摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 325 Direct Atomic Spectrometric Analysis by Slurry Atomisation Part 3.* Whole Coal Analysis by Inductively Coupled Plasma Atomic Emission Spectrometry Les Ebdon and John R. Wilkinsont Department of Environmental Sciences, Plymouth Polytechnic, Drake Circus, Plymouth, Devon PL4 8AA, UK Aqueous suspensions (10% mN) of whole powdered coal dispersed in Triton X-100 (l0/o VN) have been analysed by slurry atomisation inductively coupled plasma atomic emission spectrometry. Coals were powdered to a particle size of less than 30 pm and sprayed into a conventional torch using a high-solids type nebuliser. Line selection and operating parameters are reported and results obtained for five coals by direct aqueous calibration. Slurry atomisation results for Cu, Fe, Mn, Ni and V in a range of reference and other coals are compared with results obtained by ashing and digestion, followed by atomic absorption spectrometry, and with results for certificate values.The precision (20) for the slurry method is ca. 12% relative and recoveries relative to digestion FAAS are Cu 67-105, Fe 25-41, Mn 3G195, Ni 93-243 and V 95166%. The method also appears applicable to coal ashes. Keywords: Slurry atomisation; whole coal analysis; solid sampling; inductively coupled plasma atomic emission spectrometry; trace metal analysis Coal has been identified as an energy resource and chemical feedstock of growing future importance. Consequently there is considerable interest in the determination of trace metals in coal for reasons as diverse as pollution and corrosion control, interest in catalyst poisoning and for identification purposes.While inductively coupled plasma atomic emission spec- trometry (ICP-AES) is now recognised as a rapid multi- element analytical technique capable of determining trace metals at the levels of interest in coal samples, it is traditionally regarded as a technique for solution analysis. The first paper in this series1 discussed the possibility of using slurry atomisation to introduce aqueous suspensions of coal via a high-solids nebuliser into an ICP and the use of aqueous solutions for calibration. After optimisation Mn was deter- mined with an average recovery of 95%. These and other2 encouraging results prompted this study in which such a system has been used to investigate the possibility of determin- ing Cu, Fe, Mn, Ni and V in powdered whole coal by slurry atomisation ICP-AES.Experimental The Radyne R50P free-running plasma generator and optical system used have been described previously,lJ as has the PTFE high-solids nebuliser used,4 which is a one-piece Babington-type with a gas orifice of 0.35 mm and a solution orifice of 0.8 mm. The nebuliser was force-fed by a small peristaltic pump (Schucho Mini Pump Mark IV, 60 rev min-1, Schucho Scientific, Halliwick Court Place , London, UK). The aerosol produced was transported to the central injector tube of the plasma torch via a laboratory-constructed double- pass spray chamber.4 When high pumping-rates were used to deliver coal slurries to the nebuliser a recycling spray chamber1 was used to conserve sample solution.A modified Greenfield torch was used. An inverted image of the plasma was projected 1 : 1 on to the 25-pm entrance slit of the monochromator using a quartz lens, 7.5 cm focal length. The * For Part 1 of this series, see reference 1. i Present address: EDT Research, 14 Trading Estate Road, London NWlO 7LU, UK. spectroscopic emission lines of interest were isolated using a 0.5-m Ebert scanning monochromator [Jarrell-Ash (Europe) , Le Locle, Switzerland]. The radiation was then focused via a 25-pm exit slit on to a photomultiplier tube (Hamamatsu R106, operated at 500 V). The signal from the photomultiplier was amplified using a linear picoammeter ( L M 10, Chelsea Instruments, London, UK), and fed into a three-pen poten- tiometric chart recorder (Type MC 641-32, Watanabe Instru- ment , Japan).Certified reference material coals SRM 1632a and 1635 were obtained from the National Bureau of Standards, Washington DC, USA, and a variety of coals were kindly supplied by the National Coal Board. The coal samples were prepared in a micronising mill (grinding time of 30 min) as previously described1 and prepared as slurries (10% mlV) using 1% V/V aqueous Triton X-100 as dispersing agent. Additionally each of the coals analysed by slurry atomisa- tion were analysed by the ASTM ash - digestion procedure5 and atomic absorption spectrometry as described earlier. 1 The coal samples were ashed at 500 "C prior to dissolution in aqua regia and hydrofluoric acid.Saturated boric acid was used to neutralise any excess of hydrofluoric acid. Calibration standards were prepared by serial dilution of commercially prepared analytical standard solutions (1000 pg ml-1, BDH Chemicals, Poole, Dorset, UK). All solutions, except iron standards, were prepared to contain 1000 pg ml-1 of iron as the major matrix metal. Operating conditions selected were based on previous optimisation experience and a knowledge of the contours of the critical responses around the optimum,l but were compro- mise for a range of elements (see Table 1). Results and Discussion Line Selection Owing to the low resolution of the optical system available, problems were immediately encountered from the high levels of iron in the coal samples, which resulted in numerous intense iron emission lines. Many of the analytical lines preferred for the determination of the desired elements in typical aqueous samples were interfered with by these iron emissions. In addition, different coals containing different amounts of iron326 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 were susceptible to these spectral interferences to varying degrees. Consequently, a range of analytical emission lines was investigated for each of the elements of interest. Each line was evaluated in terms of signal to background ratio (SBR) and relative freedom from spectral interferences. Both atom and ion emission lines were examined using the plasma operating conditions given in Table 1. The analytical lines fulfilling these requirements to the greatest extent were selected for the analytical procedures to follow.Vanadium gave better results using ion lines and typical ion line operating conditions. Copper, iron, manganese and nickel gave better results using atom lines and typical atom line conditions. follow a similar order. This trend was not observed absolutely for the coals analysed but the results obtained for copper, nickel and vanadium showed that atomisation efficiencies were generally biased in favour of those slurries containing the highest concentrations of small particles. For copper and vanadium, best agreement was observed between the values determined experimentally (flame atomic absorption and slurry atomisation) and the NBS certificate values for the standard coals analysed.When the non-standard coals were analysed for copper and vanadium the agreement between the values determined using flame atomic absorption spec- Coal Analysis In addition to the coals analysed a portion of the ash obtained from SRM 1632a was used to prepare a slurry (ca. 2% mlV, which, when calculated as a mass of whole coal represented a slurry concentration of 10% mlV). The results obtained for the five coals and the ash are shown in Tables 2-6 together with the apparent “atomisation efficiencies .” These are perhaps better termed recoveries as they represent ICP slurry atornisa- tion values as a percentage of the flame AAS values. Quantitative measurements of peak intensities were taken by making replicate scans across the analytical lines used. The appropriate background contribution was subtracted to yield the net emission signal.After taking into account the relatively small variations in background level the over-all precision (20) for three replicate measurements was estimated to be ca. +12%. The uncertainty of the flame atomic absorption results was typically &5-10%. Results were obtained for all the trace metals examined, namely copper, iron, manganese, nickel and vanadium. The order of “grindability” of the coals (measured in terms of the number of particles smaller than 10 pm) is: SRM 1632a > C > B > A > SRM 1635 (see Table 7). Thus, if atomisation efficiency is simply and most directly affected by particle size, the atomisation efficiencies observed for these coals should Table 1.Preferred operating conditions for observation of atom and ion emission lines in an all-argon plasma Parameter Atom line Ion line Coolant gas flow/l min-1 . . 8.7 5.0 Plasma (intermediate) gas flow/l min-1 . . 14.0 14.0 Injector gas flow/iAin-i* . . 0.58 0.37 Power in the plasmdkW . . 0.50 0.54 Observation heightlmm aboveloadcoil . . . . 26 18 These conditions were used to determine Copper 327.4 nm Vanadium 290.9 nm Iron 372.0 nm Manganese 403.1 nm Nickel 341.8 nm Table 2. Determination of copper; atom line conditions, 327.4 nm. Values in parentheses indicate apparent atomisation efficiencies relative to values obtained using FAAS Concentration of coppedpg g-l ASTM NBS method certificate Sample (FAAS) ICPSA* value SRM1632a . . 