JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JULY 1994 VOL. 9 785 Analysis of Zirconium Alloys by Inductively Coupled Plasma Atomic Emission Spectrometry I. Steffan Institute for Analytical Chemistry University of Vienna Wahringerstr. 38 A- 7 090 Vienna Austria G. Vujicic I WM Industriestr. 59 CH-8 7 52 Glattbrugg Switzerland In the inductively coupled plasma atomic emission spectrometric analysis of real samples influences of the matrix on the analyte spectral lines could be expected especially when the ratio of the concentration of the matrix to that of the analyte is extremely high. In addition to the elimination of stray-light effects by background correction there are also spectral interferences which must be investigated. It is of great importance to obtain information concerning the spectral line overlaps before setting up an analytical programme. The experiments performed dealt with the interferences to be expected in a Zr matrix.In general because of the nature of Zr the chemistry and technology associated with this element are very complicated therefore the methods applied in its analysis have specific requirements. Where possible for all the analytes of interest (Al 6 Cd Co Cr Cu Fe Hf Mg Mn Mo Nb Ni Sn Ta V U and W) spectral lines were selected that were free from interferences caused by the matrix element. Where no interference- free analyte line could be found it was necessary to perform line interference corrections. The analytical programme developed in this way was tested with Standard Reference Materials (National Institute of Standards and Technology).The data compare well with the certified values. The confidence intervals for 95% probability were in the range of 2-12% depending on the element and the concentration. Keywords Zirconium; zirconium alloys; line selection; spectral interference The most important Zr ore is zircon a zirconium silicate plus hafnium silicate. More than 95% of the production of Zr is used in the form of zircon or zirconia ZrO for foundry moulds and refractory ceramic and abrasive materials. The metal is also used in the construction of chemical plants. It would appear that Zr is non-toxic and is compatible with body tissue and has thus become a competitor with Ti as a component of artificial joints and limbs. The good mechanical properties combined with resistance to corrosion and a low neutron absorption cross-section are the main reasons for the important role Zr plays in nuclear reactor technology.In the ore form Zr is associated with Hf (with an Hf content of up to 7%) which offers a very high neutron cross-section. This results in difficult separation chemistry from the two adjacent elements in the Periodic Table and isolation of Zr containing less than 100 mg kg-I of Hf as an impurity. In addition Zr has a very high chemical reactivity with respect to different metals and the environment in particular 0 N H and C and it is also necessary to eliminate numerous unwanted from a metallurgical point of view impurities. All of these require- ments lead to the complicated chemistry and metallurgy for this element and its compounds and the necessity of separating Zr from Hf and other elements.’ Inductively coupled plasma atomic emission spectrometry (ICP-AES) is usually reported to be almost free from inter- ferences nevertheless some interferences have to be considered.2 It is convenient to distinguish between spectral and non- spectral interferences.Spectral interferences arise from the incorrect isolation of the net analysis signal from the composite radiation that passes the ‘spectral window’ tuned to the analysis line causing a lack of selectivity for the method. Occasionally this problem can be partially overcome by appropriate choice of excitation condition^.