摘要:

Analyst November 1983 Vol. 108 PP. 1297-1312 1297 Determination of Selenium in Biological Materials with Platform Furnace Atomic-absorption Spectroscopy and Zeeman Background Correction* G. R. Carnrick D. C. Manning and W. Slavin Pevkin-Elnzev Covpovation Main Avenue Norwalk CT 06856 USA The presence of phosphate and iron provides a spectral interference for the determination of selenium a t its primary resonance line at 196.0 nm that is avoided by using Zeeman background correction. Good quality pyrolytic graphite tubes platform atomisation and integrated absorbance readings provide an acceptable quantitative environment for selenium and permit most analyses to be performed against a simple calibration graph using selenium in a nickel matrix modifier. Remaining problems appear to relate to loss of selenium prior to atomisation in the presence of certain matrices, notably the combination of sodium and sulphate.Charring in the presence of oxygen did not improve the analytical situation nor did the use of silver, molybdenum or copper as a matrix modifier. We found a 20 detection limit of 28 pg and a sensitivity of about 25 pg per 0.0044 A-s (A = absorbance units) expressed as characteristic amount. A preliminary direct method is presented for selenium in urine with nickel nitric acid and magnesium nitrate present with a detection limit in urine of about 10 pg 1-I. Keywords Selenium determination ; biological materials ; atovnic-absorption spectroscopy ; platform f w n a c e atomisation ; Zeeman background correction The determination of selenium which is very important in environmental chemistry and biochemistry was recently reviewed in considerable detail by Verlinden et a1.l The levels to be determined are almost always below those possible with simple flame atomic-absorption spectroscopy (AAS) which has forced workers to extract the selenium to concentrate it or to look for more sensitive methods.Extraction opens up the possibility of contamination. Once the analyst has opted for the time-consuming effort required for extraction lie often chooses to complete the analysis with the older fluorescence method. However in the last 5-10 years both hydride and furnace AAS have been shown to provide more rapid and reliable selenium determinations than extraction - flame AAS or the fluorescence procedure down to levels at the low end of the parts per billion (p.p.b.; pg 1-l) range.For relatively dilute solutions where the volume of sample is unlimited e.g. effluent or drinking waters the hydride AAS technique is widely used. As the absolute detection limits (in picograms) are lowest with the graphite furnace this technique has been extensively studied for biochemical analyses. Reports on the furnace determination of selenium have been contradictory and confusing, probably more so than for any other metal. Some of these reported problems we have never seen and we cannot duplicate. They must reflect special problems in particular hardware, perhaps differences between graphite tubes. Some of the problems result from the severe background effects suffered by all the volatile metals particularly when determined in an inorganic matrix especially in the presence of high concentrations of alkali or alkaline earth metals.In this work, we have tried to understand which problems apply to modern instrumentation and to make the determination of selenium more convenient using simple procedures. Many of the conflicts in the literature concerning the determination of selenium deal with the properties of the furnace tubes. For example recent papers by Vickrey and co-w o r k e r ~ ~ ~ reported considerable effects on selenium absorption using diff ereiit graphite surfaces and using tubes coated with lanthanum tantalum and zirconium. We shall compare our results with theirs. Alexander et al.* reported that the graphite tubes changed in the There are some special spectral problems that pertain to selenium.* Part of this material was presented a t a Society for Experimental Geochemistry and Health meeting at Greenville NC in November 1982 by the authors’ colleague L. A. Richardson 1298 Analyst VoZ. 108 enhancement or suppression of selenium sensitivity as they aged. They felt that carbide-forming elements such as molybdenum and zirconium altered the selenium atomisation properties. These authors found variable “enhancement” factors for different metals added at 1000 mg 1-l. Pierce and Brown5 compared furnace and hydride AAS for the determina-tion of selenium (and arsenic). They found massive interferences with furnace AAS, although the technique was very sensitive. A recent review by Matousek6 stated that selenium and arsenic showed much poorer sensitivity on pyrolytically coated tubes than uncoated tubes.He used this observation to propose that selenium is volatilised as the dimer. However Fernandez and Iannarone7 had found better sensitivity with pyrolytic graphite. Vickrey and Buren2 and others also found that pyrolytically coated tubes were advantageous for selenium. Serious spectral interferences arise from the absorption of the background correction continuum (deuterium lamp) radiation by some matrix constituents. These disappear when Zeeman background correction is used. These problems appear to be more troublesome for selenium than for most other elements and include the effect of and phosphate.lo-l2 These effects and their control are discussed below.Several workers reported that sulphate interferes with the determination of selenium even in the presence of added nicke1.l39l4 Many workers have suggested that sodium chloride interfered with selenium absorption.14-16 I t was to increase the char temperatures and remove sodium chloride that EdigeP originally developed the nickel matrix modification procedure. Beaty and Cooksey17 halved the sodium chloride background by adding hydrogen to the argon purge gas. This presumably works by binding the free chlorine as hydrogen chloride thus reducing the concentration of chlorine that can form sodium chloride. They also found that calcium probably as calcium oxide provided a large background absorbance behind the selenium determination at 196 nm. This was more than halved by adding hydrogen to the argon purge gas reducing the available oxygen.Many workers have studied the effectiveness of different matrix modifiers for the deter-mination of selenium. Many metal additives have been reported to be ~ ~ e f ~ 1 ~ ~ ~ ~ ~ 1 6 ~ although nickel remains the favoured modifier.16 The US Environmental Protection Agency (EPA) recommends22 both hydride and furnace AAS for the determination of selenium in waste waters and drinking waters and they outline methods for each. A stabilised temperature platform furnace (STPF) method for the determination of 12 elements including selenium in waste waters effluents and drinking water has been published23 with a detection limit for selenium of about 1 pg 1-l. The method was direct and used a simple nickel matrix modifier in a platform furnace system and standardisation against an aqueous working curve.Fresh and estuarine waters were analysed for selenium using a nickel matrix modifier by Stein et nZ.14 They could detect 2.5 pg 1-1 of selenium in 50 pl (125 pg) of saline water. Martin et nZ.13 also measured selenium in various environmental waters using a nickel matrix modifier but sometimes required the method of additions when sulphate was present. found suppression by various acids using furnace AAS to determine selenium in environmental waters. A recent review by Robberecht and Van Grieken25 discussed the determination of selenium in environmental waters. The paper provided very valuable information on sample collection and storage contamination and speciation.The various analytical techniques were com-pared. Versieck and Cornelis26 wrote that the normal selenium level in plasma or serum is variously reported between 50 and 100 pg 1-1 in recent publications. In recent work on selenium in serum and urine Lalonde et aZ.27 used fluorimetry and required 0.4 ml of sample. They found a mean selenium concentration in serum of 143 pg 1-l. Selenium was determined in serum directly by Saeed et aL9 and Alfthan and Kumpulainen,28 after adding nickel. Both groups found that blood selenium could not be determined at 196 nm because iron interfered. Saeed and ThomassenZg determined selenium in serum and semen after protein precipitation. Pleban et al.30 found a mean plasma level of selenium of 97 pg 1-1 using Zeeman background correction and integrated absorbance signals.They also measured selenium in erythrocytes. Verlinden3l discussed the acid decomposition of blood and plasma prior to the determination of selenium by AAS. Some authors speculate that the volatilisation of selenium or arsenic as dimers6 may be responsible. The CARNRICK et al. SELENIUM IN BIOLOGICAL MATERIALS WITH Kunselman and Selenium has proved to be particularly difficult to determine in urine November 1983 PLATFORM FURNACE AAS AND ZEEMAN BACKGROUND CORRECTION 1299 iron8 or phosphorus12 interference adds to the trouble. It is possible that some of the element is bound as an organic complex which may be volatilised and lost before dissociation in the vapour phase. Valentine et aZ.32 determined selenium in urine blood and hair samples from persons residing in an area where well water was contaminated with selenium.They found the mean selenium concentration to be 79.3 pg 1-1 in urine. They used a hydride AAS technique after wet digestion in nitric and perchloric acids and heating in 6 N hydrochloric acid. This was discussed also by Verlir~den,~~ as the hydride AAS technique that they used required that selenium be present as Se(1V). Lalonde et aL2' found a mean urinary excretion rate of 124pgd-l. Their results are equivalent to 95pg1-1 with a range from 29 to 198 Our study was undertaken to evaluate the advantages of Zeeman background correction used in conjunction with the stabilised temperature platform furnace. The interaction of a variety of problems proved to be very difficult to understand and the work extended over more than a year.The study showed why selenium has been a difficult furnace determination and why the process of assessing one parameter at a time e.g. background correction matrix modifier or furnace conditions can be misleading. Parameters that are ignored can be of much greater importance than those studied and can mask optimisation procedures. pg 1-l. Experimental Equipment All tests were performed on a Perkin-Elmer Zeeman/5000 instrument equipped with an AS40 autosampler and a Model 56 strip-chart recorder except as noted. The conditions were similar to those used earlier.33-35 A selenium electrodeless discharge lamp (EDL) was used and was run in the modulated mode usually at 6 W. All experiments were performed at 196.0 nm with a spectral slit width of 2.0 nm.Peak absorbance and integrated absorb-ances were calculated with a Perkin-Elmer Data Sytem 10 and a Hewlett-Packard 7225A Graphics Plotter was used for plotting atomisation profiles. Pyrolytically coated tubes (Perkin-Elmer Part No. B0109-322) with solid pyrolytic graphite platforms (Perkin-Elmer, Part No. B0109-324) were used. The integration time for the analytical work was generally limited to 4 or 5 s. The signals from the Data System 10 reported here are called ZAA signals for the analytical result and SB signals (single-beam) for the backgrounds. The SB signals are expressed in absorbance units (A) and the ZAA signals are usually in A s The SB signal is signal plus background but for small analyte signals the SB signal is effectively background.The actual integrated absorbance signals that were used were calculated by software on the Data System 10 from signals transmitted from the Zeeman/5000. The graphs shown in later figures show typical signals but were not used for quantitative evaluation. The furnace conditions are summarised in Table I, TABLE I FURNACE PARAMETERS Parameter Dry Char Atomise Clean Cool Temperature/"C . . . . 200 800 2100 2 600 20 Ramp/s . . . 1 1 0 1 1 Hold/s . . . . . . 60 45 5 6 20 Initial flowlml min-l . . . . 300 300 0 300 300 Record/s . . -3 Read/s . . -1 Materials Most of this work utilised a nickel matrix modifier solution prepared by dissolving the pure nickel cups supplied by Perkin-Elmer for the Delves cup technique in 5.6% heated Ultrex nitric acid.However a nickel solution prepared from analytical-reagent grade nickel nitrate was used interchangeably if the blank concentrations were equally low. Typically, 5 p1 of 1 .2y0 nickel solution were added to the 20-4 standard or diluted sample in the graphite tube. The selenium standards were diluted in 1% Ultrex nitric acid from commercial solution standards (Alfa Ventron Corp. Danvers MA USA). This resulted in 60 pg of nickel in the furnace ZAA SB 0.2 -0.1 1 I 0 2 4 0 2 4 Timels Absorbance profiles and background of a solution containing no Se and 10 pug of Ca,(PO,),. The upper curves were from the Zeernan/5000 and the lower curves from a Model 5000 using a deuter-ium continuum background corrector.The broken lines represent a solution containing 60 pg of Ni also. With deuterium continuum correction a negative signal results from structured absorbance of the continuum radiation. Fig. 1. also confirmed in Fig. 1. The correction is more complete with the Zeeman correction system. The relief from the interference when nickel is used with conventional background correction is probably due to the binding of the phosphate until the furnace vapour is too hot to permit the formation of much of the phosphorus compound that at lower temperature, absorbs background radiation. He showed that the iron interference at 196 nm was avoided by Zeeman background correction, At the less sensitive 204-nm selenium line iron did not interfere but chromium and nickel interfered by overcorrection of the continuum background and these interferences also were avoided with Zeeman background correction.This is particularly interesting as nickel is deliberately added by many workers as a matrix modifier and the interference has not been previously reported. Even small amounts of iron absorb background radiation near the 196-nm line and interfere with the determination of selenium using deuterium continuum correction. Haemoglobin iron and iron contamination of waste water streams will interfere. Hence the use of Zeeman background correction frees the selenium determination of several interferences from materials commonly associated with selenium. Selenium is probably one of the elements benefitting most greatly from the use of Zeeman background correction.High-alloy steels were analysed for selenium using Zeeman AAS by F e r n a n d e ~ . ~ ~ Char Losses In preliminary work with selenium in the presence of nickel calibration graphs were prepared by spiking de-ionised water and a urine sample diluted five-fold with known amounts of selenium. Using a char temperature of 1200 "C known to be acceptable for selenium solutions containing nickel and 20 pg of nickel for each 20-4 sample aliquot the slope of the urine calibration graph was about 40% of that for the de-ionised water standards. These results certainly implied a urine matrix effect that would prohibit the analysis of urine directly using conditions previously indicated as optimal for the determination of selenium in water solutions.De-ionised water and five-fold diluted urine were spiked with 2 ng of selenium. Little selenium was lost when nickel was added to aqueous solutions until the char temperature exceeded 1200 "C. However even in the The results of a char study are shown in Fig. 2 1302 CARNRICK et al. SELENIUM IN BIOLOGICAL MATERIALS WITH Analyst Vol. 108 0.4 7 1 I I I 600 800 1000 1200 1400 0' ' Temperature/"C Fig. 2. Char curves for 2 ng of Se in water and urine diluted 1 + 4. A matrix modifier of 60 pg of Ni was used and atomisation was at 2000 "C. I I ,Aqueous 1 0.4 1 0 2 Time/s 4 Fig. 3. Absorbance profiles from the char study in Fig. 2 showing both solutions at a char temperature of 900 "C. The effect of the urine matrix is to cause Se to be volatilised about 0.4 s earlier that is a t a lower tempera-ture.presence of nickel selenium was lost from the urine when the char temperature exceeded 900 "C. Thus it is clear that the difference in sensitivity for selenium in water and in diluted urine resulted from losses from the urine during the char step. Data System profiles (Fig. 3) show that at a char temperature of 900 "C the selenium in the urine starts to evolve in the atomisation step about 0.4 s earlier than selenium in water. We further sought to explain the loss of selenium in a urine matrix at char temperatures below 1200 "C. We studied each of the principal compounds to be found in urine to isolate that which was responsible for the loss of selenium. Each individual matrix constituent was added to a sample cup containing selenium such that the final solution contained 0.1 pg ml-1 of selenium ('2 ng per 20-4 aliquot) and a matrix concentration equal to that typically found in urine (see Table 11).The sample was not diluted therefore resulting in a concentration for each constituent that was five-fold larger than was used in the urine experiments. The data in Fig. 4 suggested that some selenium was lost at temperatures as low as 700 "C in sulphuric acid. Urea and creatinine were not studied as they decompose at low char temperatures and were not expected to be troublesome. TABLE I1 COMPOSITION OF ARTIFICIAL URINE Component Concentration % Creatinine . . 0.03 Urea . . 2.0 NaCl . . 1.1 H,SO . . 0.13 H,PO . . 0.26 While the sodium chloride reduced the maximum char temperature at matrix concentra-tions equivalent to those present in the diluted (1 + 4) urine sample neither sodium chloride nor sulphuric acid individually explained the loss of selenium.When the two compounds were added together they resulted in losses of selenium at temperatures above 900 "C. The Data System profile showed that the addition of sulphuric acid alone to an aqueous solution caused the selenium signal to begin to appear about 0.1 s earlier but there was no reduction in the integrated absorbance. Although the mechanism of this loss remains unclear it was speculated that the chloride might be providing the problem. Therefore sodium nitrate was substituted for sodium chloride in combination with 0.03% of sulphuric acid to be sure there was no chloride present but the loss of selenium persisted.A char study was repeated using sodium nitrate and sulphuric acid separately and together (Fig. 5). The results showed that there was November 1983 PLATFORM FURNACE AAS AND ZEEMAN BACKGROUND CORRECTION 1303 - 800 1000 1200 1400 Tern peratu re/"C Fig. 4. Char curves for 2 ng of Se in the pres-ence of 60 pg of Ni in several matrices associated with urine The concentrations of NaC1 H,PO, and H,SO are shown in Table 11. Atomisation was at 2000 "C. loss at 900 "C from the combination but not from the separate materials. Sodium in combi-nation with sulphuric acid provided the problem. The important result of this set of experiments was that certain matrix constituents produced char losses as evidenced by the removal of these losses at low enough char temperatures.Other workers who found "suppressions" in the presence of sodium and sulphate may have experienced char losses. We shall discuss the sulphate interference in more detail below. 0.4 u) 5 0.3 8 -2 5 -$ 0.2 lu U CI E c 0.1 a + 0 Aqueous /\. 0.3% NaN03 + 0.03% H2S04 \, 800 1000 1200 1400 Temperatu rePC Fig. 5 . Char curves for 2 ng of Se in water 0.3% NaNO,, 0.03% H,SO and in a solution containing both NaNO and H,SO together. All solutions contained 60 pg of Ni and were atomised a t 2000 "C. Matrix Modification The use of nickel or copper as a matrix modifier for selenium was part of the original matrix modification work of Ediger.16 He found that copper was almost as effective as nickel.In simple aqueous solutions selenium was lost at temperatures below 300 "C whereas in the presence of 20 pg of nickel it was not lost until 1200 "C. Welcher et aL18 showed that selenium in solutions of nickel alloys (and other heavy metal alloys) could be charred at much higher temperatures than in solutions without these metals. Hennlg added molybdenu 1304 CARNRICK et al. SELENIUM IN BIOLOGICAL MATERIALS WITH Analyst Vol. 108 and found that a char temperature of 1400 "C could be used. Szydlowsky20 found a poorer sensitivity for selenium using molybdenum than with nickel or copper. Kirkbright et aZ.21 evaluated many matrix modifiers for selenium and found nickel copper and silver to be better than others nickel and copper permitting charring at 1200 "C.Stein et aZ.14 found nickel and iron to be preferable to other metal additives for selenium in fresh or estuarine waters. Saeed et aL9 studied copper silver nickel and iron as matrix modifiers using radio-active selenium and found copper and iron were not effective in stabilising selenium in rat blood but nickel and silver were. Alexander et a,!.* used radioactive selenium and made a very systematic study of various matrix modifiers for the determination of selenium in a sodium salt in blood and urine. Manganese zinc and thallium were effective in sodium compounds nothing was very effective in blood and nickel and silver permitted charring to 1100 and 1300 "C respectively in urine. Saeed and Thomassen12 tested nickel tungsten, zirconium cerium palladium and platinum as matrix modifiers for selenium and found nickel to be as good as any.We recently found that magnesium nitrate is a valuable matrix modifier for the deter-mination of manganese aluminium and perhaps Magnesium nitrate is a familiar ashing aid,37-39 used for selenium and other metals. The role of the modifier is to bind or trap the analyte in a stable environment while the bulk of the sample matrix is volatilised. This is not very different from the requirement of an ashing aid. However experimental data showed that the addition of magnesium nitrate was not effective in increasing the permissible char temperature for the urine or for an aqueous solution if in both instances, the solutions also had 60pg of nickel present. We show later however that magnesium nitrate provides better recoveries of selenium from urine when added as an ashing aid.As other workers have suggested that other metals may be preferable to nickel as a matrix modifier and particularly Alexander et aL4 found that silver was preferable to nickel in the determination of selenium in urine we compared silver and molybdenum with nickel. Fig. 6 shows the results of a char study using silver molybdenum and nickel as matrix modifiers with undiluted urine. We found no improvement in the stabilisation of the selenium when 1000 pg ml-1 (25 pg) of silver or molybdenum were used instead of the nickel (60 pg) used previously by us. Thus we were unable to confirm the results of Alexander et aL4 In all instances selenium was lost above about 900 "C.0.2 fn \ 0 900 1000 1100 1200 1300 1400 I I I I Temperature/"C Fig. 6. Char curves conparing several matrix modifiers 25 pg of Ag and Mo and 60 p g of Ni. Atomisation was at 2000 "C and all aliquots contained 2 ng of Se. Several ~ ~ r k e r ~ ~ J ~ ~ ~ ~ ~ ~ ~ have considered copper as an alternative matrix modifier to nickel. We investigated this possibility using 6 pg of copper compared with 60 pg of nickel. We studied the char characteristics of an aqueous solution a normal urine solution diluted 1 + 4 and the same urine sample containing 0.07% of sulphuric acid. In each instance the matrix modifier contained 3.5% of nitric acid and 25 pg of magnesium nitrate which ar November 1983 PLATFORM FURNACE AAS AND ZEEMAN BACKGROUND CORRECTION 1305 TABLE I11 ABSORBANCE (Ass) OF SOLUTIONS CONTAINING SELENIUM AND COPPER OR NICKEL AS A MATRIX MODIFIER Temperature/ "C 500 700 900 1000 1100 Aqueous Urine diluted solution 1 + 4 Ni c u Ni c u 0.22 0.20 0.21 0.15 0.19 0.18 0.19 0.15 0.19 0.21 0.19 0.17 0.21 0.22 0.18 0.17 - 0.20 - 0.15 + -A- I Urine (1 + 4) + 0.07y0 H,SO, and 0.2% NaCl I + Ni c u 0.16 0.13 0.15 0.14 0.15 0.14 discussed later in addition to 1 ng of selenium.The results are reported in Table 111. The signal for the urine diluted 1 + 4 is low with the copper present especially in the presence of added sulphate and sodium shown later to be troublesome. When the absorbance profiles (Fig. 7) were compared the presence of copper in the aqueous standards caused the selenium peak to appear about 0.2 s earlier and to be narrower and therefore higher than with nickel.This may have caused workers seeking the largest peak absorbance signal to prefer copper. However the integrated absorbance values indicate a poorer recovery of the added 1 ng of selenium. Larger amounts of copper may be preferable but we did not investigate this possibility. 0.2 a, C lu -F 2 0.1 Q: 0 Aqueous I-I . . i i I +O.O7% H2S04 Urine (1+4) 1 +o.20/p NaC' 0 1 2 3 4 5 0 i 2 3 4 5 0 1 2 3 4 5 Time/s Fig. 7. Comparison of Cu (6 pg) and Ni (60 pg) as a matrix modifier for the determination of 1 ng of Se added to urine. Also present were 3.5% HNO and 25 p g of Mg(NO,),. The cool-down step was used. Oxygen Ashing of Urine Samples L'vov and Ryabchuk40 and Salmon and Holcombe41 have recently shown that the presence of oxygen during the low-temperature stages of furnace determination of volatile metals can delay the appearance temperature in the atomisation step.We looked at the possibility that the effect might be observed for selenium. Using the parameters in Table I Fig. 8 shows the typical char curve (broken) for selenium with a 5-pl matrix modifier of 60 pg of nickel and 5.6y0 of nitric acid. There is no loss of selenium below 2100 "C. We also show a char curve for five-fold diluted urine with the same nickel and nitric acid. Losses are great above 900 "C. When the same sample was charred in a 100% oxygen atmosphere the curve approached but did not equal the aqueous system. Below 700 "C the background signal exceeded 1.5 A and it preceded the selenium peak.At 900 "C there was almost no background remaining. The experiment in Fig. 8 with ashing in pure oxygen was repeated with similar results. Salmon and Holcombe41 showed that about 0.5% of oxygen in argon had the same effect as pure oxygen. Further experiments comparing charring in pure argon and argon containing 0.5% of oxygen showed that nitric acid had to be present. If less than 1% of nitric acid wa 1306 ? +! z n 2 0.1 0 0.2 0 E CD 4- -CARNRICK et aZ. SELENIUM IN BIOLOGICAL MATERIALS WITH Analyst VoZ. 108 --tJl 0.3 p-- U- Aqueous Urine (no 0 2 ) \ Urine (1+4) O2 0 1 600 800 1000 1200 1400 Tem peratu rePC Fig. 8. Char curves comparing the effect of 0.5% of 0 in the argon stream.The solutions contained 2 ng of Se in urine and 5 p1 of a matrix modifier consisting of 60 pg of Ni and 5.6% of HNO,. In both situations the urine was diluted 1 + 4. For reference the aqueous char curve is shown from an earlier figure. present charring in 0.5% oxygen caused losses of selenium. In the work of Salmon and Holc~mbe,~~ the effect of oxygen was to delay the appearance temperature up to but not exceeding about 1050 "C. Thus our data in Fig. 8 are consistent with their findings. Sulphate Interference The char loss caused by sulphate has been particularly troublesome and workers have reported varying degrees of sulphate interference on selenium. Stein et aZ.,14 using 0.2% nickel found a 70% reduction in selenium absorbance from 0.05% of sulphate charring at 1500 "C.Henn19 used charring at 1000 "C with molybdenum as a matrix modifier and found only 20% interference from 1% of sulphate. Martin et aZ.13 used charring at 1500 "C and found that the interference from 0.07% of sulphate was reduced from 80 to 15% when the nickel concentration was increased from 0.1 to 1.0%. Martin et aZ. and Stein et al. resorted to the method of additions as sulphate reduced selenium recoveries to less than half with some environmental water samples. When charring at 1200 "C with 1.25% of nickel (60 pg on the platform) there was a small depressive interference from as little as 0.0027(0 sulphuric acid. Fig. 9 shows a char curve obtained using a larger amount of sulphuric acid (0.1%). The dip at about 800 "C was reproducible in a variety of similar experiments over several days.It is interesting that charring at 800 "C provided greater losses than at 1000 "C as the same material must pass through 800 to reach 1000 "C. Without magnesium nitrate solutions high in sulphate should be charred at about 400 "C. This is feasible only if the background corrector can handle large signals a major advantage of the Zeeman correction system for selenium. The alternative of using a higher char temperature and the method of standard additions to correct for the variable char 10ss13914 is certainly risky. ~-0.1 CD 200 400 600 800 1000 Te m per at u r e/"C Fig. 9. Char curve for 0.1% H,SO, 2 ng of Se and 60 pg of Ni. Atomisation was at 2000 "C November 1983 PLATFORM FURNACE AAS AND ZEEMAN BACKGROUND CORRECTION 1307 When we tried to use our procedure for the analysis of urine we found that selenium could be quantitatively recovered in some urines but not in others.Suspecting sulphate as the problem we added 0.07% of sulphate (as sulphuric acid) higher than is normally present in urine but sulphate is variable. In a urine already showing low selenium recovery the addition of sulphate reduced the selenium recovery to 52% whereas for a urine where the selenium recovery was good the added sulphate reduced the selenium recovery to 85%. The effect of sulphate was controlled by adding magnesium nitrate to the matrix modifier solution as shown later. Perchloric Acid Interference Typically organic samples are often digested and oxidised in a mixture of nitric and perchloric acids occasionally with the addition of sulphuric acid.42 It is difficult to know the acid concentrations in the final extracts after the long boiling steps in the digestion -oxidation procedure.However an attempt was made to determine the levels of perchloric acid that could be tolerated. Little interference from perchloric acid was found up to about 0.1 N'. The absorbance profiles showed that the addition of 1 N' perchloric acid resulted in less tailing of the absorbance signal. Optimisation of the Determination of Selenium in Urine Having studied the various factors that influence the selenium determination we tried to optimise the conditions for the determination of selenium in urine. We wanted to use as large a urine sample as possible to provide the best sensitivity.We studied the relationship between the amount of urine added to the platform the amount of nickel and the amount of nitric acid needed to reduce the background sufficiently. Fig. 10 shows that whereas about 20 pg of nickel were optimal for stabilising selenium in aqueous solutions confirming the results of Ediger,16 a larger amount (about 60pg) was required to stabilise the selenium in the five-fold urine dilution. The Data System absorb-ance profiles showed that in both aqueous solutions and in urine increasing concentrations of nickel tended to delay the appearance of selenium and increased the tailing. Amounts of nickel higher than about 80 pg required integration times of longer than 4 s and probably should be avoided.The viscosity of undiluted urine made it difficult to pipette on to the platform and the background absorbances were very large. With a fresh unacidified urine sample the pipetting problems seemed to be reduced by adding 0.5% of Triton X-100 to the wash solution of the AS-40. In work with cadmium in seawater,43 we have shown that background absorbances from sodium chloride may be reduced by using nitric acid as a matrix modifier in much the same way as using ammonium nitrate. Bertenshaw et aZ.44 also used nitric acid to reduce the background absorbance at the 217-nm lead wavelength in their case. Using 20-4 aliquots 0.3 in ? 8 e 0.2 :: m 0 U 4- I 0.1 cn 4- -I I I 1 .o 10 100 N i/pg Fig. 10. Effect on the absorbance of 2 ng of Se of increasing amounts of Ni in water and in urine diluted 1 + 4.Char was at 750 "C and atomisation a t 2000 OC 2 ng Se aqueous n n -(1+4)+2 ng Se 0.3 --v) 3 0 (IJ 0.2 s n 0) 0 c h 0.1 s 03 -600 800 1000 1200 1400 Te m per at u rei"C Fig. 11. Char curve for Se in urine compared with water November 1983 PLATFORM FURNACE AAS AND ZEEMAN BACKGROUND CORRECTION 1309 TABLE IV FURNACE PARAMETERS FOR URINE DETERMINATION Parameter Dry Char Cool Atomise Clean Cool TemperaturelOC . . . . 160 800 20 2000 2600 20 Ramp/s . . 1 1 1 0 1 1 Hold/s . . . . 60 45 20 8 6 20 Initial flow/mlmin-l 300 300 300 0 300 300 Record/s . . . . . . -3 Read/s . . . . - 1 was suggested recently by Chakrabarti et aZ.45 In our experience for the determination of selenium in urine this step appeared to delay the selenium peak relative to the time when the tube reached constant temperature.Using all these factors in the urine selenium procedure and the furnace conditions in Table IV we obtained the background and char curves in Fig. 11. Actually the data were collected several times with slightly different conditions but always with essentially the same results. To obtain manageable background signals even with Zeeman correction required the char temperature to be greater than 700 "C. If the char temperature was greater than about 900 O C selenium was lost from urine although not from standards. In this work the absorbance profiles shown in Fig. 12 were typical. The effect of increasing amounts of magnesium nitrate was to reduce the amount of selenium volatilised from urine at a lower temperature than when selenium is in an aqueous solution.The small selenium peak at 1 s in the presence of 25 pg of magnesium nitrate was characteristic and was seen in all of this work. The average 0 was 0.0026 which is equivalent to 17.5 pg of selenium. This is a detection limit of about 35 pg in 4 pl of urine or about 9 pg 1-1 of selenium in the urine. The recovery of 0.5 ng of selenium against aqueous standards containing 0.09% of sulphuric acid and 0.2% of sodium chloride varied from 89 to 104%. If sodium chloride and sulphuric acid were omitted from the aqueous standards the slope was characteristically about 20% greater. The background for some of the 1 + 4 diluted urines was 0.8 A but most were about 0.5 A still too large for convenient work with the deuterium corrector.Sensitivity and Detection Limits Four urines were analysed ten times each with the results shown in Table V. 0.2 Q) C Q 4? s 2 0.1 0 -A recent review46 stated that the characteristic amount for selenium is 25 pg per 0.0044 A. 0.3 A ;' I '- A I ) -. 0 1 2 3 4 Time/s Fig. 12. Absorbance profiles for Se in urine showing various amounts of Mg(NO,) added to reduce Se losses in drying and charring. A Aqueous + 26 p g of Mg(NO,),; B + 26 p g of Mg(NO,),; C + 8.7 p g of Mg(NO,),; and D no Mg(NO,), 1310 CARNRICK et al. SELENIUM IN BIOLOGICAL MATERIALS WITH Analyst VoZ. 108 TABLE V DETERMINATION OF SELENIUM IN URINE SAMPLES Sample No. Average/A.s o/A-s Concentration/pg I-' Recovery % 1 0.0275 0.002 3 49 104 2 0.0277 0.003 3 48 96 3 0.0169 0.0022 29 89 4 0.0259 0.002 8 45 99 Average 0.002 6 From previous platform work we have reported 30 pg per 0.0044 A-s23 and a detection limit for selenium of 50 pg,46 which in a 20-4 sample is 2.5 pg 1-l.For most of this work with the Zeeman/5000 a 2-ng selenium solution yielded an integrated absorbance of 0.3 A-s or a characteristic amount of 30 pg per 0.0044 Ass. The Zeeman background correction system permits the use of brighter light sources to improve the signal to noise ratio. Using the conventional deuterium corrector the source of elemental emission and the deuterium source must have similar intensity and this requires compromise conditions. As the best detection limit was not a primary requirement in this work we mostly used a power setting of 5-6 W for the selenium lamp.Using power settings of up to 7 W (the maximum available from this lamp) the effect on the urine absorbance profiles is as shown in Fig. 13. Above 7 W the lamp became unstable. Fig. 14 compares the absorbance profiles of an aqueous standard containing 250 pg of selenium and of urine No. 4 using the lamp at 5.5 W typical of this work and at 7 W. At the maximum power the intensity is about five-fold greater and the noise is proportionately less. Using the protocol in Table IV the lamp set to its maximum intensity of 7 W and 250 pg of selenium in 20 p1 containing 60 pg of nickel 25 pg of magnesium nitrate and 3% of nitric acid on the platform the average absorbance was 0.04 and 0 was 0.0023 A-s.With the usual 20 criterion this provided a detection limit of 28 pg and a sensitivity of 27 yg per 0.0044 A-s. The relative intensity is also indicated at each setting. This is almost half the detection limit quoted previou~ly.~~ 0.2 0.1 m e s 2 0 Energy 7 w 1.00 6 W 0.4 5 W 0.05 4 W 0.03 3 w 0.002 ' 1 t I I I I I 0 1 2 3 4 5 Ti me/s Fig. 13. Absorbance profiles as functions of Se EDL intensity. An aqueous solution of 100 pg of Se 60 pg of Ni and 100 pg of Mg(NO,) was used. Discussion The reason for undertaking this study was to assess the success of the new furnace tech-The artificial environ- nology and the recent theoretical understanding of furnace processes November 1983 PLATFORM FURNACE AAS AND ZEEMAN BACKGROUND CORRECTION 131 1 0.2 (u C rrr a L 0.1 s a a 0 Urine No.4 . 7 w Std. €=5-f0Id I d v .A -- Std. 0.030 A s % 5.5 w €= 1 .O 0 1 2 3 4 5 Timels Fig. 14. Absorbance profiles of 250 pg of Se as an aqueous standard and the urine sample No. 4 containing 45 pg 1-1 of Se. Two lamp intensi-ties are compared. ments usually used for such tests often mask residual problems and theoretical inadequacies. Thus it seems necessary to use real troublesome situations. There are many examples where the usefulness of theoretical studies would be much greater if the number or variables were reduced and more reproducible conditions were used. The actual instrumentation is not an uncontrolled variable provided that (1) relatively stable thermal conditions are used (2) the electronics are faster than the atomisation processes being studied and (3) integrated absorbance signals are used.We have been able to correlate our work with that of Jenke and Woodriffg7 and with some of the recent work from Frech et aL4* using different instrumentation. We believe that close to optimum instrumental conditions for the determination of selenium are provided by the combination of the stabilised temperature platform furnace and Zeeman background correction. Spectroscopic problems associated with the presence of iron and phosphate and continuum background correction are removed by the Zeeman background corrector. Problems associated with volatilisation effects have confused earlier workers and these are controlled by the platform technique coupled with an impervious pyrolytic graphite coating of the graphite tubes and the use of integrated absorbance signals to reduce residual variations in volatilisation resulting from a variable matrix.With these instrumental conditions understood and controlled it became apparent that certain samples produced a loss of selenium during the char step even at very low tempera-tures and perhaps during the drying step also. This effect seemed most persistent when both sulphate and sodium were present together as neither matrix by itself was very troublesome. The nature of this sulphate - sodium problem was shown to result from a depression of the temperature at which char losses occur even in the presence of a nickel matrix modifier.This problem was considerably reduced by adding magnesium nitrate to the sample in its familiar role as an ashing aid. The loss of analyte during the charring step emerges as an important problem still requiring attention in certain applications. The loss of volatile forms of selenium at low temperatures is well documented in the literature of fluorescence and spectrophotometric methods for selenium. Bock and Jacobg9 found that losses were loo% simply when selenium was heated to dryness in a water- or sand-bath in the presence of certain acids. In some organic matrices, all or part of the selenium may be in a volatile form. Therefore the official methods of the AOAC warn that some plant materials contain selenium in a form that is lost at temperatures below 60 0C.42 The temperature at which selenium compounds are volatile is dependent on the oxidation state of the selenium.As a result most procedures for the determination of selenium in organic matrices use a digestion - oxidation procedure to convert Se2- to Se4+ or Se6+. Good recoveries were obtained when organics were digested by oxidation using perchloric acid in the acid mixture.37 However even with a nitric acid - perchloric acid mixture severe losses have been reported when the sample was taken to dryness.50 Our results suggest tha 1312 CARNRICK MANNING AND SLAVIN there would be little trouble from perchloric acid interference if urine samples were digested by oxidation to convert the volatile selenium into a more stable form. It appears though, that the conversion is a slow process.Jonghorbani et aZ.,51 using rat urine labelled with radioactive selenium found that recoveries approached 100% only after about 20 min of oxidation. Even then there was no assurance that losses would not occur when the sample was taken to dryness during the first step in the furnace. We thank F. J. Fernandez Sabina Slavin and B. Welz for many helpful suggestions. Discussions with G. Siess of Bodenseewerk Perkin-Elmer helped to elucidate the temperature rise on. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49.50. 51. problem with the Zeeman magnet that led to the setting of the diode with the-magnet References Verlinden M. Deelstra H. and Adriaenssens E. Talanta 1981 28 637. Vickrey T. M. and Buren M. S. Anal. Lett. 1980 13 1465. Vickrey T. M. Harrison G. V. and Ramelow G. R. Anal. Chem. 1981 53 1573. Alexander J. Saeed K. and Thomassen Y. Anal. Chim. A d a 1980 120 377. Pierce F. D. and Brown H. R. Anal. Chem. 1977 49 1417. Matousek J. P. Prog. Anal. A t . Spectrosc. 1981 4 247. Fernandez F. J. and Iannarone J. A t . Absorpt. Newsl. 1978 17 117. Manning D. C. A t . Absorpt. Newsl. 1978 17 107. Saeed K. Thomassen Y. and Langmyhr F. J. Anal. Chim. Acta 1979 110 285. Fernandez F. J. Myers S. A. and Slavin W. Anal. Chem. 1980 52 741. Fernandez F. J. and Giddings R.A t . Spectrosc. 1982 3 61. Saeed K. and Thomassen Y. Anal. Chim. Acta 1981 130 281. Martin T. D. Kopp J . F. and Ediger R. D. A t . Absorpt. Newsl. 1975 14 109. Stein V. B. Canelli E. and Richards A. H. A t . Spectrosc. 1980 1 61. Ohta K. and Suzuki M. Fresenius 2. Anal. Chem. 1980 302 177. Ediger R. D. A t . Absorpt. Newsl. 1975 14 127. Beaty R. D. and Cooksey M. M. A t . Absorpt. Newsl. 1978 17 53. Welcher G. G. Kriege 0. H. and Marks J. Y . Anal. Chem. 1974 46 1227. Henn E. L. Anal. Chem. 1975 47 428. Szydlowsky F. J. A t . Absorpt. Newsl. 1977 16 60. Kirkbright G. F. Hsiao-Chuan S. and Snook R. D. A t . Spectrosc. 1980 1 85. “Methods for Chemical Analysis of Water and Wastes,” EPA-600 4-79-020 Environmental Moni-toring and Support Laboratory Cincinnati 1979.Manning D. C. and Slavin W. Appl. Spectrosc. 1983 37 1 . Kunselman G. C. and Huff E. A, A t . Absorpt. Newsl. 1976 15 29. Robberecht H. and Van Grieken R. Talanta 1982 29 823. Versieck J. and Cornelis R. Anal. Chim. Acta 1980 116 217. Lalonde L. Jean Y. Roberts K. D. Chapdelaine A. and Bleau G. Clin. Chem. 1982 28 172. Alfthan G. and Kumpulainen J. Anal. Chim. Acta 1982 140 221. Saeed K. and Thomassen Y. Anal. Chim. Acta 1982 143 223. Pleban P. A. Munyani A, and Beachum J. Clin. Chem. 1982 28 311. Verlinden M. Talanta 1982 29 875. Valentine J . L. Kang H. K. and Spivey G. H. Enuiron. Res. 1978 17 347. Slavin W. Manning D. C. and Carnrick G. R. A t . Spectvosc. 1981 2 137. Slavin W. Carnrick G. R. and Manning D. C. Anal. Chem. 1982 54 621. Fernandez F.J. Bohler W. Beaty M. M. and Barnett W. B. A t . Spectvosc. 1981 2 73. Fernandez F. J. and Beaty M. M. Spectrochim. Acta Part B in the press. Gorsuch T. T. “The Destruction of Organic Matter,” Pergamon Press Oxford 1970. Haynes B. W. A t . Absovpt. Newsl. 1978 17 49. Tam G. K. H. and Lacroix G. J . Assoc. 08. Anal. Chem. 1982 65 647. L’vov B. V. and Ryabchuk G. N. Spectvochim. Acta Pavt B 1982 37 673. Salmon S. G. and Holcombe J . A. Anal. Chem. 1982 54 630. Horwitz W. Editor “Official Methods of Analysis of the Association of Official Analytical Chemists,” Thirteenth Edition Association of Official Analytical Chemists Washington DC 1980 p. 45. Pruszkowska E. Carnrick G. R. and Slavin W. Anal. Chem. 1983 55 182. Bertenshaw M. P. Gelsthorpe D. and Wheatstone K. C. Analyst 1982 107 163. Chakrabarti C. L. Wu S. and Bertels P. C. Paper presented at 29th Conference Spectroscopy Slavin W. and Manning D. C. Prog. Anal. A t . Spectvosc. 1982 5 243. Jenke D. R. and Woodriff R. Arnes. Laboratory August 1982 14. Frech W. Persson J.-A. and Cedergren A. Pvog. Anal. A t . Spectvosc. 1980 3 279. Bock R. and Jacob D. 2. Anal. Chcm. 1964 200 81. Stanton R. E. and McDonald A. J. Analyst 1965 90 497. Jonghorbani M. Ting B. T. G. Nahapetion A. and Young U. R Anal. Chem. 1982 54 1188. Received April 21st 1983 Accepted May 26th 1983 Society of Canada September 1982

