the results as a percentage of the grand mean for Cu, Cr and As are 12.8, 26.4 and 32% for the Rotorua data and 20.3, 14.0 and 32.9% for the Queensland data, respect- ively; if Laboratory 0 is omitted for the Rotorua As data, the range is 13.2% and similarly without Laboratory G for Queensland As data, the range is 15.8%. These ranges are very broad. Both trials appear to have produced, overall, comparable performances for As, but the Rotorua data are less spread for Cu whereas the Queensland Cr data are less variable.This behaviour is also seen in the RSD (Table 10). The 95% confidence limits (Table 10) for both data sets summarize the most likely range for the concentrations of CCA in samples. The information on variability found in this study has shown that there is room for improvement in analyses. One outlying laboratory was identified in addition to a large number of statistical outliers. Such outcomes detract from the confidence of analysis for CCA in timber treated with preservative. Uncertainties in analyses have ramifications in the timber industry both in terms of adequate preservative levels for timber protection and in the commercial operations of preservative companies and timber-treatment sites.The most satisfactory conclusion from this study would be an increased awareness of the need for laboratories to operate quality management and quality assurance programmes. The precision parameters produced in this work should be used by analysts in preservative analysis as minimum guide- lines to acceptable precision. The accuracy of analyses for CCA in timber is one area that remains a major concern. Only nominal target values were available in the Rotorua study. The mean values produced, averaged over all laboratories, provide some information on concentrations and therefore recovery of metal in the samples. There is a lack of certified reference materials in the area of timber treated with preservative generally. Such materials would facilitate an evaluation of the accuracy of different analytical methods and allow bias of various methods to be addressed directly.Recommendations on quality programmes by timber- preservation laboratories have been presented previously.2sJ6 The same general principles can be applied in the analytical procedures for the determination of CCA in timber. Conclusion The I S 0 5725 and AMC methods have been discussed with reference to analysing analytical data from two interlabora- tory trials for the determination of CCA in timber. A proposed method of analysis was supplied to laboratories in the Rotorua trial and was employed by a minority of laboratories. Results from this method did not appear to suggest improvements in precision that may have been expected from the operation of a standardized method.In the Queensland trial there was no proposed method. Values of repeatability and reproducibility for the two trials were calculated and compared and statistical outliers were identi- fied. Comparisons of the experimental data with nominal target values for CCA in the Rotorua trial showed good agreement for Cr, but less agreement for Cu and As. Finally, the continued application of quality assurance programmes is346 ANALYST, APRIL 1991, VOL. 116 J. Norton of the Queensland Forest Service is thanked for tory The participating laboratories are thanked for their Table 11 h a s t Significant difference test (5% level) for CCA-data permission to include data from the Queensland interlabora- for all concentrations combined-Rotorua data. Means with a bar in the same column are not signficantly different Laboratory Cu B Nominal target M C D F 0 N H J L P I E A Cr M J I 0 L C Nominal target D F N A H B P E As 0 Nominal target D F I N A J H L P C M E B Mean 0.250 0.249 0.243 0.243 0.242 0.239 0.239 0.238 0.237 0.229 0.225 0.225 0.224 0.223 0.220 0.496 0.495 0.471 0.459 0.458 0.449 0.438 0.427 0.424 0.416 0.414 0.411 0.405 0.399 0.380 0.351 0.314 0.300 0.298 0.295 0.295 0.294 0.291 0.289 0.284 0.282 0.281 0.280 0.273 0.262 LSD = 0.011 LSD = 0.029 LSD = 0.Od co-operation, -and P.Gray of the Analytical Chemical Laboratory, Wood Technology Division, FRI, is thanked for her help with sample preparation. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 of prime importance in continually improving the analytical performance of the participating laboratories.19 20 21 22 23 24 25 26 References Levi, M. P., Record of the Nineteenth Annual Convention British Wood Preservers Association, Cambridge, 1969, 113- 127. Annual Timber Preservation Statistics, New Zealand Forest Service, Rotorua, 1986. Williams, A. I., Analyst, 1972, 97, 104. Lambert, M. J., J. Inst. Wood Sci., 1969, 24, 27. Johanson, R., Holzforschung, 1973,27, 126. Staccioli, G., and Tamburini, U., Mater. Org., 1982, 17, 199. Williams, A. I., Analyst, 1968, 93, 611. Caulfield, D. F., and Steffes, R. A., Wood Sci., 1971, 4, 106. Visapaa, A., and Mansikkamaki, P., Pap. Puu, 1979,61, 253. Fuller, C. W., and Clare, D. J. M., Anal. Proc., 1984,21, 113. Meyer, J. A., and Siau, J. F., Wood Sci., 1972, 5 , 147. Australian Standard AS 1605-1974, Standards Association of Australia, Sydney, 1974. British Standard BS 5666 : Part 3 : 1979, British Standards Institution, London, 1979. American Wood-Preservers’ Association Standard A10-82, Stevensville, MD, 1982. American Wood-Preservers’ Association Standard A1 1-83, Stevensville, MD, 1983. American Society for Testing and Materials, ASTM: D1628-83, Philadelphia, PA, 1983. American Wood-Preservers’ Association Standard A7-75, Stevensville, MD, 1975. Drysdale, J. A., Dickinson, D. J., and Levy, J. F., Muter. Org., 1980, 15, 287. International Organization for Standardization 5725 1986(E). Precision of Test Methods-Determination of Repeatability and Reproducibility for a Standard Test Method by Interlabora- tory Tests, Geneva, 1986. Standards Association of New Zealand. MP 3640 : 1988. Specifi- cation of the Minimum Requirements of the New Zealand Timber Preservation Council, Wellington, 1988. Robertson, J. M., Technical Guide 5, Testing Laboratory Registration Council of New Zealand, Auckland, 1987. Research Report 6, Queensland Department of Forestry, Brisbane, 1986/7, p. 112. Analytical Methods Committee, Analyst, 1987, 112, 679. Analytical Methods Committee, Analyst, 1989, 114, 1489. Dawson, B. S. W., Cummins, N. H. O., Parker, G. F., Cowan, F. J., and Hong, S. 0. Analyst, 1989, 114, 827. Dawson, B. S. W., Parker, G. F., Cowan, F. J . , Croucher, M., Hong, S. O., and Cummins, N. H. O., Anal. Chim. Acta, 1990,236,423. Paper 0104551 I Received October 9th, I990 Accepted December 4th, 1990

