JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL. 9 Analysis of Electrical Arc Furnace Flue Dusts by Spark Ablation Inductively Coupled Plasma Atomic Emission Spectrometry* A. G. Coedo M. T. Dorado and 1. G. Cob0 Centro Nacional de Investigaciones Metalurgicas Consejo Superior de lnvestigaciones Cientificas Gregorio del Amo. 8,28040 Madrid Spain 223 A medium-voltage spark was used for the direct nebulization of electric arc furnace (EAF) flue dust. In order to attain the necessary sample conductivity powder pellets are briquetted after mixing the sample 1 +1 with graphite. The elutriated material was excited in an argon inductively coupled plasma (ICP). The use of cellulose as binder provides better results in terms of reproducibility. After optimization of the spark parameters (voltage 500 V; repetition rate 400 s-'; and resistance 2.2 Q) the carrier gas flow rate (2.1 I min-' of argon) and the operating power of the ICP (1.2 kW) precisions (relative standard deviation) for zinc lead cadmium and iron range from 0.8 to 2.0%.The stability of the spark sampling during a complete spark ablation (SA) ICP process ( ~ 9 0 s) was tested by plotting emission intensity versus time profiles. The similarity between the amounts of analyte obtained from different pellets was proven by collecting the spark-eroded particles and analysing their carbon contents. Five steelmaking EAF flue dusts were selected for this study using the two samples with extreme contents of the elements considered for calibration. The results obtained by SA-ICP matched the results obtained by ICP from nebulized solutions.Keywords Spark sampling; non-conductive powder; steel making; dust analysis; spark ablation inductively coupled plasma atomic emission spectrometry The environmental pollution caused by residues from electric arc furnace flue dusts which are considered to be toxic and hazardous products poses a real problem for the iron and steel industry. The use of galvanized sheets is becoming more widespread and when this material is recycled in electric arc furnaces it produces dusts containing Zn Pb and Cd which can be leached into the natural environment and can therefore accumulate in soil and water. Technological processes for treating these powders are being developed intensely at the present time.'y2 Optimization of these processes requires appli- cation of analytical control methods that ensure high sensitivity and selectivity combined with acceptable accuracy and speed. Inductively coupled plasma atomic emission spectrometry (ICP-AES) meets these requirements making possible the analytical follow-up of such processes.The system normally used is a sample dissolution step followed by ICP-AES yet this procedure is hampered by the dissolution of some of the refractory oxides that can be present. Acid digestion and alkali fusion of the insoluble residue is the conventional sample dissolution procedure. This dissolution procedure reduces the power of detection as a result of analyte dilution and high saline concentration and is time-consuming and labour inten- sive.Spark ablation has been applied for bulk analysis of conducting solids3 and powder samples4 and can simplify the ICP analytical process by eliminating the dissolution step. One possible form of preparing non-conductive powder for spark ablation (SA) sampling is to briquette the powder into pellets with a metallic powder to achieve the necessary cond~ctivity.~*~+~ This study aims to ascertain the feasibility of the SA-ICP technique for the analysis of steelmaking flue dusts. Different pellet preparation procedures were investigated to obtain physically stable test samples with adequate conductivity and the optimum proportion of the components dust conducting powder and binder. The influence of sparking operating param- eters (voltage resistance and repetition rate) was studied the total amount of eroded material was controlled and the analytical performance of the technique evaluated.A precision test was conducted to ascertain the repeatability of the ICP measurements and the reproducibility of the sample prep- * Presented at the XXVIII Colloquium Spectroscopicurn Internationale (CSI) York UK June 29-July 4 1993. aration system. Finally the results obtained for the selected samples by using the proposed method and the corresponding standard deviations (SDs) are presented. Experiment a1 Instrumentation A JY-SAS sparking unit was used as the solid sampling source. The ablated material was excited in a JY 24 ICP spectrometer. Instrumental details and working conditions are listed in Table 1. Samples Representative samples of electric arc furnace (EAF) flue dust were selected for the development of the SA-ICP analytical procedures. Consideration was first given to the information provided by different steelmaking companies on the types of steel they manufacture their steel production and the flue dust generated during the steelmaking process.These samples were initially analysed by ICP spectrometry from nebulized solu- tions obtained with the following dissolution procedure 0.200 g of sample was fused at 1200°C in a Pt crucible with 1 g of the flux mixture consisting of borax and sodium carbonate (1 1). Table 1 Optimized working conditions for SA-ICP Sparking parameters Voltagep Capacit ance/pF Inductance/pH Resistance/Q Repetition rate Electrode (cathode) Permanent carrier gas/l min- ' Analysis carrier gas/l min-' Carrier gas pressure/bar Distance/m Diameter of the transport tube/mm Power/kW Plasma gas/l min - Sheath gas) min-' Transport ICP 500 1 20 2.2 400 s-' Tungsten rod 0 = 2 mm 2.1 0.8 3 1 5 1.2 14 0.2224 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL.9 Table 2 ComDosition of EAF flue dusts (ICP pneumatic nebulization; n := 6) Zn (YO) Pb (Yo) CD (Yo) Fe (YO) - Sample X d X 0 X d X 0 DUST 1 12.52 0.07 2.76 0.05 0.04 1 0.00 1 34.75 0.18 DUST 2 24.6 1 0.15 5.33 0.112 0.47 0.005 15.51 0.15 DUST 3 36.00 0.20 11.21 0.1,o < 0.01 8.55 0.09 DUST 4 44.22 0.2 1 9.60 0. II 1 2.001 0.019 4.92 0.06 DUST 5 46.18 0.23 15.14 0.20 < 0.01 0.88 0.05 The melt was digested with water and dissolved by adding 3 ml of 65% HNO,+ 1 ml of 37% HCl.The final volume of the resulting solution was adjusted to 200ml with water. The samples selected and their composition presented as the average of six values obtained by two different operators with ICP-AES from nebulized solutions are listed in Table 2. Pellet preparation A 4 g portion of sample + 4 g of graphite + 2 g of cellulose were mixed and homogenized in a ball mixer/mill and pressed into a pellet (0=4 cm; thickness= 10 mm) with a load of 40 t cm-2 for 30 s in a hydraulic press. Results and Discussion Sample Preparation Various tests were performed to define suitable procedures for obtaining stable pellets with the necessary conductivity. The main advantage of flue dusts is their extremely fine particle size (95% have a particle size of less than 0.5 pm) which makes these products suitable for direct compaction.The matrix was modified by mixture with a conducting host material. The conducting powders tested were copper alu- minium and graphite. The criteria used to select the most appropriate diluent powder were high conductivity ease of handling low cost acquisition of bulk samples providing stable and reproducible responses to electric discharges and absence of inter-elemental spectral interferences. Copper produces spec- tral interferences in the two most sensitive analytical Zn lines (Zn 213.856nm is interfered with by Cu 213.853 and Zn 202.548 nm is interfered with by Cu 202.434-202.555 nm) and the presence of aluminium powder results in deep erosion and consequently to a massive input of sample even when using very low voltages which decreases the plasma stability.The use of graphite as diluent provides the best results and conse- quently this conductive powder was selected for further tests. The addition of a binder improves the mechanical properties of the pellets enables a better ablation to be carried out and increases the reproducibility of the eroded material. The two binders tested were N-butyl methacrylate (Elvacite) and cellu- lose. Binding with cellulose provides the best results in terms of precision of the SA-ICP results. Another essential aspect of pellet preparation is the ratio of sample to conductive powder and binder. Different proportions of the components were tested 2 + 1 + 0.5; 2 + 1 + 1; 1 + 1 + 0.5; 1 + 2 +0.5 and 2+2 +0.5. With the dilution ratio 1 + 1 +0.5 for sample graphite cellulose the optimum mechanical properties of the pellets and good conductivity were achieved. Unlike spark emission the conductivity of the pellets is not of primary importance because the spark in this