JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 823 Analysis of Aluminium Alloys Using Inductively Coupled Plasma and Glow Discharge Mass Spectrometry Xinbang Feng and Gary Horlick Department of Chemistry University of Alberta Edmonton Alberta Canada T6G 2G2 The application of inductively coupled plasma mass spectrometry (ICP-MS) and glow discharge mass spectrometry (GD-MS) to the analysis of aluminium alloys is presented. The spectral characteristics of these two techniques are compared and contrasted. For the most part GD mass spectra are simpler than ICP mass spectra in that species originating from the components of air water and solution solutes are essentially absent in GD spectra. However GD spectra show the presence of multiply charged Ar species and matrix and analyte based argide species are more prevalent than oxide species the opposite being the case with the ICP.For both techniques potential spectral interferences are evaluated and matrix effects are illustrated. For ICP-based analyses it is shown that matrix effects can be minimized by adjustment of the nebulizer gas flow rate and by the use of an internal standard which is easy to add to the dissolved samples. For GD-based analyses a matrix effect exists between low and high alloy aluminium samples and it is shown that the signal from AlAr can be used as an internal standard to minimize this matrix effect. Finally both techniques were applied to the analysis of a range of Alcan and Alcoa aluminium standards. Keywords lnductively coupled plasma mass spectrometry; glow discharge mass spectrometry; alu- minium analyses In the last decade mass spectrometry has emerged as a major technique in the area of elemental analysis.The two main systems that have seen major development in this decade are inductively coupled plasma mass spectrometry (ICP-MS)' and glow discharge mass spectrometry (GD-MS).' Both systems are applicable to the determination of trace amounts of elements in a wide variety of sample type^.^?^ The ability to directly analyse solid materials is a major asset of the GD and this ability complements the solution sample handling capa- bility of the ICP. In addition it is not necessary to have separate mass spectrometric instrumentation for each tech- nique as ICP and GD ion sources can be interfaced to the same mass ~ p e c t r o m e t e r ~ ~ ~ ~ ~ and at least one company (Finnigan MAT) is marketing a combination instrument.In this report the application of both techniques to the analysis of aluminium alloys is presented. This study was not approached as a head-to-head competition but is simply meant to provide in a single paper a presentation of these two techniques applied to the analysis of the same samples. In this way the two methodologies can be compared and con- trasted. For the most part the important areas to address during the development of a quantitative analytical method are similar for both techniques. They include semiquantitative analysis assessment of spectral overlaps assessment of matrix effects instrumental settings and choice of an internal standard.However details of each step for the two methods do differ and in particular spectral characteristics matrix effects and internal standardization have considerations unique to each technique. Other workers have used these techniques for the analysis of aluminium samples. Takeda et a1.8 have discussed the determination of ultra-trace amounts of U and Th in high- purity aluminium by the use of ICP-MS GD-MS has been used for the analysis of aluminium alloy^^*'^ and GD-MS has also been used for the routine quality control of high-purity aluminium'l and has been compared with SIMS for the determination of U and Th in aluminium.'2 Experimental ICP-MS System All measurements were carried out using a SCIEX (Perkin Elmer-SCIEX) Elan Model 250 ICP quadrupole mass spec- trometer.For the ICP-MS studies a standard MAK ICP torch was used with a Meinhard nebulizer and a Scott-type spray chamber. A sampling depth of 15mm from the load coil was used and the plasma forward power was 1.3 kW. The outer gas and intermediate gas flow rates were 12 and 11 min-'. Central (nebulizer) gas flow rates of both 1.1 and 0.9 1 min-' were used. Matrix effects were reduced at the lower central gas flow rate setting but with some sacrifice in signal inten- ~ i t y . ' ~ ' ~ The ion lens voltages selected were a compromise chosen to cover a large mass range.15 The settings used were 5 V for the