ISSN:0003-2654    

DOI:10.1039/AN9911600339

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, APRIL 1991, VOL. 116 347 Comparison of Microwave and Conventional Extraction Techniques for the Determination of Metals in Soil, Sediment and Sludge Samples by Atomic Spectrometry Joop Nieuwenhuize and Carla H. Poley-Vos Delta Institute for Hydrobiological Research (DIHO), Vierstraat 28, 440 1 €A Yerseke, The Netherlands Adrianus H. van den Akker Institute for Inland Water Management (DB W/RIZA), P.O. Box 17, 8200 AA Lelystad, The Netherlands Wouter van Delft" State Institute for Quality Control of Agricultural Products (RIKIL T), P.O. Box 230,6700 EA Wageningen, The Netherlands A description is given of an extraction method for the determination of metals in sediment and soil samples using aqua regia [HCI-HN03 (3 + I ) ] and a microwave oven. The elements Cd, Cr, Cu, Fe, Mn, Pb and Zn were measured in the solutions by means of atomic absorption spectrometry (flame or electrothermal) and inductively coupled plasma atomic emission spectrometry.The results for the analysis of six reference materials after microwave aqua regia extraction showed close agreement with the stated values. A comparison between microwave extraction and conventional reflux extraction for 30 samples showed that the former method gives the same or slightly higher results for the seven elements tested in sediments, soil and suspended matter. Keywords: Microwave extraction; reflux extraction; sediment; soil; atomic spectrometry Extraction is an important but generally very time-consuming stage in the determination of metals in sediment and soil samples by atomic spectrometry.Digestion methods are commonly used involving different mixtures of acids (HN03- HCI, HN03-H2S04, HN03-HC104, etc.) , occasionally in combination with HF in order to dissolve the silica matrix. For instance, in accordance with Netherlands NEN and German DIN standards,lJ metals in soil and sludge are extracted by heating the samples with aqua regia [HC1-HNO3 (3 + l)] in flasks fitted with reflux condensers. These methods, however, have several drawbacks. If open systems are used, there is a risk of atmospheric contamination and loss of volatile ele- ments. In order to avoid these problems, wet digestion is often carried out in polytetrafluoroethylene (PTFE) bombs.3.4 As only a small amount of material can be used for this purpose, the sample needs to possess a high degree of homogeneity.Moreover, the bombs are not fitted with valves, and only a limited amount of energy can therefore be supplied, which might influence the extraction process. A recently developed alternative5-7 is the wet extraction technique which makes use of the rapid and attractive method of microwave heating. Microwave extraction has previously been tested extensively for organic-rich plant materia1,s.g using certified reference materials and real samples, and the method had also been compared with conventional digestion methods. van de Wall et a1.10 have shown that for certain aquatic sediments and reference materials, reflux extraction in ac- cordance with the Dutch national standard NEN 64651 and microwave extraction produce comparable results. van Delft and Vos" compared five extraction methods for measuring mercury in soil.They found close agreement between the results after microwave extraction and those after extraction with HN03 in a closed PTFE bomb and with neutron activation analysis. In this paper, a microwave extraction method is described prior to the measurement of seven metals in different types of sediment and soil. These samples were taken from salt * To whom correspondence should be addressed. marshes, aquatic sediments, suspended matter (obtained with the aid of a sludge centrifuge), arable land, grassland and woodland. The experiments were carried out by three laboratories (A, B and C). Use was made of reference materials from the Community Bureau of Reference (BCR) and International Atomic Energy Agency (IAEA) and of real samples which were also digested by the conventional reflux extraction method.Experimental Instrumentation All sediment and soil samples were decomposed in a microwave oven (Model MDS-81D, CEM, Indian Trail, NC, USA), equipped with a microwave power range from 0 to 100% (600 W). The microwave oven was fitted with a rotary table on which a maximum of 12 120 ml perfluoroalkoxy (PFA) digestion vessels can be placed. The vessels were fitted with a pressure-relief valve and were sealed in a capping station. The microwave oven allows both time and energy to be programmed in a maximum of three steps. Conventional extraction was carried out in Pyrex tubes or in flasks fitted with reflux condensers.1.2 The elements Fe, Mn and Zn were measured by laboratory A with a Perkin-Elmer Model 2380 atomic absorption spectrometer fitted with an air-acetylene burner.The el- ements Cd, Cu, Cr, and Pb were measured with a Perkin- Elmer Model 3030 instrument fitted with a Zeeman-effect background correction system, an HGA-600 furnace and an AS-60 autosampler. For the measurement of Cd, Cu and Pb, pyrolytic graphite coated graphite tubes were used, fitted with a L'vov platform. The measurements of Cd and Pb were carried out in accordance with the stabilized temperature platform furnace concept .1*-14 In laboratory B the elements were determined with a Perkin-Elmer Model 600 inductively coupled plasma (JCP) atomic emission spectrometer with measurements for each element being carried out at two wavelengths.Laboratory C carried out the Cd, Cu, Pb and Zn measure- ments in an air-acetylene flame, by using a Perkin-Elmer Model 5000 atomic absorption spectrometer. Cadmium was348 ANALYST, APRIL 1991, VOL. 116 measured following complexation with ammonium pyrrolidin- 1-yldithioformate [ammonium pyrrolidinedithiocarbamate (APDC)] and extraction with isobutyl methyl ketone (IBMK). The elements Cr, Fe and Mn were measured with a Perkin-Elmer Model 6000 ICP instrument with measurements for each element again being carried out at two wavelengths. In all instances, standardization was carried out by using a calibration graph. The standard solutions contained 12 ml of 37% HCI and 4 ml of 65% HN03 per 100 ml. Recoveries of standard solutions added to the samples prior to analysis were between 95 and 105%.Table 1 gives the optimum instrumental conditions for the measurements. Reagents All chemicals were of analytical-reagent grade. Standard solutions of Cd, Cr, Cu, Fe, Mn, Pb and Zn were prepared from Merck Titrisol ampoules or BDH standard solutions (1000 mg 1-1). Nitric acid (65%) and hydrochloric acid (37%) were used for both microwave and conventional extraction. For Cd and Pb, when measured with the Perkin-Elmer 3030 instrument, a chemical modifier was used consisting of 25 g of NH4H2P04 and 1.2 g of Mg(N03)2 per litre. For the Cd measurements in the flame, the element was first complexed with 1% m/v APDC solution at a pH of 7 and extracted with IBMK saturated with water.15 Table 1 Instrumental conditions for the determination of Cd, Cr, Cu, Fe, Mn, Pb and Zn Graphite furnace conditions- - ” Wave- length/ Element” nm Cd 228.8 Cr 357.9 Cu 324.8 Pb 283.3 Flame conditions- Wave- length/ Element? nm Cd 228.8 c u 324.7 Fe 248.3 Mn 279.5 Pb 283.3 Zn 213.9 ICP conditions- Plasma flow Auxiliary flow Sample uptake R.f.power Viewing height Laboratory B Laboratory C Slit- Charring Cooling Atomiza- width/ tempera- tempera- tion tem- nm ture/’C ture/oC peraturePC 0.7 1100 - 1450 0.7 1350 - 2300 0.7 1200 30 2250 0.7 1150 - 1550 Slit- width/ nm 0.7 0.7 0.2 0.2 0.7 0.7 15 1 min-1 0.3-0.5 1 min-1 1 ml min-I 1250 W 15 mm ‘Crossflow’ nebulizer, 180 kPa ‘V-groove’ nebulizer, 170 kPa Element Wavelengthhm Cd 214.438; 228.802 Cr 205.552; 267.716 c u 223.008; 327.396 Fe 238.204; 239.562 Mn 257.610; 259.370 Pb 405.783; 220.353 Zn 202.548; 213.856 * Zeeman-effect background correction and pyrolytic graphite An air-acetylene flame and deuterium background correction coated graphite tubes fitted with a L’vov platform were used.were used. Standards Validation of the methods described here was performed by using four Certified Reference Materials (CRMs) from the Community Bureau of Reference (BCR) and two reference materials from the International Atomic Energy Agency (IAEA): BCR CRM 141 Soil-calcareous loam; BCR CRM 142 Soil-light sandy; BCR CRM 143 Sewage sludge-soil; BCR CRM 145 Sewage sludge; IAEA CRM SL-1 Lake Sediment; and IAEA CRM SOIL 7 Soil. The BCR certificates report two sets of values: total trace metal content and the aqua regia soluble fractions.Through- out this study the aqua regia soluble values are used. The aqua regia values are not classified as ‘certified’. The IAEA certificates report only the total trace metal content. Procedures Microwave dissolution A 500 or 1000 mg portion of the (freeze or oven) dried and ground material was transferred into the 120 ml PFA vessels, and 16 ml of aqua regia (12 ml of 37% HCI and 4 ml of 65% HN03) were added. Before the vessels were sealed, they were first shaken until all the C 0 2 from the CaC03 had escaped. During this study the following microwave programme was used: step 1, 30% power for 1 min; step 2, 80% power for 4 min; and step 3, 100% power for 60 min. After the samples had been cooled, the PFA vessels were opened, the decomposition products were transferred into 50 or 100 ml polypropylene calibrated flasks and the solutions made up to volume with de-mineralized water.After homo- genization, the solutions were filtered. Blanks were treated in the same way as the samples. Conventional reflux extraction A 2000 mg portion of the (freeze or oven) dried and ground material was transferred into a Pyrex tube, and 16 ml of aqua regia were added. Before placing a reflux condenser on the tube, it was first shaken until all the C 0 2 from the CaC03 had escaped. The tube was then heated for 2 h, reflux cooling being used. Afterwards the coolers were rinsed, the decompo- sition products were transferred into 100 ml calibrated flasks and the solutions made up to volume with de-mineralized water. After homogenization, the solutions were filtered.Blanks were treated in the same way as the samples. Results and Discussion Optimization of the Microwave Extraction Method Aqua regia was used as a decomposition reagent for both the microwave and conventional digestion procedures. For reflux extraction, 80-100% of the metals in soils and sewage sludges was dissolved with aqua regia compared with total diges- tion.16.17 An optimum microwave oven programme was developed for the microwave extraction method. In order to increase the pressure in the PFA vessels gradually, a micro- wave programme consisting of three stages was used. In the third stage, the power supplied is set at loo%, which then permits optimization of the decomposition time. As regards the extraction of a sludge sample from the river Maas, the duration of the final stage was varied by laboratory B from 10 to 100 min, with intermediate steps of 10 min.The results of these experiments10 showed that after 60 rnin no further significant changes took place for the seven elements studied here. On this basis the following microwave programme was chosen: 1 rnin at 30% power, 4 rnin at 80% and 60 rnin at 100%. When optimizing the microwave programme, the maximum pressure in the PFA vessels should not rise above 830 kPa otherwise the pressure-relief valve will vent and material might be lost from the vessels. When