ISSN:0003-2654    

DOI:10.1039/AN9830801297

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst November 1983 Vol. 108 $9. 1313-1317 1313 Determination of Total Mercury in Fish: An Improved Method Hon Way Louie Australian Governinent A.ualytica1 Labovatovies 1 Suakin Street Pymble New South Wales 2073 A ustvalia X simple method is described for the rapid digestion of fish for total mercury determination by cold-vapour atomic-absorption spectrophotometry a t 253.7 nm. A significant departure from previous methods is the use of concentrated hydrochloric acid together with sulpliuric and nitric acids in the open digestion of samples. The digestion of up to 3 g of fish a t 85-100 "C is completed within 30 min and good recoveries of mercury are obtained from NBS Albacore tuna and fish samples spiked with inorganic mercury. The detection limit is 0.01 pg g-l of mercury.Keywords Total mercury determination ; atomic-absorption spectrophoto-metry ; $sh Within the last three decades many methods have been developed for the digestion of fish for mercury determination. Methods using various mixtures of concentrated acids and oxidants have produced good recoveries of mercury.1-5 However G o ~ s u c ~ ~ ~ ~ has shown that hydro-chloric acid produced by the reduction of percliloric acid with organic matter causes volatilisa-tion of mercury especially at temperatures above 100 0C.89g The incorporation of hydro-chloric acid in acid mixtures was therefore not recommended6-9 unless efficient condensers were used to trap mercury vapour during the digestion of Therefore if there are many samples to be digested the use of condensers is inconvenient.Most methods involving the digestion of fish below 100 "C and where the use of mercury traps is unnecessary are time consuming and often require several hours to achieve complete digestion.l9l4 There is a need for an efficient and simple method that will allow large numbers of samples to be digested simultaneously. In this paper a method is described for the rapid digestion of fish for total mercury deter-mination by a cold-vapour atomic-absorption spectrophotometric technique based on that of Hatch and Ott.15 The stability of mercury standards in solutions of some acids is also discussed. Experimental Apparatus A twenty-hole boiling water-bath was used for the digestion of samples and Erlenmeyer flasks (125 ml) covered with glass marbles (approximately 25 mm i.d.) were used as digestion vessels.Mercury was analysed using either a Perkin-Elmer Model 306 or a GBC 901 atomic-absorption spectrophotometer in conjunction with an Instrumentation Laboratory Model 440 atomic-vapour accessory. A chart recorder was used to measure the absorption due to mercury vapour. The reaction vessel consisted of a 150-ml Corning beaker-flask and a magnetic follower. Reagents Analytical-reagent grade chemicals were used throughout. Nitric acid 16 M. Sulphuric acid 18 M. Hydrochloric acid 10 M. Tin(l1) chloride (SnC12.2H20) solution 20% m/V. Dissolve 20 g in 10 ml of hot hydro-Trace amounts of mercury Mix 100 ml of 16 M nitric acid and 50 ml of 18 M sulphuric acid chloric acid (10 M) and make up to 100 ml with distilled water.can be removed by bubbling nitrogen through the solution for 10 min. with 850 ml of water. D i h t i n g acid solution 1314 LOUIE DETERMINATION OF TOTAL MERCURY Analyst Vol. 108 Obtained from the National Bureau of Standards (NBS) Washing-ton DC USA. Dissolve 1.3535 g of mercury(I1) chloride in 1 1 of 1 M hydrochloric acid. Dilute 1 ml of the mercury stock solution to 100 ml with 1 M hydrochloric acid. This solution should be prepared weekly. Dilute 1 ml of the mercury intermediate solution to 100 ml with 1 M hydrochloric acid. This solution should be prepared fresh daily. NBS Albacore tuna.16 Standard mercury stock solution 1000 pg ml-l. Standard mercury intermediate solution 10 pg ml-l. Standard mercury working solution 0.1 pg ml-l.This solution is stable for at least 1 year. Procedure Digestion of samples (Dry samples should be moistened with a small amount of water before acids are added in order to minimise frothing.) Add 5 ml of 16 M nitric acid 2.5 ml of 18 M sulphuric acid and then 1 ml of 10 M hydrochloric acid. Cover the flask with a glass marble and when the initial reaction subsides (after approximately 15 min) place the flask on top of a boiling water-bath for 40 min. Remove the flask from the water-bath and after cooling make up to 50 ml with distilled water. Weigh approximately 2 g of homogenised fish into a 125-ml Erlenmeyer flask. A suitable aliquot of this solution is used for mercury determination. Analysis of samples A 20-ml volume of the digested fish solution or a smaller aliquot diluted to 20 ml with the diluting acid solution is poured into the reaction vessel and flushed with nitrogen for 20 s.A 3-ml volume of tin(I1) chloride solution is then added and the combined solution is stirred and allowed to react for 1 min. During this time any mercury vapour generated is passed through an absorption cell and detected at 253.7 nm by an atomic-absorption spectrophotometer. The reaction vessel is again flushed with nitrogen. Calibration graph Aliquots of the standard mercury working solution (0.1 pg ml-1) containing 0.1 0.2 0.3, 0.4 and 0.5 pg of mercury respectively are diluted to 20 ml with the diluting acid solution and their absorbances measured in the same way as the samples. Amercury standard calibration graph is then plotted.Results and Discussion NBS Albacore tuna (about 0.4 g) and fish samples (about 2 g) spiked with known amounts The results of The result obtained for NBS Albacore tuna is the mean of The results for spiked fish samples are the means of nine analyses carried out The mixed fish sample consisted of pomfret hake pollock of a standard mercury(I1) chloride solution were used in recovery studies. analyses are shown in Table I. 60 analyses. in three separate experiments. TABLE I RECOVERY OF ADDED MERCURY IN FISH SAMPLES DIGESTED BY THE PROPOSED METHOD Sample Mercury added/pg NBS Albacore tuna . . . . ot Mixed fish . . . . . . 0.10 0.20 0.30 0.40 Dried cuttle fish . . 0.10 0.20 0.30 0.40 Dried oyster . . 0.30 Mercury recovered*/pg 0.94 f 0.05; 0.10 f 0.01 0.20 f 0.01 0.31 f 0.01 0.40 f 0.02 0.11 f 0.01 0.21 & 0.01 0.31 f 0.01 0.40 f 0.01 0.29 f 0.01 * Mean & standard deviation.t No mercury added; NBS stipulated value16 = 0.95 & 0.1 pg g-l. $ Result in micrograms per gram for NBS -4lbacore tuna only November 1983 IN FISH AN IMPROVED METHOD 1315 and silver bream (all containing less than 0.03 pg g-l of mercury) which were combined in equal proportion and homogenised. A mean result of 0.94 pg g-1 of mercury was obtained for NBS Albacore tuna compared with the stipulated value16 of 0.95 & 0.1 pg 8-l. As more than 80% of the total mercury content in NBS Albacore tuna is present as methyl mercuryls the result obtained indicates the effective conversion of organomercury to inorganic mercury.Quantitative recoveries were obtained for fish samples (about 2 g) spiked with inorganic mercury over the range of 0.1-0.4 pg of mercury. Because no obvious matrix effect was detected it appears that carrying standard mercury solutions through the digestion process is not required when constructing the standard mercury calibration graph. Small amounts of undigested fat are usually present after fish samples are digested. Nevertheless the results listed in Table I show that the recoveries of mercury from fish samples are unaffected. Effect of Temperature and Hydrochloric Acid The results for fish samples digested at exactly 70 "C (using a constant-temperature water-bath) for 0.5 h were consistent with those obtained at 95 "C (Table 11).This indicates that as long as samples are digested by the proposed method at temperatures between 70 and 100 "C excellent recoveries of total mercury from fish samples are obtained. TABLE I1 COMPARISON OF MERCURY RECOVERED FROM FISH SAMPLES DIGESTED UNDER DIFFERENT CONDITIONS Mercury found*/pg g-l Sample 'HC1 present? HC1 presentt HC1 absent:' NBS Albacore tunas . . . . 0.94 f 0.05 0.94 f 0.05 0.74 f 0.06 Barramundi . . . . 0.68 & 0.02 0.70 & 0.02 0.57 f 0.02 * Mean f standard deviation. 7 Samples digested by HNO - H,SO - HC1 a t 95 and 70 "C respectively for 0.5 h. Samples digested by HNO - H,04 a t 70 "C for 0.5 h. 3 NBS stipulated value16 = 0.95 f 0.1 p g g-l. On the other hand when fish samples were digested at 70 "C in a mixture of nitric (5 ml) and sulphuric acids (2.5 ml) without the presence of hydrochloric acid recoveries of mercury were significantly lower (Table 11).Improved recoveries of mercury were obtained only when the temperatures of digestion were increased to 85-100 "C and the time extended to 2 h (Table 111). Mercury results for fish samples digested by this method were consistent with those of samples digested by the proposed method which incorporated the use of hydro-chloric acid (Table 111). The results shown in Table I11 indicate that hydrochloric acid has TABLE I11 COMPARISON OF MERCURY RECOVERED FROM FISH SAMPLES DIGESTED BY THE PROPOSED METHOD IN THE ABSENCE OR PRESEKCE OF HYDROCHLORIC ACID Mercury found*/pg g-l Sample 'HCl present? HC1 absent:' NBS Albacore tunas .. . . 0.94 f 0.05 0.93 & 0.06 Barramundi . . . . 0.68 & 0.02 0.68 f 0.03 Perch . . . . . . 0.53 f 0.02 0.51 f 0.01 Tuna . . . . . . 0.32 f 0.01 0.35 f 0.02 Gem fish . . . . . . . . 0.32 f 0.02 0.39 f 0.01 Canned tuna . . . . 0.12 f 0.01 0.13 f 0.01 * Mean f standard deviation. Samples digested by HNO - H,S04 a t 85-100 "C for 2 h. s NBS stipulated value16 = 0.95 f 0.1 p g g-l. Samples digested by HNO - H,SO - HC1 at 70-100 "C for 0.5 h 1316 LOUIE DETERMINATION OF TOTAL MERCURY Analyst Vol. 108 not caused any detectable mercury loss. It appears that hydrochloric acid is essential to speed up the digestion process and also allows the effective temperature of digestion to be lowered. The decrease in digestion time for fish samples by this new method is presumably due to the presence of chlorine and nitrosyl chloride both very powerful oxidants in the sulphuric - nitric - hydrochloric acid mixture used in the digestion process.The fact that hydrochloric acid does not cause significant mercury loss was also verified in another experiment when various amounts (0.3-0.9 g) of NBS Albacore tuna were digested by the proposed method. Effective condensers were used during thc digestion of one set of samples to prevent the loss of mercury while another set of samples was digested without using condensers. No significant difference in the recovery of mercury was detected. X -L X 1 Fig. 1. Comparison of absorbances for 0.4 pg of mercury where the standard mercury working solutions (0.1 pgml-l) were prepared in acids of different con-centrations.A 1 M HC1; B 0.1 M HCI; C 1 M HNO,; D 1 M H,SO,; E 0.5 M H,SO,; and X “pre-reduction” peak. Stability of Standard Mercury Solutions During the course of this work instability of standard mercury solutions in some acids was noted. The absorbances for 0.4 pg of mercury prepared in different acid solutions are shown in Fig. 1. The absorbance for the mercury standard prepared in 1 M hydrochloric acid was found to be significantly higher than those prepared in 1 M nitric and 0.5 M sulphuric acids, solutions commonly used for this purpose.192J2-15 In contrast to previous report^,^^^^^ which assumed that higher observed absorbances were due to mercury contamination of hydro-chloric acid no mercury was detected in 20 ml of a 1 M hydrochloric acid blank.Further, the absorbances obtained for mercury over a range of 0.1-0.5 pg in 20 ml of 1 M hydrochloric acid were consistent with those acquired in 20 ml of the diluting acid solution consisting of nitric and sulphuric acids and water. It has thus been demonstrated that hydrochloric acid does not give an enhanced signal response. In addition to showing lower absorbances the mercury standards prepared in acids other than 1 M hydrochloric yielded small “pre-reduction” peaks (Fig. 1) when small aliquots of these standard solutions were flushed with nitrogen prior to the addition of tin(I1) chloride. This would suggest that mercury(I1) chloride solutions in these acids are unstable. Such instability of dilute solutions of niercury(I1) salts has been previously notedla and ascribed to the virtual impossibility of totally excluding reducing substances.In this work we have found that tin(I1) chloride is strongly absorbed on to glass and can only be completely removed with a hot mixture of hydrochloric acid and nitric acids. The greater stability of dilute mercury standards in 1 M hydrochloric acid discussed above is possibly due to the existence of mercury as HgCl,“ at this acid concentration. This species is more stable t November 1983 IN FISH AN IMPROVED METHOD 1317 reduction than Hg2+ the mercury species predominating in mercury( 11) salt solutions pre-pared in other acids. Thus it is recommended that standard solutions for mercury determination are prepared in 1 M hydrochloric acid. Our data suggests that other dilute mercury solutions are susceptible to reduction by minute amounts of tin(I1) chloride absorbed on glass.This behaviour of dilute solutions of mercury is being investigated further. Conclusion The digestion time has been significantly reduced by inclusion of hydrochloric acid in the acid digestion mixture and preparation of mercury standard solutions in 1 M hydrochloric acid ensures greater stability of the standard solution. Two factors in the method discussed have resulted in an improved method. The author thanks the Australian Government Analyst Mr. R. C. Norris the Regional Director Mr. C. A. Young and Senior Staff of AGAL New South Wales for their encourage-ment and support. Special thanks are extended to Mr. J. O’Callaghan and Mr. A. P. Lim who carried out much of the experimental work.1 . 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. References Fricke F. L. Robbins W. B. and Caruso J. A. Pvog. Anal. A t . Spectrosc. 1979 2 257. “Official Methods of Analysis of the Association of Official Analytical Chemists,” Thirteenth Edition, Analytical Methods Committee Analyst 1977 102 769. Duve R. N. Chandra J . P. and Singh S. B. J . Assoc. 08. Anal. Chem. 1981 64 1027. Sullivan J. R. and Delfino J. J. J . Environ. Scz. Health Part A 1982 17 265. Gorsuch T. T. Analyst 1959 84 136. Gorsuch T. T. “The Destruction of Organic Matter,” Pergamon Press Oxford 1970 pp. 79-84 Analytical Methods Committee Analyst 1960 85 643. Analytical Methods Committee Analyst 1965 90 515. Rains T. C. and Menis O. J . Assoc. 08. Anal. Chem. 1972 55 1339. Feldman C. Anal. Chew. 1974 46 1606. Teeny F. M. J . Agric. Food Chem. 1975 23 668. Smart N. A. and Hill A. R. C. Analyst 1969 94 143. Thorpe V. A. J . Assoc. 08. Anal. Chew. 1971 54 206. Hatch W. R. and Ott W. L. Anal. Chew. 1968 40 2085. “Report on the Investigation for Research Material 50 Albacore Tuna,” National Bureau of Sandell E. B. “Colorimetric Determination of Trace Metals,” Third Edition Interscience New Toribara T. Y. Shield C. P. and Koval L. Talanta 1970 17 1025. The Association of Official Analytical Chemists Washington DC 1980 p. 405. and 138-143. Standards Washington DC 1971. York 1959 p. 625. Received April 12th 1983 Accepted May 26th 198