case only per- forms ablation while excitation is achieved by the ICP.Consequently the pellet preparation procedure described above was adopted. The analytical line sensitivity and spectral interferences were studied by using solutions containing approximately the back- ground equivalent concentration (BEC) of the element being investigated and the maximum expected concentrations of all the others. This study showed that by using graphite as conductive material the most sensitive lines listed in ref. 8 can be used for the determination of the elements being studied using an ICP Zn=213.856 nm Pb=220.353 nm Cd= 214.438 nm and Fe = 259.940 nm.Optimization of Spark Sampling Spark ablation conditions are defined by the following three parameters voltage (V) resistance (R) and repetition rate (f). The mildest practical condition for SA is obtained with the following parameters I/= 350 V R = 2.2 R and f= 400 Hz no application work was performed for lower frequencies owing to the poor sensitivity and repeatability obtained. Harsher conditions are obtained by decreasing resistance and increasing voltage. Lower resistance values increase the amount of analyte significantly causing material to be deposited in the injector tube of the plasma torch and signal instabilities. Consequently the sparking conditions were optimized by keeping the maxi- mum repetition rate and highest resistance available (400 s-l and 2.2Q respectively) and varying the voltage between 350 and 700 V.By increasing the condenser voltage it is possible to intro- duce more material into the ICP. This was shown by measuring the amounts of carbon from the released aerosol trapped on 47mm diameter glass microfibre filters with a pore size of 0.3 pm. Taking into consideration the intensities and precisions of the ICP measurements the voltage selected was 500V. At 350 and 400V the sensitivity and precision are poorer than those attained at 500 V; at 600 V the analytical signal becomes unstable as flickering occurs in the ICP as a result of an important increase in the amount of sample ablated which causes incomplete particle volatilization in the plasma. Table 3 shows the amount of graphite eroded at various voltages and the corresponding Zn emission intensities obtained from pellets of the sample ‘DUST 2’. The intensity uersus time curves Fig.1 show the stability of the emission for the elements considered from 10 s (pre-spark time) to at least 90 s. This is sufficient time to analyse the four elements considered in the same analytical programme. In an effort to minimize sparking times peak intensities were meas- ured employing a ‘three-point’ mode. In this mode the windows include three points with a distance between them of 0.0003 nm and the intensity value is a weighted average of these three points. Table 3 Eroded graphite and Zn emission intensities (arbitrary units) at various condenser voltages (pellets from sample ‘DUST 2’) Zn Vlv C*/mg min-’ Intensity x lo3 RSD 350 1.8 400 2.8 500 3.5 600 4.3 200 2.6 220 2.5 250 1.9 310 3.4 * Carbon determination was performed by infrared absorption after combustion in an induction furnace.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL.9 225 - ( d ) f I I I I /Dust 1 I I I I I I I Table 4 Analysis of EAF flue dusts by SA-ICP; n=6 I I I I I I I I I Zn (YO) I Dust 4 Sample DUST 1 DUST 2 DUST 3 DUST 4 DUST 5 ECRM 876-1 d X 0 Calibration ‘low’ 24.40 0.50 35.81 0.49 43.65 0.85 Calibration ‘high’ Certified Found 23.29 23.58 0.32 0.45 Pb (%) X n Calibration ‘low’ 5.00 0.11 11.52 0.25 9.43 0.10 Calibration ‘high’ Certified Found 7.82 7.54 0.23 0.20 Cd (Yo) Fe (YO) X fl 0.038 0.001 0.46 0.01 < 0.01 Calibration ‘high’ Calibration ‘low’ Certified Found 0.13 0.11 0.0 1 0.01 ~~~ ~ ~ X n Calibration ‘high’ 15.21 0.20 8.33 0.20 4.70 0.10 Calibration ‘low’ Certified Found 24.85 25.12 0.17 0.22 Fe Zn t I 1 I I I 15 30 45 60 75 Time/s Fig.1 Sparking curves (intensity uersus time) ICP Calibration After verifying the linearity of the emission intensities uersus concentrations of the elements studied within the intervals of contents considered only two of the DUST samples were used to obtain the calibration graphs ‘DUST 1’ and ‘DUST 5’ for Zn Pb and Fe determination and ‘DUST 1’ and ‘DUST 4’ for Cd analysis. The remaining DUST samples were analysed as unknowns. The scans around the selected analytical lines