carrying out the variousANALYST, APRIL 1991, VOL. 116 349 microwave dissolutions with this microwave programme, the valves did not open. It should be noted that an optimum microwave programme should be developed for each type of microwave oven and for each type of matrix.Accuracy of the Microwave Extraction Method In order to ascertain the accuracy of the optimized microwave extraction method, laboratories A, B and C each digested two different CRMs, from the IAEA (CRMs SOIL 7 and SL-1) and the BCR (CRMs 141 and 142, and 143 and 145). Each laboratory measured Cd, Cr, Cu, Fe, Mn, Pb and Zn in the digested solutions. The results generally showed close agree- ment with the stated values for all seven elements in the reference materials (Table 2). In addition, laboratories B and C each extracted two reference materials using the conven- tional reflux method. These results showed that there were no significant differences between this method and the results after microwave extraction for Cd, Cu, Mn, Pb and Zn.For BCR CRMs 141 and 142 (laboratory C), significantly higher values were found for Cr and Fe after microwave extraction (Table 3). No explanation could be given for the high Pb content found in CRM 141 after microwave extraction. In each series of 12 microwave extractions two blanks were included; these received the same treatment as the samples and were used to calculate standard deviations (SDs) and Table 2 Results of the analysis of IAEA and BCR CRMs after microwave extraction with aqua regia. Results for Cd, Cr, Cu, Mn, Pb and Zn expressed in pg g-1; results for Fe expressed in mg g-l Element Cd Cr c u Fe Mn Pb Zn Element Cd Cr c u Fe Mn Pb Zn Element Cd Cr c u Fe Mn Pb Zn IAEA CRM SL- 1 * 0.26 k 0.05 104 f 9 30.0 t 5.6 67.4 t 1.7 3400 t 160 37.7 k 7.4 223 * 10 BCR CRM 1430 31.5 k 2.0 208 k 20 236 f 13 26.3 k 0.7 935 f 100 1317 * 55 1301 t 60 BCR CRM 141 D 0.30 f 0.13 53 f 9 31.2 f 2.3 23.7 f 1.7 512 f 63 26.3 f 5.8 7 0 t 11 * Mean and confidence limit of the mean (p = 0.05).t Averages f standard deviations (n = 10). * Median and confidence interval of the median (p = 0.05). 9 Aqua regia results. mean f standard deviation. This work, laboratory A t 0.25 5 0.02 107 f 10 27.0 t 1.0 76.3 5 1.2 3770 f 80 38.5 t 3.3 227 f 19 This work, laboratory, Bt 30.8 f 0.7 192 t 7 226 f 8 25.6 f 0.6 940 f 22 1309 k 34 1234 k 26 This work, laboratory C t 0.31 f 0.01 57.1 f 3.0 33.1 f 0.5 24.5 f 0.4 507 f 8 35.4 f 1.6 76.8 f 1.6 IAEA CRM SOIL 73; 1.3 (1.U2.7) 60 (49/74) 11 (9/13) 25.7 (25.2/26.3) 631 (604/650) 60 (55/71) 104 (101/113) BCR CRM 1450 16.8 t 1.6 85.2 t 16.3 416 f 24 no data 220 f 15 332 f 22 2772 t 209 BCR CRM 1420 0.22 k 0.10 44.4 f 5.4 25.3 f 2.0 17.5 f 0.4 527 k 35 30.9 f 6.7 79.6 k 11.7 This work, laboratory A t 1.22 f 0.03 51.4k 1.8 9.6 k 0.4 25.0 t 0.2 697 * 9 62.7 f 3.0 104 f 2.2 This work, laboratory BT 15.9k 2.1 69.6 f 3.3 399 f 19 7.5 f 0.3 223 k 9 311 f 15 2608 f 96 This work, laboratory C t 0.22 f 0.02 48.6 t 2.8 26.6 f 0.4 18.5 f 0.2 550 k 14 33.9 f 2.3 85.6 f 1.2 Table 3 Avcragc values of metals in BCR CRMs, measured after reflux extraction (method A) or microwave extraction (method B). Results for Fe expressed in mg g-1, for all other elements in pg g-l (n = 10) BCR CRM 141 f SD BCR CRM 142 k SD Element Method A Method B Method A Method B Cd 0.33 t 0.04 0.31 k 0.01 0.19 k 0.02 0.22 t 0.02 Cr 49.8 f 0.6 57.1 t- 3.0 40.7 f 0.8 48.6 +.2.8 c u 32.3 t 0.4 33.1 -t 0.5 25.9 f 0.4 26.6 f 0.4 Fe 23.6 f 0.2 24.5 f 0.4 17.9 f 0.2 18.5 k 0.2 Mn 496 t 4 507 t 8 539 f 5 550 f 14 Pb 25.3 +- 0.9 35.4 f 1.6 33.2 f 2.0 33.9 t 2.3 Zn 74.2 t 0.6 76.8 t 1.6 83.7 f 0.7 85.6 f 1.2 BCR CRM 143 Element Cd Cr c u Fe Mn Pb Zn Method A 30.9 k 0.5 195 t 2 224 f 3 26.2 t 0.7 937 f 8 1266 f 20 1231 t 17 Method B 30.8 f 0.7 192 k 7 226 f 8 25.6 f 0.6 940 f 22 1309 f 34 1234 f 26 BCR CRM 145 Method A Method B 15.8 f 0.5 15.9 f 2.1 68.8 k 3.7 69.6 t 3.3 394 f 12 399 f 19 8.1 f 1.3 7.5 f 0.3 241 f 8 223 f 9 316 k 11 311 f 15 2642 k 101 2608 k 96350 detection limits.A comparison of the blanks used in both extraction methods showed that the detection limits are the same or lower if microwave extraction is used. The principal advantage of the proposed method is that extraction by means of a microwave oven is considerably faster than conventional reflux extraction (1 versus 3 h). The microwave oven also has a practical benefit: after digestion, the PFA vessels need only be rinsed with de-mineralized water before being re-used, whereas the glassware used in the conventional method must be warmed with dilute nitric acid before being re-used. Interlaboratory Calibration of Microwave Extraction Using a Sludge Sample From the River Maas The above experiments show that all three laboratories found close agreement between the stated values and results for the seven elements after microwave extraction.As each laboratory analysed different reference materials, an interlaboratory calibration was carried out with the aid of a sludge sample from the river Maas. Each laboratory performed three independent measurements of seven Table 4 Results of an intercalibration study carried out with a sludge sample from the river Maas, after microwave dissolution. Results for Fe expressed in mg g-1, for all other elements in pg g-l (n = 3 for each laboratory) Element Cd Cr c u Fe Mn Pb Zn Average 42.9 174.0 175.9 44.2 2091 496 2867 Reproducibility - SD RSD(%) 1.6 3.7 20.9 12.0 8.4 4.8 3.8 8.6 162 7.7 21 4.3 172 6.0 Repeatability SD RSD(%) 1.1 2.5 9.4 5.4 6.4 3.6 0.9 2.0 58 2.8 8.2 1.7 61 2.1 ANALYST, APRIL 1991, VOL.116 elements in the sample after extraction with aqua regia in the microwave oven. The results of this experiment are summarized in Table 4. With the exception of Cr, the relative standard deviations (RSDs) are always lower than 10%. This implies that after microwave extraction the three laboratories produced com- parable results for the seven elements. Comparison Between Microwave and Conventional Reflux Extraction Using Real Samples In order to compare the results after microwave extraction with those after conventional reflux extraction for real samples, both extraction methods were applied to real samples selected from various types of sediment and soil samples from different locations in the Netherlands. Laboratory A analysed soil samples from a salt marsh, laboratory B examined aquatic sludge and suspended matter samples obtained with the aid of a sludge centrifuge, while laboratory C analysed samples from grassland, arable land and woodland.Table 5 shows the results of regression analysis for method A (reflux extraction) versus method B (microwave extraction) over the concentration range studied. The slopes and intercepts are close to 1 and 0, respectively, showing an excellent agreement between the two methods. The results show that for Cd and Cr no significant differences were found between the results after conventional reflux extraction and those after microwave extraction. In general, no significant differences were found for Cu, Mn, Pb and Zn. In instances where differences were found between the two methods, the measurements after microwave extraction generally gave slightly higher values, which indi- cates a better yield.Conclusion Microwave extraction with aqua regia offers a good alternative to conventional reflux extraction for use in the determination Table 5 Regression analysis of results for metals in 39 different soil and sediment samples after reflux extraction (method A) and microwave extraction (method B). Results for Fe expressed in mg g-l, for all other elements in pg g-I Element Method A Method B r* a? b$ Laboratory A: results for ten salt marsh samples- Cd 4.63 4.62 0.91 0.99 0.04 Cr cu 39.9 43.3 0.99 1.03 1.18 Fe 31.7 29.8 1.00 0.94 0.11 Mn 989 105 1 1 .oo 1.01 51.37 Pb 70.8 74.5 0.97 0.94 7.55 Zn 268 281 0.99 1 .oo 13.83 - - - - - Laboratory B: results for nine aquatic sediments and suspended matter- Cd 104 103 1 .OO 0.99 -0.23 Cr 69.6 70.4 1.00 1.07 -3.65 c u 195 198 1 .oo 1.02 -0.11 Fe 26.8 27.7 1 .oo 1.05 -0.34 Mn 2365 2354 1 .oo 0.99 -23.05 Pb 197 187 1.00 0.99 -0.34 Zn 141 1 1427 1.00 1.01 -0.61 Laboratory C: results for 20 soil samples- Cd 0.36 0.35 0.99 1.02 -0.02 Cr c u 15.1 15.0 0.99 0.92 1.12 Fe Mn Pb 24.7 27.4 0.99 1.09 0.52 Zn 41.2 41.4 1 .oo 1 .oo 0.05 - - - - - - - - - - - - - - - Range 1.6-8.1 18.0-65.4 20.54.1 33.0-1 32.0 15W13 - 613-1438 0.5-680 6.7-1 90 2.8-1083 2.9-61.7 4.4-618 474496 112-1 1280 0.02-1.47 - 0.5-49.0 - - 9.5-88.2 4.2-95.2 * r = Correlation coefficient.t a = Slope. j: b = y-axis intercept.ANALYST, APRIL 1991, VOL. 116 35 1 of Cd, Cr , Cu , Fe, Mn , Pb and Zn in sediment, sludge and soil samples.Microwave extraction involves rapid extraction in closed PFA vessels, with little risk of outside contamination. The results obtained after microwave extraction for seven elements in six reference materials (from the IAEA and BCR) showed close agreement with the stated values. The results after reflux extraction also demonstrated that there were no significant differences between this method and microwave extraction. By using the microwave oven and reflux extraction methods, 30 sediment and soil samples were dissolved. The results after microwave extraction of the seven elements were equal to or slightly higher than those obtained after reflux extraction. An interlaboratory calibration between three laboratories was carried out with a sludge sample from the river Maas.After extraction in the microwave oven, the seven elements were measured; the RSD was between 3.7 and 8.6%, except for Cr, for which it was 12.0%. References 1 NEN 6465. Water and Air-sample Preparation of Sludge, Water-containing Sludge and Air Dust for the Determination of Elements by Atomic Absorption Spectrometry-Destruction With Nitric Acid and Hydrochloric Acid, Netherlands Normali- sation Institute, Delft, 1981 (in Dutch). 2 DIN 38 414 Part 7, German Standard Methods f o r the Examination of Water, Waste Water and Sludge; Sludge and Sediments (Group S ) ; Digestion Using Aqua Regia f o r Subse- quent Determination of the Acid-soluble Portion of Metals ( S 7), 1983. 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Stoeppler, M., and Backhaus, F., Fresenius Z. Anal. Chem., 1978, 291, 116. Breder, R., Fresenius Z. Anal. Chem., 1982,313, 395. Kingston, H. M., and Jassie, L. B., Anal. Chem., 1986, 58, 2534. Lamothe, P. J., Fries, T. L., and Consul, J . J . , Anal. Chem., 1986,58, 1881. Papp, C. S. E., and Fischer, L. B., Analyst, 1987, 112, 337. Nieuwenhuize, J., and Poley-Vos, C. H., At. Spectrosc., 1989, 10,148. van Delft, W., Horstman, H. J . , Lammers, H., and Vos, G., in CAS 5. Colloquium Atomspectrometrische Spurenanalytik, ed. Welz, B., Perkin-Elmer, Uberlingen, 1989, p. 603. van de Wall, C. G. J . , van den Akker, A. H . , and Stoks, P. G. M., H 2 0 , 1988, 21, 320 (in Dutch). van Delft, W., and Vos, G., Anal. Chim. Acta, 1988,209, 147. Slavin, W., Manning, D. C., and Carnrick, G. R., At. Spectrosc., 1981, 2, 137. Slavin, W., Carnrick, G. R., Manning, D. C., and Pruszkowska, E., At. Spectrosc., 1983, 4, 69. Slavin, W., and Carnrick, G. R., At. Spectrosc., 1985, 6, 157. Keukens, H. J., The Determination of Cadmium and Lead in Soil by Flame Atomic Absorption Spectrometry, RIKILT- Report 84.23., 1984 (in Dutch). Berrow, M. L., and Stein, W. M., Analyst, 1983, 108, 277. McGrath, S. P., and Cunliffe, C. H., J. Sci. Food Agric., 1985. 36, 794. Paper 0102326 D Received May 24th, 1990 Accepted December 11 th, 1990