ISSN:0003-2654    

DOI:10.1039/AN9830801313

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

1318 Analyst November 1983 Vol. 108 $9. 131 8-1322 Electrostatic Capture of Gaseous Tetraalkyllead Compounds and their Determination by Electrothermal Atomic-absorption Spectrometry Giancarlo Torsi and Francesco Palmisano Difiartimento di Chimica Laboratorio di Chimica Analitica V i a Amendola 173 70126 Bari Italy The sampling of tetraalkyllead (TAL) compounds in artificially polluted air by electrostatic capture and their subsequent determination by electro-thermal atomic-absorption spectrometry is described. This technique, coupled to a low cost portable battery-powered field sampler could provide a fast simple and sensitive method for TAL determination in environ-mental samples. Detection limits are of the order of 100 pg of lead. Keywords Tetraalkyllead determination; electrothermal atomic-absorption sfiectrometry ; electrostatic capture ; air analysis The “electrostatic accumulation furnace for electrothermal atomic spectrometry” (EAFEAS) te~hniquel-~ has been successfully used for the precise simple and fast determination of lead and mercury in air samples.I t has also been shown that the electrostatic accumulation furnace can be easily operated with a collection efficiency of approximately lOOyo with particu-late matter as well as with mercury vapour The high capture efficiency of mercury atoms, even if not ~nexpected,~ represents a remarkable indication of the potentialities of the proposed method and suggests that gaseous compounds other than mercury vapour could also be tested. The area of organometallics in general and of tetraalkyllead (TAL) in particular is of great interest because of the wide diffusion and the high toxicity of this class of compounds.Methods such as direct atomic-absorption spectr0metry,~96 flame ph~tometry,~ gas chroma-tography with electron-capture detecti~n,s,~ gas chromatography with catalytic hydrogenation pre-derivatisation and flame-ionisation detectionlOJ1 and gas chromatography with microwave plasma detection12 are unsuitable for TAL determination in environmental samples because of the lack of specificity and/or sensitivity or because of the long time required for the analysis of a single air sample. Most of the methods with high sensitivity and low detection limits presently available13-19 for the determination of TAL in air samples are based on a sampling step often performed by cryogenic trapping followed by an analysis step by gas chromato-graphy combined with atomic-absorption spectrometric (both flame and electrothermal) detection.These methods determine the TAL compounds present but are time consuming, require a gas chromatograph on-line with an atomic-absorption spectrometer and have a relatively complex sample collection step. A critical review concerning the chemical and instrumental problems associated with the measurement of TAL compounds in air can be found in reference 20. The aim of this paper is to evaluate the potential of the EAFEAS technique for the capture and measurement of TAL in artificially polluted air samples. Experimental The instrumental and experimental set-up are those described previously,2 unless specified A deuterium background corrector was used throughout the peak height being the measured otherwise.quantity. Electrostatic Accumulation Furnace I t is com-posed of a brass block A which screws on to C fixing the stainless-steel sheet B (0.2 mm thick-ness) between A and C. The brass blocks C and E are screwed on to a glass-reinforced PTFE ring D. The ring material can be substituted with high-temperature ceramic material such as Aremco ceramic Type 502-1100. Radial holes I are machined in the brass blocks C and E to An exploded view of the electrostatic accumulation furnace is shown in Fig. 1 TORSI AND PALMISANO 1319 E l D I A B I Fig. 1. Exploded view of the electrostatic For a detailed description accumulation furnace.see text. facilitate the purging and cooling of the furnace. The graphite tube G can be fixed tightly on to the furnace axis by means of the graphite rings F and H because the ring F can slide inside C against B while the ring H is screwed on to E. To obtain better data reproducibility using different furnaces a torque wrench has been used in screwing H producing in this way a reproducible force between F and H. The stainless-steel sheet B acting as a spring allows the expansion of the graphite tube G without unduly increasing the thrust against F and H. In this way the possibilities of breaking the furnace are practically eliminated. Most of our furnaces have been used for more than 100 firings without significant changes in their response owing to the low atomisation temperature needed for lead.Sampler This device has the function of accommodating the furnace described previously for field collection of air samples of known volume. The analyte accumulates on the walls of the graphite tube inside the furnace by electrostatic pre~ipitation.l-~ All the components of the sampler are contained in a small hard suitcase which during the sampling step can be held in place on a tripod for photographic use. The sampler is basically composed of the following (see Fig. 2) a battery-powered personal pump (for instance an SIPIN SP-1) ; a high voltage power source (Analog Modules Inc. Longwood FL USA) ; an electrostatic precipitator (see also Fig. 3) ; a microammeter for the control of the current flow-ing in the precipitator; a current stabiliser; a timer for the control of the sampling time; a series of batteries for the powering of the current stabiliser and the high voltage power source.Two flow lines allow a fast sampling rate for particulate matter and a low rate for volatile molecular species. Air outlet t Damping device power supply precipitator Fig. 2. Block diagram of the air sampler. Electrostatic Precipitator This device (Fig. 3) serves to accommodate the furnace and to introduce and evaporate efficiently a TAL solution under a stream of air without contamination from (inorganic) lead associated with particulate matter. In Fig. 3 A is the body (PTFE) of the furnace container cup B is a Pyrex tube C is a tightly fixed silicone-rubber septum and D is a Gelman typ 1320 TORSI AND PALMISANO ELECTROSTATIC L H I CAPTURE Analyst Vol.108 M G I I I I Fig. 3. see text. Schematic view of the electrostatic precipitator. For a detailed description filter-holder accommodating a Gelman GA-6 0.10-pm Metricel membrane filter E which ensures the practical removal of lead associated with the particulate matter. (The volume of air typically sucked was 300-400 cm3 and did not show a detectable signal for lead.) F is a PTFE tip machined in the cup A 10 mm long 2 mm 0.d. and 1 mm i.d. which protudes about 5 mm inside the graphite tube G when the cup A is screwed on to the furnace container H. This arrangement will bring the TAL vapours well inside the graphite tube thus minimising the amount of analyte absorbed by diffusion at the entrance and in the first section of the graphite tube.I is a tungsten tapered wire mounted precisely along the axis of the furnace and L is a spring contact that ensures the electrical connection with the furnace G via the metal block M. Reagents and Solutions Gasoline has been used as a convenient and easily obtainable source of TAL compounds. The total TAL content (390 mg 1-l) in gasoline has been determined according to a standard procedurels by refluxing the gasoline with concentrated hydrochloric acid to convert the organic lead into the inorganic form which was subsequently determined by polarography and atomic-absorption spectrometry. Standard solutions (0.39 ng pl-l) of TAL were pre-pared by diluting the gasoline with light petroleum (b.p. 80-120 "C) (Carlo Erba analytical-reagent grade).Amounts ranging from 1 to 6 p1 of this solution were injected with a Hamilton syringe and vaporised in tlie device as shown in lig. 3. The purity of the light petroleum used was ascertained by vaporising 6 pl with tlie electrostatic furnace on; no detectable signal for lead was found. A stock solution of lead(I1) (0.39 g 1-1) was prepared from lead(I1) nitrate (Carlo Erba, analytical-reagent grade) acidified to pH 1 with concentrated riitric acid (Carlo Erba analyti-cal-reagent grade) and stored in polyethene bottles. Dilute solutions were prepared with doubly distilled water just before use. Results and Discussion An air sample containing known amounts of TAL compounds was simulated by using the device shown in Fig. 3 and the field sampler described in Fig.2. An approximately known volume of air was sucked through the membrane E (in order to exclude lead from the particles) with the voltage on. The flow-rate (approximately 0.8 cm s-l corresponding to an average linear velocity of approximately 10 cm s-l) and the potential applied (1.8-2.0 kV) were in the range of maximum capture efficiency.2 At the same time a known volume of a standard TAL solution was slowly ejected on to the walls of the Pyrex connector B which was heated to about 60 "C (with a hot air gun) to facilitate ihe vaporisation of the solution in the air stream and tlie desorption13 of tlie TAL compounds from the walls of the connector itself. Consider-ing the amounts involved in these experiments (300-400 cm3 of filtered air sucked 0.2-2 ng of TAL injected) and the typical TAL content of laboratory air,13 it is obvious that the amount of TAL introduced into the air stream represents in the most unfavourable conditions only 176 of the injected TAL.I t can therefore be assumed that interferences from organic and inorganic lead in tlie air stream used to vaporise the solution of TAL in light petroleum can be disregarded. No attempt has been made to discriminate possible different behaviour amongst the different TAL that can be present in gasoline.1 November 1983 OF TETRAALKYLLEADS AND ELECTROTHERMAL AAS 1321 Calibration Calibration of the apparatus can be performed by vaporising as just described different volumes of a standard light petroleum solution of known TAL content. A calibration graph was plotted over the mass range 0.19-2.34ng of lead by using four different furnaces (25 points in total).A second-order regression analysis was used to fit the experimental points; the regression equation was as follows: Absorbance (a.u.) = 0.011 + 0.591C - 0.090C2 . . . . - . (1) where C is expressed in nanograms of lead. The correlation coefficient was found to be 0.9981 and the standard error of the fit was 0.020 a.u. The coefficient of variation (n = 10) at 0.19 ng was about 17%. On this basis an absolute detection limit (signal to noise ratio = 3) of about 100 pg of lead could be calculated. Capture Efficiency To evaluate the TAL capture efficiency of the proposed electrostatic precipitator t.wo series of measurements (ten replicates in each series) were performed at the same lead level (0.73 ng), the first using an aqueous Pb2+ solution introduced directly into the furnace as described in reference 3 and the second using TAL as described above.Because the TAL standard solution is prepared by diluting gasoline whose content was determined after a digestion step the possibility that a small fraction (1-4%) of the total lead introduced can be present in an ionic (non-volatilisable) form such as Pb2+ PbR,+ and PbR22+ cannot be excluded.16 This means that with the present experimental design an effective complete capture could be represented also by a result less than 100. The first and the second set of measurements gave a response of 0.450 & 0.076 and 0.415 & 0.059 a.u. respectively. The two mean responses were not significantly different according to a t-test at the 95% confidence level.This suggests that as for particulate matter1 and mercury vapour,2 the capture efficiency of the TAL species is near 100 yo and therefore the calibration graph should be independent of the relative amounts of different TAL species present in the sample. However as the precision of the measurements is poor the above statement should be taken with caution. Further experimental evidence for the near complete capture of TAL com-pounds has been gathered using an experimental design in which an additional furnace is placed downstream from the main electrostatic precipitator as shown in Fig. 4. The connec-tion between the two furnaces was made by a short piece (1-2 cm) of PTFE tubing to minirnise wall capture.With such an arrangement no signal should be observed in the second furnace, TAL solution [ - - e r r inlet C voltage -High-voltage source source Suction Pump Fig. 4. Schematic diagram of the experimental set-up used to verify the capture efficiency of TAL compounds. A Filtering membrane; B and C main and secondary electrostatic precipitators respect-ively 1322 TORSI AND PALMISANO whatever the amount of TAL introduced provided that all the analyte is captured in the first furnace. In the most unfavourable examples signals around 3 4 % of those shown by the first furnace were detected. This has been practically verified. Conclusion This method has the potential for measuring TAL (and probably any other volatile form of lead compounds) with speed low sample manipulation and low-cost instrumentation.Con-sidering the detection limits given above and the flow-rate at which the pump of the field sampler is normally operated typical sampling times should be about 2 h for urban areas21 and around 10 min for more polluted areas such as a gasoline station. However the applicability of the proposed method to real samples depends on the possibility of finding a filtering mem-brane that excludes efficiently the lead associated with particulate matter which is by far the predominant source of lead in air20s22p23; verifying that the capture efficiency is nearly quantita-tive also at the concentration of TAL normally present in real samples; and verifying that the filter membrane and the collection apparatus retain only a negligible fraction of TAL com-pounds.Work is in progress in these directions. The Minister0 della Pubblica Istruzione and the Consiglio Nazionale delle Ricerche are gratefully acknowledged for the financial support of this work. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. References Torsi G. Desimoni E. Palmisano F. and Sabbatini L. Anal. Chem. 1981 53 1035. Torsi G. Desimoni E. Palmisano F. and Sabbatini L. Analyst 1982 107 96. Torsi G. Palmisano F. Desimoni E. and Rinaldi R. Ann. Chim. 1982 72 365. Newman 0. M. G. and Palmer D. J. Nature (London) 1978 526 275. Thilliez G. Anal. Chem. 1967 39 427. Kolb B. Kemmner G. Schesler F.H. and Wiedeking E. Fresenius 2. Anal. Chern. 1966 221 166. Mutsaars P. M. and Van Steen J . E. J . Inst. Pet. London 1972 58 102. Cantuti V. and Cartoni G. J . Chromatogr. 1968 32 641. Tausch H. Ber. Oesterr Studienges. Atomenerg. 1979 SGAE No. 2636. Soulages N. L. Anal. Chem. 1966 38 28. Soulages N. L. J . Gas Chromatogr. 1968 6 356. Reamer D. O. Zoller W. H. and O’Haver T. C. Anal. Chem. 1978 50 1449. De Jonghe W. R. A. Chakraborti D. and Adams F. C. Anal. Chem. 1980 52 1974. De Jonghe W. R. A. Chakraborti D. and Adams F. C. Anal. Ckim. Acta 1980 115 89. Chan Y . K. Wong P. T. S. Bengert G. A. and Kramar O. Anal. Chem. 1979 51 186. Chakraborti D. Jiang S. G. Surkijn P. De Jonghe W. R. A. and Adams; F. C. Anal. Proc., Coker D. T. Anal. Chem. 1975 47 386. “ASTM Standards on Petroleum Products and Lubricants,” American Society for Testing and Coker D. T. Ann. Occup. Hyg. 1978 21 33. De Jonghe W. R. A. and Adams F. C. Talanta 1982 29 1057. De Jonghe W. R. A. and Adams F. C. Atmos. Environ. 1980 14 31. Harrison R. M. Perry R. and Slater D. M. Atmos. Environ. 1974 8 1187. Rohbock E. Georgii H. W. and Muller J. Atmos. Environ. 1980 14 89. 1981 18 347. Materials Philadelphia 1967 Method D 526-61. Received February Sth 1983 Accepted June 6th 198

