Effect of Acids Modifiers and Chloride on the Atomization of Aluminium in Electrothermal Atomic Absorption Spectrometry Journal of Analytical Atomic Spectrometry SHIDA TANG Department of Environmental Health and Toxicology School of Public Health State University of New York at Albany Albany N Y 12201 USA PATRICK J. PARSONS* Wadsworth Center New York State Department of Health PO Box 509 Albany N Y 12201 -0509 USA and Department of Environmental Health and Toxicology School of Public Health State University of New York at Albany Albany N Y12201 USA WALTER SLAVIN Bonaire Technologies Box 1089 Ridgejeld CT06877 USA The effect of HNO HCI and H2S04 as well as various modifiers Mg ( Pd ( Ca ( N03)2 and N)14H2P04 on the atomization of aluminium from a L'vov platform in electrothermal atomic absorption spectrometry was investigated.No interference was observed on the integrated absorbance of A1 from any acid studied. The m observed for 10 pg 1- ' (200 pg) A1 aqueous solution prepared by diluting a commercial A1 stock solution with doubly deionized water was much poorer than expected owing to a failure to transfer A1 to the furnace it could be corrected by adding as little as 0.25% v/v of a mineral acid. The expected m for A1 was unaffected by modifiers such as Mg(N03)2 Pd(N03)2 Ca(NO,) and NH.,H2P04. Addition of either calcium magnesium or palladium nitrates produced a sharper absorbance peak and A1 was delayed in appearance. Either Ca( NO& or Mg( can stabilize A1 during pyrolysis allowing a very high (> 1700 "C) thermal pretreatment temperature to be used.However multiple atomization peaks were observed when using Mg(NO,),. The multiple peaks became more troublesome as the tube aged. Calcium nitrate recommended as a better modifier for those samples (e.g. bone) in which Ca is a large component of the matrix. A serious interference from chloride salts varied with chloride concentration pH and pyrolysis temperature. This is 'suppressive' interference from chloride was overcome by using Ca(N03)2 as a modifier in HNO and a pyrolysis temperature in excess of 1400 "C. Keywords. Aluminium; mineral acid; modifier; chloride interference; electrothermal atomic absorption spectrometry Aluminium is the most abundant metal in the lithosphere and is widely used for various purposes. In its hydroxide form Al(OH) it is used as a gastric antacid.As the sulfate A1,(S04)3 it is used as a deflocculant in domestic water treatment. During the 1970s it was discovered that patients on long-term hemodialysis developed a progressively fatal neurological condition later called dialysis dementia,' that was subsequently associated with exposure to A1 from the water supply used with dialysis equipment. This neurologic condition is now clearly attributed to an increased A1 body burden specifically bioaccumulation of A1 in brain bone and other tissues.' Because of this there has been much interest in * To whom correspondence should be addressed. measuring A1 in serum/plasma and in other fluids and tissues to develop understanding of the association and for biological monitoring purposes.The added possibility that A1 may be an aetiological factor in Alzheimer's Disease is another reason for the increased interest in the biological significance of this element.2-7 In fact occupational exposure to A1 has long been associated with neurobehavioural toxicity.' Development of an accurate and precise analytical method for the determination of A1 in biological tissues is fundamental to understanding its biological role. There are many analytical techniques for the determination of Al.9 From practical con- siderations of the required sensitivity sample size throughput contamination control and cost electrothermal atomization atomic absorption spectrometry (ETAAS) is the method of choice at least for biological materials." However there are still significant problems with the determination of A1 by ETAAS as evidenced by numerous publications documenting difficulties with various modifiers and acid media.Various methods for