ISSN:0003-2654    

DOI:10.1039/AN9911600347

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, APRIL 1991, VOL. 116 353 Sequential Atomic Absorption Spectrometric Determination of Chloride and Iodide in a Flow System Using an On-line Preco ncent rat io n Tech n iq u e Fatima T. Esmadi, Maher A. Kharoaf and Abdulrahman S. Attiyat Department of Chemistry, Yarmouk University, Irbid, Jordan Aflow injection method is described for the sequential determination of chloride and iodide in a mixture of the two. The chloride-iodide mixture is precipitated from solution by silver nitrate and the precipitated silver chloride is dissolved by ammonia solution to determine chloride after which the precipitated silver iodide is dissolved by potassium cyanide solution to determine iodide. The method allows the analysis of about 15 samples h-1 and mixtures with different chloride : iodide ratios can be analysed at the pmol dm-3 level.Keywords: Sequential determination; chloride and iodide determination; atomic absorption spectrometry; preconcentration The determination of anions is a significant problem in a variety of industries and numerous techniques have therefore been developed both for the determination of individual anions and for total anionic content. Chloride has been determined indirectly using atomic absorption spectrometry (AAS) by precipitating it with silver nitrate and then either determining the remaining silver or the silver content in the precipitate after its dissolution by ammonia.' Flow injection (FI) with different detection methods such as ~pectrophotornetry~2-5 potentiometry"7 and AASg has been used for the determination of chloride.The determination of iodide has been achieved by a number of methods; the most widely used is that based on the catalytic effect of iodide on the Ce'V-As"' system.9.10 This method has also been applied to the determination of iodide in phar- maceutical and food samples using a modular stopped-flow system.'' The catalytic effect of iodide on the chlorpromazine -bromate reaction12 and on the destruction of the thiocyanate ion by the nitrite ion13 has been utilized for the determination of iodide. Owing to the similarities in the chemical properties of the halide ions, the analysis of their mixtures is of particular interest and several reactions have therefore been developed that allow their sequential determination. These methods include sequential oxidation of the halide ions14 or their conversion into organic halides followed by detection of the products by a suitable method such as gas chromatographyl5 or by potentiometric titration with silver or mercury nitrate.16 Flow injection is a technique that has greatly increased the throughput of samples and the application of this technique to the simultaneous determination of a range of analytes is expanding rapidly.17-19 Martinez-Jimenez et a1.20 have re- ported a method for the determination of chloride and iodide in a mixture of the two using FI-AAS. In this method the sample mixture is injected into a silver nitrate solution and the precipitates formed are retained on a stainless-steel filter, so that the total anion content can be determined from the decrease in the AAS signal of silver.The silver chloride precipitate is then dissolved by passing ammonia solution through the filter and the chloride content can be determined. Chloride has been determined21 using reversed FI-AAS by precipitating it with silver nitrate and then dissolving the precipitate, after a period of time, with a suitable dissolving agent. As the silver signal is proportional to the halide ion content in the sample, the latter can be determined. In this paper, the same method is applied to the sequential determi- nation of chloride and iodide based on the solubility of silver chloride and the insolubility of silver iodide in ammonia solution. The mixture of chloride and iodide is first precipi- tated from solution by silver nitrate for a given period of time to allow for preconcentration; ammonia solution is then passed which dissolves the precipitated silver chloride only, after which cyanide solution is passed which dissolves the silver iodide precipitate.The chloride content is determined from the first AAS signal of silver and the iodide content from the second. Experimental Reagents and Solutions All chemicals were of analytical-reagent grade. Solutions were prepared in distilled, de-ionized water. Standard solutions of silver (AgN03) were prepared by appropriate dilution of a 1000 ppm stock standard solution obtained from Merck. A standard iodide solution (1 g 1-1) was prepared by dissolving potassium iodide (Merck) (dried at 105°C for 2 h) in de-ionized water. Working solutions were prepared by suit- able dilution with de-ionized water.A standard chloride solution was prepared in the same way as for the iodide solution except that sodium chloride [Merck (formerly BDH)] was used instead of sodium iodide. Potassium cyanide was obtained from Merck and 30% ammonia solution was purchased from Fluka. Apparatus A Perkin-Elmer 372 atomic absorption spectrometer equipped with a silver hollow cathode lamp (4 mA) and a strip-chart recorder was used for the FI measurements. The wavelength was adjusted to 328.1 nm, the acetylene and air flow-rates were set to 1 and 8 1 min-1, respectively, and the slit-width was 2 nm. Teflon tubing of 1 mm i.d. from Beckman Altex was used in the flow system. Three Rheodyne loop injection valves were used to introduce either the washing or the dissolving solution into the precipitating loop.The length of the mixing coil was 5 cm and the i.d. 1 mm. The precipitating loop consisted of a Tygon tube (7 cm X 2.8 mm i.d.) filled with Pyrex glass beads (1.9 mm in diameter) (Thomas Scientific) which was connected vertically to the injection valve and to the nebulizer of the atomic absorption spectrometer via a Teflon tube (Fig. 1). The void volume of the precipitating loop was found to be 85 p1. A four-channel peristaltic pump was used to draw the cation and anion solutions into the precipitating loop and then to waste, and the negative pressure of the nebulizer was used to draw the washing and dissolving solutions through the precipitating loop to the nebulizer of the atomic absorption spectrometer.The flow-rate of the dissolving solution was coarsely con-354 ANALYST, APRIL 1991, VOL. 116 trolled by adjusting the nebulizer of the spectrometer. The experimental conditions of the system are listed in Table 1. Procedure Standard solutions of silver nitrate and a mixture of chloride and iodide are passed through the flow system using the manifold shown in Fig. 1. The two solutions mix in the mixing coil and are then pumped to the precipitating loop. Precipita- tion is allowed to proceed for 2 min under the conditions listed in Table 1 and the excess of silver nitrate is pumped to waste. The selecting valve 1 (injector 11) first allows a stream of the washing solution (1 x 10-5 rnol dm-3 HN03) to pass to the precipitating loop for 32 s until zero response is obtained, then ammonia is allowed to pass which dissolves the precipitated silver chloride and carries it to the nebulizer to be determined.Selecting valve 2 (injector 111) then allows a cyanide solution to pass to the precipitating loop; this solution dissolves the precipitate of silver iodide and carries it to the nebulizer. The first AAS signal obtained is proportional to the silver concentration, which is in turn proportional to the chloride concentration, and the second signal is proportional to the iodide concentration. The concentration of the silver ion is 1000 ppm. The time required for washing the precipitate is 32 s. Dissolution of the precipitated silver chloride and silver iodide requires 32 and 22 s, respectively, at a concentration of 5 x 10-5 rnol dm-3 each of chloride and iodide. Results and Discussion The determination of chloride using a continuous precipita- tion unit in a reversed FI system gives accurate, reproducible and rapid results.2* The same technique can be applied to the sequential determination of chloride and iodide by precipitat- ing these ions with silver nitrate for a given period of time, followed by dissolution of the silver chloride in ammonia and then dissolution of the silver iodide in cyanide solution.In order to obtain the most sensitive and reproducible results, several chemical and FI variables such as the concentration of the reagents, their flow-rates and the length of the mixing coil have to be optimized. The effect of the silver ion concentration on the response was studied and it was found that the sensitivity increases with an increase in the silver ion concentration up to 800 ppm for Table 1 Experimental conditions for the sequential determination of chloride and iodide Hollow cathode lamp Silver Lamp current 4 mA Wavelength 328.1 nm Acetylene flow-rate 11 min-1 Air flow-rate 8 1 min- 1 Slit-width 2 nm PL CN W - L,(i)l Waste 1 ), Waste ',\ G"2 Fig.1 Manifold used for the sequential determination of chloride and iodide: SV, selecting valve; W, washing stream; MC. mixing coil; and PL, precipitating loop iodide and 500 ppm for chloride as shown in Fig. 2. In order to ensure complete precipitation of chloride and iodide, a concentration of 1000 ppm of silver was used. The concentration of the dissolving agents (ammonia and cyanide) was optimized.For complete dissolution of the precipitates, 0.45 mol dm-3 ammonia and 0.05 rnol dm-3 cyanide solutions are required. A separate experiment indi- cated that silver iodide is insoluble in 0.45 mol dm-3 ammonia in the range studied (2-120 pmol dm-3). The effect of the flow-rate of the analyte and silver ion solution on the response was studied and it was found that the height of the signal increases linearly with an increase in the flow-rate (Fig. 3). This is to be expected as a higher flow-rate means that more reagents can be mixed and hence can produce more precipitate. A flow-rate of 1.2 ml min-1 was chosen to give sufficient sensitivity and reasonable reagent consumption. The signal is also affected by the flow-rates of the dissolving agents in that it increases with an increase in their flow-rates; however, flow-rates of 4.2 and 4.5 ml min-1 for the ammonia and cyanide solutions, respectively, gave the most reproducible results.The effect of the length of the mixing coil on the response was studied at a constant flow-rate. A change in the length of 0.25 8 0.20 m 0 5 0.15 n m -g 0.10 a" 0.05 0 100 200 300 400 500 600 700 800 900 1000 1100 Concentration of Ag+ (ppm) Fig. 2 chloride and B, iodide Effect of silver ion concentration on the response of A, 0 1.2 2.4 3.6 4.8 Flow-rate/ml min- Fig. 3 Effect of flow-rate on the determination of A, 4 x 10-5 mol dm-3 chloride; and B, 4 x 10-5 mol dm-3 iodide in a mixture of the two 0.6 8 0.5 - + 0.4 - x 0.3 - m 0.2 - C m 0 Y L 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 Concentration of chloride and iodide/ymol dm-3 Fig.4 tion Calibration graphs for A, chloride and B, iodide determina-ANALYST. APRIL 1991, VOL. 116 355 Table 2 Analysis of different chloride-iodide mixtures Amount added/ pmol dm-3 Iodide Chloride 120 2 120 5 120 10 110 10 100 10 100 20 90 30 80 so 60 50 so 50 Amount found/ pmol dm-3 Iodide Chloride 120 2.1 119 4.9 118 11 109 9.8 98 9.7 100 9.6 89 31 80 49 61 50 52 50 Amount added/ pmol dm-3 Iodide Chloride 40 60 30 70 30 60 30 80 20 90 20 100 10 100 5 100 2 100 Amount found/ pmol dm-3 Iodide Chloride 40 61 29 71 31 60 30 79 19 92 21 97 9.6 97 4.9 101 2.1 99 F CI- -Time Fig. 5 Response of different chloride-iodide mixtures: A, 2 x 10-5 mol dm-3 each of C1- and I-; B, 2 x 10-5 mol dm-3 C1- + 4 x 10-5 mol dm-3 I-; C, 4 x mol dm-3 C1- + 2 x mol dm-3 I-; D, 8 X mol dm-3 C1- + 6 x mol dm-3 I-; E, 8 x mol dm-3 each of C1- and I-; and F, 9 x 10-5 mol dm-3 C1- + 1 x 10-4 mol dm-3 I-.The chloride peak appears to the right of the iodide peak in each instance the coil had no significant effect on the signal height, which indicates that the precipitation and dissolution reactions are completed inside the precipitating loop. Under the optimum conditions described above, two calibration graphs were obtained, one for chloride and the other for iodide (Fig. 4). The linear working range is 2-100 pmol dm-3 chloride and 5-120 pmol dm-3 iodide. The upper linear limit is believed to be due to saturation of the precipitating loop. The lower detection limit, taken as that concentration which gives a signal three times the base line noise, is 2 pmol dm-3 chloride and 5 pmol dm-3 iodide.Fig. 5 shows signals for chloride and iodide standards. It can be seen that the signals are reproducible and the peak width is 30 s for chloride and 22 s for iodide, which means that a sampling frequency of about 15 h-1 can be achieved by using this method. In order to test the applicability of the method, samples containing different chloride : iodide ratios were prepared and their contents analysed (Table 2). Mixtures with chloride : iodide ratios ranging from 1 : 60 to 50 : 1 can be analysed, with a relative standard deviation (RSD) of less than 3% for a series of eight samples in each determination. Effect of Interferences The interference from foreign ions in the system was studied with a solution containing 5 x 10-5 rnol dm-3 each of chloride and iodide.The tolerance limits were taken as the largest Table 3 Effect of foreign anions on the sequential determination of 5 X 10-5 mol dm-3 each of chloride and iodide in a mixture of the two Ratio of amount of foreign ion : Ion added chloride-iodide Acetate, formate, N 0 y F- , PO4'-, C104- 50 S042-, C032-, Cr04'- 20 c 2 0 4 2 - 7 CN-, Br-, SCN- 2 amounts yielding an error of less than 5% in the peak absorbance. The results (Table 3) indicate that the most significant interferences are caused by anions that form precipitates with silver such as bromide, thiocyanate and cyanide, if they are present at a concentration which is twice that of the chloride-iodide mixture. Conclusion The simultaneous determination of chloride and iodide with an FI system at a sampling rate of about 15 samples h-1 and with an RSD of less than 3% is possible after precipitation of these ions with silver nitrate and the sequential dissolution of the precipitates with ammonia and cyanide solutions. It is possible to detect as little as 2 pmol dm-3 chloride and 5 pmol dm-3 iodide in a mixture of the two.Compared with a previously reported method,*O the proposed method is simpler, and it is not necessary to change the precipitating loop as it is cleaned by the dissolving agents, which allows a higher sampling frequency. 1 2 3 4 5 6 7 8 9 10 References Hassan. S. S . , Organic Analysis Using Atomic Absorption Spectroscopy, Ellis Horwood, Chichester, 1984.RGiiEka, J., Stewart, J. W., and Zagatto, E. A., Anal. Chim. Acta, 1976, 81, 387. Hansen, E. H., and RGiiEka, J . , Anal. Chim. Acta, 1976, 87, 353. Krug, F. J., Pessenda, L. C., Zagatto, E. A., Jacintho, A. O., and Reis, B. F . , Anal. Chim. Acta, 1981, 130, 409. Thijssen, P. C., Prop., L. T., Kateman, G., and Smit, H. C., Anal. Chim. Acta, 1985, 174, 27. Trojanowicz, M., and Matuszenski, W., Anal. Chim. Acta, 1983, 77, 151. Alexander, P. W., Haddad, P. R., and Trojanowicz, M., Anal. Chem., 1984, 56, 2417. Martinez-Jimenez, P., Gallego, M., and Valcarcel, M., Anal. Chem., 1987,59, 69. Sandell, E. B., and Kolthoff, I. M., J. Am. Chem. Soc., 1934, 56. 1426. Sandell, E. B., and Kolthoff, I . M., Mikrochim. Acta, 1937, 9.356 ANALYST, APRIL 1991, VOL. 116 11 GutiCrrez, M. C., Gomez-Hens, A., and PCrez-Bendito, D., Analyst, 1989, 114, 89. 12 Vinas, P., Hernandez Cordoba, M., and Sanchez-Pedreno, C., Talanta, 1987,34, 351. 13 Moxon, R. E., Analyst, 1984, 109,425. 14 Snell, F. D., Photometric and Fluorimetric Methods of Analysis. Nonmetals, Wiley, New York, 1981. 15 Rodriguez-Vazquez, J. A., Anal. Chim. Acta, 1974, 73, 1. 16 Williams, W. J., Handbook of Anion Determination, Butter- worths, London, 1979. 17 Hauser, P. C., Tan, S. S., Cardwell, T. J., Cattrall, R. W., and Hamilton, I. C., Analyst, 1988, 113, 1551. 18 Luque de Castro, M. D., and Valcarcel Caess, M., Analyst, 1984, 109, 413. 19 Cardwell, T. J., Cattrall, R. W., Hauser, P. C., and Hamilton, I. C., Anal. Chim. Acta, 1988, 214, 359. 20 Martinez-Jimenez, P., Gallego, M., and Valcarcel, M., Anal. Chim. Acta, 1987, 193, 127. 21 Esmadi, F., Kharoaf, M., and Attiyat, A., Talanta, in the press. Paper 0104236F Received September 18th, 1990 Accepted December I7th, 1990