ISSN:0003-2654    

DOI:10.1039/AN9830801318

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst November 1983 VoZ. 108 pp. 1323-1330 1323 Internal Standards for Simultaneous Multi-element Analysis in Inductively Coupled Plasma Atomic-emission Spectroscopy with an Electrothermal Atomiser for Sample Introduction Steven D. Hartenstein Hassan M. Swaidan and Gary D. Christian Department of Chemistry University of Washington Seattle WA 98195 USA Elemental internal standards were obtained for 14 elements in inductively coupled plasma atomic-emission spectroscopy with a modified Perkin-Elmer HGA 2000 graphite furnace sample introduction system. Internal standards were used to correct for errors due to changes in the flow-rate of the argon carrier gas the observation zone the observation period the sample volume and tube deterioration. Precisions for 10-,uI aliquots of 1 p.p.m.solutions of calcium cobalt manganese and nickel were improved by 40%. Emission responses revealed that internal standards will correct for both random and deliberate systematic operation fluctuations without re-calibration of the system. Keywords Internal standards ; multi-element analysis ; inductively coupled piasma atomic-emission spectroscopy ; electrothermal atomisation Simultaneous multi-element analysis has been achieved in part owing to inductively coupled plasma atomic-emission spectroscopy (ICP-AES) .1-5 Low detection limits and wide linear dynamic ranges make the ICP a valuable analytical tool for major minor and trace metal analysis. Several workers obtained even lower detection limits by connecting an electrothermal atomiser (EA) to a plasma source.6-12 Besides lower detection limits an EA-ICP system also offers the advantage of using micro-amounts of sample.In an EA-ICP the electrothermal atomiser produces a cloud of atoms molecules or particles that is carried to the plasma where atomisation is completed and atoms are excited in the plasma discharge. Crabi et aL9 described the characteristics of the transient emission signal from this cloud. The analytical signal showed a “piston effect” from the sudden expansion of gases in the atomiser and then a rapid pulse as the sample vapour was swept into the plasma. The emission peak was more reproducible as the carrier gas flow-rate was increased. Aziz et at?.’ showed that the analyte line to background intensity ratio increased with increased flow but results were less reproducible at higher flow-rates.Swaidan and ChristiaP studied the transient signal by varying the carrier gas flow-rate and integrating the emission signal to determine the optimum response versus flow-rate. The dependence of the net emission intensity on flow-rate and other parameters invokes the need for an internal standard to correct for such factors as the random fluctuations in argon carrier gas flow-rate. Internal standards are used in many analytical techniques to lower the random error from variations in sample introduction and instrumental para-meters.14-19 Internal standards have also been utilised in flame20 and argon inductively TABLE I ICP OPERATING CONDITIONS Instrument r 1 Jarrel-Ash Model 955 Plasma Atomcomp Model 955 direct-reading polychromator Parameter Condition Parameter Condition A I 1 Coolant argon flow-rate - 7 .. 20 1 min- Dispersion . . 0.55 nm mm-l Plasma argon flow-rate . . 0.0-0.5 1 min-f Entrance slit width (fixed) . . 25 p m Incident rf power . . 1kW Exit slit width (fixed) . . 50 pm Reflected rf power . . <2ow Observation height above coil (variable) . . 10-20m 1324 Analyst VoZ. 108 coupled plasma spectros~opy.~l-~~ Barnett et aZ.21j22 established the criteria to be considered when selecting a suitable internal standard for an analyte. Their theoretical model considered the effects of different excitation and ionisation energies partition functions and electron densities on the emission intensity of the analyte and internal standard lines. They found that the model increased the probability of choosing two compatible elements and lines but involved difficult calculations which did not guarantee success.HARTENSTEIN et a,?. STANDARDS FOR MULTI-ELEMENT As Ba I1 Ca I1 Cd c o I1 Cr I1 c u Fe I1 TABLE I1 ELEMENTS AND MONITORED WAVELENGTHS Element Wavelengthlnm 328.0 308.2 . . . . 193.6 493.4 396.8 228.8 . . . . . . 228.6 . . * . . . 205.5 . . 324.7 259.9 Element Wavelengthlnm Mg I1 . . . . 279.5 Mn I1 . . 257.6 Na . . . . . . 589.0 Ni I1 . . . . 231.6 P b I1 . . . . 220.3 Sb . . 217.5 Se . . 196.0 Si . . . . . . 251.6 211 . . . . . . 213.8 Since the development of this EA-ICP system,12 applications to multi-element analysis have been performed by integrating transient emission signals in a multi-channel mode.13 We have however noted several areas in which an internal standard could improve precision.The amount of analyte viewed during the observation in the plasma is dependent on several factors the amount of sample injected into the graphite tube the rate at which the sample is carried to the plasma the amount of sample retained on the walls of the graphite tube and connecting chamber and the time and duration of observation.13 Because of all the parameters involved in selecting an acceptable internal standard we chose to compare empirical responses of multi-element samples to determine which elements were suitable as internal standards for particular analyt e elements. 1.0 2.0 3.0 4.0 125 105 85 65 45 10 11 12 13 14 L 0 4- Argon flow/l min-' Height above coil/mm .-5 1000 800 600 z 400 200 n " 2 3 4 5 6 Arrival timeis 120 100 80 60 PO 5 6 7 8 9 Exposure ti me/s Fig 1.Emission response (arbitrary units) of aluminium, cadmium chromium manganese nicltel and zinc as the operat-ing conditions were changed (a) argon carrier gas flow-rate; (b) observation height above the induction coil; (c) time period between the furnace atomisation cycle and measurement period : and (d) iiieasurenieiit time period Novembev 1983 Apparatus ANALYSIS IN ICP-AES WITH ELECTROTHERMAL ATOMISATION Experimental 1325 The modified Perkin-Elmer HGA 2000 graphite electrothermal atomiser connected to the torch of a Jarrel-Ash Model 955 Plasma Atonicomp inductively coupled plasma spectro-meter has been previously described.12 The graphite tubes were pyrolytically coated at a charring temperature of 2 100 "C for 10 min with a loo/ methane - 90% argon gas mixture.The gas flow-rate for this was set to the lowest possible setting (approximately 0.05 1 min-l). Samples were injected into the tube with either a 10-pl Beckman Accustroke or an adjustable 5-50-pl Finnpipette with Beckman disposable polypropylene pipette tips. Operating conditions of the ICP and Model 955 direct-reading polychromator equipped with a 0.5-m Ebert monochromator are listed in Table I. The system is able to monitor the signals of 25 elements simultaneously. An ASR 33 Teletype was used to command the PDP Sla mini-computer to monitor channels corresponding to specific emission lines (Table 11).Software allowed the user to programme the exposure time (the time the PMT of a channel is exposed to the emission signal) for each sample. Digital readouts of relative intensity were recorded by the teletype. Reagents Calibration standards and sample solutions were prepared on the day they were used by serial dilution of 1 000 p.p.m. standard stock solutions from Fischer Scientific Company. De-ionised water from a Millipore MilliQ water purification system (resistance > 7 MQ) was used in all standards and blanks. Sample solutions were kept in polypropylene bottles. 01 I I I 1.0 2.0 3.0 4.0 N i/M n 100 0 u2-.-10 11 12 13 14 Argon flow/l rnin-' Height above coiI/rnrn 250 1 PI > C c 200 150 .-4- 2 100 50 250 200 150 100 50 Ah Ni I I L 2.0 3.0 4.0 5.0 6.0 5.0 6.0 7.0 8.0 9.0 Arriva I ti rn e/s Exposure tirne/s 250 I I 200 150 100 50 0 0 5 10 15 20 Sample volume/pI Fig.2. Emission response (arbitrary units) of nickel and its internal standard ratio with chromium and manganese as the operating conditions were changed (a)-(d) as in Fig. 1 ; and (e) volume of sample injected into the furnace 1326 AnaZyst VoZ. 108 Procedure The argon carrier gas flow-rate observation height above the coil arrival time and exposure time were adjusted to the optimum conditions determined previ0us1y.l~ The plasma was then allowed to stabilise. The HGA 2000 burn programme was set for a 20-s drying cycle at 400 "C a 5-s charring cycle at 800 "C and a 10-s atomisation cycle at 2400 "C.The system was then equilibrated and cleaned by running several complete burn programmes. Finally several blanks were burned and recorded until consistent intensity readings were recorded. Impurities retained on the walls of the system were cleansed by running one complete cycle with only argon carrier gas flowing through the furnace. This burn was necessary to prevent any memory effects from poisoning of the graphite tube. After a 10-pl injection of the blank the second burn was recorded. Lastly a 10-pl aliquot of sample was injected into the furnace and its emission recorded. HARTENSTEIN et al. STANDARDS FOR MULTI-ELEMENT The system was then considered ready for sample analysis.Each sample was analysed in a set of three furnace burns. 400 300 200 100 1.0 1.5 2.0 2.5 3.0 12 13 14 15 16 Argon flow/l min-' Height above coil/mm 400 1 Co/Fe 250 200 2 3 4 5 6 4 5 6 7 8 Arrival time/s Exposure ti me/s 700 600 500 400 300 200 100 0 0 5 10 15 20 Sample volu me/@ Fig. 3. Emission response (arbitrary units) of cobalt and iron and their ratio as the operating conditions were changed (a)-@) as in Fig. 2. Calculations in the blank (IB,A) from the signal of the analyte in the sample (IA) for each analyte: The net intensity of the analyte (I'A) was calculated by subtracting the signal of the analyte (1) = I A - IB,A . . The net signal intensity of the internal standard (I's) was calculated similarly November 1983 ANALYSIS IN ICP-AES WITH ELECTROTHERMAL ATOMISATION 1327 I f = I - I 1 .. . . . The internal standard ratio correction of the analyte net intensity (IR) was determined from the equation * * (3) I = IfA/Ifs . . Data were processed on an Apple I1 Plus computer with software programs Curve Fitter and Scientific Plotter (Interactive Microware Inc. State College PA). Graphs were printed on an Epson-MX 80F/T printer with graphics capability. 400 300 200 100 1.0 1.5 2.0 2.5 3.0 Argon flow/l min-' ~ 400 5 300 4- .- fn 4- .-z 200 z 100 I I I I 2 3 4 5 6 Arrival time/s 120 r 250 200 150 12 13 14 15 16 Height above coil/mm 4 5 6 7 8 Exposure ti me/s 0 5 10 15 20 Sa rn ple vo I u me/pI Fig. 4. Emission response (arbitrary units) of antimony and lead and their ratio as the operating conditions were changed (a)-@) as in Fig.2. Results and Discussion Internal standards reduce the random error from instrumental physical and chemical factors which affect the amount of analyte atoms at the measurement site. The changes in net emission intensity were studied as the amount of analyte introduced into the furnace, carried to the plasma and distributed in the observation window was varied. The amount of sample injected into the graphite tube depended on the size of aliquot precision of injection and retention of analyte. The argon carrier gas flow-rate and retention of analyte on the walls of the connecting chamber were determinants in the amount of sample reaching the plasma. Finally the sample concentration in the observation window depended on the time the signal was measured and the area of the plasma monitored.Takada and Nakano26 reported that internal standards improved results with different atomisation temperatures in a furnace but Aziz et a1.' found 2400 "C sufficient to atomise all analytes in a multi-element sample 1328 HARTENSTEIN et al. STANDARDS FOR MULTI-ELEMENT A d y s t Vol. 108 110 >-v) 0) 4- .-E 100 .-4-z 90 80 0 20 40 60 80 100 120 Burn number Fig. 5 . Variation with time of the emission response (arbitrary units) of nickel and its internal standard ratio under optimum operating conditions argon carrier gas flow-rate 1.75 1 min-l observation height above the induction coil 12 mm arrival time waiting period 4 s and measurement period 7 s.In order to determine which elements behave similarly we monitored the net intensity signals of a multi-element solution as the operating conditions were changed. The individual responses of six elements (Al Cd Cr Mn Ni and Zn) are shown in Fig. 1. Our results [Fig. l ( a ) and ( b ) ] agree with earlier work13 in wliicli the optimum conditions for these six elements were found to be an argon carrier gas flow-rate of 1.5-2.0 1 min-l and an observation height above the coil of 10-12 mni. These graphs show that generally the response increases initially with increasing flow-rate but then falls off at higher flow-rates. Flow-rates below 1.5 1 min-l were avoided because condensation accumulation in the connecting chamber extinguished the plasma.The effect on net intensity as arrival time (tlie time from the beginning of the furnace atomisation cycle to the beginning of the observation period) and exposure (observa-tion) time were varied is shown in Fig. l(c) and ( d ) . Chromium manganese and nickel appear to respond similarly suggesting that they would make adequate analyte/internal standard pairs. When chromium or manganese is used as an internal standard for nickel (Fig. Z) the response variation as operating conditions are changed is greatly reduced. Iron and cobalt (Fig. 3) and antimony and lead (Fig. 4) also niake excellent analyte/internal standard pairs the former pair agreeing with reported results which suggests cobalt to be an excellent internal standard for iron in an electrothernial atoniiser.26 Surprisingly we found barium to be a better internal standard than strontium (407.8 nm) for calcium whereas, Salin and H o r l i ~ k ~ ~ were able to use strontium for calcium in the ICP.Difficulties in delivering precise aliquots to the furnace because of damaged pipette tips and normal inconsistency were noted. The effect of aliquot size on tlie response of an element and the effectiveness of an internal standard in correcting for unintentional injection variations are shown in Figs. 2(c) 3(c) and 4(c). Finally Slavin ct aZ.27 reported that the deterioration of the graphite tube is delayed by pyrolytically coating the tube allowing substantially more firings. \I’e found that an internal standard also improved the repro-ducibility of net intensity as the graphite tube ages.Over the course of 100 firings of the TABLE I11 ELEMENTS AND INTERSAL STASDARDS Element Internal standard Element Internal standard . . Si Fe . . . . . . co Ba . . . . Ca 1111 . . . . . . Cr Ca . . Ba Ni . . . . . . Cr Cd . . . . & Pb . . . . . . Sb Pb tf * * . . . . Cd c u . . . . . . co co . . Fe Cu Sb . I Cr . . . . &In Ni Si . . . . . I A November 1983 ANALYSIS IN ICP-AES WITH ELECTROTHERMAL ATOMISATION TABLE IV PRECISION (YoRSD) OF NET INTENSITY FOR ANALYTE ELEMENTS (I’A), INTERNAL STANDARDS ( I t s ) AND INTERNAL STANDARD RATIO CORRECTION (IR) 1329 RSD % No. of r--J- 7 Analyte Internal standard determinations (n) I’A IR Ca* . . . . Ba c o t . . . . Fe Mni . . Cr Nif .. . . Cr Sbf . . . . Pb 5 4.0 9.5 2.2 6 4.8 4.0 1 .o 6 5.5 3.3 2.5 6 3.2 3.3 1.9 6 7.9 4.4 7.1 * Operating conditions flow-rate of argon carrier gas 1.75 1 min-l observation t Operating conditions flow-rate of argon carrier gas 2.0 1 inin-’ ; observation f Operating conditions flow-rate of argon carrier gas 2.5 1 min-l; observation height above coil 12 mm; arrival time 4 s; and exposure time 7 s. height above coil 14 mm; arrival time 4 s; and exposure time 6 s. height above coil 18 mm; arrival time 3 s; and exposure time 7 s. graphite tube the nickel response became more reproducible when chromium was used as an internal standard (Fig. 5). Other elements and their internal standards are listed in Table 111. Solutions of elements and their internal standards were measured and precisions determined.The results of six consecutive measurements are shown in Table IV. The internal standard ratio correction improved the precision measured as the relative standard deviation (Y’RSD) by over 40% for four of the five analyte elements with over a 300% improvement for cobalt with iron as an internal standard. In order to measure the analytical capabilities of our system we prepared five solutions of equal amounts of cobalt manganese nickel and antimony of concentrations over the range 0.050-5.0 p.p.m. Each standard solution contained 1.0 p.p.m. of iron chromium and lead as internal standards. The slope intercept and linear correlation coefficient were calculated from the net intensities for each analyte at three different sets of operating conditions.The analyte responses were linear over two orders of magnitude at all three operating conditions and the internal standard ratio responses gave slightly enhanced linear correlation ( >0.997). We then altered the matrix of our analyte solutions by preparing two sample solutions of unequal concentrations of each analyte and 1.0 p.p.m. of their internal standards. The I t A and I at each composition for each analyte in the mixtures were then used to calculate the empirical concentration from the slope and intercept of the respective calibration graphs. These empirical concentrations at each set of operating conditions are listed in Table V. The results show that an internal standard produces results closer to the true concentrations in the mixtures as we change the composition of the sample.The cobalt calibration graphs for I’A and I, with iron as an internal standard at three TABLE V SIMULTANEOUS MULTI-ELEMENT ANALYSIS Internal ( standard Analyte element element Co . . Fe Bln . . . . Cr Ni Cr Sb . . Ph True Empirical Empirical True :ompromise concentration concentration concentration concentration c operating of analyte from Z’A from ZX of analyte, conditions p.p.m. p.p.m. p.p.m. p.p.m. 1* 0.15 0.19 0.03t 0.18 f 0.03 2.35 0.15 0.17 f 0.02 0.144 f 0.003 2.35 0.15 0.17 f 0.04 0.15 f 0.03 2.35 1.79 f 0.07 1.92 f 0.04 1.25 1.85 21 3§ 2 1.85 1.78 f 0.04 1.97 f 0.08 1.25 i 1.85 1.6 f 0.1 1.73 f 0.07 1.25 1 1.42 1.27 f 0.05 1.35 f 0.02 0.42 2 1.42 1.40 f 0.01 1.54 f 0.01 0.42 7 1.4 1.8 f 0.1 1.9 f 0.2 0.42 1 0.35 0.39 i 0.03 0.39 f 0.04 2.3 2 0.35 0.40 & 0.03 0.42 f 0.03 2.3 3 0.35 0.36 f 0.04 0.36 f 0.04 2.3 Empirical :oncentration from Z‘A, p.p.m.2.19 t 0.03 2.0 f 0.1 2.2 _L 0.4 1.09 f 0.08 1.12 + 0.04 1.1 + 0.2 0.47 j_ 0.07 0.38 + 0.03 0.45 t 0.05 2 1 + 0 1 1.5 4 0.1 2.1 t 0.2 Empirical concentration from In, p.p.m. 2.46 f 0.06 2.29 f 0.07 2.32 f 0.04 1.1 f 0.1 1.34 i 0.08 1.19 & 0.05 0.45 + 0.09 0.43 f 0.01 0.46 f 0.05 2.21 f 0.03 1.6 -k 0.1 2.3 rt 0.1 * Operating conditions flow-rate of argon carrier gas 1.55 1 min-l; observation height abovc coil 12 mm; arrival time 4 s; and exposure f One standard deviation of the mean (n = 3). 1 Operating conditions flow-rate of argon carrier gas 2.0 1 min-I; observation height above coil 14 mm; arrival time 4 s; and exposure 5 Operating conditions flow-rate of argon carrier gas 2.5 I min-1; observation height above coil 18 mm; arrival time 3 s; and exposure time 7 s.time 6 s. time 7 s 1330 HARTENSTEIN SWAIDAN AND CHRISTIAN 1.0 -v) (a) 5 4.0 -w .-0.2 -0.2 ---1.0 -1.3 -0.9 -0.5 -0.1 0.3 0.7 -1.3 -0.9 -0.5 -0.1 0.3 0.7 Log(concentration p.p.m.1 Fig. 6. Log - log calibration graphs of (a) cobalt and (b) cobalt with iron as an internal standard under the following operating conditions : (1) argon carrier gas flow-rate 1.75 1 min-' observation height above induction coil 12 mm arrival time 4 s and exposure time 7 s ; (2) argon carrier gas flow-rate 2.0 1 min-' observation height above induction coil 14 mm arrival time 4 s and exposure time 6 s ; and (3) argon carrier gas flow-rate 2.5 1 min-l observation height above in-duction coil 18 mm arrival time 3 s and exposure time 7 s.different operating conditions are shown in Fig. 6. These figures show that the calibration graphs for I'* of cobalt [Fig. 6 ( a ) ] are highly dependent on the operating conditions whereas the graphs of I are almost superimposable at different conditions. To compare the differ-ence between the two sets of graphs we found the response of a 2.35 p.p.m. cobalt solution at the first set of operating conditions corresponded in Fig. 6(a) to 2.2 p.p.m. on 1 5.8 p.p.m. on 2 and 1.5 p.p.m. on 3. On the other hand the I for the same sample solution at the same operating conditions corresponded in Fig.6(b) to 2.5 p.p.m. on 1 2.4 p.p.m on 2 and 2.4 p.p.m. on 3. Therefore internal standards can be used to correct for random or deliberate changes in operating conditions without re-calibrating the system and allow for simultaneous multi-element analysis at less than optimum operation conditions for each element. Excluding sample and instrument preparation time over 70 determinations of four elements and three internal standards were accomplished simultaneously in 5 h. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. References Kniseley R. N. Fassel V. A. and Butler C. C. Clin. Chem. 1973 19 807. Winge R. K. Fassel V.A. Kniseley R. N. DeKalb E. and Hass Jr. W. J. Spectrochim. Acta, Dahlquist R. L. and Knoll J. W. Appl. Spectrosc. 1978 32 327. Boyko W. J. Keliher P. N. and Malloy J. M. Anal. Chem. 1980 52 53R. Subramanian K. S. and Meranger J. C. Sci. Tot. Environ. 1982 24 147. Gunn A. M. Millard D. L. and Kirkbright G. F. Analyst 1978 103 1066. Aziz A. Broekaert J. A. C. and Leis F. Spectrochim. Acta Part B 1982 37 369. Aziz A. Broekaert J. A. C. and Leis F. Spectrochim. Acta Part B 1982 37 381. Crabi G. Cavelli P. Achilli M. Rossi G. and Omenetto N. A t . Spectrosc. 1982 3 81. Cope M. J. Kirkbright G. F. and Burr P. M. Analyst 1982 107 611. Kirkbright G. F. and Li-Xing Z. Analyst 1982 107 617. Swaidan H. M. and Christian G. D. Can. J . Spectrosc. in the press. Swaidan H. M. and Christian G. D. unpublished data. Willis D. E. Chromatographcia 1972 5 42. McDonald D. C. Anal. Chem. 1977 49 1336. Donohue D. L. Carter J . A. and Franklin J . C. Anal. Lett. 1977 10 371. Kramer G. W. Appl. Spectrosc. 1979 33 468. Cody R. D. and Thompson G. L. Clays Clay Miner. 1976 24 224. Chi Tse-Wen and Yang Tien-Fu FenHsi Hau Hszceh 1979 7 326. Feldman F. J. Anal. Chem. 1970 42 719. Barnett W. B. Fassel V. A. and Kniseley R. N. Spectrochim. Acta Part B 1968 23 643. Barnett W. B. Fassel V. A. and Kniseley R. N. Spectrochim. Acta Part B 1970 25 139. Salin E. D. and Horlick G. Anal. Chem. 1980 52 1578. Uchida H. Nojiri Y. Haraguchi H. and Fuwa K. Anal. Chim. Acta 1981 123 57. Schmidt G. J. and Slavin W. Anal. Chem. 1982 54 2491. Takada T. and Nakano K. Anal. Chim. Acta 1979 107 129. Slavin W. Manning D. C. and Carnrick G. R. A t . Spectrosc. 1981 2 137. Part B 1977 32 327. Received May 4th 1983 Accepted June 201h 198

ISSN:0003-2654    

DOI:10.1039/AN9830801323

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, November, 1983, Vol. 108, pp. 1331-1338 1331 Electrochemical Pre-concentration Technique for Use with Inductively Coupled Plasma Atomic-emission Spectrometry Part I S. E. Long* and R. D. Snookt Trace Analysis Laboratory, Chemistry Department, Imperial College, London, S W7 2A Y The use of a wall-jet electrochemical cell for pre-concentration of trace metals from flowing streams prior to their determination by inductively coupled plasma atomic-emission spectrometry (ICP-AES) is described. The metal of interest is deposited on a glassy carbon electrode held at the reduction potential of the metal. After collection in this way the metal is stripped back into solution by applying an anodic-stripping potential. The resulting metal solution is pumped to the ICP for determination.Keywords : Electrochemical $re-concentration ; inductively coupled plasma - atomic-emission spectrometry ; wall-jet cell ; trace metals The inductively coupled plasma (ICP) has been well characterised as a sensitive source for atomic-emission spectrometry (AES) . The principal advantages of ICP-AES over other atomic spectrometric techniques are its high powers of detection, simultaneous multi-element analysis capability, freedom from many types of interferences and wide dynamic ranges of calibration, with respect to analyte concentration. For very low concentrations of trace elements (< 1 .O ng cm-3), however, conventional ICP-AES using pneumatic nebulisation for sample introduction is not adequately sensitive compared with graphite furnace atomic- absorption spectrometry,l neutron-activation analysis2 or electrochemical techniques.3 Adequate detection power can be achieved using electrothermal vaporisation for sample i n t r o d u c t i ~ n ~ ~ ~ but a number of interferences arise from the vaporisation and transport of the analyte to the plasma which have yet to be elucidated fully.6 The alternative to increasing the detection power of the ICP-AES system is to pre-concen- trate the elements of interest prior to their determination.The standard methods of pre- concentration from solutions rely upon the evaporation of solvents, complex - chelate extrac- tion, surface adsorption, ion exchange and electrodeposition procedures. It is the last that is considered most suitable for pre-concentration prior to determinations using the ICP, because the procedures are relatively fast compared with chemical pre-concentration procedures and the technique can be automated to provide continuous pre-concentration suited to the applica- tions of ICP-AES.Earlier workers have reported the use of electrodeposition techniques for pre-concentration of trace elements prior to their determination by atomic spectrometry. Thus Lund and Larsen71a have published papers concerning the electrochemical pre-concentration of trace elements before their determination by electrothermal atomic-absorption spectrometry. Similarly, Holen et aZ.9 pre-concentrated selenium electrochemically prior to its determination by vaporisation into an argon - hydrogen flame. These procedures have markedly improved the detection power of the spectrometric techniques employed but are not suitable for use with an ICP nebuliser system, which requires a continuous sample uptake rather than discrete sample introduction.Consequently, we have developed, and report here, an electrochemical pre-concentration technique for continuous sample introduction. The technique employs an electrochemical cell that is suitable for automation and is capable of unattended operation. Other advantages are that trace elements may be removed from solutions of high dissolved solids content and can be subsequently introduced into the ICP in a matrix-free solution to avoid clogging the nebuliser and to avoid the production of stray light interferences in the spectrometer of the ICP system. * Present address: School of Chemistry, Georgia Institute of Technology, Atlanta, GA, 30332, USA.7 To whom correspondence should be addressed.1332 LONG AND SNOOK: ELECTROCHEMICAL Analyst, Vol. 108 The electrochemical cell used here is of the wall-jet design.1° The solution being analysed is pumped into the cell in which an electrode is held at the reduction potential of the element of interest. Over a period of time sufficiently high amounts of the element for detection are accumulated upon the electrode. Subsequently the potential of the electrode is scanned in an anodic direction to rapidly strip the accumulated metal back into an electrolyte solution, which is pumped to the nebuliser of the ICP. Many variables influence the deposition efficiency and hence the pre-concentration factors attainable. These variables have been optimised in this study for the element copper and the feasibility of multi-element determinations using this pre-concentration technique has been assessed.Experimental The design of the wall-jet cell is shown in Fig. 1. Instrumentation The principle of operation is that a laminar flow of solution is delivered perpendicularly to the working electrode. The stream then flows radially across the electrode to the downstream counter and reference electrodes. This configuration allows the hydrodynamic conditions to be predicted and controlled. The cell was constructed in our own workshops and is demountable to facilitate periodic mechanical polishing of the electrodes. The lower part consists of the wall of the cell that has, at its centre, a glassy carbon electrode of 5-mm diameter.To one side of the glassy carbon working electrode is a platinum disc counter electrode and to the other side a silver - silver chloride reference electrode. These electrodes are therefore effectively downstream of the working electrode. It is of great importance that the plane of the electrodes is perfectly flat and free from imperfections to maintain laminar flow of the solution across the surface. The upper part of the cell consists of the solution jet forming nozzle and the solution exit ports of the cell. The diameter of the nozzle is critical and is calculated to maintain a laminar flow of solution to the working electrode (0.2 & 0.05 mm). The potentiostat for accurate control of the working electrode potential with respect to the reference electrode consists of a simple operational amplifier arrangement, which is also shown in Fig.1. It is constructed of Perspex in two parts. (a) Inlet nozzle Solution Auxi I ia ry solution input Porous plug exit I I I 1 4". AgCI) electrode [WE) (glassy carbon) Fig. 1. (a) Vertical cross-section through a wall-jet cell ; ( b ) potentiostat arrangement. ICP-AES Instrumentation The ICP-AES instrumentation has been fully described previously,ll but for convenience is described here briefly. The ICP high-frequency generator is of the crystal-controlled type operating at 27.12 MHz and has a continuously variable power output of 0-2.5 kW (Inter-November, 1983 PRE-CONCENTRATION FOR USE WITH ICP-AES.PART I 1333 national Plasma Corporation, Model 120-27). The output of the generator is fed to a manu- ally controlled impedance matching network before termination at a 1 .&turn water-cooled load coil. Sample introduction is effected using a concentric glass nebuliser (J. E. Meinhard, Model T-230-A3) and a double-pass spray chamber. Radiation from the plasma is focused on to the entrance of a 1-in monochromator equipped with a 1200 lines per millimetre plane grating blazed at 300 nm, which provided a reciprocal linear dispersion of 8 A mm-l (Rank Hilger Monospek 1000). Signal detection is facilitated by using an end window 13-stage photomultiplier (EM1 6256B) and signal registration by a potentiometric chart recorder (Servoscribe 541.20). All the measurements reported were obtained by running the plasma under compromise conditions, which are shown in Table I.TABLE I COMPROMISE PLASMA OPERATING CONDITIONS The argon plasma is sustained in a concentric demountable silica torch. Parameter Operating conditions Forward power . . . . . . . . Reflected power . . .. . . .. Viewing height . . . . . . . . Coolant flow-rate . . . . . . .. Plasma gas flow-rate . . . . . . Injector flow-rate . . . . .. . . Nebuliser uptake rate . . . . . . PMTt EHT . . .. .. . . . . Spectrometer entrance and exit slits . . 1.0 kW <2 w 25 mm ALC* 13 1 min-l < 1 .O 1 min-' 0.8 1 min-l 3.0 cms min-l 35 pm 1000 v * ALC, Above load coil. Photomultiplier tube. Procedures The wall-jet cell is connected via one of its solution exit ports and a length of Tygon tubing to the solution intake of the nebuliser and is connected to a peristaltic pump via the jet nozzle inlet.Solutions may, therefore, be pumped into the cell and out to the nebuliser after passing over the electrode surfaces. Alternatively, the solution outflow from the cell may be diverted to drain or can be recycled through the peristaltic pump and the cell via the other exit port. The trace elements of interest are deposited on to the glassy carbon electrode by holding its poten- tial with respect to the reference electrode at a suitable reduction potential for that element. After sufficient time has elapsed to accumulate the element the potential of the working electrode is scanned in the anodic direction whereupon the element is rapidly (<500 ms) stripped back into solution and subsequently pumped to the nebuliser of the ICP.This pro- cedure requires the solution to contain a strong electrolyte, which can be incorporated in the sample solution when it is prepared or can be added automatically using the peristalic pump and a mixing coil positioned before the wall-jet cell. The sample solution is introduced into the plasma by the nebuliser - spray chamber system. Emission at the analytical wavelength of interest can thus be observed. Calibration can be effected simply by subjecting matched standards to an identical analytical cycle as the sample, or the peak area of the emission signal resulting from the sample may be compared with a standard calibration graph obtained by direct nebulisation, to calculate the absolute mass of analyte determined.To do this the plating efficiency of the cell must be measured and must remain constant. Knowing the volume of solution passed over the electrode the concentration in the original solution can thus be determined. The former method of calibration has been adopted in this study. All optimisation procedures of the electrochemical cell have been performed by plating copper on to the working electrode, held at its reduction potential, for 5 min from a solution of 1.0 pg CM-~ of copper in 0.1 M potassium chloride. After plating, the copper was rapidly stripped from the electrode by applying the anodic-stripping potential, which results in a rapid transient stripping current through the cell. The magnitude of this current was measured for the purposes of optimisation.Alternatively, or simultaneously, the contents of the cell containing the stripped analyte were pumped to the ICP and the emission signal observed at the analytical wavelength of interest (Cu I 324.8 nm) was measured for the optimisation. Using the peristaltic pump, solutions are pumped into the cell at pre-determined rates.1334 LONG AND SNOOK: ELECTROCHEMICAL Analyst, Vol. 108 Reagents Wherever possible reagents of the highest purity have been used for our investigations. Solutions of 0.1 M potassium chloride were prepared using Aristar potassium chloride and de- ionised distilled water. Copper and other metal impurities were removed using an Amberlite IRC 718 ion-exchange bed. Solutions of copper were prepared as 1000 pg ~ m - ~ stock solutions in 0.1 M potassium chloride.Lower concentrations were prepared from this stock solution by serial dilution with 0.1 M potassium chloride in de-ionised distilled water. Buffer solutions to study the effect of pH were prepared as shown in Table 11. TABLE I1 BUFFER SOLUTIONS USED TO STUDY THE EFFECT OF pH ON THE EFFICIENCY OF THE ELECTROCHEMICAL CELL Reagent pH Final volume/cm3 25 cm3 0.2 M KC1 + 13 cm3 0.1 M HC1 . . . . . . 2.0 100 25 ,ma 0.2 M PHP* + 0.1 cm3 0.1 M HC1. . . . . . 4.0 100 25 cm3 0.2 M PHP* + 43.7 cm3 0.1 M NaOH . . . . 6.0 100 50 cm3 0.025 M Na2B407 + 20.5 cm3 0.1 M HC1 . . . . 8.0 100 50 cm3 0.025 M Na,B407 + 18.3 cm3 0.1 M NaOH . . 10.0 100 10 cm3 0.1 M NaOH + 25 cm3 0.2 M KC1 . . .. . . 12.0 100 * Potassium hydrogen phthalate (COOHC,H,COOK) .Results and Discussion Preliminary experiments were carried out using the wall-jet cell as a polarographic cell to establish the half-wave potential for the reduction of copper, lead and cadmium, with respect to the silver - silver chloride electrode and 0.1 M potassium chloride base electrolyte. The plating and stripping potentials were selected on the plateau regions either side of the half- wave potential and are shown in Table 111. The element chosen for optimisation and evaluation of the pre-concentration technique was copper, as this element is normally of low concentration in the sample type of interest and its determination by anodic-stripping voltammetry is well documented.l29l3 When used in the continuous-flow mode for anodic-stripping voltammetry the performance of the wall-jet cell is affected by several variables, which have been optimised in this study.TABLE I11 PLATING AND STRIPPING POTENTIALS USED WITH THE WALL- JET CELL Ion Plating potential*/V Stripping potential*/\r CU2+ . . .. + 0.20 + 0.33 Pb2+ . . . . - 0.42 - 0.15 Cd2+ . . .. - 0.50 - 0.33 * Versus silver - silver chloride reference electrode in 0.1 M potassium chloride solution (pH 3.5). Jet-wall Separation and Volume Flow-rate current, which is related to the mass of electrolyte plated on to the working electrode: The wall-jet theory1 may be used to derive the following equation for the limiting diffusion where I d = limiting diffusion current; M = mass of analyte plated; n = number of electrons transferred in the reduction process; F = Faraday's constant; D = diffusion coefficient of the depolariser; R = radius of the electrode; a = jet-nozzle diameter; V = volume flow-rate; v = kinematic viscosity of solution; and C" = concentration of depolariser (migrating species).Thus for a fixed jet-nozzle diameter, calculated to provide laminar flow to the working electrode, the limiting diffusion current and hence mass of analyte plated is dependent on the radius of the electrode and the volume flow-rate of the solution.November, 1983 PRE-CONCENTRATION FOR USE WITH ICP-AES. PART I 1335 The diffusion current is also dependent on the separation distance of the jet nozzle and the working electrode, which is assumed to be fixed in equation (1) at a position to give laminar flow over the electrode.In this study we have optimised this distance for mass transfer before attempting to verify equation (1). The results of this experiment are shown in Fig. 2. The jet nozzle to electrode separation was optimised at volume flow-rates between 1.0 and 10 cm3 min-l by measuring the total anodic-stripping current that resulted from stripping copper plated from a 1.0 pg cm-3 solution of copper in 0.1 M potassium chloride for 5 min. During the stripping cycle the cell contents were also pumped to the nebuliser system of the ICP and hence into the plasma where emission at the Cu I 324.8-nm line was observed for comparison with the data obtained by measuring the stripping current. Both methods yielded the same shaped graphs as shown in Fig. 2. a) 1 I I 2 4 6 8 10 50 > c, .- fn 45 40 al *.' .- 0 fn .- .- fn 35 s (u m al 2 30 Ot 25 (c.- 20 2 4 6 8 10 Separationhn rn Fig. 2. Optimisation of jet t o electrode separation. (a) Electrode response measured by stripping current. A, 8.0 cm3 min-1; B, 3.0 cm3 min -1; and C, 1.6 cm3 min.-L (b) Electrode response measured by emission at Cu 32.84 nm. Thus for the flow-rates studied the optimum jet nozzle to electrode separation was found to be 1.5 mm. Having established the optimum jet nozzle to electrode distance and its independence of volume flow-rate of solution, the dependence of I d on the volume flow-rate was investigated. Inspection of equation (1) reveals that I d is proportional to the volume flow-rate: IdccVi and by taking the logarithms we obtain log I d cc 2logV.Therefore, a double logarithmic plot of I d vs. V will yield a linear plot of gradient 0.75 if the equation is valid. Such a plot constructed from experimental data, obtained using the same solutions and plating times as described previously, is shown in Fig. 3. The relative emission intensity observed in the ICP at the Cu 324.8-nm line is directly proportional to the mass plated and subsequently stripped from the wall-jet cell working electrode. The emission intensity varies linearly with flow-rate between 1.0 and 10 cm3 min-1, which confirms that the cell is operating under controlled hydrodynamic conditions. The slope of this graph however is 0.82, which is higher than the predicted slope of 0.75 for a wall-jet cell. This discrepancy arises because we have optimised the jet nozzle - electrode separation for mass transfer to the electrode.At separations greater than 2 mm (see Fig. 2) the predicted volume flow-rate dependence of the stripping current is obtained. The most important proof in this experiment, however, is that the cell is operating under controlled hydrodynamic conditions, which allows us to predict the variation of plating efficiency with the variation of volume flow-rate through the cell. The nebuliser used with our ICP has a fixed solution uptake rate of 3.0 cm3 min-l. For compatibility, therefore, all further optimisations were carried out at a volume flow-rate through the wall-jet cell of 3.0 cm3 min-l. Effect of pH electrode cannot be rationalised to a direct comparison of electrode responses.The assessment of optimum pH for electrochemically monitoring species at the working The alteration1336 LONG AND SNOOK: ELECTROCHEMICAL Analyst, Vol. 108 >. 1.2 $ 1.0 - c .- ul - C .- 0.8 6 0.6 - 2 0.4 - - .- v) .- Q - 2 0.2 - 0 1 -0.5 0 0.5 1 .o Log( volume flow-rate/cm3 min- ’) Fig. 3. Influence of volume flow-rate. A, Experimental dependence ; B, theoretical dependence. of the chemical environment resulting from a change in pH modifies the positions of the deposition and stripping potentials and profoundly affects the shape of the stripping profile. For selection of optimum pH with stripping of the species into the ICP the optimisation may be simplified by adopting the pH value, which produces the maximum amount of deposited metal with minimum interference. Plating from these buffer solutions, each containing 10 pg cm-3 of copper, was effected for 5 min.Stripped analyte was then pumped to the ICP for determination. In each instance the deposition potential was selected to be on the plateau region of the current - voltage graph, determined by previously recording each particular graph polarographically. The results from this experiment are shown in Fig. 4, which shows that the optimum con- ditions at the pH values studied for the determination of copper is 4.0, which is also produced with 0.1 M potassium chloride as the supporting electrolyte. The buffer systems used for this experiment are tabulated in Table 11. 10’ 1 o2 103 1 o4 Deposition ti meis Calibrations and Wall-jet Cell Efficiency Having established the optimum operating conditions for the wall-jet cell to perform as a pre-concentration device, calibration experiments were carried out to determine the range of concentrations and plating times for which the cell’s response remained linear with respect to mass of analyte plated on to the electrode.Experiments were carried out in which different plating times were used for a range of concentrations of copper solutions. The results of these experiments are shown in Fig. 5 .November, 1983 PRE-CONCENTRATION FOR USE WITH ICP-AES. PART I 1337 For relatively high concentration solutions, ca. 50 pg ~ m - ~ , linearity is observed up to a deposition time of 200 s. For a concentration of 10 pg ~ m - ~ linearity is observed up to 1 0 3 s. Assuming the efficiency of the cell is unchanged, this result is expected, as the point of curva- ture for both calibrations occurs at the same mass of copper that has been plated.The reason that curvature is observed is that the current demand to strip relatively high masses of analyte exceeds that available from the potentiostat. The plating efficiency of the cell has been determined by observing initially the emission intensity at the Cu 324.8-nm line when a 1.0 pg ~ m - ~ copper solution was nebulised into the ICP after being pumped through the cell and then measuring the drop in intensity when the plating potential is applied. The plating efficiency determined in this way was 8% which is typical for this type of cell. The stripping process was found to be quantitative with no memory effects observed in subsequent stripping cycles using only a solution of 0.1 M potassium chloride. The quantitative nature of the plating and stripping processes is reflected in the deposition time calibrations of Fig.5 which, within experimental error, exhibit the same slope over a wide range of concentrations and deposition times. The detection limit for copper using the nebuliser only in our system is 0.01 pg CM-~. As the cell volume of the wall-jet cell is 710 pl and the plating efficiency is 8% we can calculate the theoretical pre-concentration factor for this concentration for a given deposition time. For example, plating from 0.01 pg solution at a flow-rate of 3.0 cm3 min-l we remove 0.0024 pg min-l from the flowing solution. In 10 min therefore we remove 0.024 pg, which is subsequently stripped into a volume of 710 p1 resulting in a concentration of 0.034 pg ~ m - ~ .Thus a theoretical 3-fold pre-concentration should be achievable after 10 min. A reduction in cell volume by a factor of 10 would therefore increase the pre-concentration factor to approximately 30-fold for a 10-min plating time. Clearly it would also be advantageous to develop a pre-concentration cell with a higher plating efficiency to overcome the problem of relatively long deposition times for dilute solu- tions (ca. 10 min for a solution concentration of 0.01 p.p.m. of copper) but advances in this area may only be achieved by using electrochemical cells that do not operate under controlled hydrodynamic conditions. These have the consequent disadvantage of unpredictable plating efficiencies over the wide range of concentrations that have been investigated. Possibility of Multi-element Determination The possibility of simultaneous multi-element analysis has been investigated in this study for the determination of lead, cadmium and copper.Multi-element analysis using the wall- jet cell as an electrochemical detector is only possible in a sequential manner by applying an anodic-potential scan and measuring the sequential appearance of each stripping current at the individual stripping potential of each element. In this study, however, we have plated lead, cadmium and copper at a cathodic potential lower than those tabulated in Table I11 (-0.8 V) from a solution containing 1 pg cm-3 of each element and 0.1 M potassium chloride for 5 min. By scanning rapidly to an anodic potential that is more positive than the individual stripping potentials (+ 1 .O V) we have stripped all three metals back into solution simultaneously and determined their presence using the ICP.As we were restricted by only a single-channel spectrometer it was necessary to carry out this experiment in triplicate at the wavelength of interest of each element (Pb 283.3 nm, Cd 228.8 nm and Cu 324.8 nm). With a multi-channel spectrometer, however, these determinations would be performed simultaneously. Care must be exercised when performing multi-element analysis with a wide potential range as the reduction products of dissolved oxygen in solution ( H202) can interfere with the electrochemi- chemical processes occurring at the working electrode. In such an instance deoxygenation of sample and standard solutions is recommended.Conclusions It has been demonstrated that an electrochemical cell operating under controlled hydro- dynamic conditions can be used effectively as a yre-concentration device for trace element analysis, when combined with the ICP used as a spectrochemical source. Pre-concentration times of 10 min have been found necessary to gain any significant pre-concentration advantage for trace element solutions containing 0.01 pg cm3 or less of the element of interest. However, this time could be reduced by reducing the wall-jet cell volume and by increasing the plating1338 LONG AND SNOOK efficiency of the cell, preferably without destroying the hydrodynamic behaviour of solution flow in the cell.The possibility of simultaneous multi-element analysis has also been demonstrated using the pre-concentration technique. It is envisaged that such an electrochemical pre-concentration technique will be useful for the determination of ultra-trace levels of metals in sea water as these metals could be plated relatively easily without the addition of a base electrolyte. Also, during plating the sea water could be directed away from the nebuliser to prevent clogging and the final determina- tion carried out by stripping into a weak solution of a strong electrolyte and pumping this solution to the ICP. Part I1 of this paper will describe two novel flow-through cells that have been designed to reduce the cell volume into which the analyte is stripped and to increase the surface area of the electrode in an attempt to increase the plating efficiency of these types of cell. Both cells rely upon controlled hydrodynamics and may be used as electrochemical detectors as well as for pre-concentration devices for use in atomic spectroscopy. Attempts are currently underway to achieve these aims. We thank the Science and Engineering Research Council for the support of S.E.L. and the Laboratory of the Government Chemist for the support of R.D.S. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. References L’vov, B. V., Spectrochim. Acta, Part B, 1978, 33, 153. Ward, N. I., Stephens, R., and Ryan, D. E., Anal. Chim. Acta, 1979, 110, 9. Copeland, T, R., and Skogerboe, R. K., Anal. Chem., 1974, 46, 1257. Gunn, A. M., Millard, D. L., and Kirkbright, G. F., Analyst, 1978, 103, 1066. Kirkbright, G. F., and Snook, R. D., Anal. Chem., 1979, 51, 1938. Millard, D. L., Shan, H. C., and Kirkbright, G. F., Analyst, 1980, 105, 502. Lund, W., and Larsen, B. V., Anal. Chim. Acta, 1974, 70, 299. Lund, W., and Larsen, B. V., Anal. Chim. Acta, 1974, 72, 57. Holen, B., Bye, R., and Lund, W., Anal. Chim. Acta, 1981, 131, 37. Yamada, J., and Matsuda, H., J . Electroanal. Chem., 1973, 44, 189. Gunn, A. M., Kirkbright, G. F., and Opheim, L. N., Anal. Chem., 1977, 49, 1492. Edwards, L. L., and Oregioni, B., Anal. Chem., 1975, 47, 2315. Smart, R. B., and Weber, J . H., Anal. Chim. Acta, 1980, 115, 331. Received March 7th, 1983 Accepted June lst, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801331