^ Spectral interferences can be caused by continua stray light line wings and lines or bands.The interference signals produced by a smooth and unstructured background are directly measurable in the sample spectrum at positions adjacent to the analysis lines. This enables a reliable estimate of the background below the analysis line. Interferences from lines or band components produce greater problems because their magnitude can only be determined indirectly. Background correction is one of the most difficult problems to deal with in emission spectrometry since the background under a spectral line cannot be determined directly and in addition varies with the composition of the sample. The principal method of reducing spectral interferences is selection of the appropriate analytical lines the choice of the spectrometric apparatus and the detector. For a given spec- trometer and detector line selection is based on two criteria detection limit and selectivity.A sequence of lines with decreas- ing detection limits for a certain element are usually checked for the occurrence of particular interferences. The main problem in the spectrometric analysis of Zr and Zr alloys is the line-rich emission spectrum of Zr. The object of this investigation was the selection of lines for the ICP-AES determination of trace elements present as impurities in Zr alloys in a quality control procedure for these materials. Experimental Apparatus For the studies presented an ARL 3520 ICP spectrometer was used. The technical data for this instrument are listed in Table 1. Samples In order to test the analytical procedure National Institute of Standards and Technology (NIST) zirconium Standard Reference Materials (SRM) 360a 1238 and 1239 were anafysed.Table 1 Technical data for the ARL 3520 ICP spectrometer Spectrometer Grating Torch Nebulizer Gas flows Outer Intermediate Aerosol carrier Incident power Observation height 1 m sequential Paschen-Runge 1200 grooves mm - ’ FasseI type Meinhard type 1211 rnin-l 0.8/1 min-’ i/1 min-’ 1200/w 15 mm above coil786 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JULY 1994 VOL. 9 Sample Dissolution Sample dissolution was performed according to the following procedure. A portion (2 g) of sample was treated with 10 ml of a mixture of concentrated hydrofluoric acid and 7 mol I-' nitric acid (1 + 2; added in portions) in a platinum dish. The dissolution was accompanied by a violent reaction.Then 2-3 ml of water were added. As soon as the reaction had calmed down 3 ml of hydrofluoric acid were added and the dissolution was completed by the digestion of the covered sample for 1 h on a steam-bath. The digest was subsequently evaporated to dryness and the residue was dissolved in 100 ml of 1.2 mol I-' hydrochloric acid. Reagents All the reagents used were of analytical-reagent grade (Merck Suprapur). Standard solutions were prepared from stock solu- tions containing 1 mg ml-' of each element (Merck Titrisol). Line Selection and Experiments For line selection in most cases a list of the 'most sensitive' or 'prominent' lines of the elements to be detem~ined~-~ was checked for applicability to ICP analysis. The 28 most interes- ting lines tested are listed in Table 2.For all the measurements performed solutions containing zirconium oxychloride with and without the addition of the Table 2 Spectral lines investigated Element A1 B Cd Cd c o c o Cr Cr c u Fe Fe Hf Mn Mg Wavelength/ nm 308.215 182.589 214.438 226.502 230.786 228.6 16 205.552 357.869 224.700 239.562 238.204 2 7 3.8 7 6 279.553 293.930 Element Mn Mo Mo Nb Ni Sn Ta Ta Ti V V U W W Wavelength/ nm 294.920 204.598 28 1.615 309.417 221.647 181.110 267.590 226.230 334.900 311.838 309.3 1 1 290.828 216.632 208.8 19 Table 3 DLs and BEC values for the lines selected analytes of interest were used. The concentration of the stock solution of Zr was 141.4 g I-' of ZrOC12.8H20. According to the expected concentrations of Zr in the real samples dilutions were made for the investigations.Detection limits (DL) and background equivalent concentration (BEC) values were calcu- lated for pure aqueous solutions of the analytes and for solutions of the same analyte concentrations in the presence of a Zr matrix for 11 integration^.