A1 in biological matrices have proposed Mg (NO,) ' ' *I2 NH4N0,,13 K,Cr,O7,l4 NH315 and NH4H2PO4l6 as the optimum modifier for this analysis by ETAAS. Different acid media have also been proposed includ- ing HNO3,l7 H2S0415*'8 and H3P04.19 In this paper we report the effect of several acids including HN03 HCl and H2S04 on A1 atomization and try to resolve some of the confusing and conflicting reports in the literature. In addition we report the effect of several modifiers proposed for Al including Mg(N03)2 Ca(NO,) Pd(NO,) and NH4H2P04 on the atomization of A1 from a solid pyrolytic L'vov platform.Chloride interference was also studied because C1 is a major constituent of most biological tissues and fluids. The primary aim of our work is to identify the best modifier and acid medium for the determination of A1 in serum bone and other biological samples by ETAAS. A detailed description of the final analytical method for measuring A1 in serum bone and other biological matrices will be presented elsewhere. EXPERIMENTAL Instrumentation All A1 measurements were carried out on a Model 25100 atomic absorption spectrometer (Perkin-Elmer Norwalk CT USA) equipped with an HGA-600 graphite furnace and a transverse Zeeman-effect background correction system. Pyrolytic graphite coated graphite tubes with a pre-installed forked-platform (P-E PN B0505057) made of solid pyrolytic Journal of Analytical Atomic Spectrometry August 1995 Vol.10 521graphite were used throughout the study. A Model AS-60 autosampler was used to deposit 2Opl samples onto the platform. The PE Z 5 100 was interfaced to a personal computer running Perkin-Elmer's proprietary software (version 6.0). Background and corrected absorbance peak data were extracted from the *.dat files using a customized program (peak.exe ver. 2.01) provided by Perkin-Elmer. ASCII files containing these data were transferred to an Apple Macintosh personal computer and imported into a graphical plotting package (Deltagraph Pro). Using this procedure peak profiles could be faithfully replotted from the original AA data col- lected and superimposed to reveal important details The 25100 conditions for A1 are given in Table 1 and the HGA-600 furnace programme in Table 2.The analytical wavelength selected for this study was the line at 309.3 nm. This line is about 20% more sensitive (mo= 11 pg) than the 396.2 nm line (mo = 14 pg) but is limited by a shorter linear dynamic range.20 It may be preferable to use the more sensitive 309.3 nm line for ultra-trace analytical work (e.g. to establish normal A1 concentrations in bone brain or other biological materials). However many applications (e.g. routine serum Al) are better conducted with the 396.2 nm line which has a greater linear dynamic range. The choice of analytical wavelength here did not affect the experiments conducted for the purposes of this study. Materials and Reagents Because glass contains small amounts of Al which could be a significant source of contamination for ultratrace work poly- propylene containers were used in this study.Owing to the ubiquitous nature of A1 in the laboratory environment all pipette tips autosampler cups and polypropylene containers were soaked in 2% v/v nitric acid for 24 h and rinsed thoroughly with doubly de-ionized (DI) water (Milli-Q system Millipore Corporation Bedford MA USA). Acid-washed materials were air-dried under dust-free conditions in an in-house constructed drying box. A working stock solution (100 pg 1-l Al) was prepared by serial dilution of a lo00 mg 1-l A1 standard solution (Aldrich Chemical Company Milwaukee WI USA) with DI water. Ultrapure-grade concentrated HN03 HCl and H2S04 were used (Baker Instranalyzed J.T. Baker Phillipsburg NJ USA) along with sub-boiling distilled HNO [National Institute of Standards and Technology (NIST) Gaithersburg MD USA]. Modifiers were prepared Table 1 Instrument conditions Instrument Furnace Autosampler Background correction Light source Wavelength Spectral bandwidth Measurement mode Graphite tube Injection volume