Bessel box barrel (B) - 18 V for the Bessel box plates (P) - 16 V for the einzel lens ( E l ) and - 10 V for the photon stop (S2). GD-MS System A pin sample GD ion source described by Shao and Horlick6 was used for this work.This ion source was bolted directly onto the external interface plate of the mass spectrometer (SCIEX-Perkin-Elmer Elan Model 250) in place of the normal sampling cone (full details can be found in ref. 6 ) . Typical operating pressures were 2.5 Torr (1 Torr = 133.322 Pa) for the GD device 0.1-0.2 Torr for the region between the sampling plate and the skimmer and 10-7-10-6 Torr for the mass spectrometer. These pressures were lower than the typical operating pressures for this system used for ICP-MS where the pressure is about 1-4 Torr between the sampling cone and the skimmer and the mass spectrometer operates at about Torr. The current-voltage operating values for the GD ranged from 7.5 to 9.5 mA and 800-1200 V and the fill gas was argon.The anode of the GD was biased slightly positive (+7 V) and the shadow stop" of the Elan was biased slightly negative (- 13 V). This stop is normally at ground potential when the system is used for ICP-MS. The biasing arrangement and the dependence of the ion signal intensity on these bias voltages is illustrated in ref. 6. The Elan ion lens voltage settings for the GD-MS measure- ments were 6.5 V for the Bessel box barrel (B) -30.9 V for the Bessel box plates (P) - 22.4 V for the einzel lens (El) and - 8.6 V for the photon stop (S2). Samples The aluminium samples studied were all solid metal standards obtained from Alcan and Alcoa. The five reference materials824 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL.9 obtained from Alcan (lSCXG lSWL lSWM 1SXD and 2SDZ) were all low alloy aluminium where aluminium made up over 99% of the sample. These samples were all certified for Bi Cr Cu Fe Mg Mn Ni Pb Si Sn Ti and Zn and three of the samples were also certified for Be Ca Cd Co Ga Li Na Sr V and Zr. The amount of these elements present in the samples ranged from a high of 0.5% for some elements to low values in the region of 0.005%. Eight Alcoa standards were available SA-909 SA-1170 SA-1169 SS-356-B SS-A 132AA SS-D132-A SS-319E and SS-360-C. The first three standards are low alloy aluminium standards. Aluminium made up over 99% of the sample and they were certified for Si Fe Cu Mn Mg Ti and Zn with one sample also certified for Cr Ni Pb and Sn. The amount of these elements present ranged from 0.5 to 0.001%. The last five of the Alcoa standards listed above are high alloy alu- minium materials where aluminium represents from 82 to 92% of the composition.The Si content ranges from 6 to 12% Cu from 0.03 to 4% and then at levels below 1'340 Fe Mn Mg Ni Ti and Zn round out the certified elements. For the ICP measurements these materials had to be put into solution. They were dissolved using a method described by Ward and Marciello.16 With this method 100 mg of sample were dissolved using 10 ml of 6 mol I-' HC1. After dissolution which required heating and took about 5-10 min the solution was cooled and then diluted to 100 ml providing a solution that was about 0.1% Al. For some measurements a 0.01% A1 solution was also prepared.This dissolution method was effective only for the low alloy aluminium mate- rials and therefore the high alloy aluminium samples were not analysed using ICP-MS in this study. The high alloy aluminium samples are difficult to dissolve because of their high content of Si. Aqueous standard solutions for the ICP measurements were prepared from ICP-grade standard solutions obtained from Leco Corporation. For the GD work the aluminium standards had to be machined into sample pins which were used as the cathode in the GD. These pins were about 3 mm in diameter and about 15 mm in length. Both low and high alloy aluminium standards were studied with the GD-MS system. Results and Discussion Spectral Characteristics One of the more interesting aspects of this study has been the ability to directly inter-compare some of the spectral character- istics of ICP-MS and GD-MS.The ICP and GD mass spectra are shown in Fig. 1 for an Alcan low alloy aluminium sample (1SCXG) over the m/z range 1-45. The ICP sample was a 0.01% solution and the composition of the sample is shown in Table 1. The basic background species in ICP-MS have been established for some time and they were summarized several years ago by Tan and Hor1i~k.l~ As can be seen in