17.5+ 1.0 17 f 2 (97) 16.5 5 1.0 SRM 1632a ash .. SRM 1635 . . 4.7 + 0.3 4.6 f 0.5 (98) 3.6 f 0.3 - 18 f 2 (105) - NCBA . . . . 27 21.4 18 + 2 (67) - NCBB . . . . 36 + 2 26 f 3 (72) - NCBC . . . . 52 4 3 53 + 6 (102) - * ICPSA, inductively coupled plasma slurry atomisation. Table 3. Determination of nickel; atom line conditions, 341.5 nm. Values in parentheses indicate apparent atomisation efficiencies relative to values obtained using FAAS Concentration of nickel/pg g-1 ASTM method Sample (FAAS) SRM1632a . . 42+3 SRM 1632a ash . . SRM1635 . . 7 2 0 . 5 NCBA . . . . 3 3 5 2 NCBB , . . . 6 1 5 4 NCBC . . . . 5 6 5 3 - NBS certificate value 39+5(93) 19.4 k1.0 17 f 2 (243) ICPSA 39 f 5 (93) - 38 k 4 (115) - 57 f 7 (93) - 1.74 k 0.10 - 56 + 7 (100) Table 4. Determination of vanadium; ion line conditions, 209.9 nm.Values in parentheses indicate apparent atomisation efficiencies relative to values obtained using FAAS Concentration of vanadiumlpg g-1 ASTM method Sample (FAAS) SRM1632a . . 42.554 SRM 1632a ash . . SRM1635 . . 5.7k0.5 NCBA . . . . 20 + 2 NCBB . . . . 37 + 4 NBCC . . . . 44 + 4 - NBS certificate ICPSA value 51 f 6 (120) 44 5 3 53 k 6 (125) - 20 + 2 (100) - 53 f 6 (143) - 73 .+ 9 (166) - 5.4 f 0.6 (95) 5.2 f 0.5 Table 5. Determination of manganese; atom line conditions, 403.1 nm. Values in parentheses indicate apparent atomisation efficiencies relative to values obtained using FAAS Concentration of manganese& g-1 ASTM method Sample (FAAS) SRM1632a ._. 21.5f 1 SRM1632aash. . - SRM1635 . . 28.35 1.4 NCBA . . . . 34 + 2 NCBB . . . . 73 + 4 NCBC .. . . 194 f 1 0 NBS certificate ICPSA value 42 f 5 (195) 28 4 2 18 + 2 (84) - 48 2 6 (141) - 96 +12(132) - 8.5 fl (30) 21.421.5 - 128 k 15(66) ~ ~~ Table 6. Determination of iron; atom line conditions, 372.0 nm. Values in parentheses indicate apparent atomisation efficiencies relative to values obtained using FAAS Concentration of iron/pg g-1 ASTM method Sample (FAAS) SRM1632a . . 12600f600 SRM 1632a ash . . SRM1635 . . 30705 150 NCBA . . . . 71605360 NCBB . . . . 16300f800 NCBC . . . . 20600+_1000 - NBS certificate ICPSA value 3 830 f 460 (30) 11 100 200 736f 90(25) 2390i750 3 420 f 410 (27) 2 910 k 350 (41) 6 540 k 790 (40) - - - - 7 090 f 850 (34)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 327 Table 7. Correlation between ash and moisture content and the apparent ease of grinding of the five coals analysed Ease of grinding as YO number of Moisture,* Ash,* particles smaller Coal YO Yo than 10 pmt NBS SRM 1632a (bituminouscoal) .. 2.2 22.2 45.1 NCBC.. . . . . . . 3.4 21.6 42.2 NCBB . . . . . . . . 5.9 17.5 39.6 NCBA . . . . . . 9.4 4.6 31.8 NBS SRM 1635 (sub-bituminous coal) 18.8 4.3 28.5 * Moisture and ash contents determined using NCB standard t Particle size distributions evaluated using Coulter Counter. methods, Table 8. Particle size distributions of NBS SRM 1632a and the ash produced from this coal Occurrence, % Upper limit of particle size range/pm 2.52 3.17 4.00 5.04 6.35 8.00 10.08 12.70 16.00 20.16 25.40 32.00 Coal 0.5 2.3 4.2 5.9 10.1 10.9 11.2 11.3 14.4 12.1 10.4 6.7 Total 100 Ash 3.2 7.8 11.0 12.9 14.7 13.4 11.6 9.6 8.2 4.9 1.9 0.8 100 Particles smaller than 10 pm 45.1 74.6 trometry and slurry atomisation was generally less satisfac- tory.The discrepancies were random and no single coal was identified as giving consistently high or low results using the slurry atomisation method. The results for the determination of nickel imply that there has been some contamination and/or segregation of the sample. This is demonstrated by the consistent discrepancy between the values determined experimentally for SRM 1635a (ca. 39 pg g-1 using flame atomic absorption spectrometry and slurry atomisation) and the NBS certified value (21.4 yg g-1). Overall the results obtained for nickel using flame atomic absorption spectrometry and slurry atomisation showed good agreement for all the coals except SRM 1635, which gave consistently high results using the slurry atomisation tech- nique.The results obtained for manganese are disappointing and show poor agreement between the values determined using flame atomic absorption spectrometry and slurry atom- isation. The reasons for the variable recoveries are unclear and any attempts to correlate atomisation efficiency with particle size distribution were meaningless. The low recoveries of iron were attributed initially to the high levels of iron in the atom cell causing self absorption of the signal or saturation of the photomultiplier tube. Both suggestions were discounted when dilutions of one slurry (10- and 200-fold) failed to produce results significantly different (+5%) to the parent slurry.The reasons for the poor atomisation efficiencies observed for iron (relative to other elements) are unclear. One possibility is that the iron is contained in coal as relatively large inclusions of oxides and sulphides. The iron may then be found in particles with significantly greater sizes than the rest of the coal or of greater density, alternatively these mineral particles may prove to be more refractory in the plasma. An additional possibility is that as a magnetic stirrer was used to agitate the coal slurry this may have caused migration of some magnetic iron mineral particles to the bottom of the sample container. During nebulisation of the slurries, droplets of aerosol impinging on the bottom of the spray chamber and partly nebulised slurry were returned to the sample reservoir. It has been suggested by some workers that only particles smaller than 10 pm contribute appreciably to the analyte emission signal.Furthermore, if the particle size selection effect of the spray chamber results in selective transport of very small particles to the atom cell, then the slurry recycled to the sample reservoir would be relatively depleted in these particles. In combination these two effects would be expected to cause some deterioration in signal size as successive portions of slurry nebulised into the plasma would contain progressively fewer smaller particles. Fortunately the absolute mass of slurry reaching the plasma during typical analysis times was very small and any deterioration would have been very small.No significant decrease in signal size was observed, even after prolonged nebulisation of one coal slurry. There is some evidence to suggest that, for the range of particles sizes investigated (2-20 ym), atomisation efficiencies appear to be independent of the particle size distribution of the samples. Other workers discussing these parameters have been concerned largely with the analysis of refractory geolog- ical materials.6 It may be that the small particle sizes reportedly necessary for efficient atomisation of these materials are not necessary for coal. This feature may be related both to the composition of the materials undergoing atomisation and the rates at which the processes occur. Good agreement was observed between the results obtained for the analysis of the SRM 1632a coal slurry and the ash slurry produced from this coal.This is interesting when the particle-size distributions of these two samples are considered. From Table 8 it is seen that the low-ash coals produce fewer small particles on grinding than do high-ash coals. Thus it may be postulated that the relatively large particles left after grinding are largely carbonaceous in composition. Subsequent ashing of the coal removes carbonaceous material so the number of these large particles should be reduced, i. e., the ash should contain a higher percentage of small particles than the coal from which it was produced. Particle-size analysis of the ash from SRM 1632a confirms this in that only 45% of the original coal particles were smaller than 10 pm, whereas 75% of the ash particles were smaller than 10 pm (Table 8).Hence better atomisation efficiencies might be expected for the ash slurry than for the coal slurry. A comparison of the values determined for the two slurries revealed that the manganese and iron values again showed no consistency, both elements in fact giving lower recoveries for the ash slurry (85 and 43% of the coal slurry figure for iron and manganese, respectively). For copper, nickel and vanadium (those elements giving the best results overall) the differences between the ash and coal slurry values were not significant, L5%. Conclusions Slurry atomisation ICP-AES seems to show promise for the determination of trace and minor elements in coal. In particular the results for certified reference material coals were most encouraging, e.g., for SRM 1632a (in pg g-1 with certified values in parentheses) Cu 17 (16.5 _+ 1.0) and V 51 (44 f 3), and for SRM 1635 Cu 4.6 (3.6 k 0.3) and V 5.4 (5.2 k 0.5).The agreement with the results for the ashed and digested coals by atomic absorption spectrometry is poorer. Nevertheless mean recoveries by slurry atomisation compared to the ash - digestion - AAS method