Internal gas flow Perkin-Elmer Model Zeeman 25100 PC HGA 600 Transverse ac Zeeman Aluminium hollow cathode lamp 25 mA 309.3 nm 0.7 nm Peak area absorbance 5.0 s integration Pyrolytic graphite coated graphite tube with solid pyrolytic forked platform 20 p1 AS-60 Argon 300 ml min-' Table 2 HGA-600 graphite furnace condition Dry Pyrolysis Atomization Clean TemperaturePC 230 1200 2500* 2600 Ramp/s 10 5 0 1 Hold/s 50 30 5 10 Argon gas flow rate/ml min-' 300 300 0 300 by dissolving ultrapure reagents (Puratronic Johnson Matthey Royston UK) in 1% HN03.pH Measurements Two sets of A1 solutions were prepared at pH values ranging approximately from 0.5 to 7.0 by titration with HN03 or NaOH. The pH of each solution was measured using a pH meter with a combination glass electrode (DIGI-SENSE Cole Palmer Instrument Company Chicago IL USA). RESULTS AND DISCUSSION Acids Solutions containing A1 at 10 pg 1-l (200 pg) were prepared by diluting the stock solution (100 pg 1-I) with unacidified DI water and mineral acids including HN03 HCl and H2S04. The solutions were analysed for A1 by ETAAS and the integrated absorbances (Ai) recorded at different acid concen- trations (0-8%).We did not investigate HC104 owing to its hazardous nature. The atomization of 200pg A1 from an unacidified solution resulted in sensitivity poorer than expected (mo = 19 pg expected mo= 10 pg). We found that acidification with as little as 0.25% v/v of any common mineral acid was sufficient to recover the expected mo. We found no interference from either HNO or HC1 up to the maximum concentration studied 8% v/v. This disagrees with an earlier report where an interference from HCl and HNO was explained by formation of gaseous AlC13 molecules during pyrolysis and formation of stable A~N(s).~' Matsusaki et a1.22 also failed to observe any inter- ference from HCl. It is likely that in the case of hydrochloric acid Cl is removed during pyrolysis as gaseous hydrogen chloride.Our observation that unacidified aqueous A1 solutions of relatively low concentrations (10 pg I-') gave poorer sensitivity than expected led us to consider pH as a factor in the determination of Al. Aqueous solutions containing either 200 or 600pg A1 at various pH levels were prepared by titration with either HN03 or NaOH. Each solution was analysed for A1 by ETAAS and the Ai data collected. Fig. 1 shows how pH affects A1 integrated absorbance. All analytical measurements were carried out in triplicate with blank correction and the error bars represent the range of Ai values obtained. As the pH of a 200 or 600 pg A1 solution approaches neutral the A1 integrated absorbance decreases. This is probably due to precipitation of A1 as insoluble aluminum hydroxide Al(OH) .The measured pH of aqueous 200 and 600pg A1 solutions diluted with unacidified water were approximately 5.6 and 5.1 respectively. However the pH is difficult to estimate in these essentially unbuffered solutions. The lower pH i.e. more acidic nature of the 600pg solution can be explained by the larger 0.31 * Atomization temperature of 2400°C was used for the study with Ca. Fig. 1 Effect of pH on (0) 200 pg A1 and (0) 600 pg A1 522 Journal of Analytical Atomic Spectrometry August 1995 Vol. 10aliquot of acidified stock transferred during the dilution. The more acidic 600pg solution would also explain the better m achieved than with the 200pg solution without further acidification. The chemistry of A1 in aqueous solution is highly pH dependent in that A1 solubility decreases from pH 3.0-5.0 reaches a minimum between pH 5.5 and 6.0 and increases as the pH increases further to 7.0.At very low pH A13+ is present almost entirely as the hydrated cationic complex [A1(H20)6]3'. As the pH approaches 5.0 the hydrated cationic complex undergoes base hydrolysis to form several aquahyd- roxy species and precipitates as Al(OH),(s) at pH 6.0. Thus conceivably A1 precipitation does occur to some extent in unacidified solutions and this may be the reason for the observed poor sensitivity. Possibly the poorer