ISSN:0003-2654    

DOI:10.1039/AN9911600353

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, APRIL 1991, VOL. 116 357 Determination, of Urea in Serum by Using Naturally Immobilized Urease in a Flow Injection Conductimetric System Lourival C. de Faria, Celio Pasquini" and Graciliano de Oliveira Net0 lnstituto de Quimica, Universidade Estadual de Campinas CEP 13081, C. P. 6154, Campinas, S5o Paulo, Brazil A flow injection method was developed, aimed at the determination of urea in human serum. The system makes use of the naturally immobilized urease present in Canavalia ensiformis DC (jack bean). A column is filled with small pieces of this bean, and the sample (50 pl) containing urea passes through it carried by a 1 % NaCl solution. On leaving the column the stream is merged with an alkaline reagent (0.5 mol dm-3 NaOH; 0.5% disodium dihydrogen ethylenediaminetetraacetate). The ammonium ions, arising from the enzymatic reaction that occurs inside the column, are changed into the molecular form, which permeates a polytetrafluoroethylene membrane and is received in a de-ionized water acceptor stream.The ammonia ionizes causing an increase in the conductance, which is proportional to the urea content of the sample. About 40 samples can be processed in 1 h with negligible carry-over and with a relative standard deviation of 1 % or less. The results are in agreement with those obtained by a standard spectrophotometric method. Keywords: Serum urea determination; naturally immobilized urease; flow injection Several papers describing artificial urease-immobilization processes and the construction of urea sensors have been published.14 Most of them were aimed at the development of potentiometric sensors to be used in batch procedures for the determination of urea.Artificial enzyme immobilization has also been pointed out to be one of the most suitable approaches to automating biochemical reactions in flow injection (FI) systems.5.6 However, it has been demonstrated that, sometimes, it is possible to make use of the raw material containing the enzyme, naturally immobilized inside vege- table cells, in the construction of a biosensor. Wang and Lin7 described a biosensor that was developed, from the natural occurrence of polyphenol oxidase in banana, to provide the enzyme source in the determination of dopamine. Plant tissues have also been used for constructing electrochemical sensors for glutamate,* phosphate and fluoride9 and tyro- sine.'" Meal, obtained from the jack bean (Canavalis ensi- formis DC), was used in the construction of a potentiometric biosensor for urea," and good results were obtained in terms of stability and detection limits.The use of naturally immobilized enzymes is very attractive as no immobilization process is required. However, the enzyme should exhibit sufficient activity, and the conditions under which the determination is made should allow a rapid transport of both the substratum to the inside of the cell and of the products to be detected outside of the cell membrane. The diffusion through the cell membrane is the rate-limiting factor for an FI system that is based on the use of natural immobilization. This paper describes an FI system developed for the determination of urea in human serum by using a methodol- ogy that employs the naturally immobilized urease present in the jack bean.The ammonium ion originating in the enzymic hydrolysis of urea is detected by using the conductimetric methodology previously described. 12 Experimental Apparatus The same instruments, conductimetric flow cell and poly- tetrafluoroethylene membrane previously described were used. 12 The diffusion cell was modified and had its dimensions enlarged to contain a shallow groove 0.5 mm deep, 4 mm wide and 10 cm long. * To whom correspondence should be addressed. The FI conductimetric manifold used for the determination of urea is outlined in Fig. 1. Column C1 was filled with a mixed-bed resin, to effect a final purification of the de-ionized water, and was constructed as previously described.12 Poly- ethylene tubing (0.8 mm i.d.) was used throughout.All the experiments were performed at ambient laboratory tempera- ture (near 25 "C). Column C2, in Fig. 1, was made from a Tygon tube (2.5 mm i.d.) and filled with small pieces of bean. These pieces were directly cut from a grain of the bean, from which the skin had been removed, and had the form of small cubes with edges nearly 2 mm long. Two small plugs of glass wool were used to retain the bean pieces inside the column. A 5 cm long column contained between 15 and 20 pieces, with an average total mass of 0.13 g. The column was conditioned by passing a 1% NaCl solution (flow-rate: 1.0 ml min-1) for 15 min.When not in use the column should be kept empty and stored in a refrigerator. For comparison purposes an FI manifold was assembled to work with the free urease reagent solution. This system is the same as that shown in Fig. 1, but has the bean column replaced by a 50 cm reaction coil made from a polyethylene tube (0.8 mm i.d). Reagents Standard solutions of urea were prepared daily by suitable dilution of a 1000 pg ml-1 stock standard solution. Carbonate- free sodium hydroxide solutions were made by dilution of approximately 12 mol dm-3 sodium hydroxide prepared with U p S (50 pl) ' w ' Fig. 1 Flow injection manifold for the determination of urea in serum using naturally immobilized urease and conductimetric deter- mination. P, Peristaltic pump; s, sample; C1, mixed bed ion-exchange column; C2, bean tissue column; L, thermal equilibration coil; B, water-isolated bath; DC, diffusion cell; FC, conductimetric flow cell; and W, fluid discharge.I, De-ionized water; 11, alkaline reagent: 0.5 mol dm-3 NaOH, 0.5% Na2EDTA; and 111, sample carrier fluid358 ANALYST, APRIL 1991, VOL. 116 freshly boiled de-ionized water. Disodium dihydrogen ethyl- enediaminetetraacetate (Na2EDTA) (0.05%) was used in the reagent solution. Free urease solutions were prepared by dissolving 0.30 g of the Merck product (83.35 nkat mg-1) in 5 ml of water and adjusting the volume to 500 ml with 1% NaCl solution. Portions (20 ml) of this solution were placed in 100 ml calibrated flasks and the volume of each was adjusted with a suitable 0.02 mol dm-3 tris(hydroxymethy1)amino methane (Tris)-HC1 buffer containing 1% of NaCl.The pH of these solutions was measured by a standard pH-measurement procedure using a calibrated glass electrode. All other Tris-HC1 buffer solutions were 0.02 mol dm-3 and contained 1% of NaCl. Standard solutions of urea, containing various amounts of the urease inhibitors F- and HS03-, were prepared from stock 0.100 mol dm-3 NaF and NaHS03 solutions. Analytical- reagent grade solutions and freshly prepared de-ionized water were used throughout. Serum Samples Human serum samples were obtained from the Clinical Hospital of the State University of Campinas by a single centrifugation of the whole blood. All samples were analysed for urea in the hospital laboratory and by the proposed FI method on the same day.For the proposed method, 100 pl of serum were diluted to 100 ml with water in a calibrated flask. Results and Discussion Preliminary experimental data were obtained using a 5 cm column constructed as described under Experimental. The reagent solution was 0.5 mol dm-3 NaOH containing 0.5% m/v of EDTA. If no salt is present in the carrier stream the rate of urea hydrolysis is very low and barely detectable. Adding a salt such as NaCl or KCI to the carrier causes a marked increase in ammonium ion production and a decrease in the sample washing-out time. Fig. 2 shows the dependence of the FI peak height on the concentration of NaCl in the carrier stream. The same behaviour and quantitative increase in ammonia production was observed for KCI.Additional experiments were carried out on the evaluation of the effect of the column length and on the reproducibility among columns. Fig. 3 shows how the length of the column affects the hydrolysis of the standard urea solutions. The results for urea are compared with those obtained for standard ammonia solutions prepared to contain an equivalent amount of nitrogen. The comparison shows that the hydrolysis yields are about 70 and 95% for a 3 and a 5 cm long column, respectively. Use of longer columns also increases the wash time interval, thereby reducing the sample-processing capabil- ity of the system. Five columns (each 5 cm long) were prepared as described under Experimental, and the conductimetric signals obtained 0 0.2 0.4 0.6 0.8 1 .o [NaCIl (% m/v) Fig.2 Effect of the concentration of salt in the carrier stream on the analytical signal. Urea concentration: A , 1.0; B, 3.0; and C, 5.0 pg ml-1 for each, using standard urea solutions in the concentration range 1-5 pg ml-1, were compared. A maximum relative difference of 5% was observed among signal peak heights. The effect of the concentration of sodium hydroxide in the reagent stream was investigated. The results showed that the signals for a standard solution containing 1-5 pg ml-1 of urea were slightly affected by changing the hydroxide concentra- tion from 0.1 to 1.0 rnol dm-3. Effect of pH and Inhibitors By using a 3 cm long column of beans, the effect of changing the pH of the carrier stream and of two urease inhibitors (F- and HS03-) was investigated. The results were compared with those obtained by using a solution of free urease as carrier in the manifold described under Experimental.The system employing free urease reagent showed a sensitivity that was about one-tenth of that employing the bean column. There- fore, the comparison was made with solutions that contained 3 pg ml-1 of urea for the column system and 30 pg ml-1 for that employing the free urease. No effort was made to optimize the free urease system as only relative results were required. Fig. 4 shows how the pH of the carrier stream affects the peak height for both the systems. It can be observed that the free enzyme is more affected by the activity of the hydrogen ion in solution. The pH was previously reported as a critical parameter for a potentiometric biosensor constructed with use of a meal of jack bean11 and for an artificially immobilized- urease potentiometric sensor.13 Fig.5 shows the effect of the presence of F- and HS03- in the sample solution on both free and naturally immobilized- enzyme FI systems. In the absence of pH control the effect on the free enzyme is critical while the naturally immobilized enzyme is only slightly affected. The total effect on the inhibition of the enzyme activity comes from the presence of - v) Y 1 0 u I I I I I I 0 1 2 3 4 5 Column lengthkm Fig. 3 Effect of the column length on the production of ammonium. a, Ammonium standard solutions; and 0, urea standard solutions. Nitrogen concentration of the NHICl and urea solutions: A and B, 1.0; and C and D, 2.0 pg ml-1. Carrier solution, 1% NaCl 100 I Y 0 P 25 6.5 7.0 7.5 8.0 8.5 9.0 PH Fig.4 Effect of the pH of the carrier solution on the analytical signal: A, using naturally immobilized urease (urea concentration of the test solution, 3 pg ml-1); and B, using a solution of free urease (urea concentration, 30 pg ml-1)ANALYST, APRIL 1991, VOL. 116 359 the inhibitor itself and from the change it causes in the pH of the sample solution. It is worthwhile pointing out that the ratio of inhibitor to substratum is ten times greater when the bean column is employed as the urea concentration is ten times lower than that used in the free urease experiment. If a Tris-HCI buffer of pH 7.5 is used the inhibitor effect is cancelled for the naturally immobilized-enzyme system while it is still present in the free-enzyme system.Although the response mechanism for the plant or animal tissue biosensors has not yet been well established14 the results obtained in this work suggest that the cell integrity is maintained for the pieces of bean. The presence of a salt such as NaCl or KCI, frequently neglected when biosensor metho- dologies are developed, is essential to promote the rapid transport of the substratum to the inside of the cell where the enzyme-catalysed reaction occurs. The cell membrane acts as a selective filter, which is not permeable to some substances such as the inhibitors reported here, hence making the method less prone to interference. Also, the pH of the sample does not represent a critical parameter when the bean column is used as the urease source.To an extent, this is also a consequence of the fact that the reaction will occur inside the cell where the pH should be appropriate for the enzyme action. Further- more, the conductimetric FI methodology described here does not require the enzymic reaction to occur at the same pH as that at which the detection is made, as is required for potentiometric sensors used in batch procedures. Conditions for Determination of Urea in Serum Based on the above results, the conditions for the determina- tion of urea in serum were selected. A 3 cm long column was chosen along with a carrier solution containing 1% m/v of NaCl and a reagent solution, 0.5 mol dm-3 in NaOH, containing 0.5% m/v of Na2EDTA. No buffer solution need be used owing to the high sample dilution employed.A sample volume of 50 p1 was injected. A calibration run followed by signals obtained for some samples is shown in Fig. 6. Calibration runs obtained under these conditions showed a linear dependence of the peak height of the conductimetric signal in relation to the urea concentration in the range 1-10 pg ml-1, with a typical correlation coefficient of 0.9997. The relatlbe standard deviation for the analysis of ten replicates of a standard solution containing 3.0 pg ml-1 of urea was found to be 0.8%. About 40 samples could be processed in 1 h. 100 - s - 4- L 01 a, L Y .- 50 n a, (D a, n .- 4- - l i ---- ----- 0 1 3 5 In hi bi tor/mmo I d m -3 Fig. 5 Effect of the presence of F- and HS03- in the urea sample solution on the analytical signal.0 and M, Naturally immobilized urease; and 0 and 0. free enzyme reagent solution. Solid line, F- added to the sample; and broken line. HS03- added to the sample. Square symbols used when the pH was kept at 7.5 using Tris-HCI buffer. Urea concentrations were 3 pg ml-1 for the naturally immobilized urease and 30 pg ml-l for the free enzyme solution system. respectively. All carrier solutions contained 1% NaCl Column Lifetime A 3 cm long column, prepared as described under Experi- mental, was fitted in the FI manifold and its long-term performance was evaluated for 10 h, injecting about 350 standard solutions containing 2 or 5 pg ml-1 of urea. The results show that, after this period, the activity of the column is reduced by about 10% in relation to its initial value.The rate of change in the column activity is slow. Therefore, periodical re-calibration can ensure good accuracy. In routine determi- nations a calibration involving use of three standard urea solutions was repeated every 30 min. The same column could be used to perform up to 1000 determinations, although it was found preferable to replace it every day in view of the low cost and ease of construction. Accuracy of the Proposed Method Sixty-five samples of human serum were analysed by the proposed method and by a spectrophotometric-enzymic method based on the reaction of the ammonium ion, produced in the hydrolysis of urea, with 2-oxoglutarate and reduced nicotinamide adenine dinucleotide (NADH) in the presence of glutamate dehydrogenase.15 The decrease in the NADH concentration was monitored in the ultraviolet region.The method was performed in a Cobas Mira automatic analyser. Fig. 7 shows a comparison between the results obtained by the two methods. Least-squares statistical results show that the correlation between the two methods can be expressed as: cp = (0.47 t 0.31) + (0.9942 -t 0.0237) c, 5.0 I 12 min ' Time -c Fig. 6 Typical calibration run for serum urea determination using the naturally immobilized enzyme FI system. The signals for five standard solutions and for the five samples, introduced in duplicate, are shown. The numbers above the peaks arc the urea standard solution concentrations in pg ml-1 / I 1 I I I 0 20 40 60 80 100 Urea concentration by reference method/mg dl-' Fig. 7 Correlation of the results for 65 serum urea concentrations found by the proposed method and by a standard spectrophotometric method360 ANALYST, APRIL 1991, VOL.116 where cp is the urea concentration (mg dl-1) found by the proposed method and c, is the urea concentration (mg dl-1) found by the reference spectrophotometric method. The correlation coefficient is 0.987 and the error of the estimate is 22.4 mg dl-1. These results lead to the conclusion that the proposed method compares well with the conventional spectrophotometric method. The sensitivity of the conducti- metric method permits a high sample dilution that helps in overcoming matrix effects. The method also presents advan- tages in relation to the cost and demonstrates that the naturally immobilized enzyme can be used to replace artificial immobilization in FI reactors when the raw material presents sufficient activity. The authors thank .M. A. Selleghin for providing the serum samples. References 1 Guilbault, G. G., and Montalvo, J. G., Jr., J. Am. Chem. Soc., 1969,91, 2164. 2 Anfalt, T. Graneli, A., and Jagner, D., Anal. Lett., 1973, 6, 969. 3 4 9 10 11 12 13 14 15 Alexander, P. W., and Joseph, J . J., Anal. Chim. Acta, 1981. 131, 103. Ianniello, R. M.. and Yacynych, A., Anal. Chim. Acta, 1983, 146, 249. Mottola, H. A., Anal. Chim. Acta, 1983, 145, 27. Petersson, B. A., Anal. Chim. Acta, 1988, 209, 239. Wang, J., and Lin, M. S., Anal. Chem., 1988, 60, 1545. Kuiyama, S., and Rechnitz, G. A., Anal. Chim. Acta, 1981, 131, 91. Schubert, F., Rennebarg, R., Scheller, F. W., and Kirstein, L., Anal. Chem., 1984, 56, 1677. Schubert, F., Wallenberg, U., and Scheller, F. W., Biotechnol. Lett., 1983, 5 , 239. Arnold, M. A,, and Glazier, S . A., Biotechnol. Lett., 1984, 6, 313. Pasquini, C., and de Faria, L. C., Anal. Chim. Acta, 1987,193, 19. Guilbault, G. G., and Mascini, M., AnaLChem., 1977,49, 795. Arnold, M. A., and Rechnitz, G. A., in Biosensors, Fundamen- tals and Application, eds. Turner, A. P. F., Karube, I., and Wilson, G. S., Oxford University Press, Oxford, 1987, p. 54. Sampson, E. J., and Baird, M. A., Clin. Chem., 1979,25,1721. Paper 0104841 K Received October 29th, 1990 Accepted November 22nd, I990