were for Cu 70, Fe 33, Mn 108, Ni 123 and V 125%. Evidence suggests that more uniformly ground coal samples should give results with improved accuracy and perhaps improved precision (presently estimated as typically +12% relative, which may well be328 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 acceptable for routine monitoring of trace metals in coal).Certainly the accuracy and precision of results would be improved by the availability of improved instrumentation, particularly of better resolution and hence greater freedom from spectral interferences. In this study iron interferences have been shown to be a particular problem. Work is continuing in our laboratories to find ways of further improving the accuracy and precision of this approach which certainly offers considerable savings in time over existing methodology. We would like to thank the S.E.R.C. and British Gas for the provision of an S.E.R.C. - C.A.S.E. studentship to one of us (J. R. W.), the S.E.R.C. and London Scandinavian Metal- lurgical Company for the provision of instrumentation , the National Coal Board, East Midlands, Scientific Services for the provision of samples and Dr. C. W. Fuller, CEGB, Scientific Services, Radcliffe-on-Soar, Nottingham for the provision of milling facilities. References 1. Ebdon, L., and Wilkinson, J. R., J. Anal. At. Spectrom., 1987, 2 , 39. 2. Wilkinson, J. R., Ebdon, L., and Jackson, K. W., Anal. Proc. , 1982, 19, 305. 3. Ebdon, L., Mowthorpe, D. J., and Cave, M. R., Anal. Chim. Acta, 1980, 115, 171. 4. Ebdon, L., and Cave, M. R., Analyst, 1982, 107, 172. 5. Method ANSUASTM D3683-78, American Society for Testing Materials, 1978, Philadelphia, PA, USA. 6. Fuller, C. W., Hutton, R. C., and Preston, B., Analyst, 1981, 106,913. Note-Reference 1 is to Part 1 of this series; for Part 2 see J. Anal. At. Spectrom., 1987,2, 131. Paper J6l.54 Received July I8th, I986 Accepted December 1 Oth, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200325

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 329 Preparation of Solutions for the Standard Additions Method in Flame Atomic Absorption Spectrometry Paw& Koscielniak Department of Analytical Chemistry, Jagiellonian University, ul. Karasia 3, 30-060 Krakow, Poland An apparatus for the rapid determination of analytes by the standard additions method in flame AAS is described. For the analysis of one sample the preliminary preparation of only two solutions is necessary; subsequent working solutions were prepared by stepwise changes in the concentration of one of the original two solutions in a small glass flask. By this means the analytical procedure becomes simple and rapid and requires only a small volume of sample. Examples of the determination of Ca and Zn in the presence of selected interfering elements in synthetic samples and the determination of Zn in municipal water are presented.Optimisation of the experimental conditions and both linear and non-linear approximations of the results are discussed. Keywords: Automatic calibration; flame atomic absorption spectrometry; standard additions method In a previous paper1 a simple apparatus was presented for the rapid preparation of solutions in flame atomic absorption spectrometry (Fig. 1). The method involved is briefly as follows. The glass flask A is filled to the mark with the initial solution of volume V. The aspiration capillary from the spectrometer is introduced into the flask, a volume (V - v) of the solution is aspirated into the flame and the analytical signal is detected.A piece of plastic tubing is placed on the end of the aspiration capillary tube to ensure the constant, reproducible and deep immersion of this capillary tube in the solution. After aspiration, the capillary is removed and flask A is filled again with solution from flask B. The new solution, of a different concentration, generated in flask A by this means is mixed for a few seconds. At the same time, measurement of the blank (for example, water) is carried out. The capillary tube is then introduced into flask A, the new signal is detected, etc. The succeeding processes of filling flask A, mixing and aspirating create solutions of compositions that change exponentially according to the following equation: c,= (co- c,)krn+c, . . . . . . (1) where k, = v/V and co, c, and c, are the concentrations of components in the initial solution in flask A, in the solution in flasFB and in the solution prepared after the nth dosage, respectively.To the nebuliser f 1 1 I syringe B Fig. 1. Apparatus for the preparation of solutions for flame atomic absorption spectrometry Following the above procedure, synthetic standard solu- tions were prepared for the preparation of calibration graphs for Fe, Ca and Mg and also for the examination of the interfering effect of A1 in the determination of Ca and Mg.1 In this paper, the application of the above technique to the AAS determination of Ca and Zn in synthetic and natural samples by the standard additions method is described. Theoretical Procedure In accordance with the standard additions method, the initial solution in flask A is an analysed sample with an analyte concentration c,, and the solution in flask B is a mixture of the sample and a solution of analyte of known concentration (standard addition) c d .Hence, co = c, and c, = c, + c d . The concentration of matrix components is the same in flasks A and B, and is therefore constant throughout the experiment. It follows from equation (1) that the composition of the solutions generated successively in flask A changes in this instance according to the equation c n = C x + c d ( l - k r " ) . . . ( 2 ) Using the solutions of composition co (sample) and c l , . . . , cN (the generated solutions), measurements of the analytical signals Ro, R1, . . . , RN are performed. Moreover, the signal R, for the solution in flask B should also be measured.A scheme for the presentation of the results obtained on the basis of these solutions is shown in Fig. 2. Basic Model It can be assumed in AAS that the dependence of the measured signal R on the concentrations of the analyte, c, (in this instance c, = c, + c d ) , and matrix, C M , is most often given by a linear model: l? = blc, + bZc,cM . . . . * * (3) where bl and b2 are the regression coefficients. Taking into consideration the exponential increase in the concentration c d [see equation ( 2 ) ] , the model (3) may be expressed in the following form: R, = bl [C, + c d (1 - krn)] + b 2 [C, + c d (1 - krn)]CM (4a) or = (b, + b2Ch.1) ( C x + c d ) i- [- ( b 1 + b2CM)Cd]krn (4b)330 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 4 - 3 - LL 2 - L Lt I r CX cd Cn Fig. 2. Presentation of results obtained by the standard additions method using analyte solutions of exponentially varied concentrat- ions cd Therefore, the experimental relationship between R, and krn may be approximated by the linear function where the coefficients B1 and B2 are obtained, for example, according to the least-squares method. The practical sense of these Toefficients results from the following condjtions: for n = 0, R, = B1+ 232 = Ro, andforn = 00 (kr" =O),R, = B1 " R , (see Fig. 2). Coefficients B1 and B2 are calculated on the basis of the experimental results. The concentration c, of analyte in a sample is then calculated using the equation f i = ~ 1 + ~ 2 k r n , .. . . . (5) which follows from comparison of equations (4b) and (5). Optimisation of the Experimental Conditions The random error in the calculation of the concentration c, may be found by use of the error propagation method: Relationships between the standard deviations of the re- gression coefficients B1 and B2 in the linear model and a residual variance si2 are given by well known equations.2 Hence, 1 X- X S ~ . . . . (8) (-2 + ( p + 1)2 acx = J N (k,n)2 + Z (krn - - krn)' S f l n = O N N n = O n = O where p = cx/cd, (k,n)2 = Z (krn)2/(N + 2 ) , k,. = Z k,n/(N + 2 ) and sjp = Z { 5 [(R,), - f i n ] 2 + [(Rs),-a,]2]/[(N + 2)M - 21 r n = l n = O The parameter S = -B2/cd is an estimate of the determination sensitivity; N and M denote the number of dosages and the number of readings of the signal R for one solution, respectively.Non-linear Approximation The model (5) can be expressed in the following most general form: 'L 1 2 3 4 5 6 1 I I I I 1lP Fig. 3. Dependence of the ex ression F see equation (8)] on the ratio Cdk, and on the number ofdosages N IN = 0 indicates that only two solutions (sample and sample + standard addition) are taken into account in the standard additions method] which gives a non-linear approximation for the relationship between the signal R and the concentration of analyte. In this instance the parameter ka (different from kr) should be additionally calculated together with B1 and B2 using an optimisation procedure (e.g., the simplex method). Appro- priate combination and transformation of equations (4b) and (9) leads to an equation for the calculation of the concentra- tion c, of analyte in a sample: Experimental Stock standard solutions