sensitivity may also be due to some adsorption of A1 aquahydroxy species on the vessel wall at pH 5.0-6.0. We confirmed that the poor sensitivity observed with A1 solutions at pH approaching neutral is caused by a failure to transfer A1 into the furnace.A solution containing a known amount of A1 (200 pg) in 1% HNO (pH M 1.2) was analysed for A1 in the presence of various aliquots of 0.25 moll-' NaOH injected separately by the autosampler onto the plat- form. No change in the integrated absorbance of A1 was observed although the pH of the mixture on the platform changed from 1.2 to 12.4. Thus it would appear that reports of sensitivity 'enhancement' due to acidification are only observed for A1 solutions with relatively high pH values (pH > 5.0) where a failure to transfer A1 to the furnace occurred due to precipitation of Al(OH) or adsorption of hydrolysed species on the vessel wall. 0 Modifiers Atomization curves for 200 pg A1 in 1% HNO with different modifiers were obtained by varying the atomization tempera- ture while keeping the pyrolysis temperature constant at 1200 "C [Fig.2(a)]. Having established the optimum atomiz- ation temperature for each modifier we then obtained corre- sponding pyrolysis curves [Fig. 2(b)]. The small difference in Ai of Fig. 2(b) is caused by differences in atomization tempera- ture and different tubes used. All analytical samples were prepared in 1% HNO to eliminate the pH effect described above. Using the optimized furnace parameters obtained from the data in Fig. 2 we compared the performance of Mg(N03) with Ca(NO,) as modifiers for A1 in ETAAS. Atomization profiles for 200pg A1 in the presence of various amounts of Mg(N03)2 and Ca(NO,) in new and aged tubes are shown in Fig.3. .\;\'\a 1 -A- No modifier 1 I I I Magnesium nitrate Magnesium nitrate has been the most widely used modifier for A1 determinations by ETAAS. It was first proposed for determi- nation of A1 over a decade ago.l1*l2 In that paper," it was reported that different amounts of Mg(N03) had no effect on A1 integrated absorbance and we have confirmed this obser- vation with the forked platform. Aluminium was also stabilized by Mg(N03) during pyrolysis up to temperatures of M 1900 "C [Fig. 2(h)] and the appearance of A1 in absorption was delayed. The peak absorbance was subs tantially increased as the amount of Mg deposited on the platform was increased up to 50 pg [Fig. 3(a)]. The integrated absorbance remained largely unchanged (mean Ai = 0.077 f 0.002). However as the mass of Mg deposited on the platform approached 50 pg the atomization peak split into two components.This became more troublesome as the tube/platform aged even with pre- viously recommended amounts of 50 pg Mg(N03) ( M 8 pg Mg) on the platform." Multiple atomization peaks of A1 I . I /% -E+.. Pd +- NH H PO -A- No modifier a .p 0 4 2 . . 4 E5 d.; s o . 2 1700 1900 2100 2300 2500 2700 a Fig.2 Atomization (a) and pyrolysis (b) curves for 200pg A1 with 50 pg Mg(N03)2 (0) 10 pg Ca as Ca(N03)2 (O) 15 pg Pd as Pd(N03) (0) and 125 pg H,PO,- (+ ) and without any modifier (A). For the atomization curves in (a) a fixed pyrolysis temperature of 1200°C was used; for the pyrolysis study in (b) an atomization temperature of 2500 "C was used except for Ca where 2400 "C was used increased in complexity as the amount of Mg(N03)2 was increased as shown in Fig.3(b) with a moderately aged tube of approximately 200 firings. Manning et ~ 1 . ' ~ also reported multiple atomization peaks for A1 but only in the presence of MgCl,. Although such problems make peak absorbance measurements quite unre- liable changes in Ai are of course much smaller. This kind of problem is a good example of why peak area or integrated absorbance is preferable to peak height absorbance measure- ment in ETAAS. In the case of A1 with Mg(N0,)2 modifier even using integrated absorbance measurements within-run precision is a little worse (RSD=2.2% n=120) than that obtained with Ca(N0,)2 modifier (RSD= 1.4% n= 120) with an aged tube of 100 firings. Moreover the multiple