of the two calibration samples are shown in Fig.2. The relative intensities and the absence of spectral interferences can be appreciated. The aforementioned calibrations were used to analyse the other selected steelmaking dusts. The accuracy of the proposed method was tested by analysing one electrical furnace dust reference material Euronorm Certified Reference Material ECRM 876-1. The calculated mean values obtained from the three measurements of each of three identically prepared samples and the corresponding SDs are shown in Table 4. Precision tests were performed using pellets from ‘DUST 1’ and ‘DUST 5’. The procedure described above was used to prepare six pellets of each sample. The instrumental variability was tested by measuring one pellet of each sample ten times; RSD < 1.5% for both samples with the majority of the values lying between 0.7 and 1.2%.The total variance (influenced by the difference between samples) was assessed by measuring the six pellets of each sample ten times also; RSDs ranged from 0.75-1.9 and 0.80-2.0 for DUST 1 and DUST 5 respectively. The corresponding RSD values obtained by ICP-AES from nebulized solutions were < 0.5 and < 0.7%’ respectively for DUST 1 and DUST 5. The analytical performance expressed in terms of back- ground equivalent concentration (BEC) and detection limit (DL) are given in Table 5. The DLs were calculated as the concentration of a solution giving an absorbance equal to three times the SD of the blank. Because there was no EAF flue dust that did not contain the elements being studied and 165 104 A y 44 f 2 + > 101 .- 8 3 c .- - + .- v) C Q .w - 51 2 ( a ) Dust 5 I w 0.01 nm I I I I 213.859 51 30 10 2 2 0.3 5 0 2 43 131 20 259.9 40 214.438 Wave1 eng th/n rn Fig.2 Scans for sample calibration for the analytical lines of (a) Zn; (b) Pb; (c) Cd; and ( d ) Fe. The scale given in (a) also applies to (b) (c) and ( d ) Table 5 Analytical performance Element Zn Pb Cd Fe BEC (Yo) 1.5 1 .o 0.05 1.2 DL (‘/o) 0.075 0.080 0.003 0.080 containing 100% of the remaining elements off-peak measure- ments were estimated as on-peak values. These BEC and DL values are of the same order of magnitude as those provided by ICP-AES using nebulized solutions obtained from an alkaline fusion of the sample. Conclusion With the pellet preparation procedure described it is possible to directly analyse EAF flue dusts by using SA as the solid sampling system for ICP.The method clearly simplifies the226 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL. 9 conventional analytical ICP process by removing the dissolu- tion step. By mixing the sample with graphite in a proportion of 1+1 good conductivity is achieved and the addition of cellulose permits better ablation of the material and also improves the mechanical stability of the pellets. The amounts of carbon in the released aerosol were evaluated in order to test the stability and the efficacy of SA by analysing the aerosol trapped at the end of the 0.75m long plastic tube (i.d.=5 mm) that connects the spark cell with the bottom of the torch.Transport losses take place along this tube and in order to improve the precision it is necessary to clean the plastic tube periodically using an argon or air stream in particular when a sample with much lower contents than the previous one must be analysed. The BEC and DL values are comparable to those obtained after sample dissolution ( 1 g I-' of sample) moreover the analytical precision is sufficient for application in metal recovery processes. Consequently it can be concluded that for these materials the proposed analytical procedure is a viable alternative to ICP-AES following sample dissolution. 4 5 6 7 8 References Lopez F. A. Balcazar N. Formoso A. Medina F. and Jimenez R. Rev. Metal. (Madrid) 1990 26 386. Cuadra A. and Limpo J. L. Quim. Ind. (Madrid) 1992 38 27. Gomez Coedo A. Dorado Lopez M. T. Jiminez Seco J. L. and Gutierrez Cobo I. J . Anal. At. Spectrom. 1992 7 11. Aziz A. Broekaert J. A. C. Leis F. and Laqua IS. Spectrochim. Acta Part B 1984 39 1091. Ohls K. and Sommer D. Fresenius' Z . Anal. Chem. 1979,296,241. Scott R. H. Spectrochim. Acta Part B 1978 33 123. Steffan I. ICP In$ Newsl. 1991 16 10 564. Boumans P. W. J. M. Line Coincidence Tables f o r ICP-AES Pergamon Press Oxford 2nd edn 1984. Paper 3/045 78 A Received July 30 1993 Accepted September 14 1993