Fig. 1 the GD has a considerably simpler spectrum compared with that of the ICP in this m/z range. To a great extent this results from the fact that solvent (H20 HCl) and atmospheric (02 N2) based species are generally absent in the GD spectrum. Note however the strong signal from Ar2+ in the GD spectrum [Fig.l(b)] which is absent in the ICP spectrum [Fig. l(u)]. Some spectra of this region with the vertical scale expanded are shown in Fig. 2. In the GD spectra [Fig. 2(b) and (c)] multiply ionized Ar species (Ar3+ Ar4+ Ar5+ and even Ar6+) are observed. There is certainly the possibility that A12+ could occur in the GD and at an m/z of 13.5 it could not be distinguished from Ar3+ which occurs at an m/z of 13.3. Likewise A13+ would overlap with Be at m/z 9. Finally Ar6+ constitutes a spectral interference for 7Li. The multiply charged argon and aluminium species do not appear in the ICP spectrum [Fig. 2(a)]. Finally the signals from Na and B 3.0 2.5 2.0 1.5 1 .I F I v) v) 0.5 .I- 3 0 0 0 .- c 3.0 v) a 4- 5 2.5 2.0 1.5 1 .o 0.5 5 10 15 20 25 30 35 40 45 0 m/z Fig.1 Mass spectra for (a) ICP-MS and (h) GD-MS of Alcan aluminium alloy (1SCXG) Table 1 Composition of Alcan lSCXG low alloy aluminium Composition Element (mass-%) Be Ca c o Cu Ga Mg Na Pb Sn v Zr 0.007 0.0058 0.02 0.022 0.015 0.25 0.0026 0.019 0.024 0.008 0.034 Composition Element (mass- %) Bi Cd Cr Fe Li Mn Ni Si Ti Zn 0.022 0.018 0.01 8 0.33 0.002 1 0.026 0.023 0.2 0.02 0.027 observed in the ICP spectrum and not in the GD spectrum are probably a result of contamination introduced at the dissolution step. The ICP and GD mass spectra for this sample are also presented in Fig. 3 for the m/z range 40-85. Important differ- ences exist in this spectral region. As a consequence of the HCl-based dissolution step chlorine-based species such as C10+ ClOH' and ArC1' are observed in the ICP spectrum [Fig.3(a)]. As is well known these species present spectral interference problems for the determination of V Cr and As. These chlorine-based species are not present in the GD spec- trum [Fig. 3(b)] which allows for the determination of these elements in aluminium samples by GD-MS. An important difference in spectral features between ICP-MS and GD-MS is the existence of metal argides in GD-MS? 27A140Ar+ is clearly present in the GD spectrum [Fig. 3(b)] and is not observed in the ICP spectrum [Fig. 3(a)]. The level of this argide (27A140Ar+) is about 0.06% relative to the aluminium signal. While this species does coincide with aJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 825 1 .o 0.5 0 5 10 15 20 25 30 35 40 45 m/z Fig.2 aluminium alloy ( 1SCXG) (scale expanded) Mass spectra for (a) ICP-MS (b) and (c) GD-MS of Alcan minor isotope of Zn (67Zn) its presence is actually an asset. As will be seen later AlAr+ can be used as an internal standard for GD analyses of A1 alloys. It is difficult to confirm the presence of other aluminium- based spectral features in these spectra. The small peak at m/z 43 in the ICP spectrum [Fig. 3(a)] could be AIOf as oxides are more prevalent in ICP-MS than in GD-MS but it also could arise from 43Ca+. Also AlOH+ and A12+ at m/z 44 and 54 coincide with COz+ and 40Ar14N+. Since nitrogen- based species are also absent in GD-MS spectra and since dimers are more prevalent than in ICP-MS the peak at m/z 54 in the GD spectrum is likely to be A12+.The ICP and GD mass spectra for the 80-125m/z range are shown in Fig. 4. Niobium (m/z 93) is present in the ICP spectrum but is absent in the GD spectrum. It seems that it must have entered the sample solution as a contaminant during dissolution. Also both ZrO + and NbO' species are observed in the ICP spectrum but are absent in the GD spectrum. Note that the spectra shown in Fig. 4 are for sample 1SWL and not lSCXG which was used to generate the spectra for Figs. 1-3. Although sample 1SCXG was not certified for Ag it was found to contain Ag and the lo9Agf peak obscured the 93Nb160+ signal the existence of which we wanted to illustrate. Finally the ICP and GD mass spectra for the m/z range from 190 to 220 are shown in Fig. 5. It is interesting to note the difference in relative intensities of the '08Pb and '09Bi peaks in these spectra.Two explanations are possible. Lead could have been added as a contaminant during dissolution or the differences in intensities could reflect differences in the degree of ionization between the ICP and the GD ion sources. It should be noted that relative intensity differences between the ICP and GD spectra can also be seen for other elements such as for 12'Sn and '"Sb (Fig. 4) and for Cu Zn and Ga (Fig. 3). Ti+ Mn' I j Ar,H + A 40 45 50 55 60 65 70 75 80 85 m/z Fig.3 aluminium alloy (1SCXG) Mass spectra for (a) ICP-MS and (b) GD-MS of Alcan 150 100 50 c I v) v) +d 5 . 