JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 397 Direct Analysis of Slags by Inductively Coupled Plasma Atomic Emission Spectrometry Using Slurry Sample Introduction Techniques Maria Luisa Fernandez Sanchez Ben Fairman and Alfred0 Sanz-Medel* Department of Physical and Analytical Chemistry Faculty of Chemistry University of Oviedo 33006 Oviedo Spain A practical method is described for the direct determination of Si Ca Mg Al Fe Mn Ti Na and K in slag samples by the introduction of sample suspensions into the inductively coupled plasma atomic emission spectrometer. The effect of the particle size distribution on the atomization efficiency of these elements when slurries of the slag samples were aspirated into the plasma has been studied. The slurry samples were prepared by dispersing 0.1 g of ground slag in 100 ml of 0.35% ammonia solution; calibration was achieved by the use of a reference slag prepared in the same manner.The results obtained for the direct analysis of slag slurries using such a procedure were in good agreement with those obtained by classical dissolution of the sample. The reproducibility of the results with slurry nebulization (concentration range 1.7-5.5'/0) were poorer than those observed with aqueous solutions (concentration range 0.5-4.8%) but were acceptable for slag industrial control. Keywords Elemental slag analysis; slurry atomization; inductively coupled plasma atomic emission spec- trometry An important limitation of the otherwise popular induc- tively coupled plasma (ICP) as a source for atomic emission spectrometry (AES) is that a conventional ICP technique requires the sample to be in solution.Because of this the sample preparation time in many solid sample analyses often exceeds the analysis time by unacceptable lengths. This means that the potential analytical speed of ICP-AES cannot be fully realized unless alternatives to the classical time-consuming and sometimes hazardous dissolution methodologies can be developed. Hence the considerable interest in recent years in investigations into the possibili- ties of direct analysis of solid samples. Perhaps one of the most interesting approaches to solid sampling in atomic spectrometry is slurry injection,' that is the introduction of aqueous (or organic) suspensions of fine powders or finely ground materials using a high solids nebulizer.When suspensions are nebulized into the ICP the atomization efficiency is mainly dependent on the sample transport efficiency which in turn is a function of the particle size and sample matrix. The application of slurries as a sample introduction technique to the ICP has been successfully reported by several workers for the determina- tion of a variety of elements in soil^,^^^ geological material~,~J~ clay," kaolin12J3 and refractory Watson and MoorelS used an ion-exchange resin to concentrate noble metals from a dissolved sample and the loaded resin was then mixed with water to form a slurry that was introduced into the ICP. Sugimae and Mizoguchi16 intro- duced suspensions in an organic medium into the ICP. The use of reference materials in slurry form as a calibration procedure for slurry atomization has also been rep~rted.~J' Watson17 used an internal standard of Li4B407 plus 1% Sc203 and used reference ores and slags in order to obtain calibration functions. Halicz and Brenner9 employed a similar technique using Sc as the internal standard and calibration was performed by using the ratios of the analyte to scandium internal standard reference signals as a function of the concentration of the reference material slurry.The application of a flow injection manifold for the introduction of slurries into an ICP has also been reported recently. * * To whom correspondence should be addressed. Blast furnace slag is a by-product obtained in the manu- facture of iron. The bulk of the slag consists principally of silica alumina and lime with minor amounts of magne- sium oxide iron oxide and alkali oxides.This by-product is used in road building and fertilizers but the cement industry is by far the largest user of the blast furnace slag. The slag can be used in the production of cement in two basic ways (i) as a raw material for cement manufacture; or (ii) as a cementitius material combined with portland cement. The activity of the slag used in cement or concrete is determined by its composition and by the rate at which the molten material is cooled when it comes from the furnace. The slag can be added to the cement in order to reduce the cost or to modify the performance (ie. to produce concrete that is relatively resistant to certain types of chemical attack such as those from sulphate or sea- water) .In this paper a method is described using compromise conditions for the routine multi-element determination of Si Ca Mg Al Fe Mn Ti Na and K in slag samples by slurry atomization ICP-AES. The optimum operating para- meters in the ICP and the influence of particle size on the analytical signal were investigated. The effect of different dispersants on the stability of the slag suspension and the influence of the solids content on the analytical signal were also studied. Experimental Apparatus A Perkin-Elmer Model 5000 ICP spectrometer was used. The source and spectrometer characteristics are described in Table 1. A high-solids polytetrafluoroethylene Babing- ton-type nebulizer19 with a double-pass spray chamber was used for sample introduction. The slurry sample was delivered to the ICP nebulizer through a 0.8 mm i.d.plastic tube using a peristaltic pump. The ICP operating para- meters selected for the slag analysis are also listed in Table 1. Particle size analysis was carried out using a photosedi- mentometer (Lumosed Retsch). Sample Dissolution Procedure Slag samples previously hand-ground and dried were dissolved by weighing 0.1 g of the sample into a Teflon398 