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, November, 1983, Vol. 108, $9. 1339-1344 1339 Further Diff erential-pulse Polarographic and Visible Spectrophotometric Studies of the Degradation of Permitted Synthetic Food Colouring Matters With and Without the Addition of Ascorbic Acid: Degradation in the Dark and in the Light Without the Stabilising Action of EDTA Arnold G. Fogg and Abdulhadi M. Summan Chemistry Department, Loughborough University of Technology, Loughborough, Leicestershire, LE11 3TU In general, solutions of food colouring matters have been shown by differential- pulse polarography and visible spectrophotometry to degrade faster under accelerated light degradation conditions at pH 5.5 in the presence of ascorbic acid when EDTA is absent. A full yield of ammonia from azo nitrogen atoms is obtained, and for Red 2G and Black PN the acetamido nitrogen atoms are also converted into ammonia in the absence of EDTA.Degradation of food colouring matters in the dark in the presence of ascorbic acid, but in the absence of EDTA, has been shown to occur extensively over a period of several days. The relatively high concentration of ascorbic acid used degraded in a time similar to that required for the disappearance of colour of the food colouring matters and for this reason the simple amines formed, particularly naphthionic acid, remained in appreciable yield. Red 2G was observed to degrade by deacetylation to Red 10B, which is relatively stable to attack by ascorbic acid. Black PN also was observed to degrade initially to a coloured product, which could be either the deacetylated deriva- tive of Black PN or a monoazo compound.Ascorbic acid solutions stored in the dark became yellow and the unidenti- fied product gave a polarographic reduction wave a t - 0.90 V (versus S.C.E.) at pH 5.5. Keywords : Food colouring matters ; degradation ; di-fjerential-pulse polaro- graphy ; ascorbic acid Previously, diff erential-pulse polarography was used to monitor the interaction of permitted synthetic food colouring matters with ascorbic acid in accelerated light degradation studies1 The loss of both colouring matter and ascorbic acid was shown to be accelerated when they were present together. Further, the complete conversion to ammonia of the nitrogen atoms in the azo group(s) of those food colouring matters that are azo compounds was demon- strated.The simple amine intermediates in these conversions, namely aniline, sulphanilic acid and naphthionic acid, were monitored by visible spectrophotometry using a diazotisation procedure. The rate of formation of ammonia from these amines by the action of light was shown to increase in the order aniline < sulphanilic acid < naphthionic acid. Naphthionic acid was particularly labile, as were other more complex amines formed in the degradation of the food colours. These degradation studies were performed in pH 5.5 acetate buffer solution, the ascorbic acid being protected from the oxidation by dissolved molecular oxygen by the addition of the disodium salt of ethylenediaminetetraacetic acid (EDTA) . In this paper further studies of the degradation of food colouring matters with the addition of ascorbic acid are reported.In particular, the effect of omitting EDTA in the light degrada- tion studies and the interaction of food colouring matters and ascorbic acid in the dark have been studied. Experimental The food colouring matters and ascorbic acid were determined by differential-pulse polaro- graphy as previously describedl using a PAR 174A polarographic analyser (Princeton Applied1340 FOGG AND SUMMAN : DPP AND SPECTROPHOTOMETRY OF THE Analyst, Vd. 108 Research) and a Gould H-2000 X - Y recorder. Three-electrode operation was employed using a dropping-mercury electrode, a platinum counter electrode and a saturated calomel reference electrode. Diff erential-pulse polarography was carried out with a forced drop time of 1 s, a scan rate of 5 mV s-l and a pulse height of 50 mV.Solutions were deoxy- genated by means of nitrogen that had been passed through a vanadium(I1) scrubber. Spectrophotometric measurements were made with a Pye Unicam SY8-100 spectrophoto- meter. Ammonia and the simple amines (aniline, sulphanilic acid and naphthionic acid) were determined as previously describedl using the indophenol reaction and diazotisation procedures, respectively. Accelerated light degradation studies were carried out in a specially constructed light box in which the sample solutions were held at a uniform close distance from a 500-W lamp (Philips G/74) . I The solutions were deoxygenated with nitrogen gas and were contained in tightly sealed autoclavable bottles.These bottles were also used in degradation studies carried out on solutions kept in the dark and these solutions were also deoxygenated. Results Degradation of Food Colowring Matters and Ascorbic Acid in Accelerated Light Degradation Conditions Without EDTA Stabilisation In this study the effect of omitting the EDTA from the accelerated light degradation study was investigated and the main degradation experiments were carried out in the presence of EDTA at high ascorbic acid concentrations, The results obtained are shown in Table I and should be compared with those given in Table V in the previous study.l Results for Yellow 2G have been omitted in both instances as the ascorbic acid degrades before the food colouring matter. Clearly in some instances the presence of EDTA not only protects the ascorbic acid from interaction with oxygen but also stabilises the food colouring matters with respect to reaction with ascorbic acid.Thus the times for the complete loss of food colouring matter with and without EDTA are 23 and 10 h (Red 2G), 24 and 6 h (Sunset Yellow FCF), 12 and 4 h (Ponceau 4R) and 24 and 4 h (amaranth), respectively. The formation of the full yield of ammonia from the intermediate amines is also much faster in the absence of EDTA for Red 2G, Sunset Yellow FCF and Brown FK. The formation of amines and ammonia from the food colouring matters in the absence of ascorbic acid has so far been followed only for TABLE I FORMATION OF SIMPLE AMINES AND AMMONIA ON LIGHT DEGRADATION OF A20 FOOD COLOURS I N THE PRESENCE OF ASCORBIC ACID BUT WITHOUT THE ADDITION OF EDTA The amines and ammonia were not determined until all of the food colouring matter had disappeared visibly.The additional times of light treatment before measurenient are given. All food colouring matter solutions were 10 P.p.m. initially. The initial ascorbic - acid concentration was 1600 p.p.m. Food colour Red 2G . . . . Sunset Yellow FCF Tartrazine . , Black PN , . . . BrownFK , . . . Carmoisine . . . . Ponceau 4R . . . . Amaranth . . Chocolate Brown HT Time for colour to disappear/h . . 10 .. 6 . . 42 . . 4 . . 4 , . 3 . . 4 . . 4 . . 3 Additional timeld 1 6 1 6 2 6 10 1 10 1 6 1 1 1 1 Xmine formed Aniline Sulphanilic acid Sulphanilic acid Sulphanilic acid Sulphanilic acid Naphthionic acid Naphthionic acid Naphthionic acid Naphthionic acid Molar yield of amine, yo 98.5 0 22.1 7.9 78.1 62.5 26.0 43.3 0 10.1 7.6 0 0 0 0 Molar yield of ammonia, yo 160 304 181 195 125 140 170 310 494 192 192 200 200 201 402November, 1983 DEGRADATION OF FOOD COLOURING MATTERS 1341 Chocolate Brown HT and Brown FK, because the food colouring matters are considerably more stable with, or without, the presence of EDTA under these conditions.Chocolate Brown HT had degraded visibly after 21 d when a 400% yield of ammoiiia and no naph- thionic acid were recorded. Brown FK disappeared visibly after 2 months when the yields of ammonia and sulphanilic acid were 180 and 22%, respectively. In the presence of EDTA these two food colouring matters gave final molar yields of ammonia of 200 and 4ooy0, respectively, indicating the complete conversion of one and two azo groups, respectively, to ammonia.In the absence of EDTA the yields of ammonia were 300 and 5ooy0, respectively, which indicates that in the absence of EDTA the acetamido groups in these food colouring matters are being hydrolysed and converted to ammonia, whereas in the presence of EDTA they are not. In general the acetaniido groups are stabilised in this way and this was confirmed by tests on solutions of acetanilide. After 2 d of light degradation, no ammonia was formed from acetanilide in the presence of EDTA, but a molar yield of 98.2% was obtained in its absence. An unusual effect of omitting the EDTA was observed for Red 2G and Black PN. Degradation of Food Colouring Matters and Ascorbic Acid in the Dark at Room Temperature As the food colouring matters degraded extremely rapidly under accelerated light con- ditions in the presence of ascorbic acid without EDTA, a study was made to determine whether any interaction could be observed in the dark in the absence of EDTA.The stabi- k i n g effect of EDTA on ascorbic acid in the dark is seen from the results given in Table I1 and the degradation of naphthionic acid, sulphanilic acid and aniline in the dark in the presence of ascorbic acid but without EDTA is shown in Table 111. Formation of ammonia from all three amines over a period of several days is apparent, but a feature of the degrada- tion of naphthionic acid is that after the initial degradation period some of the amine remains intact. This might be expected as the ascorbic acid has degraded completely by this stage.On the other hand, the degradation of sulphanilic acid and aniline continued to some extent after the ascorbic acid had degraded, For naphthionic acid the further degradation of the amine on the addition of more ascorbic acid later in the degradation study was confirmed (Table 111). TABLE I1 EFFECT OF EDTA IN STABILISING ASCORBIC ACID FROM OXIDATION BY AIR IN THE DARK Results are remnant of ascorbic acid as percentage of original. Original concentration of Timeld ascorbic acid, r------ L \ Conditions p .p. m. 0 1 2 3 4 5 1 6 2 2 47 13 0.7 Without EDTA . . . . 100 100 55 42 37 15 0 - - 1000 100 - - - - Timeld 0 4 30 36 50 75 82 90 I 7 L With EDTA (0.002570) . . 100 100 94 44 - 2 3 7 3 1 1000 100 - 5 5 3 0 4 - - - The food colouring matters were found generally to degrade completely within 3 weeks in the dark in the presence of high concentrations of ascorbic acid (Table IV) ; Black PN degraded rapidly within 1 d.The amounts of simple amine and ammonia formed are given in Table V. The stability of some proportion of the simple amines formed in these degradations, and in particular of naphthionic acid, is clearly seen. This would be expected from the degrada- tion results with the simple amines (Table 111). The final yield of undegraded naph- thionic acid formed from carmoisine (47y0), Ponceau 4R (45%) and amaranth (47%) are1342 FOGG AND SUMMAN: DPP AND SPECTROPHOTOMETRY OF THE AndySt, Vd. 108 TABLE I11 DEGRADATION OF SULPHANILIC ACID, NAPHTHIONIC ACID AND ANILINE IN THE DARK I N THE PRESENCE OF ASCORBIC ACID WITHOUT EDTA Amine Naphthionic acid Sulphanilic acid Aniline ..Time/d . . . . 0 20 70 77t y t .. .. 10 40 70 80 150 .. .. 0 10 35 50 70 80 Amine remaining, yo 100 41.1 41.1 17.3 12.2 76.9 30.7 20.2 20.2 13.4 100 100 100 50.2 34.8 22.4 22.4 Molar yield of ammonia, yo 12.6* 70.5* 70.9" 94.4* 99.7* 0 25.3 72.6 81.8 82.6 88.9 0 0 51.4 67.2 81.2 81.5 * These results indicate that some ammonia is obtained on heating naphthionic acid with sodium hydroxide solution. With a further addition of 2000 p.p.m. of ascorbic acid a t 70 d. similar to the percentage remnant when naphthionic acid is treated similarly (41 yo), whereas the remaining yield from Chocolate Brown HT (84%) is considerably larger.This latter yield is consistent, however, as one molecule of Chocolate Brown HT gives two molecules of naphthionic acid. The further degradation of the naphthionic acid on further addition of ascorbic acid was confirmed for carmoisine and Ponceau 4R (Table V). In previous heat degradation studies at intermediate temperatures (e.g., 80 "C) deacetyla- tion of the acetamido group of Red 2G was observed and Red 10B was formed.2 In this study Red 10B was formed during the interaction of Red 2G and ascorbic acid in the dark. The hue of the degrading solution and the half-wave potential of the polarographic wave were found to be changed in the same way as in the heat study. Black PN, which also has an acetamido group, degrades rapidly in the dark in the presence of ascorbic acid to give a pale yellow solution (at pH 5.5).When this solution was acidified to pH 2, the solution became purple. The solution at pH 5.5 exhibits a new differential-pulse polarographic peak close to that of Black PN but the wave is more clearly defined at pH 2 (EP = 0.OOV). Better separation of the diff erential-pulse polarographic peaks of Black PN and this product can be achieved by the addition of tetraphenylphosphonium chloride in an analogous manner to that with Red 2G and Red 10B.2 This product, formed from Black PN, may be the deacetylated derivative of Black PN, but sulphanilic acid and ammonia are also present in these solutions, which indicates that the product could be a monoazo dye. Some degrada- tion of the azo bond that yields sulphanilic acid has clearly occurred in the solution.The formation and subsequent degradation of the new product can be followed polarographically. Solutions of the triphenylmethane food colours, Green S, Patent Blue V and Brilliant Blue FCF, turned green on interaction with ascorbic acid in the dark and new differential polaro- graphic peaks were observed. At pH 5.5 the peak potentials for Green S, Patent Blue V and Brilliant Blue FCF are at -0.64, -0.73 and -0.80 V, respectively. The corresponding peaks for their products are at -0.40, -0.56 and -0.40 V, respectively. The final observation made was that ascorbic acid solutions stored in the dark at pH 5.5 became yellow even in the absence of the food colouring matters. Further, a new cathodic polarographic wave appeared with a half-wave potential of -0.90 V and increased in size as the anodic wave of ascorbic acid at +0.05 V became smaller. The new compound gave an improved polarographic wave at pH 2 (EP = -0.54V). Neither the colour nor the new wave was observed in the accelerated light degradation studies.November, 1983 DEGRADATION OF FOOD COLOURING MATTERS 1343 Discussion The previous study of the interaction of food colouring matters and ascorbic acid was carried out in the presence of EDTA.l Ascorbic acid is very rapidly oxidised by dissolved molecular oxygen and EDTA was added to stabilise the ascorbic acid.This allowed the degradation of the food colouring matters in the presence of relatively low concentrations of ascorbic acid to be studied in order to obtain an indication of the stoicheiometry involved.EDTA is used extensively as an antioxidant synergist - stabiliser3y4 and it is therefore not surprising that not only is the reaction of ascorbic acid with molecular oxygen inhibited but also its reaction with the food colouring matters. The degradation rate of food colouring matters with ascorbic acid has been shown here to be generally more rapid in the absence of EDTA and the results obtained under these conditions may be considered to be more relevant to actual food processing conditions, although other antioxidants are frequently present in foodstuffs. EDTA was also shown to protect the acetamido groups in Red 2G and Black PN from hydrolysis and the subsequent formation of ammonia. This effect seems to have a wider application in the stabilisation of drugs, such as paracetamol, and studies on this will be reported separately.TABLE IV DEGRADATION OF FOOD COLOURING MATTERS AND ASCORBIC ACID I N ADMIXTURE I N THE DARK Food colour Black PN . . . . .. Amaranth . . .. .. Ponceau 4R . . . . .. Brown FK .. * . . . Carmoisine . . .. .. Chocolate Brown HT . . Sunset Yellow FCF . . .. Tartrazine . . .. . . Red 2G . . . . .. .. Green S . . . . . . . . Brilliant Blue FCF . . .. Patent Blue V . . .. . . .. Indigo Carmine . . . . Erythrosine . . .. .. Quinoline Yellow . . .. Yellow 2G . . .. 1 . Approximate time for complete loss of 100 p.p.m. of ascorbic acid*/d 30 32 60 40 46 40 56 54 42 42 44 35 60 45 57 16 Time for loss of half of food colour in the presence of 100 p.p.m. of ascorbic acid*/d 15 25 14 15 25 17 44 54 40 16 45 32 > 60 13 > 60 > 60 Time for loss of all food colour in the presence of 1000 p.p.m. of ascorbic acidt/d 1 12 15 16 18 21 23 24 251 25 25 30 31 > 325 >32§ > 32s * In the presence of 25 p.p.m.of EDTA. 7 In the absence of EDTA. $ Time for complete change of Red 2G to Red 10B and other products. 8 Ascorbic acid degrades completely first. The rate of degradation of food colouring matters in the dark in the presence of ascorbic acid was found to be faster than had been expected, but such degradation in blackcurrant health drinks, for example, must be well known in the soft drinks industry. A feature of these degradations in the dark under the conditions used here is that, as the ascorbic acid tends to degrade before the food colouring matter, simple amines such as naphthionic acid remain as final degradation products.For Red 2G a major degradation product is Red 10B and an intermediate coloured compound, possibly the deacetylated derivative, or a monoazo compound, is formed from Black PN. Black PN is particularly susceptible to interaction with ascorbic acid even in the dark. Coloured intermediates showing characteristic polaro- graphic half-wave potentials were also observed for the triphenylmethane colours, Green S, Patent Blue V and Brilliant Blue FCF. The identity of the coloured product formed from ascorbic acid in the dark is being investigated. This study is being extended to include a study of the degradation of food colouring matters1344 FOGG AND SUMMAN TABLE V FORMATION OF SIMPLE AMINES AND AMMONIA ON DEGRADATION OF AZO FOOD ASCORBIC ACID (1 000 p.p.m.) WITHOUT EDTA COLOURING MATTERS IN THE DARK IN THE PRESENCE OF The amines and ammonia were not determined until all of the food colour had disappeared visibly.Food colour Red 2G* Sunset Yellow FCF Tart razine .. Black PN . . .. Brown FK Carmoisine .. Ponceau 4R . . Amaranth .. Chocolate Brown HT Time for colour to Additional disappearld timeld . . 23 .. 24 .. 1 . . 16 .. 18 . I 15 .. 12 .. 21 1 10 30 1 10 20 1 15 30 1 5 30 5 10 30 50: 1 10 20 4!§ 10 20 1 20 50 Amine formed Sulphanilic acid Sulphaiiilic acid Sulphanilic acid Sulphanilic acid Naphthionic acid Naphthionic acid Naphthionic acid Naphthionic acid Molar yield of Molar yield of amine, yo ammonia, % 13.1 7.3 7.2 12.8 7.8 7.6 41.2t 22.4t 13.2 7.5 7.4 46.9 46.9 41.9 11.8 44.8 44.9 44.9 11.8 46.7 46.7 41.7 83.6 83.6 83.6 12.lt 190 195 196 190 194 195 306 369 381 183 191 195 166 167 168 200 165 166 166 200 162 164 162 330 330 330 * The colour changed to pink, which was found to be due to the formation of Red 10B.2 After 25 d Determination of aniline t The purple degradation product was present in these solutions but was dcstroyed by the addition of 25.7% of the Red 2G remained and the yield of Red 10B was found to be 72.5’7;.and ammonia were not made owing to the stability of the Red 10B formed. the permanganate. With a further addition of 2000 p.p.m. of ascorbic acid after 30 d. 3 With a further addition of 2000 p.p.m. of ascorbic acid after 20 d. and the formation of simple amines and ammonia from them, in accelerated heat degradation conditions, with and without the addition of ascorbic acid. The original intention was to include the results here, but preliminary indications are that the system behaves in a markedly different way in that under some conditions very low yields of ammonia and simple amines are obtained, despite full visible degradation of the food colouring matters. The authors thank Pointing Limited and Williams (Hounslow) Limited for providing samples of food colouring matters, Dr. N. T. Crosby (Laboratory of the Government Chemist) for helpful advice, the Government of Saudi Arabia for financial support and Unim Al-Qura University for leave of absence for A.M.S. References 1. 2. 3. 4. Fogg, A. G., and Summan, A. M., Analyst, 1983, 108, 691. Fogg, A. G., and Whetstone, RI. R., Analyst, 1982, 107, 456. “Kirk-Othmer Encyclopaedia of Chemical Technology,” Third Edition, \‘olume Eleven, John N’iley, Reynolds, J. E. F., Editor, “hlartindale, the Extra Pharmacopoeia,” Twenty-eighth Edition, Pharma- Received J u n e 3vd, 1983 Accepted June 30th, 1983 New York,” 1980. ceutical Press, London, 1982, pp. 383 and 1281.