of Ca, Zn, Fe, Cu and Mg containing each metal at a concentration of 1 mg ml-1 were prepared by the use of commercial standards (Titrisol, Merck).Working solutions were obtained by dilution of the stock solutions. The measurements were carried out with an AAS-1 flame spectrometer (Carl Zeiss, Jena, GDR) under standard condi- tions for each analyte. An air - acetylene flame was used, with a constant aspiration rate of 3.5 ml min-1 maintained during the measurements. The absorbance was recorded ten times for each solution, averaged and the mean absorbance was taken as the signal R. Volumes of V = 10.149 ml and v = 6.817 ml for flask A (Fig.l), with standard deviations of 0.921 x 10-2 and 1.012 x 10-2 ml in V and v, respectively, were determined experimentally by ten-fold weighing of the flask with the appropriate portions of distilled water. The solution added to this flask was transported from a calibrated flask (of volume 50 ml) through polyethylene tubing (the required pressure was applied with the aid of a syringe). Results Taking into account equation (9, the dependence of the expression F = ,/[(kriy+ (p + 1)2]/[(kr">2 + (krn - m2] (11) n = O for kr = 0.672 on llp = cdk, and N has been determined in order to select the optimal experimental conditions. The results are presented in Fig. 3. On the basis of these results, it was decided both that the ratio of the concentrations c&, should be about 4 and that five dosages of the solution fromJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 331 Table 1.Results of the determination of Ca and Zn obtained by the conventional and proposed standard additions methods Concentration of analyte/pg ml-1 Analyte Ca . . . . . . Zn . . . . . . Accompanying element and concentration/ pg ml-1 Ti, 0 Ti, 5 Ti, 20 Ti, 50 Ti, 100 Fe, 100 c u , 100 Ca, 100 Mg, 100 Complex matrix Complex matrix Complex matrix Complex matrix Obtained by conventional True value method 5 5.07 5 4.93 5 4.95 5 5.05 5 5.01 0.5 0.5 0.5 0.5 ? ? ? ? 0.509 0.485 0.485 0.492 0.853 0.682 0.735 0.512 Obtained by proposed method 5.08 5.01 4.98 4.98 4.91 0.532 0.512 0.491 0.498 0.792 0.653 0.760 0.525 r- 0.4 1 0.3 CM y o 9” 5 CO c1 CZ c3 c4 c5 L w h “ 1 c, = 5 pg ml-1 cd = 20 pg m1-l C” Fig.4. Determination of 5 pg ml-1 of Ca in the resence of Ti accordin to the proposed (x) and conventional (&) techniques. Model (8 is represented by the full lines. k, = 0.672. cM = 0,5,20,50 and 100 pg ml-1 flask B into flask A should be performed during the analysis of each sample. First the proposed method was applied to the determination of Ca in the presence of Ti. For this, five synthetic samples of composition c, = 5 pg ml-1 of Ca and CM = 0,5,20,50 and 100 pg ml-1 of Ti, respectively, were prepared. The concentration of Ca in flask B was cd = 20 pg m1-I. In Fig. 4 the experimental results obtained are indicated by crosses; the model (5) fitted to these points (for k, = 0.672) is represented by full lines.The determination of Ca in the above samples was also carried out according to the conventional standard additions method (using for each sample seven solutions in separate calibrated flasks). The results obtained are represented in Fig. 4 by full points. The next experiments concerned the determination of Zn in (a) four synthetic samples containing Zn of concentration c, = 0.5 pg ml-1 and Fe, Cu, Ca and Mg of the same concentrations CM = 100 pg ml-1, and (b) four samples of municipal water. The concentration cd was 2 pg ml-I of Zn. In each instance the measurements revealed a non-linear dependence of absor- bance on the concentration of Zn. The results of the determination of Zn in the presence of Fe are shown in Fig. 5 .In this instance the best fit of the mod51 (9) to the experimental results was found for k, = 0.623, R = 0.3306 - 0.2496 X 0.623,. This is seen by the linear distribution of the experimen- tal points (full points) in the co-ordinates R and c, for k,. In the “real” co-ordinates R and c, for k,, the non-linear (istribution of these points (crosses) is described by the model R = 0.3306 - 0.2496 x 0.927, x 0.672,. The determination of Zn in the samples investigated were also performed by the conventional standard additions method. In this instance for each sample seven separated solutions were prepared; the maximum addition of Zn to these solutions was 2 pg ml-1. The calculated results for the determination of Ca and Zn in all samples by both the conventional and proposed standard additions methods are presented in Table 1.Discussion In the above experiments five solutions were generated from the initial solution during the analysis of one sample ( N = 5). This is seen to be the optimum number by taking into account on the one hand the determination random error [see equation (7) and Fig. 31 and on the other hand the accuracy of the preparation of the solutions in the proposed method. It is worth emphasising that the errors in the fixing of the volumes and consequently of the compositions of the generated solutions in flask A are accumulated.’ Therefore, with too many dosages the determinations may be significantly influen- ced by these errors. Random error analysis also indicated that (a) the concentration cd should be about 3-4 times greater than c, (an increase in the ratio Cd/C, above this value does not substantially improve the precision of the determination) and (b) the error depends strongly on the fitting of the linear function to the experimental points (s;), on the number of absorbance readings for one solution ( M ) and on the sensitivity of the determinations (S).The suggested method allows the simple preparation of working solutions in the standard additions method. The preliminary preparation of only two solutions (sample and sample + addition cd) is necessary; subsequent working solutions are made by stepwise changes in the concentration of one of these initial solutions. Therefore, the procedure is very rapid. For example, the analysis of one sample investigated in this paper (including the preparation of seven solutions, measurements and calculations) required only about 20 min.Moreover, relatively small volumes of the sample are con- sumed. In contrast, for the preparation of seven solutions according to the conventional standard additions method in332 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 0.3 . / co Cl c2 q c4 cs c, Cl c2 G3 c4c5 C, = 2 pg ml-’ C n (for k,) C n (for k.1 I I I I I I , 1 I I d 1 c, = 0.532 pg ml-1 Fig. 5. Determination of 0.5 pg ml-1 of Zn in the presence of Fe using the co-ordinates R and c, for k, (e) and R and c, for k, (X); k, = 0.672 and k, = 0.623. Line A, R = 0.3306 - 0.2496 X ( - :::;: ) x 0.672n Line B, = 0.3306 - 0.2496 X 0.623. 50-ml calibrated flasks a sample volume about 10 times greater than that in the proposed procedure is needed.The above advantages are important from the practical point of view. The accuracy of the determination of Ca and Zn by the proposed and conventional methods are comparable. This can be seen especially for the determination of Ca using the linear approximation. A slightly greater error in the determination of Zn in both methods results from the fitting of the non-linear model (9) to the experimental points. In general, this model should be applied very carefully, especially when considerable non-linearity exists in the experimental points. In such instances, the selection of the appropriate non-linear function becomes very important.’ The proposed apparatus for the preparation of the solutions may be automated and controlled by a microcomputer and could be successfully applied to routine analyses in flame AAS and AES. The author thanks Doc. dr. hab. A. Parczewski for helpful discussions during the preparation of this paper. The investi- gations were supported financially within the scope of project CPBP No. 01.17. References 1. KoScielniak, P., Analyst, 1986, 111, 991. 2. Czerminski, J . , Iwasiewicz, A . , Paszek, Z . , and Sikorski, A., “Metody Statystyczne w Doiwiadczalnictwie Chemicznym,” PWN, Warsaw, 1970, p. 163. 3. KoScielniak, P., and Parczewski, A . , Fresenius 2. Anal. Chem., 1985,321, 572. Paper J6l.50 Received July 7th, 1986 Accepted November loth, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200329