ISSN:0003-2654    

DOI:10.1039/AN9911600357

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, APRIL 1991, VOL. 116 361 Nitrate Ion-selective Electrodes Based on Quaternary Phosphonium Salts in Plasticized Poly(viny1 chloride) and Influence of Membrane Homogeneity on Their Performance* Manassis G. Mitrakas and Costas A. Alexiades Laboratory of Analytical Chemistry, Department of Chemical Engineering, School of Engineering, Aristotle University of Thessalo niki, Thessaloniki 54006, Greece Vissarion 2. Keramidas laboratory of Soil Science, School of Agriculture, Aristotle University of Thessaloniki, Thessaloniki 54006, Greece Nitrate ion-selective electrodes based on quaternary phosphonium salts in plasticized poly(viny1 chloride) (PVC) were found to be superior in stability and sensitivity to those based on quaternary ammonium salts in the same matrix and to Orion 93-07 electrodes.The difference in the characteristics of the electrodes tested was attributed to the different extent of homogeneity existing in the sensing membrane. Scanning electron microscopy revealed that quaternary phosphonium salts were homogeneously dispersed in the sensing membrane because of some specific interaction with the plasticized PVC. On the other hand, the other sensors tested were segregated to various extents. This specific interaction of quaternary phosphonium salts with the plasticized PVC, resulting in a homogeneous sensing membrane, was considered to be the cause for the improved characteristics of the nitrate electrodes. In line with this, additional rules for constructing ion-selective electrodes in non-porous polymer membranes are proposed.Keywords: Nitrate ion-selective electrode; quaternary phosphonium salt; membrane homogeneity; scanning electron microscopy; poly(viny1 chloride) membrane The development of ion-selective electrodes (ISEs) for the determination of nitrate began with the introduction of liquid-membrane electrodes, which utilized a porous mem- brane, holding a mixture of sensor and solvent. Sensors that have been commonly used included quaternary ammonium salts (QAS) , triphenylmethane dyes, Fez+- or NiZ+-o-phen- anthroline complexes and quaternary phosphonium salts (QPS). The commercially available electrodes of this type, however, suffered from known drawbacks, namely, short- term stability and relatively poor selectivity and sensitivity. One of the major contributions to the design of ISEs was the introduction of a non-porous poly(viny1 chloride) (PVC) matrix, into which the sensor was embedded.This technique was first introduced for the construction of calcium ion- selective electrodes and subsequently extended to nitrate ISEs. 1-5 Nielsen and Hansen6 conducted a thorough study concern- ing the optimum combination of sensor-plasticizer-PVC for constructing nitrate ISEs based on QAS, which they found to be superior in stability and sensitivity to earlier electrodes. These workers considered that the electrode characteristics depended primarily on the distribution coefficient, q , of the sensor between the aqueous and the membrane phase. By using thermodynamic relationships and the concept of solubil- ity parameter, d, they were able to relate q to the solubility parameters of water (daq), sensor ( d s ) , membrane matrix (d,) and the volume fractions of sensor (V,) and membrane matrix (Vm), through the equation: 4 = exp{VsVm[(ds - daJ2 - (4 - dm)2]>(RT)-1 (1) where, R = universal gas constant and T = temperature in K.Equation (1) helped Nielsen and Hansen6 to establish certain rules for selecting the membrane constituents in order to obtain optimum electrode characteristics. The derivation of eqn. (l), however, apart from the equilibrium condition * Part of the PhD Thesis of M. G. Mitrakas. imposed, rests also on two inherent assumptions: (i) the sensor maintains its monomeric form in both the aqueous and the membrane phase; and (ii) the sensing membrane constitutes a true solution (no separation of phases).However, it is difficult to ascertain in practice whether these two assumptions are met when the sensor is embedded in the plasticized PVC. A smooth, transparent membrane is a good indication of homogeneity. Homogeneity, a prerequisite for constructing good quality ISEs, is considered to be a measure of the compatibility of the sensor with the PVC matrix and the extent to which the sensing membrane approximates to a true solution. During experiments in this laboratory aimed at constructing ISEs based on QAS and Fez+- or Ni2+-o-phenanthroline complexes in a PVC matrix, it was observed that the smooth and transparent appearance of the sensing membrane was not an infallible proof of its homogeneity. Indeed, the sensors were not homogeneously dispersed in the non-porous medium as preliminary observations with an optical microscope revealed. This led us to believe that Nielsen and Hansen,6 who also used QAS in plasticized PVC, had overlooked this fact and, therefore, their attempt to interpret the experimental results by means of eqn.(1) is open to question. It was also felt that a more detailed examination of the homogeneity achieved in the sensing membrane was needed and that further improvement of nitrate ISEs was still feasible by choosing other sensors, such as QPS, presumably more compatible with PVC. In view of the above, the objectives of this study were: (i) to develop nitrate ion-selective electrodes based on QPS in a PVC matrix and to examine the degree of homogeneity achieved, when the sensor is embedded in this matrix, by means of optical and scanning electron microscopy; (ii) to check the characteristics and the performance of these electrodes in the determination of nitrate and compare them with well known electrodes; and (iii) to formulate simplified rules, if possible, for selecting the membrane constituents for optimum electrode characteristics.362 ANALYST, APRIL 1991, VOL.116 Experimental Membrane Material Sensors. Tetradodecylammonium bromide (TDDAB) and tributylhexadecylphosphonium bromide (TBHDPB) were Fluka purum-grade reagents. Tetraoctylphosphonium iodide (TOPI), tetradodecylphosphonium iodide (TDDPI) and tetrahexadecylphosphonium iodide (THDPI) were syn- thesized as described by Feshchenko et al. ,7 and TDDAB was purified by successive recrystallization .g All QPS were puri- fied by recrystallization from ethanol-diethyl ether, followed by preparative thin-layer chromatography with chloroform (Silica gel 60 F254, RF 0.2) and recrystallization from chloro- form-diethyl ether.The purities of QPS were checked by elemental analysis and melting point determination with differential scanning calorimetry (DSC) and verified by mass spectrometry. Plasticizers. All plasticizers were of analytical-reagent grade. The plasticizers used were dibutyl phthalate (DBP), dioctyl phthalate (DOP) and trixylyl phosphate (TXP). PVC. Breon S 110/10 (BP Chemicals) was used. Membrane and Electrode Preparation For QAS, the procedure of Nielsen and Hansen6 was followed, with the sensing membrane containing 29% PVC.For QPS, the amount of PVC was increased to 35% in order to render the membrane more rigid and durable as it was observed that dissolution of these sensors to a membrane containing 29% PVC made it softer. The membrane was prepared by dissolving 0.35 g of PVC and the appropriate amounts of plasticizer and sensor (see Table l ) , making a total of 1 g, in 20 ml of tetrahydrofuran (THF). This solution was poured into a 55 mm i.d. flat Petri dish and left at room temperature for the slow evaporation of the solvent. The resulting membrane was conditioned for 48 h in a 0.1 mol dm-3 KN03 solution in order to replace iodide or bromide with nitrate. Therefore, all sensors are denoted with a suffix N henceforth in order to represent the nitrate in the anionic; form.From this membrane (0.30 mm thick) a 10 mm diameter disc was cut and glued to the end of a PVC tube, with 4% m/m PVC in THF. The inner reference electrode was an Ag-AgC1 wire (Metrohm, 6.0711 .OOO) immersed in a solution of KN03-KCI (each 1 x 10-2 mol dm-3). The Orion 93-07 nitrate ISE was provided commercially. Photomicrographs The appropriate amounts of PVC, plasticizer and sensor, making a total of 0.5 g, were dissolved in 50 ml of THF. This solution was poured into a 90 mm i.d. flat Petri dish, which contained a piece of silicon foil ( 5 x 5 x 0.5 mm) and a glass plate (20 x 20 x 1 mm), and left at room temperature for the slow evaporation of the solvent. After conditioning, the membrane attached to the silicon foil was spattered with gold and photographed with a Jeol JSM 840A scanning electron microscope, while the membrane attached to the glass plate was photographed with an Axioma Zeiss optical microscope.Membranes consisting of PVC and the plasticizer only (without sensor) were also prepared in the same way and examined under the scanning electron microscope (see under Results and Discussion). Photographs of the sensing membrane of the commercially available Orion 93-07 nitrate ISE were taken only with the optical microscope. Apparatus Measurements of the e.m.f. were performed with an Orion 901 microprocessor ionalyser, and pH measurements with a Metrohm E 520 pH meter. An Orion 90-02 double-junction external-reference electrode with Orion 90-00-02 and 0.04 Table 1 Response characteristics of nitrate ISEs based on quaternary phosphonium salts in a membrane containing 35% PVC, as influenced by the composition of the sensing membrane.The same characteristics of well known nitrate ISEs are also shown PN03 d,l(cal E"tl Plasticizer Sensor (%) cm-3)1/** mV DBP TBHDPN (8) 9.4 97 DBP TBHDPN(15) 9.4 94 DBP TBHDPN(20) 9.4 94 DBP TOPN(4) 9.4 95 DBP TOPN(10) 9.4 97 DBP TOPN(12) 9.4 97 DOP TOPN(14) 9.1 95 TXP TOPN(14) 9.8 97 DBP TDDPN(3) 9.4 97 DBP TDDPN(4) 9.4 95 DBP TDDPN(6) 9.4 94 DOP TDDPN(3) 9.1 97 DOP TDDPN(4) 9.1 94 TXP TDDPN(4) 9.8 97 DBP THDPN(2) 9.4 102 DOP TDDAN(4) 9.1 88 DOP TBHDPN (8) 9.1 96 DBP TOPN(7) 9.4 96 DBP TOPN(15) 9.4 98 DBP TDDPN(1) 9.4 98 DBP TDDAN(4) 9.4 98 DBP TDDAN(3) 9.4 85 Orion 93-07 * cal = calorie, 1 cal = 4.184 joules. t The theoretical value of E" at 20 "C and pN03 of 2.06 is 96 mV.$ Calculated according to IUPAC definition. 11 § Calculated statistically. l2 SlopeImV decade-' 57.3 58.5 58.5 56.8 57.6 57.9 58.1 58.5 58.6 58.8 57.1 56.9 57.2 55.8 55.0 57.2 56.8 56.4 56.3 56.2 56.0 56.0 76 Lower linear - limit 3.95 4.05 4.00 3.90 4.30 4.38 4.46 4.46 4.42 4.35 4.36 4.46 4.80 4.70 4.50 4.63 4.45 4.45 4.70 4.45 4.30 4.30 56.0 Detection limit IUPACS Stat.§ 4.70 5.25 4.75 5.46 4.75 5.32 4.70 5.20 5.15 5.80 5.18 5.90 5.25 5.93 5.25 5.92 5.22 6.03 5.16 5.85 5.13 5.85 5.36 5.95 5.72 6.30 5.60 6.28 5.50 6.10 5.50 6.20 5.45 6.15 5.58 6.18 5.67 6.35 5.45 6.10 5.25 5.72 5.35 5.80 4.40 5.15ANALYST, APRIL 1991, VOL. 116 363 rnol dm-3 ammonium acetate inner and outer filling solutions, respectively, and a Metrohm E 402 saturated calomel elec- trode (SCE), were used.Measurement Techniques The characteristics of the electrodes (Table 1) in standard potassium nitrate solutions were determined. The solutions were not adjusted to identical ionic strength, and the concentrations were converted into activities by means of the Debye-Hiickel equation. Electrode lifetimes and nitrate content of waters, soil and plant extracts were assessed in an ionic strength adjustor (ISA) solution. The ISA solution was composed of lead acetate (0.02 rnol dm-3), PbO (0.01 rnol dm-3), potassium acetate (0.02 mol dm-3) and sulphamic acid (0.01 mol dm-3),9 and is denoted as ISA-Pb henceforth. p N 0 3 response The e.m.f. was measured in solutions thermostated at 20 "C, under constant stirring, against the double-junction reference electrode.All slopes were calculated by means of regression analysis on the linear part of the calibration graph. The equation of the non-linear part of the calibration graph was computed by Marquardt 1 0 non-linear regression analysis. The lower linear limit was determined as the common point of the linear and non-linear calibration graphs. Potential-pH curves These were obtained manually, point-wise over the pH range 3-1 1. Selectivity coefficients These were obtained by the fixed interference method,ll where the concentration of the interfering anion (X) in all instances was fixed at 0.01 rnol dm-3. In practice, they were computed as the intercept between the linear calibration graph and the lower limit of detection in the solution used and expressed as pkT& = -log kkg,,,.Detection limits Nitrate detection limit, cLA, as defined by IUPAC,** was determined as the nitrate concentration which gave a differ- ence of 18 mV in the potential calculated from the linear and non-linear calibration graphs. Statistically, for nitrate ISEs the response of which is not limited by the solubility of the sensor, the detection limit was calculated by the equation12 cL = (1OL/S - 1)cb, where, L = 2 X 1.645 (aA2 + aB2)1/2for the95% confidence level, S = the slope, cb = the nitrate concentration of the blank (calculated) and oA, oB, the e.m.f. standard deviations, are calculated from ten replicates of standard 1 x 10-3 rnol dm-3 NO3- and blank solution, respectively. E" values These were measured in a 1 x 10-2 rnol dm-3 N03- solution versus the SCE (20 "C), as the NO3- concentration in the internal reference solution of the electrodes was also 1 X 10-7 rnol dm-3.Solubility parameters The solubility parameters of QPS, calculated from the relationship d = pXG/M, (p = density of the material, G = molar attraction constant and M , = relative molecular mass) as suggested by Small,13 were found to be: ~TBHDPN = 9.8, Those of QAS were taken from Nielsen and Hansen6 and those of PVC and plasticizers from Brydson:14 dpVC = 9.5, diethyl phthalate, dDEP = 9.9; dibutyl phthalate, dDUp = 9.4; dioctyl phthalate, dDop = 8.9; and trixylyl phosphate, dTxp = 9.9. dTopN = 9.7, d l - ~ ~ p ~ = 9.5 and dTHL)pN = 9.4 (Gal Cm-3)1/2. The solubility parameter of the resulting membrane (d,) was calculated from the equation: where w refers to the percentage mass and the subscript 'plast' refers to the plasticizer.Electrode lifetimes These were measured by storing the electrodes in 100 ml of a mixture (1 + 1) of 1 x 10-3 mol dm-3 KN03 and ISA-Pb solutions under constant stirring. The mixture was renewed twice a week, and the calibration graph assessed at appro- priate intervals with measurement of the slope and the detection limit. Nitrate determination All the solutions used were of analytical-reagent grade. Nitrate in water, soil and aqueous extracts of plants was determined by the electrodes using ISA-Pb9 in a 1 + 1 mixture. Nitrate was also determined by ion chromato- graphy's to serve as an independent method for comparison. Results and Discussion Characteristics of Nitrate ISEs Based on QPS The optimum membrane composition (PVC-plasticizer-sen- sor), experimentally assessed, was: ( a ) 35 + 50 + 15-17 TDDPN; and ( d ) 35 + 64 + 2 THDPN; where the plasticizer could be DBP, DOP or TXP.This optimum membrane composition was established by a large set of experiments involving various combinations, representative examples of which are shown in Table 1. Diethyl phthalate, a commonly used plasticizer with a relative molecular mass (M,) of 222.2, was rejected because it dissolved significantly in water. All electrodes with TBHDPN, TOPN and TDDPN as sensors showed no potential-pH dependency within the pH range 3-11. When THDPN was used as a sensor, however, the electrodes showed such a dependency above pH 6.This was attributed to insufficient purification of this sensor (melting point 93 "C) as verified by DSC, which showed the presence of an impurity with a melting point of 90 "C. None of the cations NHS+, K+, Na+, Ca2+, Mg2+, Al3+ and Pb2+ interfered. The response time was less than 1 min in pure nitrate solutions whose concentration was lower than 1 x 10-5 rnol dm-3 NO3- and less than 20 s in concentrations higher than 1 x 10-5 rnol dm-3 NO3-, and it was inversely related to the volume fraction of the sensor, V,. The detection limit decreased (Table 1) in the sequence TBHDPN > TOPN > TDDPN > THDPN as expected, as the M , of the sensors increased progressively. The selectivity coefficients of nitrate ISEs TBHDPN; ( b ) 35 + 51 +13-15 TOPN; ( c ) 35 + 61 + 3-4 Table 2 Selectivity coefficients of nitrate ISEs based on quaternary phosphonium salts in comparison with well known electrodes. They were measured by the fixed interference method and are expressed as PkRO:,.X 15% 15% 4% 4% Interfering TBHDPN- TOPN- TDDPN- TDDAN- Orion anion* DBP DBP DBP DBP 93-07 I - Br- NOI- c1- F- HCO3- H2POI- sop CH3COO- -1.20 - 0.90 1.25 1 .80 2.20 2.20 2.20 2.20 2.30 * 1 x 10-2 rnol dm-3.1.15 -1.15 - 0.92 0.94 1.35 1.35 2.10 2.25 3.00 3.35 2.80 3.25 3.30 3.70 2.70 3.20 3.00 3.50 -1.15 -1.15 0.93 0.88 1.20 1.30 2.30 2.25 3.25 2.70 3.40 - 3.70 2.80 3.30 2.80 3.20 2.70364 w ANALYST, APRIL 1991, VOL. 116 I \ h\, I I based on QPS are presented in Table 2. It can be seen that, for interfering anions usually present in water, soil and plant extracts, these selectivity coefficients are comparable to those exhibited by the well known electrodes.Comparison of Nitrate ISEs Based on QPS With Well Known Electrodes Effect of d, The determined solubility parameters of QPS (R4P+) were similar to the corresponding QAS (R4N+) with the same chain length of R. This is attributed to the very bulky R substituents, which decrease the contribution of the central atoms (P or N) to the cohesive energy density of the molecule to a minimum. Consequently, any difference in performance observed (and discussed below) between the electrodes based on QPS and their QAS counterparts could not be attributed to differences in the value of d, alone. Effect of V, and dplast The electrode characteristics should be influenced by the value of V, and the type of plasticizer (as dplast influences the value of d,) as predicted by eqn.