^ With such solutions spectral scans were performed over a range of kO.06 nm to both sides of the lines investigated. The integration time was 2 s per step and the step size 0.004 nm. These scans were used both for selection of the analyte lines and for evaluation of the appropriate background correction position(s). The correction coefficients for the spectral inter- ferences observed and identified in the scans were determined by concentration measurements using synthetic Zr solutions of increasing concentrations.For lines suffering from inter- ferences caused by the Zr matrix correction coefficients A and B were calculated according to the equation cEl= Bczr + A where cZr is the concentration of Zr (%) and cEl the concen- tration (pg ml-l) measured at the analyte line in the presence of the Zr matrix. Calibration was performed using the Zr matrix stock solution with addition of increasing amounts of the analytes. The concentrations of Zr and the analytes were adjusted to the concentrations expected in the samples. The analytical programme established was tested by its application to the SRMs. Each standard sample was dissolved 12 times and analysed. The mean values of the concentration and the confidence intervals for 95% probability were calculated.Results and Discussion The concentrations of the unwanted elements in Zr materials are very low and therefore the influence of the Zr spectrum on the analyte lines was investigated systematically. For this purpose spectra were measured using synthetic standard samples. For line selection the 28 spectral lines listed in Table 2 were checked and eventually 19 lines were selected for analytical purposes (Table 3). Also shown in Table 3 are the detection limits and BEC values obtained for pure aqueous solutions and for comparison purposes in the Zr matrix used for the investigations. For better comparison the ratios of the DLs in water and in the Zr solution are also listed. It is evident that the DLs in the Zr matrix are worse than those obtained for pure aqueous solutions of the single-element standards.Element A1 B Cd c o Cr cu Fe Hf Mg Mn Mo Nb Ni Sn Ta Ti V U W Wavelength/ nm 308.21 5 182.589 214.438 230.786 205.552 224.700 239.562 273.876 279.553 293.930 204.59 8 309.41 7 221.647 181.110 267.590 334.900 311.838 290.828 216.632 DL*/ pg ml-l 0.028 0.016 0.006 0.042 0.019 0.014 0.027 0.055 0.001 0.009 0.174 0.034 0.030 0.071 0.160 0.044 0.004 0.084 0.293 BEC*/ pg ml - 2.73 0.59 0.3 1 3.51 0.9 1 0.81 1.38 3.60 0.06 1.24 4.78 1.14 0.93 3.54 11.56 0.67 0.70 7.62 21.76 DL-F/ pg ml-' 0.061 0.021 0.01 5 0.065 0.032 0.04 1 0.028 0.110 0.001 0.029 0.268 0.068 0.032 0.123 0.323 0.083 0.020 0.112 0.743 BECtIl m1- 3.97 1.12 0.68 3.91 1.70 2.34 1.72 4.01 0.06 1.77 8.80 1.93 1.61 7.54 12.17 1.02 1.17 8.81 22.90 DL* DLt 0.46 0.76 0.40 0.65 0.59 0.34 0.96 0.50 1 .00 0.31 0.65 0.50 0.94 0.58 0.50 0.53 0.20 0.75 0.39 * Water.t Matrix.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JULY 1994 VOL. 9 787 Some of the scans obtained are shown in Figs. 1-6. In each figure the measurements performed with a pure Zr matrix are compared with scans using the same matrix solution but with the addition of 10 pg ml-I of the elements of interest. The A1 line at 308.215 nm (Fig. 1) shows a simple background shift and a complex line overlap. This simple 'flat' background can be corrected by appropriate selection of the background correc- tion position (see Table 4). Because of the complex line overlap calibration should be performed in the Zr matrix. The B line at 182.587 nm (Fig. 2) was selected because the more sensitive B line (182.641 nm) suffers from additional interference caused by the S line at 182.62 nm.Therefore the background correc- tion position was also selected at the lower wavelength side of the peak. For this line matrix interference was observed. The Cd line at 214.438 nm represents an interference-free line for the Zr matrix. The Co line at 230.786nm was chosen for the determination of Co in the Zr matrix. The Cr line at 205.552 nm (Fig. 3) was chosen for the determination of this trace element in the Zr matrix after application of matrix interference correc- tion. The Cu line at 224.700 nm is an interference-free spectral line for the Zr matrix. For the determination of Fe the line at 239.562nm was selected. The selection of a suitable Hf line 308.14 308.18 308.22 308.26 308.30 308.16 308.20 308.24 308.28 Wavele ngt hln rn Fig. 1 Scan around the 308.215 nm A1 line A HCl; and C Zr matrix+ 10 mg 1-l of A1 50 I -3j 40 i i I I I B ' I '7 ' I I / I I I \ I I \ I I \ I I 3 Zr matrix; 182.52 182.56 182.60 182.64 182.54 182.58 182.62 182.66 Wavelengt hln m Fig.2 Scan around the 182.589 nm B line A Zr matrix; and B Zr matrix + 10 mg 1-l of B 25 1 .