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 333 Removal of Phosphate and Silicate Interferences in the Determination of Magnesium, Calcium and Strontium by Atomic Absorption Spectrometry M. M. El-Defrawy, M. E. Khalifa, A. M. Abdallah" and M. A. Akl Department of Chemistry, Faculty of Science, University of Mansoura, PO Box 30, Mansoura, Egypt A number of ions affect the absorbance of magnesium, calcium and strontium in a fuel-rich air - acetylene flame and this effect is removed by the presence of catechol or pyrogallol. An investigation of the removal of the interfering effect of silicate or phosphate using the continuous titration technique revealed a stoicheiometric relationship due to the formation of anionic complexes between pyrogallol and silicate or phosphate, i.e., 4 : 1 and 3 : 1, respectively.This stoicheiornetry has been confirmed by spectrophotometric measurements. Keywords: Atomic absorption spectrometry; removal of phosphate and silicate interferences; magnesium, calcium and strontium determination; p yrogallol In recent years analyses by atomic absorption spectrometry (AAS) have led investigations into various interferent sup- pressors,l-9 such as sulphosalicylic and 4-aminosalicylic acids.68 The aim of this paper is to study the applicability of two other phenolic compounds, i. e., catechol and pyrogallol, as suppressors for the interfering effects of foreign species on the atomic absorption signals of magnesium, calcium and strontium in a fuel-rich air - acetylene flame. Experimental A Unicam SP-90A Series 2 atomic absorption spectrometer was used with Pye Unicam magnesium, calcium and strontium hollow-cathode lamps and a conventional 10-cm slot burner head for an air - acetylene flame.A continuous titration devicelo was attached to the instrument; absorbance values were recorded with a single-pen recorder at a chart speed of 30 cm min-1. The evaluation of the titration plots was carried out on a programmable pocket calculator. The instrumental parameters used are given in Table 1. The emission studies on the molecular MgO and HPO band systems were also performed with the same instrument. The molecular absorption spectra were measured on a Jobin-Yvon UV - visible double-beam spectrophotometer, using 1-cm quartz cuvettes. All chemicals were of analytical-reagent grade.The water used was usually freshly de-ionised, doubly distilled. Results and Discussion Tables 2-4 show the interfering effects of foreign species on the atomic absorption signals of magnesium, calcium and Table 1. Instrumental conditions for the determination of magnesium, calcium and strontium by AAS. Monochromator dispersions are 1.6 and 5.4 nm mm-1 at 310 and 450 nm, respectively Parameter Magnesium Calcium Strontium Lampcurrent/mA . . 5 10 12 Wavelengthhm . . . . 285.2 422.7 460.7 Slitwidth/mm . . . . 0.08 0.1 0.1 Observation heightkm 1.0 1 .o 1.0 Air flow-rate/l min-1 . . 5.0 5.0 5.0 Acetylene flow-rate/ 1min-l . . . . . . 1.5 1.5 1.5 * To whom correspondence should be addressed. strontium, respectively, in a fuel-rich air - acetylene flame.The normalising action of catechol or pyrogallol on the absorption signals of the analytes can be seen clearly. Pyrogallol is more effective than catechol in normalising the absorbances of the three analytes. The main features of the data that distinguish pyrogallol and catechol from other already established releasing agents, such as lanthanum and strontium chlorides, are that they eliminate mutual interfering effects of the alkaline earth elements and control the enhancement effect by some cations, e.g., Cd2+, Co2+ and In3+, on the analytes' absorption signals because catechol or pyrogallol prevents the formation of a eutectic mixture11 with the oxides of these metals. The depressive effect of phosphate or silicate due to the formation of involatile compounds12 can be overcome by the addition of catechol or pyrogallol.The effects of both AP+ and aluminate, [Al(OH),]-, species on magnesium absorbance are the same, resulting in the formation of magnesium aluminate. Table 5 indicates that the depressive effect of aluminium is greater in the presence of sodium due to the increased background of the latter13J4 in the neutral and alkaline solutions. However, the addition of Table 2. Interference of various added species (200 pg ml-1) on the absorbance of 5 pg ml-1 of magnesium and the effect of catechol or pyrogallol Added substance Li+ . . . . . . Sr2+ . . . . . . Ba2+. . . . . . GI2+. . . . . . Cd2+ . . . . Ni2+ . . co2+ Sn2+ . . La3+ . . Ce3+. . Fe3+ . . S042- Si032- I++ . . ~ 0 ~ 3 - . . . . . . . . . . . . . . .. . . . . . . . . * . . . . . . . . . . . . . . . B4072- . . . . HC02- . . . . C4H4062- (tartrate) PdC1,Z- . . . . Magnesium recovery, % . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Without treatment 140 140 135 122 143 113 130 120 130 118 155 125 142 19 30 86 158 121 115 With 1 M catechol 98 98 98 103 98 100 93 98 89 95 100 93 100 91 91 100 135 100 119 With 1 M p yrogallol 100 106 106 95 106 100 100 109 100 93 105 106 100 100 92 98 100 100 100334 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 Table 3. Interference of various added species (200 pg ml-1) on the absorbance of 10 pg ml-1 of calcium and the effect of catechol or pyrogallol Added substance Ba2+ . . . . . . . . sr2+ . . . . . . . . Hg2+ . . . . . . co2+ Cd2+ .. . . . . . . . . . . Sn2+ . . Ce3+ . . Fe3+ . . 1n3+ . . . . . . . . . . . . 9 . . . . . . . . . . . . . Br- . . . . . . . . N03- . . . . . . N02- . . . . . . SO4*- . . . . . . S20& (peroxy- disulphate) . . . . - . 103- . . . . . . . . po43- . . . . . . SQ2- . . . . . . B4W- . . . . . . W042- . . . . C4H4042- (siccinate) c4H4062- (tartrate) . . HC6H5072- (citrate) . . PdC142- . , . , . . ptcl62- . . . . . . Calcium recovery, Yo Without treatment 121 123 118 123 125 115 136 123 159 70 65 75 78 65 65 15 20 75 10 78 60 75 114 159 With 1 M catechol 109 106 106 106 102 112 100 100 106 100 100 100 82 75 100 94 100 108 90 100 75 100 106 106 With 1 M pyrogallol 100 100 100 100 100 102 100 100 100 100 100 100 96 98 100 100 100 100 100 100 100 100 100 100 Table 4.Interference of various added species (200 pg ml-l) on the absorbance of 10 pg ml-1 of strontium and the effect of catechol or pyrogallol Strontium rcovery, Yo Added substance Mg2+ . . . . . . Ba2+ . . . . . . . . Cd2+ Ni2+ . . co2+ La3+ . . Ce3+ . . Fe3+ . . Cr3+ . . NO3- S042- 1n3+ . . N02- SzOg2- 103- . . Si032- w o p HC02- P043- PtC16- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ Without treatment 89 127 124 113 120 120 122 118 76 76 150 128 151 128 128 10 10 39 82 122 With 1 M catechol 100 102 102 109 113 106 98 109 102 102 100 100 148 100 100 93 93 100 100 92 With 1 M p yrogallol 100 100 100 100 100 100 100 100 100 100 100 100 102 98 100 100 100 100 100 100 pyrogallol removes the depressive effect of both sodium background and the aluminium involatile compounds when the solution is made alkaline.Action of Pyrogallol The following equilibria are the basis of the discussion to follow: MX + 2L * LX(Sci1) + ML(SO1) + M(g) + 2L(,) + X(g) where M = Mg, Ca or Sr, L = pyrogallol, X = phosphate or silicate, This equation indicates that pyrogallol co-ordinates 100 75 $? 25 I a b 4 d, I I 1 I 1 I 1 0 10 20 30 40 50 60 70 80 Pyrogallol concentration/mM Fig. 1. Effect of pyrogallol on recovery (%) of 5 pg ml-1 Mg2+, 10 pg ml-l Ca2+ and Sr2+ from A, 6.9 mM and B, 17.9 mM sodium silicate matrices. Absorbance readings of Mg, Ca or Sr alone are assumed to be 100; all other readings are referred to this value.For a and b see text 8 75 25 I I I I I I I I I I I I 0 4 8 12 16 20 24 28 32 36 40 44 Pyrogallol concentration/mM Fig. 2. Effect of pyrogallol on the recovery (YO) of 10 pg ml-1 of A, Ca2+ and B, Sr2+ from 8.1 m M phosphoric acid matrix. For a and b see text with both the analyte (M) and anion (X) in the condensed phase. Fig, 1 illustrates the effect of pyrogallol on the atomic absorption signals of magnesium, calcium and str ntium in the presence of sodium silicate. Pyrogallol plays an ffective role in releasing the analytes from the interfering sil’cate. At the pyrogallol to silicate ratio of 4: 1 (points a and b, which represent this ratio at different silicate concentrations), the normal absorption signals of the analytes are obtained.In the same context, Fig. 2 shows that pyrogallol normalises the analyte calcium or strontium atom population in the flame in the presence of phosphoric acid at a pyrogallol to phosphate ratio of 3 : 1. The effect of pyrogallol on releasing magnesium from the phosphoric acid matrix is shown in Fig. 3. The shapes of the graphs A, B and C of Fig. 3 are somewhat peculiar. Repeating the experiments five times over a number of weeks gave a reproducibility of ca. 10% for the results. It is assumed that this phenomenon is due to the variety of compoundsls formed between magnesium and electron donating species, such as OH- and phosphates, generated from the disintegration products of pyrogallol and phosphoric acid. The flat regions at the beginning of the titration graphs are related to the complexing action of pyrogallol on the excess of phosphate present in the aspirated solution.This is followed by a marked increase in the absorbance of magnesium up to the point of its normal absorption. It is relevant to point out that the difference in magnesium absorbance at the beginning of the graphs (points a, b and c) and the maxima at 8, 6 and S. is ca. 0.19 absorbance unit, i.e., an increase in magnesium atom population of 260% is obtained. However, the absorbance does not level out on continually titrating pyrogallol as is so for calcium and strontium (Fig. 2). This is due to the greater range of compounds formed by magnesium than calcium or stron- tium. Magnesium atoms, being less electropositive, are more likely to react with the species produced at high pyrogallol BJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 335 Table 5. Interference of sodium and/or aluminium on the absorbance of 5 pg ml-1 of magnesium, and the effect of pyrogallol at different pH values Without pyrogallol With 1 M pyrogallol Added substance/ pg ml-1 PH 1000Na+(asNaCl) . . . . 