0 0 > -. Y Sn + I Zr+ Nb' I I (a' I I I (b) Zr' I 7 - + Sb' 100 - 50 - m/z Fig.4 Mass spectra for (a) ICP-MS and (b) GD-MS of Alcan aluminium alloy (1SWL)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL.9 826 800 600 400 200 I v) v) 4- 5 0 + Pb' Pb' 1 190 195 200 205 210 215 220 m/z Fig.5 aluminium alloy (1SWL) Mass spectra for (a) ICP-MS and (6) GD-MS of Alcan ICP-MS Analyses Many analysts who utilize ICP-MS are initially only experi- enced in ICP atomic emission spectrometry (ICP-AES) and some time is required to develop the proper analytical intuition for the development of quantitative analytical methods for ICP-MS. For the most part the important areas to address during the development of an ICP-MS method have been delineated and include semi-quantitative analysis assessment of spectral overlaps assessment of matrix effects instrument settings and choice of internal standard.A discussion of the consideration of these areas for the analysis of steels using ICP-MS has been presented by Vaughan and Horlick14 and an analogous approach is taken here for the analysis of aluminium alloys. Because of the similarity to the methodology presented in ref. 14 the discussion of each area is brief. The samples studied in this section include all the low alloy aluminium standards from both Alcan and Alcoa. Qualitative spectral scans In the last section spectral scans of the Alcan low alloy aluminium standard 1SCXG were presented. The presence of all the certified elements could be verified and in addition Sr Ag and Sb were observed in some samples. Because of the possibility of contamination during dissolution and also because of the existence of certain spectral interferences unique to ICP-MS the GD-MS results are a significant help in verifying the presence and/or absence of some elements (i.e.V and Cr are present Nb is probably not present). Evaluation of potential spectral interferences The basic isobaric and background spectral interferences in ICP-MS have been well d o c ~ m e n t e d . ' ~ . ~ ~ Some of the major problem species are summarized in Table2 for the common elements determined in aluminium alloys. Only the major background species are listed for water and dilute nitric acid sulfuric acid and hydrochloric acid solutions and the tables of Tan and Horlick17 should be consulted for more details. Some of the elements of interest in aluminium metallurgy (such as Si and P) cannot be easily identified because of spectral overlaps with species basic to the ICP discharge.These molecular background ions include 14N2' and 14N160H+ which affect the major isotopes of Si and P. Iron and Ca are also problems. For iron its major isotope (56Fe) is affected by ArO' its next two most abundant isotopes 54Fe and 57Fe are affected by ArN+ and ArOH' respectively and the least abundant isotope 58Fe has an isobaric overlap with the major isotope of Ni. The most serious problem concerns Ca where all of its six isotopes are affected. With its major isotope (40Ca) affected by 40Ar and its third most abundant isotope (42Ca) affected by ArH,' both isotopes of 46Ca (0.004) and 48Ca (0.187) suffer from the interference of NO2+ and isobaric overlap from 46Ti and 48Ti.Even the next least abundant isotope 43Ca (0.14) and the next most abundant isotope 44Ca (2.09) are affected by the oxide A10' and the hydroxide AlOH' as well as the background species CO,'. Molecular background species can also be formed from the components of the overall sample solution matrix. Sulfur and chlorine containing components can be particularly trouble- some. For example the use of HCl in the dissolution step presents some major potential problems. Two key elements that can be affected are V and Cr. A chloride containing matrix results in the formation of 35C10+ 35C10H+ and 37C10+ which interfere with both "V and ',Cr as well as with 53Cr. These are serious problems because 'OV is the only other naturally occurring isotope of V with a natural abundance of only 0.25% and "Cr is the only remaining isotope of Cr with a natural abundance of 4.34% and both 'OV and "Cr suffer from isobaric overlaps from 50Ti 36ArN 35C115N and 34S160 background species.In