atomization peak problem cannot be overcome by increasing pyrolysis temperatures or by the addition of a cool-down step before atomization. Calcium nitrate The 'enhancement' effect of Ca(N03) on A1 absorbance reported in the l i t e r a t ~ r e ~ ~ - ~ ~ is most likely due to use of peak height measurements rather than integrated absorbance.In agreement with Manning et we also found no enhance- ment of A1 on integrated absorbance from Ca(NO,),. The pyrolysis curve for A1 in the presence of Ca(NO,) in Fig. 2(b) shows that Ca(NO,) stabilizes A1 during pyrolysis up to a temperature of about 1700°C. An interesting feature of the atomization curve for A1 in the presence of Ca(N03)2 [Fig. 2(a)] is the significantly lower temperature required to atomize Al with Ca( NO3) compared with other situations.The optimum atomization temperature with Ca(NO,) is some 200°C less than that required with either Mg(N03) or other modifiers. Like Mg the use of Ca altered the A1 atomization Journal of Analytical Atomic Spectrometry August 1995 Vol. 10 52350 c19 0.47 0.2 - 0- Mg-new tube ( d ) Ca-aged tube 10 c19 d 0 1 2 3 4 :S Mg-aged tube 0.2 4 0 1 2 3 4 5 Ca-new tube Time/s Fig. 3 Atomization profiles for 200 pg A1 with different amounts of Mg or Ca modifier in new and aged (200 firings) tubes profile with delayed appearance time and increased peak absorbances. As the amount of Mg(NO,) was increased the A1 peak appearance time was delayed [Fig. 3(a)]. But as Ca(N03) was increased the appearance time of the A1 peak remained constant [Fig.3(c)]. In an aged tube with Ca(N0,)2 the A1 atomization peaks obtained were still symmetric and sharp [Fig. 3(d)] although not as sharp as with a new tube. The A1 appearance time as with new tubes was also delayed but independent of the amount of Ca deposited on the platform. The conclusion based on these results is that Ca as Ca(N0,)2 is a good modifier for the determination of A1 by ETAAS especially in bone or other biological samples that have relatively high Ca content. Addition of Ca(N03) as a modifier to the standards is necessary to ensure that A1 atomization will be similar to that in bone samples. Liang2' avoided adding a modifier directly to A1 standards but had to pretreat the L'vov platform with bone digestate several times prior to calibration. The memory effect from depositing a Ca-rich matrix on the platform resulted in an absorption profile that was very similar to bone samples.We also observed a memory effect from Ca when studying the modification effects on Al. However the memory effect from Ca gradually disappeared after several firings. Thus we found it preferable to add Ca(NO,) as a modifier to A1 standards and ensure identical atomization behaviour with samples. Palladium nitrate Palladium nitrate mixed with magnesium nitrate has been proposed as a universal modifier28 and this approach has been applied for the determination of more than 20 elements by ETAAS.,' In this work however we investigated palladium nitrate alone as a modifier for the determination of A1 by ETAAS. Pyrolysis and atomization curves for A1 in the presence of Pd are shown in Fig.2(a)-(b). We found that using Pd(N03) modifier provided no stabilization for A1 during pyrolysis [Fig. 2(b)]. However Pd did delay the A1 appearance time and increased the peak absorbance but integrated absorbances remained unchanged (data not shown). As with Mg(N03)2 the appearance time was increasingly delayed with increasing mass of Pd(N03) deposited on the platform. However we observed multiple A1 atomization peaks with Pd(N03)2 too especially with aged tubes. Thus we could find no advantage in using Pd(N03) as a modifier for Al. Ammonium dihydrogen phosphate In an early study Garmestani et aL3' observed an enhancement effect for A1 using K2HP04. However since they used peak absorbance measurements the reported enhancement was probably due to changes in atomization profile.Manning et ~ 1 . ' ~ found no effect on A1 absorbance from up to 200 pg Na2HP04 when a tungsten wire atomizer was