6.5 3000Na+(asNaCl) . . . . 6.5 5000Na+(asNaCl) . . . . 6.5 7000Na+(asNaCl) . . . . 6.5 7000Na+(asNaOH) . . . . 12.2 200 Al3+ (as AlC13) . . . . 4.7 200~13+ + lOOONa+(asNaCl) . . . . 4.7 200~13+ + 3000Na+(asNaCl) . . . . 4.7 200~13+ + 5000Na+(asNaCl) . . . . 4.7 200~13+ + 7000Na+(asNaCl) . . . . 4.7 200~13+ + 7000Na+(asNaOH) . . , . 11.9 Magnesium recovery, YO 83 78 76 71 78 80 48 47 43 41 58 PH 4.6 4.6 4.6 4.6 8.2 3.9 3.5 3.5 3.6 3.6 9.1 Magnesium recovery, YO 101 101 101 109 102 94 102 91 82 75 102 C L I 1 I 0 25 50 75 1 Pyrogallol concentration/mM Fig. 3.Effect of pyrogallol on absorbance of 5 pg ml-1 Mg2+ in the presence of: A, 11.3; B, 14.5; and C, 16.15 mM phosphoric acid matrix. 1, Signal from 5 pg ml-1 Mg2+ in stoicheiometric flame condition (air flow-rate 5 1 min-1; acetylene flow-rate 1 1 min-1); 2, signal from 5 pg ml-1 Mg2+ in fuel-lean conditions (air flow-rate 5 1 min-1; acetylene flow-rate 0.5-0.6 1 min-1). For a-b and f i 4 , see text concentrations, for example, by reacting with the excess of OH produced from the disintegration of excess pyrogallol in the flame, to form the stable MgO.16 Mg + OH+ MgO + H More interesting is the reaction between H and PO under such flame conditions to give HPO17 and the depressive effect which magnesium atoms exhibit upon the HPO emission intensity.18 Fig. 4 illustrates the effect of increasing the concentration of pyrogallol on the emission intensities of both the complex molecular band, mainly from MgO,19 at 500.7 nm (Fig. 4, graph A) and the HPO band system20 at 511 nm (Fig. 4, graph B). Fig. 4, graph C indicates that the introduction of magnesium (chloride) to the phosphoric acid and pyrogallol solution reduces the HPO emission intensity. However, although magnesium atoms reduce HPO emission when they 100 80 - m & 60 .- v) C 0 v) .- .- E u40 20 I I 1 I 1 1 I 0 20 40 60 80 100 120 140 160 Pyrogallol concentration/mM Fig. 4. Effect of pyrogallol on emission intensity of molecular MgO and HPO band systems: A, emission of MgO at 500.7 nm from 5 pg ml-.' M62+; B, emission of HPO at 511 nm from 14.5 mM phosphonc acid; and C, emission of HPO at 511 nm from 14.5 mM phosphoric acid in the presence of 5 pg ml-1 Mg2+ are oxidised to MgO by the OH species from the degradation of excess of pyrogallol, the pyrogallol also exhibits a releasing effect by breaking magnesium phosphate bonds to release magnesium atoms with an accompanying increase in the magnesium absorption.The latter effect appears to be much greater than the former or else the releasing effect to give 100% magnesium recovery would not be obtained. Fig. 5 iqdicates that the mole ratio [pyrogallol]/[phosphate], at the three inflection points a, b and cis the same, i.e., 3 : 1. In other words, the restoration of the absorbance of the analyte magnesium to the reference value is essentially occuring at the same proportionality of [pyrogallol]/[phosphate] . Fig. 6 shows the flame profiles of 5 pg ml-1 of magnesium in different matrices. The graphs confirm the ability of pyrogallol to improve the atomic absorption signal of magnesium in all portions of the flame, giving signals as if there were no interference present.336 I I I I I JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 0.30 - 0) C m e g 0.20 2 - 0.10 x vl Fig. 5. Magnesium absorbance from 5 pg ml-1 Mg2+ as a function of [pyrogallol] to [phosphate] ratio: A, 11.3; B, 14.5; and C, 16.15 mM phosphoric acid matrix 0 !/ C I I I I Fig. 6. Distribution of magnesium atoms as a function of observation height, from different Mg2+ matrices (each solution 5 pg ml-1 Mg2+): A, magnesium chloride; B, magnesium chloride in 32.3 mM phos- phoric acid matrix; C, magnesium chloride in 35.7 mM sodium silicate matrix; D, magnesium chloride in 1 M pyrogallol; E, magnesium chloride (A) plus 1 M pyro allol; and F, magnesium chloride in 35.7 mM sodium silicate matrix $C) plus 1 M pyrogallol 0 0.2 Observation heightkm Spectrophotometric Measurements The UV spectra for the pyrogallol solution were obtained by mixing 2 ml of fresh 2 mM aqueous pyrogallol solution and 2 ml of borate buffer solution of pH 11 (4 g Na2B407. 10H20 + 10 g NaOH in 1 1 of water) and diluting with doubly distilled water to 10 ml.The solution was kept in the dark for 3 h.The spectra were measured and the maximum absorbances were found at 271 and 335 nm. The phosphoric acid or sodium silicate buffered solution showed no characteristic absorbance in the 170400 nm region of the spectra. The blank solution was borate buffer alone. The spectra of pyrogallol plus phosphate or silicate in the borate buffer were determined by mixing the same volumes of pyrogallol and borate with a series of phosphate or silicate solutions (1-6 pg ml-1 of Po43- or SiO32-) in two sets of 10-ml calibrated flasks. The absorbance was measured using buffered pyrogallol solution as a blank to P) 6 0.3 . e :: a 0.2 ' 0.1 ' 0 ,. /-+. 1 b. - 270 290 310 330 350 370 390 410 430 450 Wavelengt hln m Fig. 7. Absorption spectra of A, 0.4 mM pyrogallol solution; B, solution in A lus 4 pg ml-1 phosphoric acid; and C solution in A plus 4 pg m~-1 soctum silicate 0.9 0.8 0.7 0.6 ? 0.5 0) -e 0.4 Q 0.3 0.2 0.1 0 270 280 290 300 310 Wavelengthhm Fig.8. Absorption spectra of: A, 0.4 mM pyrogallol plus 5 pg ml-l Mg2+ solution at room temperature; B, solution in A plus 4 p~ ml-l phosphoric acid after mixing; C, solution in B heated to ca. 80 C and cooled; and D, solution in B heated to boiling and cooled to room temperature avoid any oxygen interference. Fig. 7 indicates the absorption spectra of the compounds formed between pyrogallol - phosphate or pyrogallol - silicate after 3 h of preparation. The maximum absorbances are at 282 and 362 nm, respectively. The mole ratios of pyrogallol to phosphate or silicate in the solution were found to be 3 : 1 and 4: 1, respectively. These ratios were determined by plotting the absorbance at h,,,, versus the volume of pyrogallol added; at the point where the extrapolated lines meet, all of the phosphate or silicate has been complexed by the added pyrogallol.The number of millimoles of pyrogallol needed to complex all the phosphate or silicate may be read off on the abscissa of the graph; the amount of phosphate or silicate present is also known. From these amounts, the mole ratios can be calculated.21 The spectra of pyrogallol - silicate buffered mixtures in the presence of magnesium, calcium or strontium (chloride) solution at metal ion concentrations of 1-6 pg ml-1 showed the same maximum absorbance as the pyrogallol - silicate buffered solution, i.e., at 362 nm, but the magnitude of theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 337 absorbances was increased and the bands became sharper. Similar experiments were conducted in the presence of phosphate instead of silicate. Fig. 8 shows the absorption spectra of: A, pyrogallol and magnesium; B, pyrogallol, magnesium and phosphate directly after the preparation; C and B heated to ca. 80 “C and cooled; and D and B heated to boiling and cooled to room temperature. All solutions were buffered by the borate buffer. It can be seen that the rate of formation of the complex increases as the temperature increases. Fortunately, the reaction of pyrogallol with oxygen, from the ambient air and that which may be present in the investigated solution, only takes place in the highly alkaline medium of ca.9 M KOH solution and the absorbance is in the visible region of the spectrum at 450 k 20 nm.22723 1. 2. 3. 4. 5. 6. References El-Defrawy, M. M., Posta, J., and Beck, M. T., Anal. Chim. Acta, 1980, 115, 155. Kodama, M., and Miyagawa, S., Anal. Chem., 1980,52,2358. Abdallah, A. M., and Mostafa, M. A., Ann. Chim. (Rome), 1980, 70, 1. Mostyn, R. A., and Cunningham, A. F., Anal. Chem., 1966, 38,121. El-Defrawy, M. M., Abdallah, A. M., and El-Asmy, A. F., Anal. Chim. Acta, 1985, 174, 343. Abdallah, A. M., El-Defrawy, M. M., Mostafa, M. A,, and Sakla, A. B., Talanta, 1985, 32, 19. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Abdallah, A. M., El-Defrawy, M. M., Mostafa, M. A., and Sakla, A. B., Anal. Chim. Acta, 1985, 174, 347. Abdallah, A. M., El-Defrawy, M. M., and Mostafa, M. A., Anal. Chim. Acta, 1984, 165, 105. Van Loon, J. C., “Analytical Atomic Absorption Spectro- scopy, Selected Methods,” Academic Press, New York, 1980. Posta, J., and Lakatos, J., Magy. Kem. Foly., 1980, 86, 284. Halls, D. J., and Townshend, A., Anal. Chim. Acta, 1966, 36, 278. Dinnin, I. J., Anal. Chem., 1960, 32, 1475. Allan, J. E., Analyst, 1958, 83, 466. Phifer, L. H., Anal. Chern., 1957, 29, 1528. Durrant, B. J., and Durrant, B., “Introduction to Advanced Inorganic Chemistry,” Second Edition, ELBS and Longmans, London, 1977. Kirkbright, G. F. , and Sargent, M., “Atomic Absorption and Fluorescence Spectroscopy,” Academic Press, London, 1974. Thanh, M. Lam, and Peyron, M., J . Chem. Phys., 1964, 61, 1531. Kerber, J. D., Barnett, W. B., and Khan, H. L., At. Absorpt. Newsl., 1970,9,39. Gaydon, A. G., “The Spectroscopy of Flames,” Second Edition, John Wiley, New York, 1974. Syty, A., and Dean, J. A., Appl. Opt., 1968, 7, 1331. Brewer, S . , “Solving Problems in Analytical Chemistry,” John Wiley, New York, 1980. Williams, D. D., Blachy, C. H., and Miller, R. R., Anal. Chem., 1952, 24, 1819. Duncan, I. A., Harriman, A., and Porter, G., Anal. Chem., 1979, 51, 2206. Paper J6l99 Received October 24th, 1986 Accepted December 3rd, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200333