addition to N C Ar C1 and Al other sample components can also cause MO and MOH spectral interference problems. For example it can be seen from the spectra shown in Fig. 6(a) that some isotopes of Cd suffer spectral overlap from ZrO' species. This problem is illustrated in detail by the bar graph simulated spectrum shown in Fig. 6(b). Overall however for low alloy aluminium samples and for 0.01% solutions the analytes are normally at concentrations of less than 0.03 yg ml-' and thus MO and MOH spectral interference problems among analytes are minimal. Finally as mentioned above and in the last section alu- minium itself does not contribute in a major way to spectral interference problems for ICP-based determinations.It is a monoisotopic element and species such as A10+ AlOH' and Al,' are relatively minor or coincide with minor isotopes (A10' with 43Ca') or other plasma species (AlOH' with COz' Al,' with ArN'). Matrix efects and internal standardization In ICP-MS a high concentration of a matrix element is known to affect analyte sensitivity and analyte signal suppression is most commonly ~ b s e r v e d . ' ~ ~ ~ A set of data illustrating the matrix effect of excess aluminium (1000 pg ml-l) on the signal for Ni (0.1 yg ml-l) as a function of central gas flow rate is presented in Fig. 7(a). As shown by the plot in Fig.7(b) the Ni signal is seriously suppressed by aluminium if the central gas flow rate is set to the value that yields the maximum Ni signal. The matrix effect can be minimized [Fig. 7(c)] by a reduction in the central gas flow rate. All analytes exhibited behaviour similar to that shown for Ni in Fig. 7 . This approach to reducing matrix effects does result in a reduction of sensi- tivity. At the flow rate required for a reduced matrix effect (0.9 1 min-l) signal sensitivity was typically down 30-45%.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 827 Table 2 Basic spectral interferences for elements of major interest in aluminium alloys [natural abundances (YO) in parentheses] Background species Element 7Li (92.5) 'Be (100) 23Na (100) 24Mg(78.8) 25Mg (10.15) 28Si (92.21) 32S (95.02) 40Ca (96.94) 46Ti (8.01) 48Ti (73.98) "V (99.76) "Cr (83.76) 53Cr (9.51) 54Fe (5.82) "Mn (100) 56Fe (91.66) 57Fe (2.19) 58Ni (67.77) 60Ni (26.16) 62Ni (3.66) 63Cu (69.1) 64Zn (48.89) 65Cu (30.9) 69Ga (60.16) 71Ga (39.84) "Sr (82.58) "Zr (51.45) '"Sn (32.59) "*Pb (52.35) '09Bi (100) 2 7 ~ 1 ( i o o ) 31P (100) 5 9 ~ ~ (100) Isobaric overlap H20 HNO H2S04 H Cl - 40Ar (99.60) 46Ca (0.004) 48Ca (0.19) - - 54Cr (2.38) - - 58Fe (0.33) - - 64Ni (1.16) - - lZ0Te (0.1) - Internal standardization is used in almost all quantitative ICP-MS determinations.Not only does it compensate for multiplicative noises but because of the generally similar nature of matrix effects among the elements it also compen- sates for matrix effects.17 In this work Co Y and Rh were found to be effective as internal standard elements.during dissolution. The GD samples could of course be contaminated during the fabrication step. It is felt that the pre-sputtering step that is common in GD methodology would remove any surface transferred contamination. Internal stan- dards however cannot be added to solid samples and one must rely on sample components to provide an internal standard. In the analysis of steels a minor isotope of Fe (57Fe) can be used.6 This is not possible for aluminium analyses as aluminium is monoisotopic. An important aspect of the GD-MS analysis of aluminium samples presented here is the use of AlAr to function both as an internal standard and to compensate for matrix effects. Analytical results Aqueous calibration curves for both internal standard and matrix-matching methods were employed. For matrix match- ing standard solutions were prepared with the addition of aluminium at a concentration of 100 pg ml-' in order to match the 0.01% A1 sample solutions.In order to determine some of the lower level analytes 0.1% solutions of sample were also measured with matrix matched (1000 pg ml-' Al) standards. Results are shown for one of the Alcoa standards in Table 3 and for one of the Alcan samples in Table4. These results were obtained with the use of matrix-matched standard solu- tions but did not involve the use of an internal standard. Evaluation of background spectral interferences The spectral characteristics of GD-MS and ICP-MS with respect to the analysis of aluminium samples were presented earlier in this paper.It was seen