used. Radunovic et ~ 1 . ' ~ used NH4H2P04 as a modifier for determining A1 in various biological samples using similar equipment. Those authors claimed NH4H2P04 was preferable to Mg(NO,) for measuring A1 in bone and sera because the former permits pyrolysis temperatures of up to 1400°C without any loss and because the ammonium salt facilitates chloride removal from NaCl as volatile NH4Cl. However we found Ca(NO,) preferable to NH4H2P04 for determination of Al especially in those matrices which are either Ca-rich and/or contain Cl (discussed below). We found that using NH4H2P04 produced a similar pyrolysis curve to the situation without any modifier [Fig.2(a)] and did not change the A1 peak absorbance (data not shown) as did Ca( NO,) and Mg( NO3) (Fig. 3). 524 Journal of Analytical Atomic Spectrometry August 1995 Vol. 10Chloride Interference Interferences from chloride on A1 absorbance have been well documented and widely r e p ~ r t e d . " - ~ ~ * ~ ~ A suppressive effect from chloride on A1 is generally attributed to the loss of A1 during pyrolysis because of the volatile nature of AlCl (subli- mation point 177.8 "C). Matsusaki et ~ 1 . ~ ~ investigated chloride interference on Al and observed different levels of absorbance suppression by several chloride salts. They also reported that adding HNO eliminated the suppression effects of both NaCl and KCl but a similar effect from CaCl could only be eliminated by adding ( NH4)4EDTA. Using new pyrolytic graphite coated graphite tubes with a L'vov platform and small (< 50 pg) amounts of CaCl Slavin et aL3' were unable to replicate the suppressive effect of CaC1 reported by Matsusaki.The interference from CaCl was found to increase with tube age and it was suggested that this could have been caused by erosion of the pyrolytic coating. It was concluded that in aged tubes a large proportion of the CaCl matrix is still present during atomization and then causes typical vapour-phase interferences between A1 and C1. The interference from chloride in the determination of A1 was studied using CaCl and NaCl as sources of chloride. We found no interference on 200 pg of A1 from small (<20 pg) amounts of CaCl but an interference from CaCl was observed when the amount of CaCl was greater than 20pg using a pyrolysis temperature of 1200 "C.This interference worsened with increasing amount of CaCl deposited on the platform. The pyrolysis characteristics of 200 pg A1 in the presence of 330 pg (6 pmol C1-) CaCl at different concentrations of HNO are shown in Fig. 4(a). The interference on the integrated absorbance of A1 from CaCl is clearly evident at pyrolysis temperatures of <1400°C. However it is interesting to see that the interference disappears at pyrolysis temperatures above 1400 "C until the temperatures exceed 1700 "C when A1 is lost. This result suggests that the CaCl interference with A1 is due to formation of a stable molecular species containing A1 during the atomization rather then the pyrolysis step.If A1 were lost as AlCl during pyrolysis the interference would be expected to be more serious at higher pyrolysis temperatures (> 1400 "C). But the experimental results are just the opposite. That means that large amounts of C1 must still be available during the atomization step if pyrolysis temperatures of <1400"C are used and this results in the vapour-phase interferences between A1 and Cl. Reason(s) for the remaining large amounts of C1 after pyrolysis at temperatures of < 1400°C might be that chloride was left on the platform as undecomposed salt(s) because of the low pyrolysis temperature used or some Cl was trapped at the cold ends of the tube.What is really happening requires further studies. Our experiments also show that this inter- ference can be reduced by increasing HNO concentrations to 5% v/v at pyrolysis temperatures of < 1400 "C [Fig. 4(a)]. This observation supports the suggestion that the CaCl interference is due to inefficient removal of C1 during pyrolysis because of the low pyrolysis temperatures (< 1400 "C) used. If C1 