- 3 201 3 P B :, * I 0 I I I t 1 I I 205.48 205.52 205.56 205.60 205.50 205.54 205.58 205.62 Wavelengthlnm Fig. 3 Scan around the 205.552 nm Cr line A Zr matrix; and B Zr matrix+ 10 mg I-' of Cr 600 ; 500 Y .- 5 3 L. 400 E e .- < 300 w .- m al .- E 200 - m C 0 i7j 100 0 309.34 309.38 309.42 309.46 309.50 309.36 309.40 309.44 309.48 Wavelengthlnm Fig.4 Scan around the 309.417 nm Nb line A Zr matrix; and B Zr matrix+ 10 mg 1-1 of Nb 25 I ; I \ ' I i l l 0 221.58 221.62 221.66 221.70 221.60 221.64 221.68 221.72 Wavelengt hln m Fig. 5 Scan around the 221.647 nm Ni line A Zr matrix; and B Zr matrix+ 10 mg 1-' of Ni788 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JULY 1994 VOL. 9 was of special interest because of the importance of the efficient separation of Zr and Hf (as discussed previously). The line at 273.876 nm was found to be suitable for the determination of Hf in Zr matrices. The Mg line at 279.553 nm is a very sensitive 0 181.04 181.08 181.12 181.16 181.06 181.10 181.14 181.18 Wavelengthlnm Fig.6 Scan around the 181.110nm Sn line A Zr matrix; B Zr matrix+ 10 mg I-' of Sn; and C 250 mg 1-' of Sn Table 4 Background correction positions for the lines selected Element A1 B Cd c o Cr c u Fe Hf Mn Mo Nb Ni Sn Ta Ti v U W Mg Wavelength/ nm 308.215 182.589 214.438 230.786 205.552 224.700 239.562 273.876 279.553 293.930 204.598 309.417 221.647 181.110 267.590 334.900 311.838 290.828 216.632 Background correction position/ nm + 0.060 - 0.025 - 0.030 + 0.048 + 0.036 + 0.048 + 0.036 - 0.036 + 0.036 + 0.036 + 0.036 - 0.030 + 0.040 - 0.048 + 0.036 - 0.048 - 0.030 + 0.034 - 0.0182 Table 5 Spectral lines tested in the Zr matrix but not selected for the analysis line without spectral interferences in the matrix investigated.The Mn line at 293.930nm was selected as it offers the possibility of background correction and compared with other Mn lines investigated the surroundings are relatively inter- ference free.Similar to the Mn 293.930nm line the intensity of the surroundings of the Mo line at 204.598 nm are almost constant. This offers a good possibility for background correc- tion on both sides of the peak Of all the Nb lines investigated the line at 309.417 nm (Fig. 4) shows the best spectral charac- teristics independent of the possible influences of the peak- wing interference produced by a Zr line at 309.507 nm. This interference had to be corrected for. Other Nb lines exhibit lower detection power or suffer from even stronger inter- ferences. The scan around the Ni line at 221.647 nm is shown in Fig. 5. The small peak on the right side of the Ni peak was identified as a spectral line of Si at 221.669 nm. This interference was corrected for in the analytical programme.The Sn line selected (181.110 nm Fig. 6) shows only weak interferences and relatively poor detection power. Of all the Ta lines investigated the line at 267.590 nm showed the best detection limit and only weak interferences. The V line at 311.838 nm is practically free from interferences. Of all the U lines investi- gated only the spectral line at 290.828 nm shows relatively good properties suitable for analytical purposes. The results of the investigations of W lines were similar to those of U. The W line at 216.632 nm was finally selected for determination in the Zr matrix. The scans were also used to select the background correction positions of the spectral lines chosen (Table 4). The behaviour of the lines tested for their applicability for measurements in the Zr matrix but not selected is shown in Table 5.For the A1 (308.215 nm) B (182.589 nm) Cr (205.552 nm) Nb (309.417nm) Ni 221.647nm and Sn 181.110nm lines interferences caused by the Zr matrix were observed The necessity for on-peak corrections appears to be due to