(1). However, for nitrate ISEs based on QPS, the experimental results (Table 1) showed that the type of plasticizer and the value of V, (within the range used in this study) did not significantly influence the slope, linearity or detection limit, while in all instances the asym- metry potential was less than 3 mV. On the other hand, these factors had some effect on the electrodes based on QAS as was also reported by Nielsen and Hansen.6 This difference in behaviour between the two groups of sensors and the apparent non-conformance to eqn. (1) of the electrodes based on QPS is elaborated in the next section. Lifetime studies Nitrate ISEs based on 4% of TDDPN-DBP and 3% of TDDPN-DOP exhibited an almost constant slope (Fig. 1) and detection limit (Fig.2) for a period of 18 weeks, whereas this period for 15% TOPN-DBP was 13 weeks. It should be noted that the term lifetime implies the time up to which the electrode slope does not decrease by more than 2 mV. Beyond that time the electrode slope, and its characteristics in general, changed at such a rate that the precision and accuracy of the measurements were influenced. Consequently, it can be stated that the lifetime of the electrodes based on TOPN and TDDPN were 16 and 20 weeks, respectively. On the other hand, the slope and the detection limit (Figs. 1 and 2) of Orion 93-07 and TDDAN nitrate ISEs were continuously changing, giving lifetimes of 5 and 9 weeks, respectively. The lifetime of the latter electrode was in agreement with that reported by Wright and Bailey,16 while the 16 week lifetime for the TOPN electrode far exceeded the 2 week lifetime of the correspond- ing ammonium-based electrode (TOAN) reported by the same workers.Dispersion of the Sensors into the Membrane Matrix The dispersion of the sensors into the membrane matrix was observed with both optical and scanning electron microscopy. This was necessary in order to explain the above mentioned improved characteristics of the electrodes based on QPS, on the one hand, and the apparent non-conformance of our experimental results to eqn. (l), on the other. With optical microscopy (Fig. 3 ) the sensors appear as black spots on a grey background (membrane matrix), while with scanning electron microscopy (Fig.4 ) the sensors appear white on a grey-black background. Optical microscopy revealed that the sensing membranes in the Orion 93-07 and 3% TDDAN-DBP electrodes were not homogeneous [Fig. 3(a) and (b)]. Light photographs of the Orion electrode were considered to be sufficient evidence for the existence of membrane heterogeneity, and scanning electron microscopy was judged not to be necessary. However, for the other QAS electrodes (4% TDDAN-DBP and 4% TDDAN-DOP), scanning electron microscopy was considered necessary to verify the membrane heterogeneity [Fig. 4(a) and (b)]. The term heterogeneity is used in the sense that two or more separate phases exist. The asymmetry potential, EAS, observed in certain of the electrodes was the first point that could be accounted for by means of the photographs taken; E" includes e.m.f.contribu- tions from the internal reference electrode, inner phase boundary diffusion potentials and membrane asymmetry. The membrane asymmetry, owing to its heterogeneity, explains mainly: (i) the E*S of -20 mV (Table 1) of the Orion 93-07 ISE [Fig. 3(a)]; and (ii) the dependence of E" and other electrode parameters on V,, and on the type of plasticizer for TDDAN electrodes, found in this study and reported by Nielsen and Hansen6 (see Table 1 and Figs. 3(b), and 4(a) and (b)]. Nielsen and Hansen6 interpreted their results in terms of eqn. (l), believing that the observed differences in behaviour of the various electrodes tested reflected variations in the parameters V, and d,, which in turn influenced the value of the distribution coefficient, q .However, the findings of the 60 I 6.0 I 5.8 m 5.6 a 1 +- .- E 5.4 C 0 .- +- 5.2 8 5.0 *.... c *A 4.8 I 1 1 I I 1 1 0 2 4 6 8 10 12 14 16 18 20 Electrode agelweeks Fig. 2 Detection limits with ISA-Pb obtained during the lifetime studies of the nitrate ISEs, stored in a mixture of 1 X 10-3 mol dm-3 NO3- and ISA-Pb solutions (1 + 1). A, 4% TDDPN-DBP; B, 3% TDDPN-DOP; C, 15% TOPN-DBP; D, 4% TDDAN-DBP; and E, Orion 93-07ANALYST, APRIL 1991, VOL. 116 365 Fig. 3 Photomicrographs taken with an optical microscope ( X 100 magnification) of two nitrate ISEs: ( a ) Orion 93-07; and (b) 3% TDDAN-DBP present study suggested that the dependence of E" and other electrode parameters on V, and d,, when QAS were used as sensors, reflected the membrane heterogeneity rather than the variation in the values of V, and d,.Heterogeneity violates the theoretical assumptions on which eqn. (1) is based and in practical terms might cause differences in the electrical properties across the membrane and deterioration of its stability, owing to the incongruous dissolution of its com- ponents in the contacting solution (see also the following discussion on QPS). The 4% TDDAN-DBP ISE showed no asymmetry potential (Table 1; E*S = +2 mV), in spite of its membrane heterogeneity, possibly because of the symmetrical distribution of the sensor in the matrix [Fig. 4(a)]. Optical microscopy was unable to reveal the extent of the dispersion of the sensor in the PVC matrix for QPS.Therefore, scanning electron microscopy was necessary in order to unravel the fine details of the whole structure and to distinguish the extent of disturbance, if any, of the PVC structure on addition of the sensor. The structure of the PVC matrix without the sensor (actual photographs not given) was almost the same as that shown in Fig. 4(c), ( d ) and (e) for the plasticizers DBP, DOP and TXP, respectively. It was, therefore, evident that, on addition of the sensor (QPS) and up to the point of the membrane saturation, no discrete phases were discernible and the structure of the PVC matrix was not significantly disturbed [Fig. 4(c), (d) and (e)]. In addition, no distinct particles were discernible and no contrast existed, even when using the ~ 4 0 0 0 magnification power of the scanning electron microscope (where its resolution is 0.2 pm).Both facts implied the absence of localized accumulation of the sensor. All these findings clearly demonstrated that the sensing membrane was homogeneous in nature (no separation of the phases) and that QPS were more compatible with the plasticized PVC than the corresponding QAS and o-phenan- throline complexes [cf. Fig. 4(c) and ( d ) with Fig. 4(a) and ( b ) and Fig. 3(a)]. One can, therefore, postulate, with a reasonable degree of certainty, that the sensing membrane approximated to a true solution, i.e., that the sensors (QPS) were dissolved in the plasticized PVC. The homogeneous dispersion of QPS in the plasticized PVC, irrespective of the type of plasticizer and the value of V, used in this study, had the result that the electrode pzrameters were also independent of these factors as can be seen from the data of Table 1 and of Figs.1 and 2. It also accounted for the following experimental observation: (i) the lower detection limit observed for the 4% TDDPN-DBP ISE compared with that of the 4% TDDAN-DBP electrode (Table 1 and Fig. 2); (ii) the longer lifetimes of QPS electrodes owing to the stronger retention of the sensor in the PVC matrix, as the major factor limiting the lifetime of a solvent-polymeric membrane is the loss of the sensor and/or the plasticizer in the solution analysed; and (iii) the presence of no asymmetry potential for the QPS-based electrodes, as the homogeneity of the sensing membrane ensures its symmetry and the congruous dissolution of its constituents into the contacting solution.It was evident, therefore, that when homogeneity of the sensing membrane was achieved, small differences in the value of d, could not influence significantly the electrode parameters. This was also dictated by the mathematical form of eqn. (1). Also, a wide variation in the value of V, up to the point of membrane saturation did not significantly change the electrode parameters, in disagree- ment with what eqn. (1) predicts. This non-conformance to eqn. (1) might be due to the fact that the state of equilibrium, mandatory for eqn. (1) to hold, is rarely, if ever, attained during the usually short periods of measurement. It seems, therefore, that, although eqn. (1) provides thermodynamic information as to which parameters influence the quality of nitrate ISEs, its practical usefulness is indeed open to question.The isolated spots on Fig. 4(c) and (d) represent small undissolved particles of the sensor as the point of the membrane saturation is approached, a fact that was also verified at higher magnification. This explains the similar detection limits for 4% TDDAN-DBP and 4% TDDPN-DBP ISEs observed at the beginning of the lifetime studies (Fig. 2). When the point of saturation was exceeded (supersaturation) the initial membrane matrix structure was disturbed [Fig. 4 0 1 . This super concentration of the sensor, depicted by the short white lines (contrast), had an adverse effect on the electrode parameters (see the detection limits of the 3 and 6% TDDPN-DBP ISEs given in Table 1) because of the mem- brane heterogeneity.It also explains the same detection limits observed for 4% TDDAN-DBP and 6% TDDPN-DBP ISEs (Table 1). The higher compatibility and, in fact, dissolution of QPS into the plasticized PVC in comparison with the corresponding QAS and o-phenanthroline complexes deserves further atten- tion. Plasticized PVC is an amorphous compound,l4 while all the sensors used possess crystalline properties at ambient temperature. Therefore, for dissolution of the sensor to occur, there must be hydrogen bonding or some other specific interaction, either with the plasticizer or with the supporting medium (PVC) or both. This specific interaction is necessary to overcome the tendency of crystalline compounds to form separate phases on solidification.Corroborating evidence for the existence of some specific interaction between the sensor and the supporting medium was obtained by separate experi- ments in which the dissolution of TDDAN and TDDPN in the plasticizers was examined. Neither sensor dissolved in the plasticizers alone. This observation coupled with the fact that TDDAN was segregated in the membrane matrix [Figs. 3(b) and 4(6)], whereas all QPS seemed to be dissolved, supported our belief that QPS must exhibit some specific interaction with the PVC, whereas QAS does not. As PVC is a weak proton-donor polymer,14 one possible mechanism governing366 ANALYST, APRIL 1991, VOL. 116 Fig. 4 ( c ) 4% TDDPN-DBP; (d), 4% TDDPN-DOP; ( e ) 14% TOPN-TXP; and cf) 6% TDDPN-DBP Photomicrographs taken with an electron scanning microscope of various nitrate ISEs : (u) 4% TDDAN-DBP; (b) 4% TDDAN-DOP; QPS and QAS interaction with the PVC stems from the configuration of the QPS and QAS molecules and the distribution of the negative fractional charge, 6-, in the polar bonds P-I, P-Br, N-I and N-Br.These bonds are considered because the sensors were dispersed in the matrix in the form of bromide or iodide salts. The molecule of the sensor consists of four organic chains at the apices of a tetrahedron, co- ordinated to one P or N atom at the centre. Bromide and iodide ions lie outside the organic tetrahedron and, consider- ing the electronegativity values of P = 2.1, I = 2.5, Br = 2.8 and N = 3.0, it seems reasonable to postulate that, for the N-I or N-Br bond, the negative fractional charge 6- is concen- trated towards the central (N) within the tetrahedron and, hence for steric reasons, any interactions with the positive charge of the PVC molecule is hindered./ \ H+-C--Cl However, for the P-I or P-Br bond, 6- is more concentrated towards the I- and Br- outside the tetrahedron. This external concentration of 6- favours van der Waal’s attraction forces with the PVC molecule, rendering the two compounds compatible with each other and preventing the formation of separate phases. Applications The nitrate ISEs based on QPS in a PVC matrix were used to determine nitrates in waters and in soil and plant extracts (Table 3). Nitrate values assessed with these electrodes were closely correlated with those obtained with 4% TDDAN- DBP, Orion 93-07 nitrate ISEs and ion chromatography, while the regression lines had a slope and an intercept not statistically different from 1 and 0, respectively, at the 5% probability level (Table 3).The precision of the measurementsANALYST, APRIL 1991, VOL. 116 367 Table 3 Determination of nitrate with nitrate ISEs based on quaternary phosphonium salts. The results are compared with those obtained with TDDAN and Orion 93-07 nitrate ISEs and the ion chromatography (IC) method Sample Tap water Tap water Tap water Well water Well water Well water Plant Plant Soil Soil 15% TOPN- DBP 2.9 21.9 42.1 55.5 69.0 107.3 159.1 210.1 92.5 76.9 (Yd 4% TDDPN- DBP 2.8 21.3 40.6 55.8 69.4 104.9 157.9 211.5 89.4 79.0 (Y2) 3 yo TDDPN- DOP 2.9 21.3 42.2 56.7 71.0 105.3 158.5 213.0 92.1 78.7 (Y3) 4 yo TDDAN- DBP 2.5 20.6 41.2 56.7 69.7 103.8 155.2 210.5 88.5 78.4 0-1) Orion 93-07 2.7 21.0 41.5 56.0 69.8 104.9 156.6 209.2 88.4 79.0 (YS) Correlation coefficient ( r ) Regression equation 0.9995* yl = 0.14 + 0.989~ 0.9997* y2 = -0.58 + 0.993~ 0.9996" y3 = -0.01 + 0.997~ 0.9999* yz = -0.24 + 1.009~4 0.9999* y2 = -0.64 + 1.012~5 * Significant at the 5% probability level.IC 2.6 21.5 42.5 57.8 70.1 109.5 157.1 214.1 90.4 78.9 (X) (oA) and blank values (oB) obtained with the electrodes were assessed from ten replicates of a standard solution of 100 mg dm-3 of NO3- and were found to be: (i) 15% TOPN-DBP or DOP, 0.60 2 0.15, 1.30 k 0.20; (ii) 4% TDDPN-DBP or DOP, 0.50 iz 0.10, 0.60 k 0.20; (iii) 4% TDDPN-TXP, 0.60 k 0.15,0.70 k 0.20; (iv) 4% TDDAN-DBP, 0.90 k 0.15,l.OO k 0.20; and ( v ) Orion 93-07, 0.90 & 0.15, 1.00 k 0.20 mV.Conclusions The homogeneous dispersion of QPS in a PVC matrix resulted in nitrate ISEs, superior to earlier electrodes in both stability and sensitivity. It also helped in clarifying the following practical aspects for the development of nitrate ISEs based on non-porous PVC membranes: (i) the solubility parameter of PVC is 9.5 (cal cm-3)1/* and the solubility parameters of the plasticizers range from 8.5 to 10.5 (Brydson14) resulting in a membrane solubility parameter range of 9-10; consequently, according to the general rule that d, = d,, the solubility parameter of the sensor used in a PVC matrix must lie within the range 9-10; (ii) the point of membrane saturation with the sensor must not be exceeded; and (iii) the homogeneous dispersion of the sensor (a crystalline compound) in a plasticized PVC matrix is ensured when the sensor is either dissolved in the plasticizer or linked by some mechanism to the supporting medium (PVC).It is our opinion that the sensor must exhibit proton-acceptor properties and must, by necess- ity, be linked to the PVC. In general, the choice of the membrane constituents (polymer and/or sensor) should be made in such a way that they should have the same solubility parameters and exhibit some specific interaction, favouring the homogeneity of the sensing membrane. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 References Moody. G . J . . Oke, R. B., and Thomas, J . D. R.. Analyst, 1970, 95. 910. Davies, J . E. W., Moody. G. J., andThomas, J . D. R., Analyst. 1972, 97, 87. Hulanicki, A., Maj-Zurawska. M., and Lewandowski, R., Anal. Chim. Acta, 1978, 98, 151. Siemroth, J., Henning, I . , and Hatmann, P., GDR Pat. 127327, 1977; Chem. Abstr., 1978.89, 122624q. Viellmeyer, H. P., Dahse, I., and Gruenke, U., GDR Pat. 210352, 1984; Chem. Abstr.. 1985, 102, 207683'. Nielsen, H. J., and Hansen, E. H., Anal. Chim. Acta, 1976, 85, 1. Feshchenko, N. G., Mazepa, I. K., Makovetskii, Yu. P., and Kirsanov, A. V., Z h . Obsh. Khim., 1969, 39, 1886. Eriksen, S. P., Tuck, L. D . , and Oneto, J . F., J . Org. Chem., 1960, 25, 849. Mitrakas, M., and Alexiades, C.. Mikrochim. Acta, Part I . 1990. 7. Marquardt, D. W., J. SOC. Ind. Appl. Math., 1963,431. Recommendation f o r Nomenclature of Ion-Selective Electrodes: Appendices on Provisional Nomenclature, Symbols, Units and Standards, Number 43, IUPAC Secretariat, Oxford, 1975. Midgley, D., Analyst, 1979, 104, 248. Small, P. A.. J. Appl. Chem., 1953, 3, 71. Brydson, J . A., Plastic Materials, Chapel River Press, Andover, 2nd edn., 1970, p. 68. Tsitouridou, R., and Puxbaum, H . , Int. J. Environ. Anal. Chem., 1987, 31, 11. Wright, J. A., and Bailey, P. L., in Ion Selective Electrodes Conference, Budapest, 1977, ed. Pungor, E . , Elsevier, Amster- dam. 1978, pp. 603-609. Paper 010461 4 K Received October 15th, 1990 Accepted December 4th, I990