is trapped at the cold ends of the tube increase of HNO concentration would be expected to have no effect on the interference produced by C1. Increasing the HNO concentration helps to remove Cl most likely as volatile HCl gas during the pyrolysis step since the Cl interference is removed. Thus the interference from CaCl can be eliminated by employing a high HNO concentration (5% in this case) or by keeping the pyrolysis temperature between 1400-1700 "C.We investigated the Cl interference further using higher concentrations of NaCl. Pyrolysis curves for 400pg A1 in the presence of 585 pg NaCl(l0 p o l C1-) with 1% HNO 10% HNO and 10% HNO,+ 10 pg Ca as Ca(N03) are shown in 0. e Q) 0 c 800 1200 1600 2000 0.054 * 10% HN03 I I 8' . . \ o ! c G c $ m l 200 600 1000 1400 1800 I 2200 1 I I ndicated tern pe rat u re/' C Fig. 4 Pyrolysis curves for A1 showing interferences from two chloride salts (a) 200 pg A1 plus 330 pg CaC1 (6.0 pmol 1-' C1-) in 1% and 5% HN03; and (b) 400 pg A1 plus 585 pg NaCl(l0 pmol 1-I C1-) in 1% and 10% HN03 and 10% HN03 plus 1Opg Ca as Ca(NO,),. Atomization temperature = 2400 "C Fig. 4(b). A similar but more pronounced suppressive effect on A1 is observed from NaCl compared with CaCl,.Furthermore this interference is only partially reduced by increasing the HNO concentration to 10%. Increasing the HNO concen- tration does not eliminate C1 interference completely in this case because without a modifier such as Ca which can stabilize A1 as in the case of CaCl some A1 is lost at pyrolysis temperatures > 1400 "C [Fig. 2(b)]. However a complete recovery is possible by using 1Opg Ca as Ca(N03)2 with 10% HNO,. This agrees well with the CaCl results in that Cl interference is due to formation of a stable molecular species containing A1 during atomization because of inefficient removal of Cl during pyrolysis. This conclusion is also supported in part by the atomic and background absorption profiles of 400 pg A1 in the presence of 585 pg NaCl shown in Fig.5. A large background absorbance signal (Ai = 0.429) but very small atomic absorbance signal (Ai = 0.004) is observed when a pyrolysis temperature of 800°C is used (Fig. 5). The back- ground absorbance signal is greatly reduced (94.4%) by raising the pyrolysis temperature to 1400 "C (Fig. 5) and the atomic absorbance signal is greatly recovered (Ai = 0.103) at the same time. CONCLUSIONS In this study we investigated the effect of mineral acids modifiers and chloride interference on the determination of A1 by ETAAS using forked platforms. Poorer than expected sensitivity or characteristic mass (mo) for A1 solutions of low pg 1-' concentrations was observed if the solutions were prepared by serial dilution of a commercial A1 stock solution with DI water.This is most likely caused by a failure to transfer A1 to the furnace owing to precipitation of A1 as Al(OH) or adsorption of hydrolysis species on the vessel wall at pH>5. Acidification of aqueous A1 solutions with any mineral acid studied results in a recovery of the expected mo Journal of Analytical Atomic Spectrometry August 1995 Vol. 10 525( a 1 o'61 BK (A = 0.429) 0*4i ' Pyrolysis at 800 "C \ \ I I 1 \ 0.2 i AA (4 = 0.004) ,' 1 2 3 I y \ Pyrolysisat 1400°C 0.1 1 BK (Ai = 0.024)/ \ 0 1 2 3 Time/s Fig. 5 Atomization and background absorption profiles for 400 pg A1 in 1% HNOJ plus 585 pg NaCl (10 pnol I-' Cl-) at a pyrolysis temperature of 800-1400 "C for A1 by ETAAS.Discrepancies in the literature reported for effect of acids on A1 can be explained by use of standards of different pH values. Reported enhancement effects due to modifiers may also be due to acidification of their aqueous solution since modifier solutions are generally prepared in dilute acid. We recommend using a 1% HN03 acid medium for standards and samples (if possible) in the determination of A1 by ETAAS. Integrated absorbance of A1 was unaffected by using either Ca(N03)2 Pd(N03)2 Mg(N03)2 or NH4H2P04 as modifiers in 1% HN03. Many A1 interference problems have been reported in the early literature