ISSN:0003-2654    

DOI:10.1039/AN9830801339

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, November, 1983, Vol. 108, $9. 1345-1348 Polarography of Palladium and Indium at a Dropping-mercury Electrode Using E-Caprolactam as a Complexing Agent 1345 Bal Krishan Puri and Ashok Kumar Department of Chemistvy, Indian Institute of Technology, New Delhi-110 016, India The polarographic behaviour of palladium(I1) and indium(II1) has been studied using E-caprolactam as the complexing agent a t a constant ionic strength (0.1 M potassium nitrate). Well defined diffusion-controlled, irreversible waves were obtained for both metals. The forward rate constant (K:.,) and the transfer coefficient (nar) were calculated. The diffusion current constants were calculated to be 4.19 for palladium and 6.41 for indium, and were constant over wide concentration ranges of 1.0-3.0 and 1.0-3.5 mM, respectively.Based on the large difference in their E ; values, a method is suggested for the simultaneous determinations of these metals when they are present together in pure solutions. The method has also been applied to the determination of these metals in some synthetic samples and certain alloys. The error did not exceed -j=0.25y0 in any instance. Keywords : Palladium determination ; i n d i u m determination ; alloy analysis ; Polarography ; ecaprolactam Nitrogenous organic complexing agents have theoretical and experimental significance in polarographic studies of metals. These compounds are important mainly because they are specific and selective, whereas ordinary supporting electrolytes are not. Among the most important organic supporting electrolytes are pyridinium chloride,l EDTA,, ethanolamineS and gly~ine.~ A survey of the literature revealed that ecaprolactam has only been used as a complexing agent in the extraction of some rare earth elements, into mixed solvents such as butyl phosphate and isoamyl alcoh01.~ In this work the polarographic characteristics of palladium and indium, using ecaprolactam as the complexing agent, have been studied.Conditions have been developed for the determinatiion of these metals individually and in the presence of each other in some synthetic samples and certain alloys. Experimental Polarograms in all instances were recorded at 30 & 0.1 "C with a manual polarograph (Toshniwal CLO,) in conjunction with a Polyflex galvanometer (Toshniwal PL50) ; pH values of the solutions were measured with an Elico pH meter.A saturated calomel electrode was used as the reference electrode and was connected to the polarographic cell through a potas- sium chloride - agar bridge; oxygen was expelled by passing a stream of purified nitrogen through the test solution for 5 min. The dropping- mercury electrode had the following characteristics: m = 2.90 mg s-l; t = 2.9 s; mW = 2.43 mg3 s-4; and h,,,, = 58.7 cm in 0.1 M potassium nitrate solution with open circuit. The number of electrons (n) involved in the reduction process was determined by the milli- coulometric method of De Vries and Kroon.6 Doubly distilled mercury was used. Results and Discussion A number of polarograms were recorded at various pH values, keeping other factors con- stant.At lower pH the wave height was considerably increased owing to reduction of the H+ involved, and at higher pH a slight turbidity appeared, resulting in a decrease in the current. The diffusion current in both instances did not change in this pH range. The linear dependence of the limiting current on the square root of the height of the mercury column indicates that the rate of reduction of these metal - 6-caprolactam complexes is a diffusion-controlled process. A graph of &(i/id-i) vemm Ed.&.. is a straight line in both A well defined wave was obtained in the pH range 2.54.0.1346 PURI AND KUMAR: POLAROGRAPHY OF Pd AND In AT A Analyst, Vd. 108 instances. The values of the slopes and E , for various concentrations of ecaprolactam are given in Tables I and I1 and indicate that these metal - 6-caprolactam complexes are reduced irreversibly under these conditions.The formation of the complex is evident as the half-wave potential is shifted to more negative values (Tables I and 11) with an increase in the ecapro- lactam concentration. In view of the irreversibility of the wave, values of kinetic parameters such as the forward rate constant (k$) and charge-transfer coefficient (ncc) were calculated by Koutecky's treatment as extended by Meites and Israel,' and the values are recorded in Tables I and 11. The observed decrease in the value of noc implies that the transfer of electrons is made increasingly difficult. In other words, the electrode reaction becomes more irrever- sible with increasing concentration of the complexing agent. It is also evident that the values of k t decrease as the concentration of ecaprolactam increases.This observation is in agreement with the conclusions derived on the basis of nor (or cc) values. KaEena and MatouSek's method8 for the diffusion coefficient was applied, showing dynamic equilibrium in the system. TABLE I POLAROGRAPHIC DATA FOR THE PALLADIUM - E-CAPROLACTAM SYSTEM Pd(I1) concentration = 0.5 x ~ O - , M ; pH = 3.1; p = 0.1 M KNO,; n = 2. r-Caprolactam - E l us. Slope of D$ x 1031 concentration/M i d / p A S.C.E./V log plot/mV cm2 s-l nu kfh/cni s-1 0.2 5.30 1.450 117 3.09 0.463 4.64 x 10-13 0.3 5.09 1.468 118 2.97 0.459 3.91 x 10-13 0.4 5.02 1.490 120 2.93 0.451 3.84 x lo-', 0.6 4.53 1.523 123 2.64 0.441 3.19 x 10-13 0.6 4.04 1.558 127 2.35 0.427 3.19 x lo-', The shape of the waves does not change much when the E-caprolactam concentration is increased from 0.1 to 0.5 M, although the half-wave potential was shifted to more negative values.This signifies an irreversible process but, because of irreversibility, no deduction could be made regarding the composition of the complexes. However, on the basis of n values determined coulometrically (n = 2 for palladium and 3 for indium), the electrode reaction may be represented as follows : Pd2+ + 2e -+ PdO In3+ + 3e -+ In0 Effect of Metal Ion Concentration As these metals give well defined waves in E-caprolactam alone, without any other additives present, for their determination 0.1 M e-caprolactam (pH 3.1) was used as the base electrolyte.The diffusion current constants ( I ) calculated under these conditions using the IlkoviE equation, I = &/c rn%tb, were 4.19 for palladium and 6.41 for indium, and were constant over wide concentration ranges of 1 .O-3.0 and 1 .O-3.5 mM in 20 ml of solution, respectively. Therefore, the polarographic results can be used for the quantitative determination of these metals and also for their differentiation, as their half-wave potentials differed by more than 0.2 v. TABLE I1 POLAROGRAPHIC DATA FOR THE INDIUM - E-CAPROLACTAM SYSTEM In(II1) concentration = 0.5 x 10-3 M ; pH = 3.1; p = 0.1 M KNO,; n = 3. r-Caprolactam -E* vs. Slope of D$ x lo3/ concentration/M &/PA S.C.E/V log plot/mV cm2 s-l na K&/cm s-1 0.1 7.79 0.550 106 3.03 0.511 2.84 x 0.2 7.44 0.600 113 2.89 0.480 1.55 x 10-6 0.3 7.15 0.626 125 2.78 0.434 1.85 x 10-6 0.4 6.86 0.670 131 2.67 0.414 1.16 x 10-6 0.5 6.00 0.725 138 2.33 0.393 6.21 x 10-7November, 1983 DROPPING MERCURY ELECTRODE USING E-CAPROLACTAM 1347 General Procedure for the Determination of Palladium( 11) and Indium( 111) conditions.graphs were constructed in each instance. solution, the polarogram was obtained under the above conditions. of this wave was referred to its calibration graph (which was linear). Table 111. Polarograms for different concentrations of each metal were obtained under the above Calibration To determine the metal contents in an unknown The diffusion current (id) The results are given in The diffusion currents were measured by the extrapolation method.TABLE I11 DETERMINATION OF METAL IONS WITH E-CAPROLACTAM AS COMPLEXING AGENT ecaprolactam concentration = 0.1 M ; pH = 3.1; sensitivity = i6. Amount of metal ion P Metal Takenlmg Found/mg Error, % Palladium . . . . 0.968 0.970 + 0.2 1 1.563 1.561 -0.13 2.008 2.004 -0.20 2.795 2.800 +0.18 3.120 3.113 -0.22 Indium .. . . 0.501 0.502 +0.20 0.919 0.917 -0.22 1.560 1.558 -0.13 2.015 2.012 -0.15 2.750 2.750 +0.22 3.310 3.306 - 0.12 Mixed Polarograms of Palladium and Indium From the individual polarograms of palladium and indium, it is possible to differentiate the two metals in E-caprolactam solution because their half-wave potentials differed by 0.9 V from each other. Consequently, a series of polarograms were recorded with synthetically mixed solutions of the cations and the id values measured for each metal were referred to the respec- tive calibration graph.The concentrations of palladium(I1) and indium(II1) present in the mixed solution are given in Table IV. The results obtained are accurate and reproducible. TABLE IV POLAROGRAPHIC DETERMINATION OF PALLADIUM(II) AND INDIUM(III) I N A MIXTURE E-Caprolactam concentration = 0.1 M ; pH = 3.1; sensitivity = is, Amount added/mg Amount found/mg Error, yo No. Palladium Indium Pzm-m Palladium Indium 1 0.708 0.517 0.707 0.516 -0.20 -0.19 2 1.017 0.800 1.019 0.798 -+ 0.20 - 0.25 3 1.791 1.573 1.789 1.569 -0.11 -0.25 Sample ,.-h---- 7-w 4 2.345 1.940 2.348 1.945 f0.13 +0.26 5 3.010 2.501 3.014 2.506 $0.13 +0.16 Determination of Palladium in Alloys A 0.1-g amount of the alloy was dissolved in concentrated hydrochloric acid in the presence of a few drops of nitric acid.The solution was heated until the volume had reduced to about 5 ml, then cooled, 10 ml of hydrochloric acid were added and the volume was made up to 100 ml in a calibrated flask with distilled water. An aliquot of this solution was taken and palladium was determined as described above. The results are given in Table V.1348 PURI AND KUMAR TABLE V DETERMINATION OF PALLADIUM IN ALLOYS Amount of palladium Certified 7- Alloy composition, % Takenlmg Found/mg Platinum - iridium . . . . Pt, 55; Ir, 28; 2.750 1 2.753 1 Rh, 7; Cu, 3; 2.751 9 Fe, 3.5; Pd, 3.5 2.7482 2.7500 2.752 8 Oakay .. .. . . Ni, 60; Pt, 20; 3.017 1 3.0170 3.019 8 3.020 0 3.0169 3.0177 v, 9.5; P, 10.5 Average found/mg Error, yo 2.751 2 t0.040 3.0183 -t0.039 Determination of Indium in a Synthetic Sample indium in a synthetic sample.The results are given in Table VI. As no suitable alloy of indium was available, the method was applied to the determination of TABLE VI DETERMINATION OF INDIUM IN A SYNTHETIC SAMPLE Composition of sample: Ni, 12.50; Co, 5.50; Mg, 16.4; Ca, 9.0; Zn, 1.60; Fe, 51.50; In, 3.5%. Amount of indium Sample t--u7 No. Taken/mg Found/mg Error, yo 1 0.678 0.677 -0.21 2 0.917 0.916 -0.22 3 1.375 1.378 + 0.22 4 2.071 2.074 + 0.14 5 2.554 2.550 -0.16 Effects of Diverse Ions The effects of various anions and metal ions at a fixed concentration of palladium and indium ions (0.5 m M ) were ascertained. For nitrate, acetate, sulphate, oxalate, formate, salicylate and citrate, the E , values remained almost constant in all instances. Among the metal ions examined, cobalt(I1) and nickel(I1) could be tolerated in equal amounts of palla- dium; manganese( 11) interfered in the determination of palladium( 11) but fortunately it never occurs together with palladium in alloys. In the determination of indium, cadmium( 11) interfered, whereas cobalt(I1) and nickel(I1) could be tolerated in a three-fold excess and lead(I1) in an equal amount. Alkaline earth metal ions [magnesium(II), calcium(II), barium(I1) and strontium(II)] did not interfere in either determination for up to a ten-fold excess. 1. 2. 3. 4. 5. 6. 7. 8. References Mikula, J. J., and Codll, M., Anal. Chim. Acta, 1953, 9, 467. Willis, J. B., Anal. Chim. Acta, 1948, 21, 7. Rao, A. L. J., and Puri, B. K., Analyst, 1971, 96, 364. Kopa, M., and Palogal, J., Collect. Czech. Chew Commztn., 1958, 23, 50. Fremin, Yu. G., and Valkoval, I . , Z h . Anal. I C ~ Z P I Z . , 1971, 26, 1310. De Vries, T., and Kroon, J. L., J . Anz. CIicm. Soc., 1963, 75, 2484. Meites, L., “Polarographic Techniques,” Interscience, New York, 1965. KaEena, V., and MatouSek, L., Collect. Czcli. Cheiii. Commun., 1953, 18, 294. Received January 5th, 1983 Accepted June 7th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801345

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, November, 1983, Vol. 108, pp. 1349-1356 1349 Differential Electrolytic Potentiometry with Periodic Polarisation Part XXIX." Anhydrous Acetic Acid Precipitation and Complexation Titrations in E. Bishop Univevsity of Exeter, Chemistvy Department, Stocker Road, Exetev, EX4 4QD and Abdalla M. S. Abdennabi University of Petroleum and Minerals, P.O. Box 144, Dhalivan Iiztrvnational A irpovt, Dhahvapz, Saudi A vabia The applications of direct current and mark-space biased square wave differential electrolytic potentiometry alongside zero-current potentiometry to single and mixed halide precipitation titrations with silver(I), to single halide precipitation or complexation titrations with mercury( I) and to single halide complexation titrations with mercury(I1) in anhydrous acetic acid media have been examined, using silver, silver amalgam and gold amalgam electrodes.Mercury(1) offers no advantage over silver, which gives excellent results for single halides, particularly with silver electrodes, and chloride and bromide mixtures are fully resolved in this medium as well as chloride and iodide, particularly with silver amalgam electrodes, while bromide and iodide titrate to total halide only : chloride, bromide and iodide mixtures are resolved, but not accurately. The mercury(I1) titrations are much improved in the anhydrous medium with respect to water ; gold amalgam electrodes are favoured . Keywords : Difleerential electrolytic potentiowetry ; periodic polarisation ; non- aqueous argentimetry ; non-aqueous wercurirnetry ; anhydrous acetic acid solvent Direct current differential electrolytic potentiometry (d.c.DEP) has been applied successfully to precipitation reactions using silver and silver halide electrodes in aqueous media,l and showed enhanced response speed and precision.It also allowed determination of very low amounts and concentrations of halides not otherwise acce~sible.~,3 Periodic polarisation in argentimetry4-' offers yet faster response and prolonged electrode activity. Non-aqueous media have received little attention in precipitation titrimetry. Leblanc and McFadden recommended ethylene glycol for silver chloride titrations,* whilst Bishopg and Coghill and Kirklandlo have used alcohol. Mixtures of organic solvents and water have been ~ s e d , ~ ~ ~ and a series of solvents has been examined,ll benefits accrue provided that the solvent does not impair electrode function.Completely anhydrous media do not appear to have been examined. Little has been published on mercury( 11) precipitations, despite the more favourable chloride solubility product, and nothing on polarised electrodes or non-aqueous media. Neither mercury( I) nor mercury( 11) give satisfactory results in precise zero-current potentiometry,ll but the defined reactions enforced by polarisation yield better performance. Polarised electrodes have been used in aqueous complexometry, for the determination of EDTA,12 calcium and copper( 11) using liquid mercury e1ectr0des.l~ Monk and Steed14 found zero-current gold amalgam electrodes to be too sluggish and used large direct currents, so producing very broad peaks.Nartshorn used gold amalgam electrodes in d.c. and mark-space biased differential electrolytic potentiometry (m.s.b.DEP).' Many determinations by d.c. and m.s.b.DEP were made by East,l6 who found the latter to give faster response, improved stability of potentials and an alleviation of the tendency of amalgamated electrodes to lose mercury by stripping. No reference has been traced to complexation titrations in non-aqueous media. In this paper an exploration of precipitation of halides by silver and of complexation of chloride and bromide by mercury(1) and mercury(I1) in anhydrous acetic acid using d.c. and m.s.b.DEP is described. Huber and Tallant15 used platinum electrodes. * For Part XXVIII of this series, see reference list, p.1355.1350 BISHOP AND ABDENNABI : DIFFERENTIAL ELECTROLYTIC Analyst, Vol. 108 Experimental The apparatus, general procedure and preparation of anhydrous acetic acid and of solutions have been described.17~ls The supporting electrolyte was 0.05 mol 1-1 lithium perchlorate in anhydrous acetic acid. Silver and amalgamated silver and gold wire electrodes were used.lg Standard Solutions Sodium chloride (volumetric standard) was dried at 500 "C for 1 h and the required amount was dissolved in 0.1 mol 1-1 Aristar nitric acid and diluted to volume. This solution was used as the primary standard. Silver, mercury ( I ) and mercury (11) acetate in anhydrous acetic acid. Solubility limitations prevent preparations of solutions stronger than 4 x mol 1-1 of the silver salt and 0.01 mol 1-1 of the mercury salts.The solutions were standardised by titration of the aqueous sodium chloride using DEP to locate the end-p~ints.~ Sodium chloride solution, 0.01 moll-l. A 2 C 2oo i I ( I , 7.0 7.2 7.4 7.6 7.0 7.2 7.4 7.6 Volume of titranvml Fig. 1. Titration of 3 ml of 0.01 moll-' lithium chloride with 0.0038 moll-' silver acetate in anhydrous acetic acid using (a) gold amalgam and (b) silver amalgam electrodes. Supporting electro- lyte, 0.05 moll-' lithium perchlorate in anhydrous acetic acid. A, Anode - S.C.E. ; C, cathode - S.C.E. ; 2, zero-current electrode - S.C.E. ; EA, differential potential; solid line, d.c. DEP, A-C, current density = 8 x 10-7Acm-2; broken line, m.s.b. DEP, bias 15%, 60Hz. 500 4 2 v) 400 3 $ > E 300 200 500 4 400 c! cn v) 2 300 > E G 2 200 7.2 7.4 7.6 7.0 7.2 7.4 Volume of titranvml Titration of 3 ml of 0.01 moll-' (a) lithium chloride and (b) lithium bromide, using silver electrodes. Conditions and symbols as in Fig.1. Fig. 2. Lithium halides in anhydrous acetic acid, 0.01 moll-l. The anhydrous chloride and bromide and the monohydrate iodide were dried and the required amounts dissolved in anhydrous acetic acid and diluted to volume. The iodide reacts with anhydrous acetic acid, liberating iodine,20 and cannot therefore be prepared determinately.November, 1983 POTENTIOMETRY WITH PERIODIC POLARISATION. PART XXIX 1351 Results and Discussion Silver Titration of Individual Halides With an aliquot of 3 ml of 0.01 mol 1-1 halide in 40 ml of supporting electrolyte, the initial titrand concentration is 7 x 10-4mo11-1, the Q of the titration (the numerical value for quality of a titration)21 is negative for chloride assuming the aqueous value of the formation constant so that conventional methods fail.Consequently, DEP has been used in the standardisation. The low dielectric constant and ion-pairing characteristic of the chosen solvent were expected to depress the solubility of the silver halides and so permit conventional potentiometry to afford a locateable end-point. This was so, as the zero-current curves in Figs. 1-3 reveal, but, although silver amalgam electrodes gave better results than gold amalgam, the amalgam electrode curves are not good, and silver metal gave much more satisfactory results. This is contrary to expectation based on aqueous media and arises from the ion-pairing effect of the solvent being attenuated by the already pronounced covalency of the mercury-halide bond.Confirmation is given by the mercury( I) titrations as described below. r 100 4 3 0.0 2 P > E iij > -100 100 > E . 50 LU" v 6.0 6.2 6.4 Volume of titranvml I 7.2 7.4 7.6 0-1 ' 2.2 2.4 2.6 Volume of titrant/mI Fig. 3. Titration of 3 ml of Fig. 4. Titration of a mixture of 2 ml of 0.01 moll-' lithium iodide (and 0.01 moll-' lithium chloride and 1 ml of 0.01 moll-' iodine), using silver electrodes. lithium bromide, using silver electrodes. Conditions Conditions and symbols as in and symbols as in Fig. 1. Fig. 1. Not evident from the curves, except from the breadth of the differential peaks, is the very sluggish response of the amalgam electrodes, also observed in aqueous media, the titrations requiring 3 h for completion.Silver electrodes showed negligible drift, fast response and excellent titration quality. Gold amalgam shows purely the second-order response to chloride of the mercury phase, whereas silver amalgam gives a mixed response, and silver is purely a first-order response to cation. The titration of bromide, only just possible at these concentra- tions in aqueous media, is excellent in glacial acetic acid using silver electrodes (Fig. 2). Iodide, in the presence of released iodine,18 yields satisfactory titration curves, Fig. 3, but the1352 BISHOP AND ABDENNABI : DIFFERENTIAL ELECTROLYTIC Analyst, VoZ. 108 potential change and the differential peaks are compressed by the mixed oxidation - reduction potential control.The reaction of iodide with the solvent, producing iodine, removes the usefulness of the method. All the titrations, especially with silver electrodes, are free from distortions or disturbances attributable to solvent molecule adsorption ; anhydrous acetic acid is therefore without effect on electrode behaviour. Silver Titration of Halide Mixtures In aqueous media, unfavourable formation constant ratios render titrations of mixtures, other than of chloride with iodide, inaccurate except for total halide determination. In anhydrous acetic acid, mixtures of chloride and bromide proved to be accurately resolvable. With silver electrodes (Fig. 4) only the m.s.b.DEP curves are precisely interpretable, preferably by the geometrical method. In contrast to the chloride titrations above, the silver amalgam electrodes gave better results (Fig.5), possibly because of the depressed activity of silver in the amalgam. Anhydrous acetic acid evidently acts as a discriminating solvent in this reaction. 500 400 . 300 w x s 3 200 I I' I ' A / / .- '. I 2.2 2.4 2.6 7.2 7.4 Volume of titrant/mi 200 150 > E . 100 Lu" 50 Fig. 5. As Fig. 4, but using silver amalgam electrodes. Mixtures of chloride and iodide gave good titration curves (Fig. 6), although the iodide result is not meaningful on account of the reaction with the solvent. The presence of iodine in the solution is without effect on the chloride reaction. Mixtures of iodide and bromide were not resolved ; a single high-quality end-point was obtained corresponding to total halide.The effect of iodine on the potentials may have prevented resolution. Mixtures of all three halides are resolved as shown in Fig. 7, in contrast to the mixture of bromide and iodide, but, although the total halide end-point is good and the other m.s.b. end-points can be interpreted, the iodide and bromide end-points are not correct.November, 1983 POTENTIOMETRY WITH PERIODIC POLARISATION. PART XXIX 1353 Mercury( I) Titration of Individual Halides With a 10-ml aliquot of titrand in 40ml of supporting electrolyte, the initial concentration of analyte was 2 x mol l-l, approaching failure of conventional methods in aqueous media for the chloride titration. In anhydrous acetic acid, abnormally high current densities (10 pA ern-,) were required to resolve and stabilise the electrode behaviour and the DEP peaks were low and broad.The zero-current potentiometric curves for both chloride and bromide were smooth but of low quality, giving poor precision; the m.s.b. peaks were the most precisely interpretable (Fig. 8). Clearly, the solvent is exerting a levelling effect on these reactions. The response speed of the silver amalgam electrodes is also poor. Mercury(1) evidently offers no advantage over silver for halide titrations. Gold amalgam electrodes gave little response. 4 200 c! cn 3 6 ; 100 E &l 0 Fig. 500 4 400 $ v) 2 2 a > 300 E 200 %-7?%+-O 3.6 3.0 Volume of titranvml 6. Titration of a mixture of 1 ml of 0.01 kol l-1 lithium iodide and 1 ml of 0.01 moll-' lithium chloride, using silver amalgam electrodes. Conditions and symbols as in Fig.1. No precipitation of mercury(1) halide was observed. The ion-paired clusters of (Hg,Cl,), may simply fail to reach the critical size for separation in the solid phase at these low concentra- tions in this solvent, or the reaction may be converted to complexation. The stoicheiometry precludes incorporation of acetate in the product. It is most likely that the solvent transforms the reaction to discrete ion pairing. Mercury( 11) Titration of Individual Halides Complexometric titrations of chloride and bromide appear to be facilitated by anhydrous acetic acid. The ion-pairing effect of the solvent appears to improve the formation constants of the 1 : 2 complexes and, at the same time, the electrode processes are much better regulated1354 BISHOP AND ABDENNABI : DIFFERENTIAL ELECTROLYTIC ArtabSt, vd.108 200 100 4 c! s" P $ 0 E a v) > C 0.0 0.2 ml , A 500 4 400 v) I? 2 Lo" > 300 200 250 200 1 ' 0 Volume of titranthl Fig. 7. Titration of a mixture of 1 ml of 0.01 moll-' each of lithium chloride, bromide and iodide, using silver amalgam electrodes. Conditions and symbols as in Fig. 1. 4 300 200 G A z C 'Oat-/ 100 E 50 . Lo" I l l 1 I , I 9.6 9.8 10.0 10.2 9.8 10.0 10.2 Volume of titrantlml Fig. 8. Titration of 10 ml of 0.01 mol 1-1 (a) lithium chloride and (b) lithium bromide with 0.01 mol 1-1 mercury(1) acetate in anhydrous acetic acid. Medium, 0.05 moll-' lithium perchlorate in anhydrous acetic acid.Con- ditions and symbols as in Fig. 1 ; silver amalgam electrodes; d.c. current density = A cm-*, m.s.b. bias = 43%.November, 1983 POTENTIOMETRY WITH PERIODIC POLARISATION. PART XXIX 1355 50( 40C 4 c! c! 2 E m 30C 3 > 20a 150 100 =. E . G 50 0 4.8 5.0 5.2 4.6 4.8 5.0 Volume of titrantlml Fig. 9. Titration of 10 ml of 0.01 moll-' (a) lithium chloride and (b) lithium bromide with 0.01 mol 1-1 mercury(I1) acetate, using gold amal- gam electrodes. D.c. current density = 2 x A cm-2, m.s.b. bias = 19% : and symbols as in Fig. 1. than in aqueous media, perhaps by suppression of disproportionation of the dimercury(1) species or by prevention of their formation. Gold amalgam electrodes assume their proper function, potential equilibration becomes fast and charge transfer overpotential is greatly diminished.All titration curves shown in Fig. 9 are of excellent form and quality. Analytical Validity In all the titrations with silver, mercury(1) and mercury(I1) the m.s.b. and d.c.DEP end- points coincided with the zero-current end-point and were more, or much more, sharply defined. In the silver titrations of lithium bromide, a series of six under similar conditions gave a standard deviation of 0.0093 ml, for a mean of 26.37 ml, while for mercury(I1) a standard deviation of 0.012 ml for a mean of 19.28 ml and n = 6 was found, This is well within the experimental error of careful volumetric titrimetry. References 1. 2. 3. 4. 5. 6. 7. Bishop, E., and Dhaneshwar, R. G., Analyst, 1962, 87, 207. Bishop, E., and Dhaneshwar, R.G., Analyst, 1962, 87, 845. Bishop, E., and Dhaneshwar, R. G., Anal. Chem., 1964, 36, 726. Frank, V. F., 2. Elektrochem., 1954, 58, 348. Laitinen, H. A., and Hall, L. C., Anal. Chem., 1957, 29, 1893. Webber, T. J. N., Proc. SOC. Anal. Chem., 1967, 4, 161. Hartshorn, L. G., PhD Thesis, University of Exeter, 1972.1356 BISHOP AND ABDENNABI 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. Leblanc, R. B., and McFadden, R. T., Talanta, 1960, 5, 78. Bishop, E., Analyst, 1952, 77, 672. Coghill, E. C., and Kirkland, J. J., Anal. Chem., 1955, 27, 1611. Bishop, E., Darker, D., Jones, M. D., Stewart, P. M., and Sultan, S. M., Analyst, 1983, 108, 1007. Knight, W. S., and Osteryoung, R. A., Anal. Chirn. A d a , 1959, 20, 481. Martin, A. E., and Reilley, C. N., Anal. Chem., 1959, 31, 992. Monk, R. G., and Steed, K. C . , Anal. Chim. A d a , 1962, 26, 305. Huber, C. O., and Tallant, R. D., J . Electroanal. Chem., 1968, 18, 421. East, G. A., PhD Thesis, University of Exeter, 1975. Abdennabi, A. M. S., and Bishop, E., Analyst, 1982, 107, 1032. Abdennabi, A. M. S., and Bishop, E., Analyst, 1983, 108, 1227. Bishop, E., and Dhaneshwar, R. G., Analyst, 1963, 88, 424. Bishop, E., and Abdennabi, A. M. S., Analyst, 1983, 108, 1260. Bishop, E., “Indicators,” Pergamon Press, Oxford, 1972, p. 43. NOTE-References 1, 2, 17, 18 and 20 are to Parts V, VIII, XXV, XXVII and XXVIII of this series, respectively. Received April 6th, 1983 Accepted May 9th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801349