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 339 Determination of Sulphur Compounds by Fully Automated Molecular Emission Cavity Analysis Nikolaos P. Evmiridis" and Alan Townshend Chemistry Department/ University of Hull, Hull HU6 7RX, UK The performance of an automated molecular emission cavity analysis (MECA) instrument for the determination of sulphur compounds was investigated. Thiourea, sulphuric acid, ammonium thiocyanate and promethazine were chosen as model compounds. The control of the time required for the various sectors of the MECA operation (time spent in flame, cooling time before sample injection and solvent pre-evaporation time) allows conditions of greatest sensitivity or reproducibility to be established. If complete solvent evaporation is possible before cavity introduction into the flame, sensitivity and calibration linearity are greatly improved.The automated instrument also allows an assessment to be made of the simultaneous determination of thiocyanate and sulphate. Keywords: Molecular emission cavity analysis; hydrogen-based flame; sulphur compounds; automated device Molecular emission cavity analysis (MECA) is a well known flame spectroscopic technique used mainly for the determina- tion of non-metals and metalloids, especially sulphur and phosphorus.1 The technique normally involves the manual injection of a few pl of sample solution into the MECA cavity followed by manual insertion of the cavity into a pre-deter- mined position in a hydrogen-based flame. The operation must be precise, reproducible and sufficiently fast for emission to begin only when the cavity has reached its pre-determined position.These manual operations require operator skill and can be a major cause of imprecision. A completely automated instrument has recently been described2 which eliminates the manual injection and manual cavity insertion stages. It involves automatic sample dispens- ing into the cavity and mechanical insertion of the cavity into the flame. Both operations are microprocessor controlled. The instrument was used for the determination of phosphorus 0x0-anions, based on the generation of HPO emission in the cavity. The automated procedure had a relative standard deviation (RSD) of 0.9% compared with 4.5% for the manual procedure. For phosphate, the detection limit was 2.5 ng of phosphorus.Molecular emission cavity analysis is more sensitive for sulphur compounds than phosphorus compounds,1 and con- ventional MECA has been widely applied to sulphur com- pounds in solution, based on the generation of blue S2 emission. A problem that can arise in such determinations is the effect of the solvent, especially when it is volatilised in the cavity at the same time as the analyte.1 The automated instrument again should remove the imprecision associated with manual operation, and microprocessor control of the timing of the various stages of the procedure, including the time available for pre-evaporation of solvent, can also lead to improved analytical performance. This paper describes some investigations that have been carried out with the automated system in order to optimise conditions for the determination of a number of inorganic and organic sulphur compounds, especially with regard to minimising solvent effects.Experimental Preparation of Stock Solutions All chemicals except promethazine were AnalaR reagents (BDH Chemicals) and distilled water was used throughout. * On study leave from the University of Ioannina, Chemistry Department, Doboli 30, Ioannina, Greece. Stock thiourea, thiocyanate and sulphuric acid solutions containing 1000 pg ml-1 of sulphur were prepared by dissolving 0.238 g of thiourea and ammonium thiocyanate and 1.70 ml of 96% sulphuric acid in 100 ml of water. In addition an aqueous thiocyanate solution of 200 pg ml-1 of sulphur was prepared by dissolving 0.4760 g of ammonium thiocyanate in exactly 1 1 of water.Promethazine solution (1000 pg ml-1 of sulphur) was prepared by dissolving 0.250 g of promethazine hydrochloride in water or methanol and diluting to 25 ml with the same solvent. Apparatus and Experimental Conditions The components and mode of operation of the automated spectrometer and sample dispenser were as described previ- ously.2 The automated instrument was used with a cylindrical cavity which, unless stated otherwise, was made of aluminium of dimensions 4 x 4 mm (depth x diameter). It was pre-treated once by leaving it briefly in dilute phosphoric acid, followed by insertion into the flame. No water-cooling system was used to cool the cavity. The flame composition was H2 2.5, N2 6.0 and air 5.0 1 min-1 and the sample size was 5 pl. The spectrophotometer slit width was 0.6 mm, the wavelength 384 nm and the photomultiplier sensitivity 2 (arbitrary units).The cavity position was optimised and the solvent drying period was set to give the most intense S2 emission. The recorder range was set to 500 or 100 mV full-scale deflection (FSD) and the chart speed to 10 cm min-1. The cavity residence time in the flame (ff), the cavity cooling time, i.e., the time from the removal of the cavity from the flame to injection of the next sample (fc) and the solvent evaporation time in the cavity before insertion into the flame (fe) were optimised. These times were controlled by the microprocessor-controlled timer, by setting values therein for TI, T2 and T3 for the cavity controller and T4 for the sampling cycle time of the ASD-53 sampler control unit.2 Fig.1 illustrates the relationship between the timing of the various operations and the T values. Optimisation Apart from the necessary experiments for the optimisation of flame composition, cavity position in the flame, centring of the cavity in the optical system, dimensions and material of cavity and sample volume, the optimisation of ff, tc and te was important for achieving high sensitivity and good reproduci- bility. A set of optimisation experiments for these times was carried out, the results of which are shown in Tables 1-3. Each340 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 result is the mean of the results of a series of automated measurements under the conditions shown.Table 1 shows that if <SO s are allowed for a cavity to cool after a measurement before injecting the next sample, there is a decrease in peak height when no solvent pre-evaporation time (te) is allowed. The standard deviation is not affected significantly by cooling time variations. Table 2 shows the results of an experiment intended to optimise Tl and T2 for aqueous sulphuric acid when the solvent pre-evaporation time is zero. When Tl is large and T2 is small, the peak height is small and the standard deviation is very large. Under these conditions the cavity is relatively hot at the time the aqueous solution is introduced into the cavity, so that some sputtering and irreproducible loss of sample occurs. A longer cooling period, or a shorter previous period in the flame, so that the cavity is cooler at the time of injection, results in constant higher and more reproducible signals.Into o u t of the the flame flame Sample in cavity Fig. 1. Relationship between the cavity movement time cycle ( t ) and the microprocessor timing schedule ( T ) Table 1. Effect of cooling time on peak height from 5 pl containing 10 pg ml-l of sulphur as sulphuric acid (tp = 15 s , t, = 0 s) Mean peak Standard deviation Cooling time/s height/mV (n = lO)/mV . 140 252 6.6 100 230 5.0 80 242 4.2 60 151 6.3 Table 2. Effect of change in TI and T2 on the peak height ( T3 = T4 = 60 s, 5 pl containing 100 pg ml-l of sulphur as sulphuric acid) Mean peak Standard deviation T,/s T2/s height/mV (n = 10)/mV 10 10 284 1.2 20 10 228 16.6 10 80 285 0.0 20 80 270 5.0 Table 3 gives the results of similar experiments on pro- methazine hydrochloride solutions in water and methanol, allowing 120 or 180 s for solvent pre-evaporation.The performance characteristics of new cavities and those which had been used many times for phosphorus and sulphur determination were compared. For aqueous promethazine in an aged cavity under conditions of small TI (short cavity residence time in flame), large T2 (long cooling time) and small T3 (short pre-evaporation time), two peaks are obser- ved, a common indication of incomplete solvent pre-evapora- tion.1 Other combinations of conditions give single peaks of reasonably similar intensity, although the intensity is slightly less for shorter flame residence times at the long solvent pre-evaporation time.Standard deviations are relatively small 300 > E 200 , - m C 0, v, .