which are now understood to be the result of using peak height absorbance to quantitate the A1 signal. This work shows the sensitivity of the peak shape (thus peak height absorbance) to the matrix composition.Most of the modifiers evaluated here result in increased peak height absorbance because the peak width at half height is smaller i.e. the peaks are sharper but the integrated absorbance remains unchanged. As the integrated absorbance is a better indicator of the extent of free atom formation it is almost always preferable in quantification instead of peak height absorbance. The aluminium atomization profile suffers from multiple peaks with the most widely used modifier Mg(N03)2 especially in aged tubes. Our results support Ca as a modifier preferable to Mg for the determination of A1 in samples rich in Ca such as bone and many other biological matrices. A substantial interference from chloride on A1 was found to depend upon Cl concentration HN03 concentration and pyrolysis temperature.Our results have shown that this inter- ference is caused by loss of A1 during atomization rather than during pyrolysis. After pyrolysis at temperatures of < 1400 "C large amounts of Cl remain owing to undecomposed chloride salt@) on the platform or trapping of some C1 at the cold ends of the tube. The chloride interference was effectively eliminated by using Ca(N03) as modifier in HN03 and a pyrolysis temperature 1400-1700 "C. We are grateful to Glen Carnrick of Perkin-Elmer for helpful discussions and John McCaffrey of Perkin-Elmer for providing the computer program for extracting the atomization data from peak files. REFERENCES 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 Alfrey A.C. LeGendre G. R. and Kaehny W. D. N. Engl. J. Med. 1976 294 184. Crapper D. R. Krishnan S. S. and Quittkat S. Brain 1976 99 67. Perl D. P. Pendlebury W. W. and Munoz-Garcia D. in Proceedings of the First International Symposium on Geochemistry and Health ed. Thornton I. Science Reviews Northwood UK McLachlan D. R. Neurobiol. Aging 1986 7 525. Katzman R. N. Engl. J. Med. 1986 314 964. McLachlan D. R. Lukiw W. J. and Kruck T. P. Can. J. Neurol. Sci. 1989 16 490. Eichhorn G. L. Exp. Gerontol. 1993 28 493. McLaughlin A. T. G. Kazantzis G. King E. Teare O. Porter R. J. and Owen R. Brit. J. I d . Med. 1962 19 253. Savory J. and Wills M. R. Kidney. Int. Suppl. 1986 18 S24. Cedergren A. and Frech W. Pure Appl. Chem. 1987 59,221. Slavin W.Carnrick G. R. and Manning D. C. Anal. Chem. 1982 54 621. Manning D. C. Slavin W. and Carnrick G. R. Spectrochim. Acta Part B 1982 37 331. Smeyers-Verbeke J. and Verbeelen D. Anal. Chem. 1988,60,380. Shan X. Luan S. and Ni Z. J. Anal. At. Spectrom. 1988 3 99. Gitelman H. J. and Alderman F. R. Clin. Chem. 1989,35/7,1517. Radunovic A. Bradbury M. W. B. and Delves H. T. Analyst (Cambridge U K ) 1993 118 533. Oster O. Clin. Chim. Acta 1981 114 53. Toda W. Lux J. and Van Loon J. C. Anal. Lett. 1980 13 1105. Craney C. L. Swartout K. Smith F. W. I. and West C. D. Anal. Chem. 1986 58 656. Manning D. C. and Slavin W. At. Spectrom. 1986 7 123. Persson J-A. Frech W. and Cedergren A. Anal. Chim. Acta,. 1977 92 95. Matsusaki K. Yoshino T. and Yamamoto Y. Tulanta 1979 26 377. Martin R. B. in Aluminium and Renal Failure eds. De Broe M. E. and Coburn J. W. Kluwer Academic Publishers Thompson K. C. Godden R. G. and Thomerson D. R. Anal. Chim. Acta 1975 74 289. Dolinsek F. Stupar J. and Spenko M. Analyst (Cambridge UK) 1975 100 884. Liang L. D'Haese P. C. Lamberts L. V. and De Broe M. Clin. Chem. 1991 37 461. Liang L. Spectrochim. Acta Part B 1992 47 239. Schlemmer G. and Welz B. Spectrochim. Acta Part B 1986 41 1157. Welz B. Schlemmer G. and Mudakavi J. R. J. Anal. At. Spectrom. 1992 7 1257. Garmestani K. Blotcky A. J. and Rack E. P. Anal. Chem. 1978 50 144. Slavin W. Manning D. C. and Carnrick G. R. Anal. Chem. 1981 53 1504. 1985 pp. 227-235. 1990 pp. 7-26. Paper 410 7 3 0 7 J Received November 11 1994 Accepted April 11 1995 526 Journal of Analytical Atomic Spectrometry August 1995 .Vol. 10