ISSN:0003-2654    

DOI:10.1039/AN9911600361

出版商:RSC    

年代:1991

数据来源: RSC

摘要:

ANALYST, APRIL 1991, VOL. 116 369 Adsorptive Stripping Voltammetric Behaviour of Copper(i1) at a Hanging Mercury Drop Electrode in the Presence of Excess of Imidazole F. Nil Ertas, Josino C. Moreira and Arnold G. Fogg* Department of Chem istr yf Lo ug h bo roug h University of Techno log yf L oug h bo roug h Leiceste rsh ire LEI1 3TUf UK In the presence of excess of imidazole (1.0 x 10-3 rnol dm-s), copper(ii), at pH 8.5, adsorbs at a hanging mercury drop electrode to give two adsorptive stripping voltammetric peaks at -0.36 and -0.46 V. The peak at -0.36 V is only present at accumulation potentials more negative than -0.05 V versus Ag-AgCI: it increases in height as the accumulation potential becomes more negative up to and beyond -0.6 V, the voltammetric sweep being started at -0.20 V.This peak appears to be due to the adsorption of polymeric [C~lI(lm)~] or its reduced copper(i) form. The peak a t -0.46 V is only present at high imidazole concentrations (>5.0 x 10-4 rnol dm-3): the accumulation is uniform from 0.0 to -0.36 V but is negligible at potentials more negative than -0.46 V. This peak appears to be due to adsorption of [Cu(lm),]2+. On cycling between 0.0 and -0.6 V this latter complex is converted into the polymeric complex and only the peak at -0.36 V remains. Copper(ii) can be determined by using the peak at -0.46 V after accumulation at 0.0 V, or a t -0.36 V after accumulation at -0.6 V. The latter method is more sensitive: the detection limit is about 2.0 x 10-9 rnol dm-3 after accumulation for 3 min. Keywords: Adsorptive stripping voltammetry; copper(1i); imidazole complex The imidazole ring, which is present in the amino acid histidine, functions as a ligand towards transition metal ions in a variety of biologically important molecules including the iron-haem system, vitamin B12 and its derivatives and several metalloproteins. 1 The imidazole nitrogen atoms of the his- tidine residues provide one of the primary means by which metal ions can be bound to proteins.The relationship between the structural property of the imidazole ring, its function in biological systems and its complexes with a number of transition metal ions have been reviewed.2 Imidazole is amphoteric, being a moderately strong organic base capable of accepting a proton at N-3 (the pyridine-like nitrogen, pK, = 7.1) and also a very weak acid capable of losing a proton from N-1 (the pyrrole-like nitrogen, pK, = 14.3).In neutral solutions the unprotonated imidazole molecule usually functions as a ligand by using the unshared pair of electrons on N-3. In sufficiently basic media the conjugate base of imidazole, Im-, is formed and its complexes with dipositive metal ions are considered to have a stoichiometry [M(Im)2]. These complexes have been considered to be polymeric, and are, in general, insoluble. The imidazole salt of copper(1) can be prepared also and it has been considered to have a polymeric bridge structure.3 Li et al.4 have investigated the polarographic behaviour of the copper( 11) complex formed at high concentrations of imidazole in water-ethanol mixtures and showed that the complex is reduced in two steps giving two waves of approximately equal height, the first at -0.19 V and the second at -0.57 V versus a saturated calomel electrode (SCE).Both waves are due to a one-electron reduction. They also found the copper(I1) and copper(1) complexes with the highest Im : Cu ratio to be [C~(Im)~12+ and [ c ~ ( I r n ) ~ ] + , respectively. Recent work in this laboratory5.6 has indicated that histidine can be determined at the nanomolar level at a hanging mercury drop electrode (HMDE) as its copper(I1) complex, and that copper(I1) can be accumulated rapidly and selectively at an HMDE modified by adsorption of a poly(L-histidine) film. In view of the biochemical importance of compounds containing the imidazole ring, and the affinity of this ring for coordinating metal ions,7 a study has been made, and is * To whom correspondence should be addressed.reported here, of the adsorptive stripping voltammetric behaviour of copper(I1) in the presence of excess of the parent imidazole molecule. Copper(I1) can be determined by this means. Experimental Adsorptive stripping voltammetry was carried out by using a Metrohm 626 Polarecord with a 663 VA stand in conjunction with a multi-mode electrode in the HMDE mode. The three-electrode system was completed by means of a glassy carbon auxiliary electrode and an Ag-AgC1 reference elec- trode. All potentials given are relative to this Ag-AgC1 electrode. A pulse amplitude of 50 mV was used with a scan rate of 10 mV s-1 and a forced pulse interval of 1 s.A Princeton Applied Research Model 174A polarographic analyser in conjunction with a VA 663 stand was used for cyclic voltammetry. The pH measurements were made with a Corning combined pH/ reference electrode by using a Radiometer PHM 64 pH meter. Imidazole [ultraviolet(UV) spectroscopic grade, specially prepared for use in UV spectrophotometric beta-lactam assays] was obtained from Merck (formerly BDH) and the other chemicals were purchased from Sigma. All were used without further purification. A 0.2 mol dm-3 solution of imidazole was prepared by dissolving 0.1361 g of imidazole in water acidified with 6 drops of 6 rnol dm-3 HCl in a 10 ml calibrated flask. Procedure The general procedure used to obtain adsorptive stripping voltammograms was as follows.A 20 ml aliquot of 0.01 rnol dm-3 hydrogen carbonate buffer solution was placed in a voltammetric cell and the required amounts of standard imidazole and copper(1r) solutions were added. The stirrer was switched on and the solution was purged with nitrogen for 6 min. Subsequently, de-oxygenation was carried out for 15 s between adsorptive stripping cycles. After forming a new HMDE, accumulation was effected for 2 min at the required potential while stirring the solution. At the end of the accumulation period the stirrer was switched off, and, after 20 s had elapsed to allow the solution to become quiescent, a370 ANALYST, APRIL 1991, VOL. 116 negative potential scan was initiated between the accumula- tion potential and -0.7 V. In the study of the influence of accumulation time, the adsorptive accumulation was carried out at potentials more negative than -0.4 V, and immediately after this step, the potential was changed to -0.1 V from where a negative potential scan was initiated.Cyclic voltammetry was preceded by accumulation at 0.0, -0.1 or -0.6 V for 2 min. A scan rate of 50 mV s-1 was used. The initial sweep was to more negative potentials when accumulation was performed at 0.0 or -0.1 V. Accumulation at -0.6 V was followed by an anodic sweep to 0.0 V. Second and third scans were made on the same drop immediately after the first scan without further accumulation. Direct current voltammetry was performed by using the HMDE and starting at 0.18 V with a scan rate of 2 mV s-1. Results and Discussion The shapes of the differential-pulse stripping voltammograms and of the cyclic voltammograms obtained after accumulation in a 1.0 X 10-7 mol dm-3 solution of copper(I1) in the presence of excess of imidazole were found to be dependent on the pH, accumulation potential, accumulation time, and copper(r1) and imidazole concentrations.The influence of the pH on the differential-pulse stripping peak current of a 2.0 x 10-7 mol dm-3 solution of copper(I1) in the presence of 1.0 x 10-3 rnol dm-3 imidazole is summarized in Table 1. At pH 4.5 (0.1 mol dm-3 acetate buffer) no significant adsorption of the complex at the electrode surface was observed. In neutral or basic media two different peaks, at -0.36 and -0.46 V versus Ag-AgC1, were observed and the heights of both of these decreased with increasing pH above 8.5.This effect can be attributed to the formation of hydroxo complexes with copper(n) .4 As the highest currents were obtained in 0.1 mol dm-3 hydrogen carbonate buffer (pH 8.5), this buffer was chosen for use in further studies. A very small shift (only a few mV) was observed in the peak potentials when the pH was varied from 7.0 to 10.5, showing that there is no loss or gain of protons in the reduction process. The influence of the accumulation potential on the cyclic voltammograms of a 3.0 x 10-7 rnol dm-3 solution of copper(Ii), at various imidazole concentrations, is shown in Fig. 1. The cyclic voltammograms obtained when accumula- tion was carried out at 0.0 V for 120 s [Fig. l(a)], in the presence of 1.0 x 10-3 mol dm-3 imidazole, gave a single peak at -0.42 V in the cathodic scan. Two small, broad peaks at -0.40 and -0.35 V were observed in the subsequent anodic sweep.No peak was observed when the imidazole concentra- tion was significantly lower. When accumulation was perfor- med at -0.1 V, an increase in the peak current and a shift in the peak potential were observed when the imidazole concen- tration was increased from 1.0 X 10-6 to 1.0 X 10-3 mol dm-3. A single peak at -0.17 V was observed in the presence of 1.0 x 10-6 mol dm-3 imidazole. At imidazole concentrations between 1.0 x 10-4 and 1.0 x 10-3 mol dm-3 [Fig. l(b)], two peaks were observed in the cathodic scans: the first at about -0.36 V and the second at -0.46 V. Associated with them, Table 1 Effect of pH on the peak current obtained for the differential-pulse adsorptive stripping voltammetry of the copper(I1)- imidazole complex after accumulation at -0.1 V for 120 s.[Cop- per(~~)] = 2.0 x 10-7 mol dm-3; [imidazole] = 1.0 x mol dm-3 Peak current at PH -0.46 VInA 4.5 0 7.0 12.2 8.5 29.2 9.5 9.0 10.5 5.0 peaks at -0.40 and -0.30 V were observed in the anodic scans. Small shifts in the peak potential of the first peak were observed with increasing imidazole concentration. Decreases in the height of the peak at -0.36 V and increases in the height of the peak at -0.46 V were observed with increasing imidazole concentration. At an imidazole concentration of 25 2 .> 0 - 25 25 P .> 0 - 25 r I I I -0.2 -0.4 -0.6 nA I I 1 I -0.2 -0.4 -0.6 D A E 25 -25 IVB L A I I I I -0.2 -0.4 -0.6 PotentiaiN Fig.1 Effect of the imidazole concentration and the accumulation potential on the cyclic voltammograms. obtained at an HMDE, for a 3 x 10-7 rnol dm-3 copper(i1) solution in 0.1 rnol dm-3 hydrogen carbonate buffer at pH 8.5 with an accumulation time of 120 s. ( a ) Accumulation at 0.0 V; (b) accumulation at -0.1 V; and (c) accumulation at -0.6V. Concentration of imidazole [Im]: A, 0; B, 1.0 x 10-6; C, 1.0 x 10-5; D , 1.0 x rnol dm-3 for (a) and ( c ) . For (b) A, 1.0 x 10-4; B, 3.0 x D , 7.0 x 10-4; E, 1.0 x 10-3; and F, 1.5 X 10-3 mol dm-3 and E, 1.0 x C, 5.0 xANALYST, APRIL 1991, VOL. 116 371 2 x 10-3 rnol dm-3, only the peak at -0.46 V was present in the cyclic voltammogram. The presence of a single peak in both the cathodic and anodic scans was observed in the voltammograms when accumulation was carried out at potentials more negative than -0.4 V and scanning from 0.0 V [Fig.l(c)]. In this instance the difference in the peak potentials of the cathodic and anodic waves was found to be 60 mV and the value of the width at half-height observed was about 60 mV. This suggests a one-electron reduction. The influence of the accumulation potential on the peak currents for a 3 x 10-7 rnol dm-3 solution of copper(i1) in the presence of 1.0 x 10-3 rnol dm-3 imidazole is shown in Table 2. The potential of the peak at the less negative potential was shifted significantly in the negative direction with an increase in the imidazole concentration, as expected for complex formation (see Table 3). Cyclic voltammograms of a 3 x 10-7 rnol dm-3 solution of copper(i1) in the presence of 1.0 x 10-3 rnol dm-3 imidazole obtained by successive scans at the same drop are shown in Fig. 2.Initial accumulation was performed for 120 s. No further accumulation was carried out between the scans. When accumulation was performed at -0.1 V [Fig 2(a)], the two cathodic peaks at -0.36 and -0.46 V, and associated with them two anodic peaks at -0.40 and -0.30 V, were observed. A decrease in the height of the peak at -0.46 V and a simultaneous increase in the height of the peak at -0.36 V were observed with increasing scan number. An isosbestic point was obtained at -0.41 V. The appearance of two cathodic peaks appears to be due to accumulation of two different copper(i1)-imidazole complexes at the electrode surface.The subsequent predominance of the peak at -0.36 V must be caused by the complex which is reduced at -0.46 V being converted into that responsible for the peak at -0.36 V. In fact, Nozaki et al.8 observed four different copper(I1) complexes in the copper(Ii)-4-methylimidazole system, the zone of predominance of each depending on the 4-methyl- 'Fable 2 Influence of the accumulation potential (Eacc) on the peak currents (ip) of a differential-pulse adsorptive stripping voltarnmo- gram of a 3 x rnol solution of copper(i1) in presence of 1.0 X 10-3 rnol dm-3 imidazole in 0.1 rnol dm-3 hydrogen carbonate buffer (pH 8.5). Accumulation time: 120 s iplnA E a c P 0.0 -0.05 -0.10 -0.15 -0.20 -0.30* -0.33* -0.37* -0.40* -0.42* -0.45* -0.50* * Potential scanned from -0.36 V - - 5.0 26.0 37.0 35.5 35.5 37.0 42.5 61.5 72.5 84.0 89.0 -0.20 v -0.46 V 37.5 37.5 42.5 44.0 41 .O 40.0 39.0 33.5 19.0 12.0 - - Table 3 Effect of the imidazole concentration on the cathodic peak potential ( E p ) of the copper(ii)-imidazole complex in 0.1 rnol dm-3 hydrogen carbonate buffer (pH 8.5).[Copper(ii)] = 3 x rnol dm-3. Accumulation step, -0.6 V for 120 s [ Imidazole]/ rnol dm-3 EdV 1.0 x 10-6 -0.18 1.0 x 10-5 -0.22 1.0 x 10-4 -0.27 1.0 x 10-3 -0.36 imidazole concentration. Only the cathodic peak at -0.37 V and the anodic peak at -0.30 V were observed in the voltammograms when accumulation was carried out at -0.6 V [Fig. 2(b)]. A small increase in the peak current with successive scans suggests film formation on the electrode surface.The influence of the addition of copper(I1) on the cyclic voltammograms obtained for a 1.0 x 10-3 rnol dm-3 solution of imidazole in 0.1 rnol dm-3 hydrogen carbonate buffer (pH 8.5) is shown in Fig. 3. With accumulation at -0.1 V [Fig. 3(a)], a single peak at about -0.42 V was observed at copper(i1) concentrations lower than 1.0 x 10-7 rnol dm-3. The height of this peak increased with increasing copper(i1) concentration. At copper(i1) concentrations higher than 1.0 X rnol dm-3, a second peak was observed at -0.36 V. 25 .-a 2 0 -25 -0.2 -0.4 -0.6 -0.2 -0.4 -0.6 PotentialN Fig. 2 Effect of successive scans on the cyclic voltammograms of a 3 x 10-7 rnol dm-3 copper(i1) solution in the presence of 1 x 10-3 mol dm-3 imidazole in 0.1 rnol dm-3 hydrogen carbonate buffer at pH 8.5.Accumulation for 120 s at (a) -0.1 V and ( b ) -0.6 V. The numbers indicate scan number 25 5 \ .-a , - 25 Fig. 3 G I I 1 -0.2 -0.4 -0.6 Y -0.2 -0.4 -0.6 PotentialN Effect of the copper(i1) concentration on the cyclic voltarnmo- grams at the HMDE, for a 1 X 10-3 rnol dm-3 irnidazole solution in hydrogen carbonate buffer at pH 8.5. Accumulation at ( a ) -0.1 V and ( b ) -0.6 V for 120s. Copper(i1) added: A, 0; B, 5 x D, 2 x lo-'; E, 3 x F, 4 x 10-7; and G, 5 x 10-7 rnol dm-3 C , 1 X372 The height of this peak increased with a further increase in the copper(i1) concentration. When this peak was present in the voltammogram, only a small increase in the height of the peak at -0.42 V was observed with increasing copper(1i) concentra- tion.A similar behaviour was observed in the anodic scans and, in this instance, a peak at -0.30 V, associated with the cathodic peak at -0.36 V, predominated. These results agree with those obtained by increasing the imidazole concentration and suggest that the complexes responsible for the two peaks are strongly dependent on the ratio of copper(i1) : imidazole. These experiments suggest that the complex undergoing reduction at -0.36 V is more stable and more easily formed at the electrode surface during the scan time. The nature of these peaks is still not clear. One hypothesis for the cyclic voltammograms obtained is that the reduction observed at -0.46 V could be due to the reduction of a higher order complex of copper(i1) present in the solution, possibly [ C U ( I ~ ) ~ ] ~ + , according to the reaction [Cu(Im),]2+ (ads) + e- +.[Cu(Im),]+(ads) + 21m with the subsequent formation of a polymeric copper(1)- imidazole complex at the electrode surface. The peaks at -0.30 V (oxidation) and -0.36 V (reduction) would be due to the reduction or oxidation of the metal ion in the polymeric film formed at the electrode surface. In fact, copper(i) was reported to react with imidazole at pH values higher than 4.5, forming [Cu(Im),]+ which, at pH >6.5, polymerizes. In this instance imidazole acts as a bidentate ligand for copper(r) .9 The formation of a polymeric complex between copper(i1) and imidazole containing two imidazole molecules per metal atom in aqueous hydrogen carbonate solution has also been described.3 The peak current, obtained when accumulation was per- formed at -0.6 V and the scan was started at 0.0 V, increased rectilinearly with the square root of the scan rate.This relationship suggests diffusional behaviour probably caused by multimolecular film formation.10 A shift in the cathodic peak potential from -0.32 to -0.36 V was observed when the scan rate was varied from 10 to 100 mV s-1. A shift from -0.30 to -0.26 V was observed for the anodic peak under the same conditions. Similar results were obtained for both peaks (i.e., those at -0.36 and -0.42 V) when accumulation was carried out at -0.1 V. These results suggest that the system is irreversible to a small extent. From the above results it is apparent that copper(t1) can be determined by using an excess of imidazole as an accumulation reagent.When the accumulation was performed at -0.6 V, the height of the copper(I1)-imidazole peak at -0.36 V, obtained by using a solution containing 6 X 10-8 mol dm-3 copper(1i) and 1 x 10-3 rnol dm-3 imidazole, showed a rectilinear relationship [correlation coefficient ( r ) = 0.9981 with the accumulation time up to 6 min. Rectilinear calibra- tion graphs were obtained for copper(i1) in the presence of 1.0 x 10-3 rnol dm-3 imidazole when accumulation was carried out at 0.0 or -0.6V for 3 min. When accumulation was performed at 0.0 V for 3 min, a rectilinear calibration graph ANALYST, APRIL 1991, VOL. 116 was obtained from 5 X 10-9 to 1 X 10-7 rnol dm-3 copper(i1) ( r = 0.9998) with a slope of 2.84 x 108 nA mol-1. At copper(i1) concentrations higher than 1 x 10-7 rnol dm-3, deviation from linearity was observed probably owing to saturation of the electrode surface.Better sensitivity (5.2 X 108 nA mol-1) and range [5 X 10-9-1.5 X 10-7 rnol dm-3 copper(ii)] were obtained when accumulation was performed at -0.6 V. The limit of detection was 2 x 10-9 rnol dm-3 for accumulation at -0.6 V for 3 min. Several reagents have been suggested for the differential- pulse adsorptive stripping voltammetric determination of copper(ii).*1.12 The limits of detection for determinations with catechol and 8-hydroxyquinoline are given as 3 x 10-10 and 1 x 10-10 rnol dm-3, respectively,13 based on an accumulation time of 1 min from a stirred solution: the use of catechol has the disadvantage that its solutions are readily oxidized by air and the reagent must be freshly prepared." The detection limit using imidazole does not appear to be as good.Further studies are in progress on the adsorptive stripping voltammetric behaviour of a range of imidazole derivatives, including several of pharmaceutical importance. The determi- nation of these compounds by using this technique will be investigated; those compounds with reducible groups, e.g., the nitroimidazoles, might be determinable directly or as copper(i1) complexes. The use of imidazoles to determine copper and other metal ions will be studied in greater detail in order to find the most suitable reagent(s) for these metals. It is also intended to study the adsorptive stripping voltammetry of imidazole derivatives of beta-lactam antibiotics. J. C. M. thanks the Conselho Nacional de Desenvolvimento Cientifico e Tecnologico (CNPq, Brazil) for financial support. 1 2 3 4 5 6 7 8 9 10 11 12 13 References Tanford, C., J . Am. Chem. SOC., 1952, 74, 211. Sundberg, R. J . , and Martin, R . B . , Chem. Rev., 1974,74,471. Brown, G. P., and Aftergut, S . , J . Polym. Sci., 1964, 2, 1839. Li, N. C., White, J . M., and Doody. E., J. Am. Chem. SOC., 1954, 76, 6219. Moreira, J . C., and Fogg, A. G., Analyst, 1990, 115, 41. Moreira, J. C., Zhao, R . , and Fogg, A. G., Analyst, 1990, 115, 1561. Pettit, L. D., Pure Appl. Chem., 1984, 56, 247. Nozaki, Y., Gurd, F. R. N., Chen, R. F., and Edsall, J . T.. J . Am. Chem. SOC., 1957, 79, 2123. Sigwart, C., Kronek, P., and Hemmerich, P., Helv. Chim. Acta, 1970, 53, 177. Murray. R. W., in Electroanalytical Chemistry, ed. Bard, A. J . , Marcel Dekker, New York, 1984, vol. 13. van den Berg, C. M. G., Anal. Chim. Acta, 1984, 164, 195. van den Berg, C. M. G.. J . Electroanal. Chem., 1986,215,111. van den Berg, C . M. G., Analyst, 1989, 114, 1527. Paper 0104404 K Received October lst, 1990 Accepted December I7th, 1990

ISSN:0003-2654    

DOI:10.1039/AN9911600369

出版商:RSC    

年代:1991

数据来源: RSC