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, November, 1983, Vol. 108, pp. 1357-1364 1357 Flow Injection Analysis with Tubular Membrane Ion-selective Electrodes in the Presence of Anionic Surfactants Anthony J. Frend, Gwilym J. Moody and J. D. R. Thomas and Brian J. Birch Applied Chemistry Department, U WIST, Cardig, CF1 3XA Unilever Research Laboratory, Bebington, Wirral, Merseyside, L62 4XN Calcium ion-selective electrodes based on calcium bis(di[4-( lI1,3,3-tetra- methylbutyl)phenyl]phosphate) with trioctyl phosphate in poly(viny1 chloride) have improved resistance to interference by anionic surfactants when used in flow injection analysis. Tubular membrane electrodes have been used in flow injection analysis for determining calcium ion levels in selected potable and raw water and the data agree well with determinations by atomic-absorption spectroscopy and EDTA titration.Tubular membrane electrodes with improved membrane sensor have the potential for the determination of free calcium ions for a concentration as low as 10-4 M in the presence of a moderate background concentration of detergent, but this is poor compared with the concentration of M, which can be attained by the conventional type electrode with the same membrane system under static conditions. Keywords Calcium ion-selective electrodes ; anionic surfactant interference ; flow injection analysis The operation of flow injection analysis (FIA) with potentiometric ion-sensors in which calibration standards and samples are allowed to flow over the membrane surface of the sensing electrode coupled to a reference electrode is well e~tablished.l-~ The FIA principle has been used in this work for assessing the possibility of using the method for measuring ionic calcium in the presence of anionic surfactants that are otherwise known to have a detrimental effect on calcium ion-selective electrode^.^-^ The electrodes used are based on the poly(viny1 chloride) matrix membrane principle with a calcium bis (di[4-(1,lJ3,3-tetra- methylbutyl)phenyl]phosphate } sensor and either a dioctyl phenylphosphonate (membrane A) or a trioctyl phosphate (membrane B) solvent mediator mainly in the tubular flow-through mode, but with some initial experiments for the conventional “cascade flow” mode.The membrane A system is the type used normally in calcium ISEs of superior calcium ion selectivity,B but they suffer from gross interference by anionic surf act ant^.^,^ The membrane B system does not have such good calcium ion selectivity, but it is more resistant to anionic surfactant interference than other calcium ISEs so far examined.’ Experimental Electrodes Master membranes (A and B) were fabricated by the general procedure previously described*-10 using 0.17 g of poly(viny1 chloride) , 0.04 g of sensor and 0.36 g of solvent medi- ator, that is, in percentage ratios of 29.8, 7.0 and 63.2, respectively.Discs of the membranes were used to assemble conventional mode calcium ISEs in the normal way.8-10 For FIA, the exposed membrane was modified by blocking off most of the exposed area with a 1-mm layer of poly(viny1 chloride) - tetrahydrofuran cement and leaving a central calcium ion- sensing area of 2 mm wide across the diameter.Flow-through mode electrode tubular membranes were made from poly(viny1 chloride) , sensor and solvent mediator components of the same composition as membranes A and B. The tubular membranes were prepared as previously described,ll by repeatedly dipping a 0.7 mm diameter nickel wire (approximately 5 cm long) into a solution of the membrane components in tetrahydrofuran until the wire plus membrane layer was 3 mm in diameter.1358 FRENII et aE. : FIA WITH TUBULAR ION-SELECTIVE Analyst, VoZ. 108 This decreased to 2 mm diameter upon evaporation of the excess of tetrahydrofuran. The tubular membrane was carefully stripped from the wire by pulling the wire and membrane in opposite directions and a 2 cm length was cut.A poly(viny1 chloride) - tetrahydrofuran glue was used to seal the calcium ion-sensitive tube to the ends of poly(viny1 chloride) tubes installed in a Perspex flow-cell body, which was fitted with a silver - silver chloride internal reference electrode. The internal reference filling solution of the flow-through electrode was concentric to the tubular electrode. Electrodes were conditioned overnight in 10-1 M calcium chloride solution, which flowed through the tubular membrane of the flow-through electrode. r I S .- E - m Eo (0 FIA Assemblies and Electrode Calibrations The conventional mode calcium ISEs were set up for FIA in the normal “cascade flow’’ arrangement2 with a carrier solution reservoir, an Ismatec, Model MP 13G J-4, peristaltic pump, a flow injection head (stainless-steel high-performance liquid chromatograph valve type, of 50 mm3 sample volume) and calcium ISEs fitted for cascade flow of the carrier/ sample solution, which were arranged in series with the appropriate Solva-Tube inter- connecting tubing.For the cell assembly set in the overflow vessels (kept at a constant level by differential pumping), the EIL, Model RJ 23, ceramic junction saturated calomel reference electrode was set to have the carrier solution flow on to the ceramic junction from a loop of the main streame3 A similar sequence was used for FIA with the flow-through tubular electrodes, except that the EIL, Model RJ 23, ceramic junction saturated calomel electrode was set in a complete by-pass loop of carrier solution (Fig.1). For this, the reference electrode was fitted with a fabricated polyethene cup over the ceramic junction and through which the by-passing carrier solution flowed. The carrier stream flow-rate was 6 cm3 min-l. (a) To millivoltmeter t Carrier stream S 1 -F Waste ( b) To mi I livol t meter t dn Ag - AgCl internal reference Flow stream Perspex body Fig. 1. FIA manifold for the determination of calcium ions using a flow-through tubular membrane calcium ion-selective electrode with 0.1 M calcium chloride internal reference solution. Solva-tube (Gradko Scientific K116-0533-05) of 0.20 mm inner diameter was used between the injection valve and electrode for which the distance was kept short (5 cm) in order to minimise dispersion. The carrier solution (pH = 9.4) consisted of M disodium tetraborate(II1) decahydrate, Electrode calibrations were carried out by injecting calcium chloride standard solutions over the 10-6-10-1 M concentration range in ascending and descending order of concei tra- tion.A 1.7 x M aCar+ (where a is the activity of Ca2+) calcium buffer in sodium tri- polyphosphate was also injected. A typical run is shown in Fig. 2; Fig. 3 summarises calibration graphs from these runs. M sodium hydroxide and 0.14 M sodium chloride in de-ionised water.November, 1983 ELECTRODES IN THE PRESENCE OF ANIONIC SURFACTANTS 1359 T 0 W U W .- w - ’c L B 8 a 3E 4 t i n _-- . . 90 > 6o E Ci a 30 0 -6 -5 -4 -3 - 2 Log ( a,,? + 1 Fig. 3. Calibration graphs for calcium ion- selective electrodes based on FIA data.A, “Cascade flow cell” [electrode with mem- brane A) ; B, tubular flow cell (electrode with lo-’ I 1 0 - ~ I1.7~10-~ membrane A); and C, tubular flow cell 1 0 - ~ I O - ~ (electrode with membrane B) . [Calcium ion] standardslhn t Time FIA recorder peaks produced by the injec- tion of standard calcium chloride solutions for a tubular electrode (membrane A). Fig. 2. Procedures Three sets of experiments were carried out, namely, analysis of potable and natural water to evaluate the system, studies of the effect of anionic surfactants and analysis of calcium in wash liquors. Analysis of natural and potable watey Water samples (occasionally with background anionic surfactant) were injected into the tubular electrode FIA system alternately with calcium ion standards.Calcium ion levels calculated for the waters from calibration graphs were compared with those obtained by atomic-absorption spectroscopy and EDTA titrations. The atomic-absorption spectrophotometer employed was a Perkin-Elmer, Model 107, instrument set for calcium measurements at 422.7 nm with an air - acetylene flame at 12.5 or 14.0 dm3min-l. The EDTA titrations for calcium were conducted to a calcichrome end- point. Studies of the efect of anionic surfactant For these FIA studies with tubular membrane flow-through electrodes calcium chloride standard solutions contained backgrounds of and M anionic surfactant, namely, sodium dodecylsulphate (SDS) and sodium tetradecylbenzenesulphonate as appropriate. Analysis of calcium in wash liquors Wash liquors were prepared by dissolving “model soap powder” in de-ionised water.This had the following composition (yo mass in parentheses) with water added to 100%: sodium tetradecylbenzenesulphonate (ABS) (7.0) ; tallow soap (2.0) ; sodium silicate (10.0) ; sodium sulphate (10.0) ; sodium tripolyphosphate (STP) (35.0) ; and sodium perborate(V) (25.0). M. The liquors, alternated with M calcium chloride solution, were injected with the 50 mm3 injection valve into the FIA system with the tubular flow-through electrode fitted with membrane B. Calcium chloride was added in concentrations of 1, 2, 3, 4 and 5 x and1360 Analyst, VoZ. 108 Reagents and Materials Calcium bis(di [4- (l,l,3,3-tetramethylbuty1)phenyl]phosphate}12 and dioctyl phenylphosplio- nate13 were prepared as previously described, Specially pure sodium dodecylsulphate (SDS) was obtained from BDH Chemicals and sodium tetradecylbenzenesulphonate (ABS) was obtained from Unilever Research Laboratory.All other materials were of the best laboratory-reagent grade available. FREND et aZ. : FIA WITH TUBULAR ION-SELECTIVE 0 al -0 al .- c - + L E 8 a l IJ 5 min H 10-3 M 10-3 M Ca2 + Ca2 + + Time Fig. 4. Effect of SDS M) on FIA peak of tap water for tubular flow-through electrode (membrane A). Results and Discussion Figs. 2 and 4 demonstrate both of the FIA systems to be fast and reproducible, although there will be small differences between one membrane and another because of variations in thickness. The calibration graphs in Fig. 3 show that the slopes of the linear ranges are similar (about 30mV decade-l) for the systems.However, although the peak height (expressed as AE) for the conventional FIA assembly is consistently larger (see Fig. 3) than for the corresponding solutions with the tubular membrane it is the latter system that has been used in the work discussed below, because it is closed from the atmosphere. Regular injection of standards maintains a check on diurnal variations and electrode deterioration. TABLE I FIA RESPONSE TIMES FOR TUBULAR ELECTRODES WITH MEMBRANES OF COMPOSITIONS A AND B Response times/s r 1 [Ca2+] injected/M A B 10-4 4.9 (f0.3) 5.1 (h0.4) 10-3 3.7 ( f O . l ) 3.6 (hO.2) 10-2 3.0 (f0.5) 3.1 (k0.4) 10-1 3.0 (f0.3) 3.2 (h0.4)November, 1983 ELECTRODES IN THE PRESENCE OF ANIONIC SURFACTANTS 1361 Tubular membrane B gives lower AE values than tubular membrane A, attributable to the increased sodium ion response by the trioctyl phosphate based rnembraiiel4 and raising the base line e.m.f.produced by the sodium of the carrier stream. Electrode response times were measured by setting the chart recorder at a speed of 12 cm min-l and noting the time from the first shift of e.m.f. from the base line to the peak maximum. It can be seen from Table I that the response times of the tubular membrane A and membrane B electrodes are similar and fast. Water Analysis Data The data for calcium analysis of various potable and natural waters of the Cardiff area are summarised in Table 11; the four methods produced similar results, but the electrode data, which are expressed as activities, are slightly lower than the other data.TABLE I1 COMPARISON OF CALCIUM LEVELS FOUND IN WATER SAMPLES BY FIA (TUBULAR ELECTRODE), EDTA TITRATION AND ATOMIC-ABSORPTION SPECTROSCOPY (A&) Samples were taken on 6 separate days in each instance. Calcium content (s.d. for n = 6), p.p.m. r 1 FIA (tubular electrode) Water sample r m (Cardiff area) Membrane A* Membrane R t EDTA AAS Tap water 1 . . . . 33.5 (0.4) 32.0t (0.4) 34.2 (0.7) 34.5 (0.8) Tap water 2 . . . . 29.6 (0.5) 29.2 (0.3) 31.4 (0.9) 32.0 (0.7) Tap water 3 . . . . 36.7 (0.6) 37.1 (0.4) 39.1 (0.7) 37.6 ) 1.0) River water . . . . 35.9 (0.4) 36.1 (0.6) 36.9 (0.6) 37.8 (0.8) Lake water . . . . 31.4 (0.3) 31.6 (0.6) 31.7 (0.8) 32.7 (1.0) * Calcium content expressed as ucaz+ for electrodes with membranes A and B.t No explanation can be offered for this result being significantly different from that of membrane A, although the reason seems to lie with membrane B. Of greater significance in this context is the effect of SDS added to samples under analysis. Thus, Fig. 4 shows that the response by membrane A to calcium in tap water is considerably less when the tap water contains a background of 1 0 - 4 ~ SDS (just 27 p.p.m. of DS-). Further, despite minimum contact of the contaminated tap water with the electrode mem- brane, subsequent injections of the same original sample of tap water and of calcium ion standards give only a proportion of the original AE deflection (Fig. 4). As for the static experiments reported previously the trioctyl phosphate based membrane (B) is much more resistant to the presence of SDS.This is emphasised by the comparative data for membranes A and B in Table 111, which demonstrate that added SDS for membrane TABLE I11 EFFECT OF EXPOSING TUBULAR MEMBRANE ELECTRODES TO M SDS DURING FIA AE (mean of 6 cleterminations)/mV Membrane A Membrane B 7- Y 7- Sample injected Before After* Before After* M Ca2+ . . .. .. . . 57.3 46.9 (81.8) 52.5 52.4 (99.8) M Ca2+ . . . . .. . . 28.2 24.0 (85.1) 26.0 24.9 (95.8) Tap water . . . . . . . . 29.0 24.9 (85.9) 26.1 25.4 (97.3) Tap water + lo-* M SDS . . . . 16.7 (59.6)t 20.8 (79.7)t * Data in parentheses are percentages of AE observed before exposure to t Data in parentheses are percentages of A E observed for tap water without background M SDS. 10-4 M SDS.1362 FREND et aZ.: FIA WITH TUBULAR ION-SELECTIVE Analyst, VoZ. 108 B has a smaller influence on the AE deflection of tap water and is also recovered much more quickly after exposure to Such resistance to the effect of added surfactant by membrane B confirms the trends observed for the trioctyl phosphate solvent mediator membranes in static solution studies. M SDS. Studies on Added Anionic Surfactants to Calcium Ion Standards Fig. 5 summarises the effects of adding lo-* and 1 0 - 3 ~ SDS and ABS to the injected calcium standards for tubular membranes (A and B). Again, the relative resistance of membrane B to the anionic interference is demonstrated in all instances. An interesting phenomenon was observed for membrane A with M SDS background. This was the M calcium, that is, the response to calcium (+ swing) occurred before that to surfactant (- swing).This indicates that the surfactant interference is not immediate, but needs time for the complexes (calcium - surfactant or surfactant - membrane) involved to become established. The phenomenon is probably different from the transient increases in e.m.f. observed for membranes with CI2--Cls alkan-1-01 solvent mediators in static solution studies' at 25 "C. For the data of membrane A (Table IV) it can be seen that the effect is emphasised for the more concentrated calcium solutions and that the first peak showing increases in e.m.f. from the base line of the carrier stream is absent for solutions with M SDS background con- taining less than 10-3 M calcium chloride solution.-shape of the response for solutions containing > 60 30 > E 2 0 60 30 0 I -6 -5 -4 -3 -2 -6 -5 -4 -3 - 2 Log ( ac,z- 1 Fig. 5. Calibration graphs for tubular membrane calcium ion-selective elect- rodes based on FIA data illustrating the effect of added anionic surfactant. (a) Membrane A for SDS; (b) membrane B for SDS; (c) membrane A for ABS; and ( d ) membrane B for ABS. A, Pure calcium chloride solutions; B, calcium chloride in M anionic surfac- tant. M anionic surfactant; and C, calcium chloride in Calcium Ion Analysis in Wash Liquors Table V summarises data obtained for free calcium ion analysis in wash liquors. Although the tubular membrane FIA arrangement is sensitive to low calcium ion activities, it can beNovember, 1983 ELECTRODES IN THE PRESENCE OF ANIONIC SURFACTANTS TABLE IV RESPONSE TIMES OF POSITIVE AND NEGATIVE FIA PEAKS OF TUBULAR ELECTRODE (MEMBRANE A) TO CALCIUM AND DODECYL SULPHATE IONS I N SINGLE AND MIXED SOLUTIONS 1363 Solution composition Ca2+ solutions with no added SDS .. Ca2+ solutions in M SDS. . . . Concentration of Ca2+/w . 10-1 10-2 10-3 . 10-1 10-2 5 x 10-3 10-3 10-4 10-5 10-6 Nil Response time from injection/s Positive peak Negative peak 5.9 - 6.5 - 6.7 - 6.5 29.2 6.4 20.1 5.0 12.7 5.1 8.6 5.6 5.6 5.4 - 5.4 f A \ - - - seen from Table V that the calcium ISE with a membrane of calcium bis{di[4-(1,1,3,3-tetra- methylbutyl)phenyl]phosphate} and trioctyl phosphate in poly(viny1 chloride) can be used for calcium ion activity determinations in wash liquors with moderate levels of the “model soap powder” detergent.However, the tubular membrane flow-through FIA elec- trode is not sufficiently sensitive for the low calcium ion levels that occur in solutions con- taining >3 g dm-3 of “model soap powder” and in solutions containing lower levels of total calcium for 3 g dm-3 “model soap powder” solutions. Therefore, data for these solutions are not quoted in Table V, but it is emphasised that conventional electrodes of membrane B composition appear to be sufficiently sensitive for use in the solutions under static conditions.7 TABLE V COMPARISON OF CALCIUM ION ACTIVITIES FOUND BY TUBULAR ELECTRODE BY FIA AND BY THE STATIC METHOD FOR MEMBRANE B ELECTRODES Calcium added/mM 1 2 3 4 5 4 5 Concentration of model soap powderlg dm-3 1 1 1 1 1 3 3 Experimental acaz+ (s.d.for n = 6 ) / ~ Static mode 1.7 (0.5) x 2.4 (1.4) x 6.4 (2.7) x 1.2 (0.4) x 10-3 1.6 (0.3) x 1.0 (0.4) x 10-4 2.1 (0.6) x 10-4 FIA Calculated UC~O+/M 6.3 (0.5) x 1.7 x 10-5 3.4 (0.3) x 10-4 2.6 x 10-4 6.3 (0.7) x 7.6 x 10-4 1.4 (0.1) x 10-3 1.6 x 10-3 2.2 (0.2) x 10-3 2.7 x 10-3 1.6 (0.3) x 1.6 x 10-4 2.5 (0.1) x 10-4 1.9 x 10-4 Conclusion These studies confirm that calcium ISEs based on calcium bis (di [a-( 1,1,3,3-tetramethyl- butyl)phenyl]phosphate} and a trioctyl phosphate solvent mediator are able to resist inter- ferences from SDS and ABS anionic surfactants. The electrodes have the potential for use in free calcium ion analysis in flow injection modes in the presence of moderate levels of detergent as long as the free calcium ion level does not fall below loM4 M, but this falls short of the free calcium ion level determinable ( M) by the electrodes under static con- dit ions.The authors thank the Science and Engineering Research Council for a studentship (to A. J.F.) under the scheme for Cooperative Awards in Science and Engineering in conjunction with Unilever Research Laboratory, Port Sunlight. References 1. 2. RbiiCka, J., and Hansen, E. H., Anal. Chirn. Acla, 1975, 78, 145. Hansen, E. H., Ghose, A. K., and RbiiEka, J.. Analyst, 1977, 102, 705.1364 3. FREND, MOODY, THOMAS AND BIRCH 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Hansen, E. H., and RbiiEka, J., RSC International Symposium on Electroanalysis in Clinical, Birch, B. J., and Clarke, D. E., Anal. Chim. Acta, 1973, 67, 387. Llenado, R. A., Anal. Chem., 1975, 47, 2243. Craggs, A., Moody, G. J., Thomas, J . D. R., and Birch, B. J , , Analyst, 1980, 105, 426. Frend, A., Moody, G. J., Thomas, J . D. R., and Birch, B. J., AnaZyst, 1983, 108, 1072. Moody, G. J., and Thomas, J. D. R., Ion-Sel. Electrode Rev., 1979, 1, 3. Moody, G. J., Oke, R. B., and Thomas, J. D. R., Analyst, 1970, 95, 910. Craggs, A., Moody, G. J., and Thomas, J. D. R., J . Chewz. Educ., 1974, 51, 541. Cockroft, R. N., PhD Thesis, University of London, 1977. Craggs, A., Delduca, P. G., Keil, L., Key, B. J., Moody, G. J., and Thomas, J. D. R., J . Inorg. Nucl. Craggs, A., Delduca, P. G., Keil, L., Moody, G. J., and Thomas, J . D. R., J . Inovg. Nucl. Chem., Moody, G. J., Nassory, N. S., and Thomas, J. D. R., Analyst, 1978, 103, 68. Environmental and Pharmaceutical Chemistry, 13-16 April 1981, IJWIST, Cardiff, paper 31. Chem., 1978, 40, 1483. 1978, 40, 1943. Received April 20th, 1983 Accepted June 9th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801357