both partial spectral overlap and changes in the sensitivity of the analytical calibration with respect to changes in concentration of the matrix. Correction coefficients were calculated by a linear approach. The correlation coefficients vary in the range The detection limits calculated for solids using the DLT values from Table 3 and the mean values of the concentrations obtained for the NIST SRMs after 12 determinations are shown in Table 6. Unfortunately not all of the 19 elements investigated were certified in the SRMs.The DLs calculated vary between 1.4 pg g-' for Fe and 13 pg g-' for Mo. For W the DL is 37 pg g-' which is extremely high compared with the other elements nevertheless a value of 44 pg 8-l with a confidence interval of 6.8 % was obtained for the determination of W in SRM 1239. The detection limit for U was not sufficient for a direct determination in SRM 360a. Similar to the determination of U in many complex matrices a preconcen- tration step could be advantageous. For the determination of 0.985-0.999. Element Cd c o Cr Fe Mn Mo Ta v W Wavelength/nm 226.502 230.786 357.869 238.204 294.920 28 1.61 5 226.230 309.3 11 208.819 Type of interference Not serious Two interfering peaks Intense line wing Line overlap Line wing and line Matrix Line Complex line overlap Line Comment Usable Not usable Usable (limited) Not usable Not usable overlap Usable (limited) SBR* poor Not usable Difficult to correct * SBR signal-to-background ratio.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JULY 1994 VOL.9 789 Table 6 Analytes determined in the Zr SRMs from NIST Element c u Cr Fe Hf Mn Mo Ni Sn (YO) Ti U W DL*/ 2.1 3.3 1.4 5.5 1.5 1.6 6.2 4.2 5.6 Pg 8-l 13 37 SRM 360a SRM 1238 Certified 140.00 1060.00 144 1 .OO 3.00 554.00 1.42 27.00 0.15 - - - ~ Found 138.0+ 6 1048.0 & 44 1450.0 f 30 2.8 k 0.1 538.0 f 18 1.4 & 0.1 5 24.0 + 2 t D L - - - Certified 60 580 2500 178 60 120 100 100 90 - - Found 62k4 592 f 30 2420 & 100 180f6 58$2 124$-5 110i-5 95f5 95+4 - - Certified 130 1055 2300 77 50 45 45 40 45 - - Found 129&8 1050 & 42 2180&80 74&4 48k2 40+4 40+5 38+3 44f3 - - * Detection limits calculated in the solid form using DLf values from Table 3.all other elements certified in the NIST SRMs the confidence intervals for 95% probability vary from 2 to 12% depending on the elements and the concentration. The data obtained were in good agreement with the certified values. Conclusion Spectral interferences can seriously limit the performance of emission spectrometric analyses. The idealized DLs should not be over-emphasized. Detection limits conventionally deter- mined on blank matrices are not necessarily applicable to the samples themselves. The most important decision in AES is the selection of appropriate analysis lines and wavelength@) position(s) for the background measurements. Even with the use of modern computer technology these decisions must still be made by the analyst. In some cases the choice of the interference-free analyte lines is not possible and these diffi- culties must be overcome by applying interference corrections in the course of the analytical programme or even by appli- cation of separation techniques for the analytes. This work was sponsored by the Fonds zur Forderung der Wissenschaftlichen Forschung Vienna Austria project no. L j795. References Schemel J. H. Metal Handbook American Society for Metals OH USA 9th edn 1980 vol. 3. Inductively Coupled Plasma Emission Spectroscopy. Part I ed. Boumans P. W. J. M. John Wiley New York 1987. Boumans P. W. J. M. in Inductively Coupled Plasma Emission Spectroscopy. Part I ed. Boumans P. W. J. M. John Wiley New York 1987 p. 22. Boumans P. W. J. M. Line Coincidence Tables for Inductively Coupled Plasma Atomic Emission Spectrometry Pergamon Press Oxford 1984. Winge R. K. Fassel V. A. Peterson V. J. and Floyd M. A. Inductively Coupled Plasma Atomic Emission Spectrometry. An Atlas of Spectral Information Elsevier Amsterdam 1984. Parsons M. L. Forster A. and Anderson D. An Atlas of Spectral Interferences in ICP Spectroscopy Plenum New York 1980. Paper 4/00171 K Received January 11 1994 Accepted March 31 1994