that although the two sources do share some spectral features many differences exist both with respect to the nature and level of certain background species. Thus spectral interference problems must be specifi- cally evaluated for the GD source. In short for GD-MS in contrast to ICP-MS multiply-charged Ar species are present (Ar2+ Ar3+ Ar4+ etc.) background species originating from the components of air water and solution solutes are minimal or completely absent and matrix and analyte-based argide species are more prevalent than oxide species. Many specific examples were presented earlier during the discussion of Figs. The argide problem is further illustrated in Fig. 8. In the high alloy Alcoa aluminium standard SS-319E Cu is 3.83% and CuAr can be seen in the spectrum shown in Fig.8. The 63Cu argide presents a spectral overlap problem for the mono- isotopic element Rh and the 65Cu argide overlaps with one of 1-6. GD-MS Analyses For conductive metal alloys the d.c. GD can be used for the direct analysis of a sample although some sample fabrication (ie. machining) is often required. All samples in this study were formed into pins 3 mm in diameter and 15 mm in length. In the ICP-MS work difficulty was experienced in achieving complete dissolution of the high alloy aluminium samples probably because of their high Si content and they were not analysed by ICP-MS in this study. In contrast both the high and low alloy aluminium samples could be analysed by GD-MS.Also because the solid sample can be analysed directly GD-MS is not prone to contamination that can occur828 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 I 800 ( a ) Zr+,Nb+ - 600 v) v) C 3 .I- 8 400 \ c > v) Q .- c 5 200 0 80 85 90 95 100 105 110 115 120 125 105 110 115 120 125 4 ( C) I NbO r 3 v) C a 13 .- .I- .- c 2 l a > .- .I- - a 0 106 108 110 112 114 116 118 120 122 124 107 109 1 1 1 113 115 117 119 121 123 125 m/z Fig. 6 ICP mass spectra of an aluminium alloy sample (Alcan 1SWM) illustrating the overlap of elemental isotopes and metal monooxides. (a) Actual spectra; (b) enlargement of (a) m/z 105-125; and (c) simu- lated spectrum the isotopes of Pd. Manganese Fe Ni and Zn are present at 0.58 0.68 0.20 and 0.35'/0 respectively and low signal levels of their argides are also present.Matrix ejfects and internal standardization Two types of aluminium alloys were analysed by GD-MS low alloy aluminium and high alloy aluminium. The signal levels for several elements (normalized to the 1 % concentration level) in one alloy from each sample type are listed in Table 5. The normalized signal levels are 45-50% lower for the low alloy aluminium compared with the high alloy aluminium. This indicates that a matrix effect is present in that the low and high alloy aluminium samples could not be used to establish a single calibration curve for elements common to both alloy types. However it was noticed that the signal for AlAr' was affected in approximately the same way. Thus it appears that the AlAr signal can be used in the manner of an internal standard to compensate for this matrix effect.One should be U 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 Central gas flow rate/l min-' (6) 50 25 t -25 t - 50 0.1 10 100 1000 50 25 51 0 1 -25 1 - 50 0.1 1 10 100 1000 IAll/pg m l ~ ' Fig. 7 Effect of A1 concentration on the Ni signal as a function of (a) central gas flow rate for 0 pg ml-' of A1 (solid line) and 1000 pg ml-' of A1 (broken line) showing A flow rate for maximum signal and B flow rate for reduced matrix effects; (b) A1 concentration at the central gas flow rate yielding maximum signal intensity; and (c) at a reduced central gas flow rate Table 3 Results of the analysis of Alcoa standard SA-909 Element 24Mg 55Mn 58Ni 60Ni 63cu 65cu (j4Zn 66Zn Certified value( O/O) 0.030 0.03 1 0.034 0.034 0.03 1 0.03 1 0.030 0.030 Result(%) 0.0316 0.03 19 0.0353 0.0296 0.03 16 0.0302 0.0321 0.03 15 RSD(%) (n=4) 4 2 3 5 4 5 3 5 cautioned however that an argide may not under all GD operating conditions mimic analyte signal behavi0~r.l~ The difference in signal intensities between the low and high alloy aluminium samples can in part be accounted for by differences in sputtering rate.Each sample was sputtered twice for a 1 h period for the first time and a 2 h period for the second time. The difference in mass lost between these two sputtering periods was obtained and the sputtering rate calcu- lated. This gave a sputtering rate for the second hour. ThisJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 829 Table 4 Results for the analysis of Alcan standard 1SWL by ICP-MS Element 24Mg 48Ti "Mn 60Ni (j3CU W U 64Zn 66Zn 69Ga "Zr "Sn Certified value(%) 0.01 5 0.025 0.023 0.023 0.03 0.03 0.023 0.023 0.012 0.013 0.024 Result ( YO) 0.0142 0.0243 0.0229 0.0225 0.0340 0.0335 0.0260 0.0243 0.01 11 0.01 30 0.0173 RSD(Yo) (n=4) 3 2 2 2 2 2 2 2 3 2 2 CuAr' 800 11 '''Nb+ I 1 ZnAr' r I 600 c C 3 8 \ 2 400 .