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, November, 1983, Vol. 108, PP. 1365-1373 1365 Accuracy of Determination of Total Suspended Solids in River Waters : Analytical Quality Control in the Harmonised Monitoring Scheme Analytical Quality Control (Harmonised Monitoring) Committee* Water Research Centre, P.O. Box 16, Henley Road, Medinenham, Mavlow, Buckinghamshire, SL7 2HD The Department of the Environment, in collaboration with the Regional Water Authorities, has initiated a Scheme for the Harmonised Monitoring of the Quality of Inland Fresh Waters in England and Wales. The Scottish Development Department has introduced a similar scheme in Scotland in collaboration with the River Purification Boards. To achieve the required comparability of results from all laboratories involved, each laboratory takes part in an Analytical Quality Control (AQC) scheme; this work is co-ordinated by the Water Research Centre.The general approach adopted to AQC has been described and this paper presents the tests made and results obtained in the determination of total suspended solids in river waters. Broadly, each of the 1 1 participating laboratories achieved total errors not greater than &20% of the determinand concentration or 2 mgl-1 (whichever was the larger for different sample concentrations). Keywords : River-water analysis ; total suspended solids determination ; accuracy of results ; inter-laboratory comparability ; analytical quality control The scheme for the Harmonised Monitoring of the Quality of Inland Fresh Water has been described in detai1.l It is intended to provide objective data on river water quality so that accurate assessments can be made of long-term trends in the qualities of rivers and of the amount of materials discharged by them to the sea.The scheme complements monitoring carried out for regional or local purposes and one of its essential aims is to achieve compar- ability of results from all the participating laboratories. To that end, special investigations have been made to establish suitable locations for sampling and to define the necessary sam- pling frequencies. Sampling procedures have been recommended and, to ensure that subsequent analyses do not introduce unacceptably large errors, each participating laboratory carries out a specially designed programme of tests to ensure that their analytical results are of adequate accuracy for the scheme.The Water Research Centre (WRC) is under contract to the Department of the Environ- ment to advise on and co-ordinate this Analytical Quality Control (AQC) programme. The need for, and details of an approach to, a planned AQC system for this and similar schemes have been discussed in some detail elsewhere.2 In view of the growing interest in achieving comparable results from each of a number of laboratories, it was thought useful to describe the AQC work for the Harmonised Monitoring Scheme and to present the results obtained for different determinands. This paper considers the determination of total sus- pended solids, earlier papers describe the work for ~hloride,~ ammoniacal nitrogen4 and total oxidised nitrogen and nitrate5 ; subsequent papers will deal with other determinands of general importance in rivers.Some of the earlier work presented in this paper was originally published as a WRC Technical Reporte6 Organisation of the Work A Committee was formed to plan the collaborative work, and has representativest from the Department of the Environment, Scottish Development Department, each Regional Water Authority (RWA), Scottish River Purification Boards and the WRC. This Committee * All correspondence should be addressed to S. Blake, a t the Water Research Centre. t The names of representatives a t the time this work reported here was carried out are given in the Appendix.1366 AQC COMMITTEE: ACCURACY OF DETERMINATION Analyst, VoZ. 108 decided to adopt the approach to the AQC described elsewhere,2 each determinand being studied in two phases.One laboratory in each of the ten RWAs and one in Scotland participated, the WRC acting as the co-ordinating laboratory2 ; 11 laboratories were generally involved. Having obtained satisfactory results in Phase (i), those laboratories act as co-ordinators of tests within RWAs and Scotland. Certain RWAs are not involved in this phase because all analyses for Harmonised Monitoring are made by one laboratory. Phase (i). Phase (ii). This paper deals only with Phase (i). Required Analytical Accuracy The following requirements were agreed by the Committee to represent the targets at which to aim for total suspended solids2: maximum tolerable bias, 10% of the determinand con- centration or 1 mgl-l, whichever is the greater; and maximum tolerable total standard deviation, 5% of the determinand concentration or 0.05 mg l-l, whichever is the greater.Analytical Quality Control The approach followed was exactly as presented previously2; no attempt is made here, therefore, to explain the reasons underlying the various activities described below. The participating laboratories were as follows : Anglian Water Authority, Welland and Nene River Division Laboratory,* Oundle; Northumbrian Water Authority, Headquarters Laboratory, Gosforth ; North West Water Authority, Rivers Division Laboratory, Warrington ; Severn- Trent Water Authority, Regional Laboratory, Finham ; Southern Water Authority, Resource Planning Laboratory, Winchester ; South West Water Authority, Rivers and Marine Labora- tory, Exeter ; Thames Water Authority, Thames Conservancy Division Laboratory, Reading; Welsh Water Authority, Chester Area Laboratory, Chester ; Wessex Water Authority, Bristol Avon Divisional Laboratory, Saltford ; Yorkshire Water Authority, Headquarters Laboratory, Leeds ; and Forth River Purification Board, Headquarters Laboratory, Edinburgh.The sequence of participating laboratories in the above list does not relate to the order of number- ing of laboratories in the tables. Choice of Analytical Methods All laboratories used the 1972 DOE filtration procedure,’t together with the following refinements, which it was agreed by the Committee should be applied for this exercise but which are not specified in the DOE method. (i) To ensure complete transfer of suspended matter to the filter, the measuring cylinder should be rinsed carefully three times with 5-10 ml of water and the washings transferred into the filter.(ii) A guide to the sample volume used for the test is as follows, where the values refer to the expected suspended solids concentration (mg 1-l) and sample volume (ml), respectively : <20, 500; 20-100, 250; and >loo, 100. The results of preliminary checks carried out by the Committee suggest that errors can arise using this method when samples with a high content of total dissolved solids (> 1000 mg 1-l) are examined. Investigations into the magnitude and the exact cause of this effect are being carried out by the Committee. An adaptation to the basic method sometimes used by laboratories involves the insertion of an inert gauze or net between the filter-paper and the funnel base.This arrangement is intended to provide uniform filtration over the whole paper rather than over the regions * The original laboratory participating in the scheme for the Anglian Water Authority, the Welland and Nene River Division Laboratory, was replaced, after the precision and main inter-laboratory bias tests, by the Regional Standards Laboratory a t Cambridge. The results of the precision and main inter-laboratory bias tests reported here are those of the Welland and Nene River Division Laboratory. Those for “follow- up” tests were obtained by the Regional Standards Laboratory, after i t too had satisfactorily completed the tests of precision.All laboratories used Whatman filters. These filters are named here for identification purposes only and this should not be taken as an endorsement or otherwise of these filters, relative to any others of equivalent type, by the AQC (HM) Committee, the Water Research Centre or the Department of the Environment. t The use of Whatman GF/C or equivalent filters is specified in the analytical method.November, 1983 OF TOTAL SUSPENDED SOLIDS IN RIVER WATERS 1367 immediately above the holes in the base of the funnel. The adoption of this procedure, how- ever, can lead to poor results compared with the normal method for real samples. The magni- tude of the difference and the effects of the gauze insert on other performance characteristics of the method are still being investigated.In this work only laboratory 1 used the gauze insert and its results give no indication of a consistent inter-laboratory bias. Within-laboratory Precision Tests On each of 5 d each laboratory made triplicate analyses, in random order, of two standard kaolin suspensions and a local river water. Triplicate blank determinations were also made each day by applying the analytical procedure to aliquots of de-ionised or distilled water used to prepare the standard kaolin* suspensions. The concentrations of kaolin in the two standard suspensions were 5 and 50 mg 1-1, the suspensions being prepared in all laboratories from samples of the same batch of Light Kaolin BP circulated by the WRC. Each laboratory collected its own river water sample daily from a local source.The concentrations of natural suspended solids in the river waters varied from one laboratory to another and from day to day. On completion of the tests each laboratory analysed its results for the blank and standard solutions statistically to obtain estimates of the within-batch (sw), between-batch (sb) and total (st) standard deviationsg where st = (sw2 + sb2)*. It was considered that these calcula- tions would be inappropriate for the river water because of possible day to day changes in the determinand concentration in real samples. Such changes would lead to falsely high values for sb and st. No attempt was made therefore to calculate either sb or st for the river water. The sw values are thus the only information directly available for the precision of real samples.TABLE I RESULTS FROM WITHIN-LABORATORY PRECISION TESTS (NOT BLANK CORRECTED) All results are in milligrams per litre except where specified as a percentage. The target total standard deviation is 5% of the determinand concentration, or 0.5 mg 1-1 (whichever is larger). sw, sb and st refer to the estimates of within-batch, between-batch and total standard deviation, respectively, and have 10, 4 and 4-14 degrees of freedom, respectively. s t is the estimate of standard deviation of any one result in any one batch of analyses. A local river water chosen by each laboratory Relative Within-batch total relative Blank determination Kaolin standard (5 mg 1-l) Kaolin standard (50 mg 1-I) standard standard Labora- ,-A--q r------h-----~ * 7 deviation, deviation, tory No.Mean sw s t Mean sw sb st Mean* sw Sb st % Mean sp % 1 0.21 0.24 0.28 4.69 0.23 N.s.t 0.27 47.9 0.46 N.s. 0.59 1.2 7.8 0.43 5.5 t$ -0.12 0.19 0.19 4.60 0.15 0.20 0.25 46.3 0.77 N.s. 0.82 1.6 18.7 0.56 3.0 -0.63 0.05 0.95* 4.08 0.15 0.61 0.635 43.2 0.57 1.75 1.84 3.7 1.0 0.21 20.7 4 -0.87 0.47 0.81; 4.23 0.31 0.47 0.566 47.0 1.33 1.74 2.19 4.4 24.1 0.64 2.7 5 -0.07 0.54 0.688 5.61 0.21 0.22 0.30 49.9 1.20 N.s. 1.61 3.2 42.7 2.81 6.68 6 -0.39 0.48 0.48 3.95 0.42 N.s. 0.539 43.5 0.97 1.99 2.228 4.4 6.8 0.588 8.5 7 0.00 0.05 0.07 4.52 0.17 0.00 0.17 47.2 0.68 0.82 1.06 2.1 5.9 0.29 4.9 8 -0.37 0.21 0.23 4.61 0.32 N.s. 0.35 47.6 0.56 0 0.56 1.1 60.1 0.87 1.4 9 -0.29 0.46 0.46 3.99 0.23 0.26 0.35 36.0 0.88 0 0.88 1.8 14.3 0.29 2.0 10 0.00 0.06 0.07 3.32 0.27 N.s. 0.36 43.0 1.46 2.45 2.859 5.78 25.0 1.43 6.75 11 0.25 0.14 0.16 5.24 0.15 N.s.0.20 49.4 0.37 0.40 0.55 1.1 5.0 0.20 3.9 * Does not meet targets. t N.s Not significant. $ Labbratory 2 repeated its tests on the 5 mg I-’ kaolin standard because of an initial marginal failure to achieve the target total standard 8 Not significantly greater than the target value. deviation. Blanks were also repeated and these were used to calculate the corrected result given in Table 11. The values of st for the two standard suspensions were compared with the appropriate target using the F-test and were accepted as satisfactory provided st was not significantly greater (0.05 probability level) than the appropriate target.2 Most of the laboratories do not perform blank correction in routine analysis.The results presented in Table I were, therefore, calculated without blank correction. The possibility was noted, however, of between-batch variability in the results for standard * Kaolin was chosen to prepare standard suspensions for the tests because the material, being a clay, represents a type of solid often found in natural river water samples but, unlike solids from such samples, can readily be obtained in a pure form complying with a pharmacopoeia specification. See reference 8 for further details.1368 AQC COMMITTEE ACCURACY OF l>ETEIIMINATION A?zalyst, 'Vd. 108 suspensions arising from day to day variation in any particulate matter present in the de- ionised or distilled water used to prepare these standards.Results were, therefore, also calculated with blank correction using the mean value of the three blanks analysed in each batch, to eliminate the effects of any such variations. These results are presented in Table 11. Three laboratories (6,9 and 11) do undertake blank correction routinely, and the precision data, presented against these laboratories in parentheses in Table 11, apply to their normal blank correction procedure. Nearly all laboratories achieved standard deviations that were not significantly (0.05 probability level) greater than the target value for both blank corrected and uncorrected results. Only with laboratories 3 and 4, for the blank determination, was the precision target not achieved. However, these st values for the blank determination may not reflect directly the precision for real samples in routine analysis, because, as stated above, they would involve any variation in the suspended solids content of the de-ionised or distilled water used.Both laboratories achieved the precision target in all subsequent work carried out on samples. All the estimates of total standard deviation quoted refer to kaolin suspension and not to river waters. However, it should also be noted that none of the estimates of within-batch standard deviation for the river waters significantly (0.05 probability level) exceeded the target for total standard deviation (see Table I). Taken together, these findings suggest (but do not prove) that acceptable analytical pre- cision can be achieved for real samples of river water.It should be noted, however, that very low recoveries of kaolin from the standard suspen- sions used in these precision tests were obtained in some laboratories. This aspect of perform- ance was markedly improved between the precision and inter-laboratory bias tests, although complete recovery of kaolin in all laboratories has remained unattainable. This is discussed further in subsequent sections. The satisfactory completion of these tests in all laboratories indicated that precision was adequate and the next stage of the AQC work was started. Both laboratories were informed and the discrepancies investigated. TABLE I1 RESULTS FROM WITHIN-LABORATORY PRECISION TESTS (MEAN BLANK CORRECTED) All results are in miligrams per litre except \\.here specified as a, prrcentage.The target total standard deviation is 5% of the determinand concentration, or 0.5 mg 1-l (whichever is larger). sw, sb and st refer to the estimates of within-batch, between-batch and total standard deviation, respectively, and have 10, 4 and 4-14 degrees of freedom, respectively. st is the estimate of standard deviation of any one result in any one batch oi analyses. Values in parentheses apply to the laboratories' routine blank correction procedure (see text for details). Relative total standard A A 7 deviation, Kaolin standard (50 mg 1-1) 7 7--- Kaolin standard (5 mg 1-1) Laboratory , No. Mean sw sb S t Mean* SW sb S t % 1 4.48 0.23 0.21 0.31 4.57 0.15 N.s.f: 0.16 ? 4.71 0.15 0.34 0.37 47.7 0.46 N.s. 0.53 1.1 46.4 0.77 N.s. 0.82 1.7 43.8 0.55 1.45 1.55 3.5 4 5.09 0.31 0.31 0.43 47.9 1.33 N.S.1.69 3.5 5 5.68 0.21 0.65 0.696 49.9 1.20 1.38 1.83 3.7 6 4.33 0.42 (0.63) N.s. (N.s.) 0.47u(0.635) 43.9 0.97 (1.25) 1.84 (1.86) 2.08 (2.245) 4.7 (5.15) 7 4.52 0.17 N.s. 0.18 47.2 0.68 0.88 1.11 2.3 8 4.99 0.32 N.s. 0.39 48.0 0.56 0 0.56 1.2 9 4.28 0.23 (0.52) N.s. (N.s.) 0.26 (0.525) 36.3 0.88 (0.84) 0.88 (0.84) 2.4 (2.3) 10 3.32 0.27 N.s. 0.36 43.0 1.46 ;.!? 2.855 6.65 11 4.99 0.15 (0.25) N.s. (0) 0.18 (0.25) 49.1 0.37 (0.39) 0.35 (0.34) 0.51 (0.52) 1.0 (1.1) Does not meet targets. t See the third footnote to Table I. f: N.s., Not significant. 5 Not significantly greater than the target value. Establishment of Quality Control Charts Each laboratory set up a preliminary statistical quality control chart2 based on the analysis of a standard solution in each subsequent batch of analyses. These charts are intended to aid the continuing, long-term assessment of accuracy in each laboratory and are not discussed further here.November, 1983 OF TOTAL SUSPENDED SOLIDS IN RIVER WATERS 1369 Tests of Between-laboratory Bias To complete this initial phase of AQC, direct checks of between-laboratory bias were made as follows : ten standard kaolin suspensions were prepared at accurately known concentrations in separate Pyrex glass bottles by the WRC.The concentration of each suspension was close to 50 mg 1-l. Laboratories were told only the sample volume to be used for the tests. The results are summarised in Table 111. The bottles of suspension were distributed, one to each laboratory. The suspensions were each analysed on each of 4 d.TABLE I11 RESULTS FROM THE SUSPENDED SOLIDS INTER-LABORATORY BIAS TEST The results give a mean recovery of 97.17%. determinand concentration or 1 mg 1-1 (whichever is the larger). The target is a maximum tolerable bias of 10% of the Laboratory I 1 did riot take part in this test but gave acceptable results in subsequent “follow-up” tests of between-laboratory bias. Laboratory No. 1 2 3 4 5 6 7 8 9 10 Nominal kaolin concentration/ mg 1-l 50.45 50.36 50.36 50.27 49.55 49.55 50.55 50.27 50.36 50.55 Observed mean kaolin concentration / mg 1-1 48.6 48.7 47.1 48.8 50.5 48.8 50.5 48.4 49.4 47.3 Maximum possible Recovery, bias from nominal % value, yo 96.3 -7.1 96.7 -6.8 93.5 - 10.3* 97.1 -6.5 101.7 +6.1 98.5 - 3.4 99.9 -1.0 96.3 -9.1 98.1 -5.3 93.6 -9.5 Maximum possible bias assuming 9 7.1 7 yo recovery, % - 4.4 -4.0 -7.7 -3.8 +9.1 $- 3.3 + 3.7 - 6.5 +4.4 - 6.9 * Does not meet target.To assess whether or not the bias of any laboratory exceeded the target value, the following procedure was adopted, Let the mean result and its 90% confidence interval for laboratory i be denoted by Zi & Li, and let the true concentration of the distributed suspension for laboratory i be denoted by Ti. The value of the maximum possible bias at laboratory i (95% confidence level) was then calculated as follows : 100(Zi + Li - Ti)/Ti if Zi > Ti or 100(Zi - Li - T,)/T, if Zi < Ti Ti was obtained from the mass of kaolin (after drying at 105 “C) used to prepare the suspen- sion. The results in Table I11 show that all laboratories met the target with the exception of laboratory 3.However that laboratory fell outside the target so marginally (maximum possible bias = 10.3%) that the result is regarded as acceptable. The results presented in Table I11 reveal a tendency for the over-all mean recoveries to be low. Recoveries of less than lOOyo were obtained by nine out of the ten laboratories. The reason for this tendency is not known with certainty and the problem is examined in more detail in the Discussion. In view of this slightly low recovery, the bias of each laboratory from the mean of all laboratories was calculated using this equation with TiF/lOO in place of Ti, where F = mean recovery of all the laboratories (as a percentage). These results are also given in Table 111, which shows that no laboratory obtained a recovery differing by more than the target of 10% from the mean recovery of all laboratories.Routine AQC To attempt to ensure that the required accuracy of results is maintained, AQC is now an integral part of the routine analyses for the Harmonised Monitoring Scheme. As mentioned above, prime reliance for this purpme is placed on within-laboratory AQC using statistical1370 AQC COMMITTEE : ACCURACY OF DETERMINATION AnaZyst, VoZ. 108 quality control charts. However, to obtain direct checks of between-laboratory bias, standard kaolin suspensions, prepared at accurately known concentrations by the WRC, are distributed at intervals to all laboratories. The results of such tests (over a 3 year period) are of value in indicating the efficiency of the AQC work and are summarised in Table IV.Each of these tests was carried out as described under Tests of Between-laboratory Bias, except that the four replicate analyses were all made in one batch of analyses. TABLE IV RESULTS FROM ROUTINE BETWEEN-LABORATORY BIAS TESTS The mean kaolin concentration refers to the concentration of suspended solids found by that laboratory in the suspension distributed by the WRC. Each laboratory received a suspension with an accurately known concentration of suspended solids, but this concentration varied from laboratory to laboratory. Standard 1 : Standard 2: Standard 3: Standard 4: Standard 5: April 1978 March 1980 October 1980 March 1981 October 1981 r-r A 7 ~ - - - - - - - 7 -----, - Mean upper Laboratory Mean/ Upper limit Mean/ Upper limit Mean/ Upper limit Mean/ Upper limit Mean/ Upper limit limit for NO.mg 1-l for bias,* % mg I-' for bias,* % mg 1-l for bias,* yo mg 1-' for bias,* % mg 1-l for bias,* yo bias, 7 yo 1 10.70 +16.4$ 2 10.30 +8.9 3 9.40 -4.8 4 8.95 -12.5t 5 9.50 -4.9- 6 10.05 +7.1 7 9.53 -4.4 8 8.93 -16.4$ 9 9.15 -8.4 10 9.75 +5.0 11 10.12 +6.0 Mean 26.20 +3.1 22.00 +9.7 31.10 25.50 -5.9 20.80 -5.9 30.80 25.98 -6.0 20.00 -5.7 27.20 24.60 -8.1 21.00 +2.3 29.90 25.66 -6.4 21.20 +7.2 31.30 24.70 -7.5 19.00 -9.0 29.60 27.75 +6.5 19.80 -10.54 31.50 25.20 -4.5 21.00 $5.8 30.50 26.75 +4.9 22.75 +11.4$ 34.25 26.90 +6.6 20.80 +7.2 28.34 26.70 +9.3 20.90 +2.2 32.20 -6.4 34.10 +3.7 -13.74 29.75 -15.5$ -4.4 34.60 +4.4 +9.3 30.50 -14.14 -8.3 34.40 $1.7 -6.5 32.308 -11.34 -1.5 32.40 -8.3 +4.9 35.20 4-4.7 +17.4$ 36.00 +9.0 +6.8 35.00 +7.9 +7.6 36.30 +8.3 recovery, % -.96.70 95.26 91.85 94.84 92.56 Maximum possible bias (90% confidence limits) calculated with respect to mean recovery of laboratories. t The mean upper limit for bias is the mean of the individual upper limits for bias from the five tests. 1 Target value for bias exceeded. 5 Mean not included in mean of all laboratories. The laboratory's own control chart indicated that problems had occurred on analysis. After extensive investigations no apparent cause for the result was found. + 5.3 + 1.3 - 9.1 + 6.0 - 2.1 + 0.6 + 0.6 - 10.1 + 0.5 - 0.7 6.7 this particular The bias of each laboratory is, as in the initial inter-laboratory bias test, measured from the mean of all laboratories, as the over-all mean recoveries for the tests tended to be low. Table IV shows that the upper limit (95% confidence level) for bias is usually less than 10% and that the mean of the individual upper limits is less than 10% for each laboratory (except laboratory 8, for which it is only 10.lyo).Specifically the target was exceeded on only ten out of a possible 55 occasions and, of these ten, in only four was the upper limit for bias greater than 1.5 times the target value, i.e., laboratory 1, sample 1 ; laboratory 8, sample 1 ; laboratory 4, sample 4; and laboratory 3, sample 5. Whilst the use of a 1-d test* has shown that freedom from bias has been maintained, there has been no direct check on the total standard deviation.Thus, the routine control of pre- cision (along with certain sources of bias) is maintained by the use of control charts. It appears that satisfactory results have been maintained, although the occasional devi- ations emphasise the need for continuing AQC. In conjunction with the most recent test of between-laboratory bias, a standard made up from micro-crystalline cellulose (first used for this purpose by Severn-Trent Water Authority) was also distributed for analysis using the same method adopted for kaolin suspensions. The results from this test are presented in Table V. It can be seen that the mean recovery of micro-crystalline cellulose was just over loo%, with only one laboratory (laboratory 8) failing to meet the target of 10% for the upper limit for bias.Discussion The targets for systematic and random error imply that the tolerable total error of individual analytical resulfs is 20% or 2 mg l-l, whichever is the greater. Although the former target may * Initially the tests were to be conducted over a 4-d period but owing to the lack of available time in the laboratories, after the initial bias test all subsequent tests were performed in one batch of analyses on 1 a.November, 1983 OF TOTAL SUSPENDED SOLIDS IN RIVER WATERS TABLE V RESULTS OF THE SUSPENDED SOLIDS INTER-LABORATORY BIAS TEST- MICRO-CRYSTALLINE CELLULOSE (MCC) STANDARD SOLUTION (OCTOBER 1981) The results give a mean recovery of 101.50%. The target is a maximum tolerable bias of 10% of the determinand concentration or 1 mgl-1 (which- ever is the larger).Nominal MCC Observed mean MCC Maximum possible Laboratory concentration/ concentration/ Recovery, bias from nominal No. mg 1-1 mg 1-l % value, yo 1 35.82 36.6 102.2 3.8 2 37.27 38.4 103.0 4.1 4 36.09 37.0 102.5 2.5 5 36.73 37.1 101.0 1.7 7 36.73 37.2 101.3 1.3 8* 36.82 32.7 88.8 - 15.4t 9 36.91 38.5 104.3 9.8 10 37.36 37.2 99.6 - 1.6 11 36.73 38.1 103.7 4.4 3 36.18 35.6 98.5 -3.1 6 36.91 36.5 98.9 - 6.3 * Mean not included in mean of all laboratories-statistical outlier. t Does not meet target. 1371 seem lax, it should be noted that only at the expense of substantially greater effort could targets smaller than the above be achieved routinely in most laboratories. The target values used seem, therefore, to represent a reasonable compromise between the degree of accuracy and the work required to achieve it. The preliminary tests of precision and between-laboratory bias demonstrated that the laboratories were capable of achieving adequate accuracy.The results of subsequent tests indicated that this standard of accuracy was usually maintained, with relatively few failures which were not merely marginal. Nevertheless, the frequency and seriousness of the failures that were encountered are some- what greater than with the other determinands r e p ~ r t e d , ~ - ~ emphasising the greater difficult- ies with this (non-specific) determinand and the need, as with the others, for continuing AQC. The sequential approach followed in the AQC programme involves a relatively large amount of work in each laboratory and a relatively long period of time is necessary to complete all the tests. However, these very points provide many opportunities for unsuspected errors to reveal themselves, thereby facilitating recognition and elimination of problems so that a permanently sound basis is established for routine achievement of the required accuracy.2 Mention has already been made of the tendency for over-all mean recoveries of kaolin to be slightly low and of the practice of judging the acceptability of results against the mean recovery of all the laboratories in consequence of this.The reason for the low recovery is not certain, but possibilities include the difficulty in ensuring complete quantitative transfer of suspended materials in the original sample bottle to the filter-paper,B and a fraction of the finer kaolin particles passing through the filter-paper. Prior to this work an investigation8 of the suitability of various substances as standard materials for these tests was conducted.Kaolin was finally selected, having suitable physico- chemical properties (insolubility, stability, ability to form and maintain a good suspension) and because it is a clay mineral and in that sense representative of the main type of material usually found in riverine suspended matter, Although particle-size analysis showed8 that a portion of the particles were smaller than the nominal pore size of the filters employed in the analytical method used by the laboratories, microscopic examination of filtrates failed to disclose the presence of kaolin that had passed through the filter.On balance, therefore, it was not considered that the low recoveries (encountered in the preliminary work as well as in this work) could be attributed to the passage of fine kaolin particles through the filter. More- over, other tests8 showed that some loss of kaolin could arise when the sample to be filtered was measured using glass measuring cylinders, and then filtered, as in the method routinely applied. It was therefore concluded that the slightly low recoveries obtained in the early tests reported here were probably also attributable to incomplete transfer.1372 AQC COMMITTEE : ACCURACY OF DETERMINATION Analyst, VoZ. 108 The occurrence of over-all mean recoveries in the range 92-95% in recent follow-up tests (using the same, homogenised bulk kaolin sample) has prompted re-examination of the prob- lem.Although the observation of higher and essentially quantitative recoveries with micro- crystalline cellulose (Table V) does not prove that passage of kaolin through the filter is occurring,* it was felt desirable to try to produce a sample of kaolin with a smaller proportion of particles of less than the nominal pore size of the filters, in an attempt to eliminate any doubt concerning the origin of the low recoveries. Such work is in progress and meanwhile follow-up tests will involve the use of both kaolin and micro-crystalline cellulose. It is recognised, however, that the latter is less representative of riverine suspended matter than is kaolin, and it is for this reasQn that micro-crystalline cellulose will not be used as the sole standard material, in spite of its more uniform particle size.The stability of suspensions is a further factor affecting the suitability of materials as standards. Stability has been checked and found to be satisfactory for the kaolin suspensions used in the tests reported here. In routine determinations, however, sample instability could also be a source of error and tests are currently being carried out by the laboratories in con- junction with the WRC to investigate this aspect of suspended solids determination. Sampling is a possible cause of considerable error in suspended solids determinations because of the difficulties in taking representative samples from a river. Such factors as, for example, the orientation of the sample bottle during sampling would be expected to have a considerable effect on the representativeness of the sample, This aspect of suspended solids determination requires further investigation.Conclusions For the precision and bias tests, all participants achieved the required accuracy for suspended solids determinations on kaolin suspensions. In the first test, however, there was an over-all tendency for the results to be about 3% lower than the true concentration and subsequent bias tests have continued to show low recoveries of kaolin. Work to determine the cause of this is currently being performed by the WRC. There is no direct evidence on the accuracy of determination of a real sample-sampling problems and sample instability prevented the use of suspended solids from a river water in both inter-laboratory bias tests and the estimation of the total standard deviation of results.However, the results for the within-batch standard deviations for river waters and the results for the kaolin suspensions together suggest that acceptable analytical accuracy can be achieved for river waters. Subse- quent analytical quality control is, in addition to normal precautions, based mainly on the use of quality control charts and the analysis of samples distributed at regular intervals by the WRC. The results of the work show that the sequential approach proposed by WRC and adopted by the Analytical Quality Control (Harmonised Monitoring) Committee for achieving com- parable analytical results has now proved successful for the determination of suspended solids.This is a non-specific determinand, being defined by the analytical method adopted, and this work has provided valuable experience of the special problems that arise with such determi- nands. Previously the approach had only been applied to specific determinands, i.e., ~hloride,~ ammoniacal nitrogen,* and total osidised nitrogen and nitrate.5 I t is now being applied in successive studies of other determinands, both specific and non-specific, the results for which will be reported in subsequent papers. This work has checked the control of analytical errors in laboratories, but has not included checks on sampling errors and errors due to sample instability, both of which may cause an analytical result for a sample not to represent the suspended solids content of the river sampled.The investigation of both these types of error is under consideration by the Committee and work has started on tests of sample stability in some of the participating laboratories. Continuing care is needed to ensure that this position is maintained. Although the analytical work reported was performed by the Regional Water Authority and River Purification Board Laboratories, the co-ordination of this work was carried out by the WRC under contract to the Department of the Environment whose permission to publish is acknowledged. * Micro-crystalline cellulose has very different physico-chemical properties from kaolin, and may be less prone to adsorption on to the walls of containers, such as measuring cylinders.November, 1983 OF TOTAL SUSPENDED SOLIDS I N RIVER WATERS 1373 Appendix The following are (or have been) members of the Analytical Quality Control (Harmonised Monitoring) Committee in the period during which this work was performed: Department of the Environment, Dr. P.R. Hinchcliffe, Mr. P. Tan, Mr. R. Donachie, Mr. P. Kingslan, Mr. P. H. Garnett, Mr. N. Taylor, Dr. E. A. Simpson and Mr. J. G. Flint; Water Research Centre, Mr. A. L. Wilson, Mr. R. V. Cheeseman, Dr. D. T. E. Hunt, Mr. D. J. Dewey and Mr. M. J. Gardner; Anglian Water Authority, Dr. B. T. Croll; Forth River Purification Board, Mr. I. R. M. Black and Mr. T. Williamson; Northumbrian Water Authority, Mr. W. Wollers and Mr. B. D. Ravenscroft; North West Water Authority, Mr. P. Morries and Mr. J. B. Allcroft ; Severn-Trent Water Authority, Dr. K. C. Wheatstone; Southern Water Authority, Mr. M. J. Beard; South West Water Authority, Mr. B. Milford; Thames Water Authority, Mr. D. V. Hopkin; Welsh Water Authority, Mr. B. E. P. Clement and Mr. J. R. Borland; Wessex Water Authority, Mr. J. G. Jones; Yorkshire Water Authority, Mr. M. G. Firth and Mr. N. Croft; Water Data Unit, Mr. D. Rodda and Mrs. C. Browne; ScottishDevelopment Department, Mr. J. C. Lambie and Mr. T. Hooton; and Welsh Office, Mr. J. Saunders. 1. 2. 3. 4. 5. 6. 7. 8 . 9. References Simpson, E. A., J . Inst. Water Eng. Sci., 1978, 32, 45. Wilson, A. L., Analyst, 1979, 104, 273. Analytical Quality Control (Harmonised Monitoring) Committee, Analyst, 1979, 104, 290. Analytical Quality Control (Harmonised Monitoring) Committee, Analyst, 1982, 107, 680. Analytical Quality Control (Harmonised Monitoring) Committee, Analyst, 1982, 107, 1407. Analytical Quality Control (Harmonised Monitoring) Committee, “Accuracy of Determination of Total Suspended Solids and Ash (Non-Volatile Suspended Solids) in River Waters,” Technical Report TR 163, Water Research Centre, Medmenham, Buckinghamshire, 1981. Department of the Environment, “Analysis of Raw, Potable and Waste Water,” HM Stationery Office, London, 1972. Crane, L., and Dewey, D. J., “The Determination of Suspended Solids and Ash in Waters by Filtra- tion and Ignition,” Technical Report TR 127, Water Research Centre, Medmenham, Buckingham- shire, 1980. Wilson, A. L., Talanta, 1970, 17, 31. Received November lst, 1982 Accepted June 6th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801365

出版商:RSC    

年代:1983

数据来源: RSC