- 100 0 20 40 60 80 150 100 50 0 2 4 6 S u I p h u r concentration i pg m I - Fig. 2. Calibration graphs: (a) for aqueous solutions of A, thiourea and B, ammonium thiocyanate without solvent pre-evaporation; ( b ) for the same solutions with solvent pre-evaporation 100 (a) > E 1 m S m v) .- 0 20 40 60 80 0 2 4 6 Sulphur concentrationiug r n - 1 Fig. 3. Calibration graphs for promethazine hydrochloride: (a) without and ( b ) with solvent pre-evaporation for A, aqueous solution and B, methanol solution Table 3. Effect of changes in T I , T2 and T3 (T4 = 60 s ) on the response from promethazine hydrochloride ( 5 yl containing 5 pg ml-1 of sulphur) Mean peak height k standard deviation (n = 10)/mV Aqueous solution Methanolic solution TIIS T ~ I s T~Is 5 10 120 10 10 120 5 80 120 10 80 120 5 10 180 10 10 180 5 80 180 10 80 180 * Two peaks obtained, height of first peak given. Aged cavity 52 2 1.8 53 f 0.0 23* k 0.0 50 4 0.7 45 k 2.5 52 k 1.5 46 4 1.0 51 k 2.2 New cavity 40* + 3.5 55* k 1.7 20* 4 2.2 37* 2 0.7 63* k 2.2 45* k 5.0 34* k 1.6 40* k 5.0 Aged cavity 26 k 4.4 23 k 0.7 28 2 1.5 28 k 2.2 24 k 0.0 19 k 2.7 27 k 1.8 27 2 1.0 New cavity 58 f 3.1 40* k 0.7 81 t 3.5 73 * 10.0 63 k 2.1 28* _C 3.0 77 k 4.5 75 k 8.1JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 341 Table 4. Characteristic analytical parameters for the determination of some sulphur compounds by MECA, obtained with and without solvent pre-evaporation Solvent RSDt Substance Solvent pre-evaporation tm*ls (n = lo), '10 Thiourea .. . . . . . . Water Yes 2.2 2.0 No 2.4 0.5 NH4SCN . . . . . . . . Water Yes 2.5 2.0 No 2.6 0.5 Sulphuricacid . . . . . . Water No 5.8 0.5 Dilute H3P04 No 5.2 0.5 No 1.5 1.5 Methanol Yes 1.2 4.3 No 1.5 1.5 Promethazine hydrochloride Water Yes 1.2 3.5 * From insertion of the cavity into the flame to maximum S2 emission intensity. t For a concentration in the middle of the calibration range. Detection limit/ pg ml-1 0.5 1 .o 1 .o 1 .o 20.0 2.5 1.0 5.0 0.5 5.0 Slope of linear part of calibration grapWmV per pg ml-1 S 7.5 5.0 4.2 34 20 10 10 25 1.6 2.0 Table 5. Peak-height intensities obtained for mixtures of ammonium thiocyanate and sulphuric acid in water NH4SCN/ H2S04/ Peak height/ H2S04/ Peak pg ml-1 S pg ml-1 S mV pg ml-1 S height'/mV 5.0 0.0 16 0.5 15 10.0 0.0 47 1 .o 50 20.0 0.0 103 2.0 105 40.0 0.0 197 4.0 200 60.0 0.0 250 6.0 265 80.0 0.0 285 8.0 285 Peak height/mV Peak height/mV ygml-lS pgml-lS SCN- S042- pgml-lS SCN- S042- 10.0 5.0 45 * 10.0 42 12 20.0 10.0 92 20 20.0 90 40 40.0 20.0 195 57 40.0 183 135 60.0 30.0 237 100 60.0 225 208 80.0 40.0 257 150 80.0 250 290 NH4SCNI H2S04/ H2S04/ * 5.0 2.5 15 5.0 15 * * No H2S04 peak observed.~ ~ ~~ for all combinations of T values. Methanolic solutions all give single peaks in aged cavities, all with similar low intensities. For the aqueous promethazine solution in a new cavity, double peaks are observed in all cases. This could indicate that complete evaporation of water was not achieved at any of the T-value combinations.The variation in peak height is mainly attributed to the extent of peak separation. As a new aluminium cavity is free from possible coatings of condensed phosphates or polymerised sulphur formed during long usage, interaction of the promethazine with the aluminium surface during heating might speed up or delay part of the emission, thus splitting the peak. At the shorter evaporation time ( T3 - T4 = 60 s) and the longer cooling period ( T2 = 80 s), the peaks are smaller than any others. When the sample is injected into a hotter cavity (T2 = 10 s) the peaks are larger, especially after a longer previous residence time in the flame (10 s), because greater solvent evaporation can occur before the sample is inserted into the flame. The peak height obtained is very similar to that obtained in the aged cavity. When methanol is used as a solvent in experiments with new cavities, the intensities are different to those from aqueous solutions, but the trends are similar. When the cavity is held in the flame for the longer period (10 s) and the cooling time is only 10 s, the emissions tend to be less intense.These are the only conditions under which double peaks were obtained. This indicates that the solvent has been evaporated completely under all other conditions, as would be expected from greater volatility of methanol compared with water. A longer cooling time generally is necessary to achieve greater sensitivity. From the results of these experiments it is possible to select optimum T values for the determination of particular sulphur compounds by automated MECA.Results and Discussion Molecular emission cavity analysis has often been carried out by injection of a solution into a cavity and introduction of the cavity into the hydrogen-based flame before the solvent has evaporated. However, if the solvent and analyte subsequently volatilise so that both are present in the cavity space simultaneously, this can lead to decreased sensitivity and multi-peaked responses.' This can arise because the solvent or its decomposition products can consume radicals involved in the chemiluminescent reaction, hydrogen atoms may form inactive hydrogen molecules in the presence of carbon-con- taining species and the rapid production of solvent vapour dilutes the analyte species in the cavity space and removes them more quickly therefrom.The sensitivity difference with and without solvent pre- evaporation is demonstrated by comparing the calibration graphs for aqueous solutions of thiourea and ammonium thiocyanate obtained under both conditions [Fig. 2(a) and (b)]. When the water is evaporated before inserting the cavity into the flame, the calibration graphs are linear and show much greater sensitivity than when there is no pre-evapora- tion. Calibration graphs for promethazine hydrochloride in aqueous solution obtained without and with solvent pre- evaporation may be compared in Fig. 3(a) and ( b ) , respec- tively. They again show increased sensitivity and linearity when solvent pre-evaporation is used.Often, of course, it is necessary to use organic solvents. Fig. 3(a) and (b) also show calibration graphs for promethazine hydrochloride in methanol, without and with solvent pre-evaporation, respec- tively. Much greater sensitivity is again achieved by solvent pre-evaporation. The characteristic analytical parameters for the compounds investigated, obtained with and without solvent pre-evapora- tion, are given in Table 4. In addition to underlining the observations described above, the data for sulphuric acid confirm the enhancing effect of phosphoric acid on the sulphuric acid response whilst having little effect on the fm value .3 Molecular emission cavity analysis is capable of specia- ti0n1~3.4 because the tm value increases as the volatility of the analyte decreases.Thus, a mixture of ammonium thiocyanate and sulphuric acid gives rise to two peaks, the earlier one from thiocyanate, the later one from sulphuric acid (Table 4). The peak heights obtained from various mixtures of the two compounds are given in Table 5 . The thiocyanate peak height342 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 is not significantly affected by the presence of sulphuric acid until the sulphuric acid concentration approaches that of the selected. This enables the quantification of binary mixtures of sulphur compounds to be placed on a more secure footing. thiocyanate, i.e., 10-20 pg ml-1 of sulphur. At the highest sulphuric acid concentration the thiocyanate peak is slightly diminished. The sulphuric acid response is less sensitive than that for thiocyanate, as has been observed previously,l>3 and is more affected by changes in the thiocyanate concentration; partly because of peak overlap at higher thiocyanate concen- trations. Greater resolution could have been achieved by variation of the flame conditions. 1. 2. 3. 4. Conclusion The history of the cavity used and the time allowed for each stage in the MECA procedure (residence in flame, cooling and solvent pre-evaporation) have an important influence on the performance of MECA. The reproducible control of these and other operations by the automated device allows the optimum conditions with respect to sensitivity or reproducibility to be References Burguera, M., Bogdanski, S. L., and Townshend, A., CRC Crit. Rev. Anal. Chem., 1981, 10, 185. El-Hag, I. H., and Townshend, A., J. Anal. At. Spectrom., 1986, 1,383. Al-Abachi, M. Q., Belcher, R., Bogdanski, S. L., and Townshend, A., Anal. Chim. Acta, 1976, 86, 139. Belcher, R., Bogdanski, S. L., Osibanjo, O., and Townshend, A,, Anal. Chim. Acta, 1976,84, 1. Paper J6l87 Received September 22nd, 1986 Accepted October 27th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200339

出版商:RSC    

年代:1987

数据来源: RSC