- tn al 4- - 201r 0 85 90 95 100 105 110 115 120 125 m/z Fig.8 Glow discharge mass spectrum of a high alloy aluminium sample (Alcoa SS-319E) approach was used as these aluminium samples required significant pre-sputtering time ( 15-20 min) before stable signals could be obtained.In this experiment the sputtering rate for the low alloy aluminium (SA-909) was about 40+ 3 pg min-' while the sputtering rate for the high alloy aluminium (SS-319E) was about 57 & 3 pg min-'.This is a 30% difference and partially explains the signal differences shown in Table 5. Further explanation may lie in differences in the degree of ionization in the GD for these two sample types but this is just speculation. Whatever the complete explanation may be for the signal differences observed for these two alloy types use of AlAr as an internal standard does compensate for this matrix effect. The AlAr+ also functions as a classic internal standard to compensate for signal changes as a function of time. These samples required a considerable pre-sputtering time in the range of 15-20 min. This is illustrated in Fig. 9(a) for the Cu signal. The AlAr' signal [Fig.9(b)] follows the analyte signal over this time and the ratio (Cu:AIAr) stabilized (zl-2% relative standard deviation) in about 20 min [Fig. 9(c)]. This 8000 (a) 6000 4000 2000 0 0 10 20 30 40 50 Ti me/m in Signal intensity (GD-MS) as a function of time for (a) Cu; and (b) AlAr; and (c) the signal intensity ratio of Cu to AlAr pre-sputtering time is probably required to remove an oxide coat from the aluminium samples. An attempt was made to shorten this pre-sputtering time by operating the GD device at higher currents but sparking often ensued and this approach to shortening this time has not yet proved to be reliable. On the other hand Marcus and Duckworthg have shown that by the use of an r.f.-powered GD rapid analysis of aluminium samples should be possible.Analytical results A group of calibration curves was established by using the Alcan low alloy aluminium standards lSCXG lSWL lSWM Table 5 Comparison of relative signal intensities (GD-MS) of analytes (normalized to 1 YO level) between high.and low alloy aluminium samples Normalized intensity* Mass 24 48 55 58 63 64 67 Analyte Mg Ti Mn Ni c u Zn AlAr High alloy (90% Al) 2 10000 280000 290000 140000 110000 126000 8000t Low alloy (99yo Al) 120000 15oooO 150000 70000 60000 64000 4000 Loss in intensity for low alloy Al(%) 43 46 48 50 45 49 50 * Signals normalized to the 1% concentration level. t Normalized intensity of AlAr for the high alloy is 7200/0.90= 8000.830 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 Table 6 Detection limits of trace elements directly in Alcan aluminium solids by GD-MS Elements Ti V Mn Ni c o c u Zn Ga Zr Sn Pb Bi Mg Mass 24 48 51 55 58 59 63 64 69 90 120 208 209 Detection limits (ppb) 108 42 31 30 59 30 63 58 44 10 41 87 56 Table7 Results for the analysis of lSWL low alloy aluminium by GD-MS Element Mg Ti V Mn Ni c o c u Zn Ga Zr Sn Pb Bi Certified value( %) 0.015 0.025 0.019 0.023 0.022 0.00 1 0.030 0.023 0.012 0.0 13 0.024 0.018 0.018 Result (%) 0.0145 0.0230 0.0160 0.0202 0.0205 0.001 3 0.0278 0.0209 0.0107 0.0129 0.0225 0.0195 0.0193 RSD(%) 2 1 1 2 2 6 3 3 3 2 3 1 2 2.5 2.0 1.5 (a) - - -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5 1.0 - 2.5 2.0 1.5 1 .o 0.5 0 -2.5 -2.0 -1.5 -1.0 -0.5 0 0.5 1.0 Log [concentration (%)I Fig.10 Calibration curves for (a) 24Mg and (b) 63Cu in both high and low alloy aluminium standards with AIAr' (m/z 67) as the internal standard Table 8 Results for the analysis of SS-360-C high alloy aluminium by GD-MS Certified Element value( Yo) Result(%) R S D (Yo) Mg 0.52 0.503 4 Ti 0.079 0.076 1 Mn 0.22 0.224 4 c u 0.3 1 0.308 3 Zn 0.25 0.248 4 lSXD and 2SDZ for the determination of Mg Ti V Mn Ni Co Cu Zn Ga Zr Sn Pb and Bi; AlAr' was the internal standard.The concentrations for these components in the standards were in the range of 0.001-0.030% except for one of the standards which contained 0.25% Mg. The slopes of the log-log plots for the calibration curves were in the range of 0.95-1.06. Detection limits (3s) for these elements were evalu- ated. They were obtained based on the standard deviations (n= 16) of background noise and the sensitivity of analyte signals.The detection limits (mass-%) calculated are summar- ized in Table 6. The low alloy aluminium 1SWL was analysed for all the above components as an 'unknown'. The results are listed in Table 7 and the standard deviations for the results were calculated based on n=6. Another group of calibration curves was established by using both low and high alloy aluminium standards from Alcoa for the determination of Mg Ti Mn Ni Cu and Zn. The seven standards were SA-909 SA-1170 SA-1169 SS-356-B SS-A132AA SS-D132-A and SS-3 19E. The composition of the components for the Alcoa standards was in the range of 0.009-1.3% for Mg 0.03-2.5% for Ni and 0.03-3.8% for Cu and the slopes of the log-log plots for the standard curves were usually in the range 0.85-0.95.Calibration curves for 24Mg and 63Cu with AlAr+ as the internal standard are shown in Fig. 10. The slopes of the log-log plots in Fig. 10 are 0.90 (Mg) and 0.89 (Cu) and the correlation coefficients are 0.998 (Mg) and 0.996 (Cu). Note that with the use of AlAr' as the internal standard both low and high alloy aluminium standards can be used to establish a single calibration curve. An Alcoa high alloy aluminium sample (SS-36O-C) one not used to establish the calibration curves was analysed as an 'unknown' sample and the results are listed in Table 8. Conclusions Clearly both ICP-MS and GD-MS can be successfully applied to the analysis of aluminium alloys. The GD-MS technique has the advantage that sample dissolution is not required and that air solvent and solute do not contribute to the spectral background.However potential spectral interferences unique to each technique still abound and care is still required in assessing their presence or absence. It is also important to point out that even though dissolution involves dilution of the analyte 100-10 000-fold when 1-0.01 YO sample solutions are prepared ICP-MS still has comparable or superior detection limits when referenced back to the solid composition; a conse- quence of the 1-10pgml-' range of detection limits for ICP-MS uersus the 1-10 ng g-' detection limits currently typical for GD-MS. Finally ICP-MS analyses typically require dissolution of a sample. This step is prone to contamination and for many samples is difficult to quantitatively complete.In fact many new materials are simply difficult to dissolve. On the other hand GD-MS analyses do have time-consuming steps involving sample form fabrication and pre-sputtering and certainly some contamination is possible during fabrication although the pre-sputtering step should minimize surface trans- ferred contaminants. References Horlick G. and Shao Y. Inductively Coupled Plasma Mass Spectrometry for Elemental Analysis in Znductiuely Coupled Plasmas in Analytical Atomic Spectrometry ed. Montaser A. and Golightly D. W. VCH Publishers 2nd edn. 1992 pp. 551-612. Harrison W. W. J. Anal. At. Spectrom. 1992 7 75. McLaren J. W. At. Sectrosc. 1992 13 81. Koppenaal D. W. Anal. Chem. 1990,62 303R.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 83 1 5 6 7 8 9 10 11 12 Kim H. J. Piepmeier E. H. Beck G. L. Brumbaugh G. G. and Farmer 0. T. Anal. Chem. 1990 62 639. Shao Y. and Horlick G. Spectrochim. Acta Part B 1991,46 165. Kawaguchi H. Paper presented at the 1992 Winter Conference on Plasma Spectrochemistry San Diego CA USA January 6-11 1992 (paper No. IL14). Takeda K. Yamaguchi T. Akiyama H. and Masuda T. Analyst 1991 116 501. Marcus R. K. and Duckworth D. C. Pittcon Atlanta GA USA 1989 (paper No. 657). Vieth W. and Huneke J. C. Spectrochim. Acta Part B 1991 46 137. Vassamillet L. F. J . Anal. At. Spectrom. 1989 4 451. Shilomatsu H. M. and Iyer S . S. Nucl. Instrum. Methods Phys. Res. Sect. A 1988 280 488. 13 Tan S. H. and Horlick G. J . Anal. At. Spectrom. 1987 2 745. 14 Vaughan M. A. and Horlick G. J. Anal. At. Spectrom. 1989,4,45. 15 Vaughan M. A. and Horlick G. Spectrochim. Acta Part B 1990 45 1301. 16 Ward A. F. and Marciello L. F. Anal. Chem. 1979 51 2264. 17 Tan S. H. and Horlick G. Appl. Spectrosc. 1986 40 445. 18 Vaughan M. A. and Horlick G. Appl. Spectrosc. 1986 40 434. 19 King F. L. McCormack A. L. and Harrison W. W. J. Anal. At. Spectrom. 1988 3 883. Paper 3/0665OI Received November 5 1993 Accepted March 22 1994