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 2.0 1.5 1 .o 0.5 Table 1 Instrumentation and plasma operating conditions Spectrometer- Sequential scanning (0.03 nm bandpass Holographic ultraviolet grating Fassel-fused silica torch 2 kW 27.12 MHz High-solids Babington-type nebulizer Coolant flow rate Auxiliary flow rate Injector flow rate Nebulizer pressure Observation height Plasma power ICP source- Plasma operating conditions- 16 1 min-' 0.8 1 min-I 0.6 1 min-I 193 kPa 16 mm 1.25 kW I -12 I / B I - 0 ' I 1 1 ' 0 0.9 1.0 1.1 1.2 1.3 1.4 1.5 R.f. power/kW bomb and adding 5 ml of water 1 ml of HF and 3 ml of aqua regia (3 + 1 HCl + HN03). The bomb was heated in an oven at 140 "C for 45 min and then left to cool. Ten ml of a saturated solution of boric acid were added and the bomb was re-sealed and heated to 120 "C for another 20 min. After cooling the solution was transferred into a 250 ml calibrated flask and diluted to volume with distilled de- ionized water. Slurry Sample Preparation The original slag samples were dry-ground in a ball-mill (Pulverisette Model 6 Fritsch) to a size that allowed them to pass through a 37 ,um sieve.The particle size of the sieved sample was further reduced down to (10 pm by wet- grinding* as follows 0.1 g of the sample was added to 30 ml polyethylene bottles containing 1 g of 3 mm zirconium oxide beads and 0.5 ml of water and ground by shaking for 3 h in a laboratory flask shaker. The resultant slurry was transferred into a 100 ml calibrated flask and diluted to the mark with 0.35% v/v ammonia solution. The slurry was then placed in an ultrasonic bath for 15 min. A magnetic stirrer was used to agitate the sample while it was pumped into the ICP in order to maintain a uniform slurry of the slag. - 6 8 10 12 14 16 18 20 22- Observation height/mm 25 20 - 15 - 10 1.0 - 0.5 - - 5 B o c o s 4 - 0 " ' a ' ' ' a * r Q 0 20 22 24 26 28 30 32 34 36 38 40 42 Gas pressure/psi Results and Discussion Plasma Operating Conditions for Slag Slurries The analytical emission lines selected (see Table 2) for the determination of Si Ca Al Mg Fe Mn Ti Na and K in slags by ICP-AES were selected from the literature20 and from previous preliminary work with slag s1urries.2*~22 The influence of the more critical operating parameters of the ICP on each analytical line when using slurry samples prepared as above were studied by investigating the signal- to-background (S/B) ratio observed as a function of the viewing height above the load coil; the plasma power Fig.1 Effect on A the emission intensity of Mn B the background emission intensity and C the signal to background ratio of a ground slag slurry as a function of (a) r.f.power; (b) observation height; and (c) nebulizer gas pressure applied; and the nebulizer gas flow. This last parameter should be the most critical as the injector gas flow not only affects the efficiency of the nebulization but also determines the sample residence time in the plasma channel. The results obtained showed that generally the signal and the background increased with increasing power and decreased with increased viewing height and nebulizer gas flow. As an example the results for Mn are illustrated in Fig. l(a)-(c) which also show the effects of these parameters on the S/B ratios. The best SIB ratios for Si Ca Mg and Mn were observed at low observation heights low gas flows (longer residence time in the plasma) and high radiofrequency (r.f.) power.In contrast the best SIB ratios for Na and K were observed for high observation heights and low power whereas the optimum results for Al Fe and Ti were obtained using an intermediate observation height and r.f. power and high gas flows. The optimum operating parameters observed for Table 2 Optimum experimental conditions for each element of interest Nebulizer Wavelength/ Observation gas flow/ Plasma Element nm height/mm ml min-' power/kW Si 288.15 10 0.64 1.30 Ca 3 17.93 12 0.5 1 1.40 279.55 9 0.47 1.40 A1 396.15 12 0.64 1.10 Mg Fe 259.94 14 0.64 1 .oo Mn 257.61 10 0.5 1 1.30 Ti 334.94 14 0.69 1.10 Na 589.59 26 0.64 0.90 K 766.49 26 0.60 1 .ooJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL.6 399 1.0 0.8 0.6 0.4 0.2 the excitation of a slurry of a sample in an ICP are given in Table 2 for each element studied. A simplex technique was used to determine the optimum conditions for each element.23*24 For multi-elemental analysis a set of com- promise conditions were chosen and the values finally used are given under plasma operating conditions in Table 1. - - - - - A n- Particle Size The most critical parameter in slurry atomization is probably the particle size distribution of the ~ample.~*'*~J~ In the particle size investigations 15 g of sample were ground in the ball-mill for increasing periods of time and different size fractions were isolated using a series of sieves (by manual agitation of the sieves). It was observed as expected that suspensions prepared from different particle size ranges (180-63,63-37 and (37 pm) produced drama- tically different emission intensities for all of the elements studied.This effect for three important elements viz. Si Ca and A1 is shown in Fig. 2. Only when the particle size of the slag was reduced to <37 pm could analytically useful signals be observed in the system. It was verified that 2.5-3 h of grinding in the ball-mill ensured that all the sample passed through a 37 pm sieve. Williams et aZ.* described an effective method for the rapid reduction of the particle size by shaking the coarse slurry samples with Zr02 beads in polyethylene bottles. The use of such a technique was tested by shaking the sieved slag sample for 0.5 and 3 h with a laboratory flask shaker.The measured particle size distribution of these two fractions is shown in Fig. 3. The figure shows how effective the use of Zr02 beads during grinding can be in reducing the mean particle size of the slag samples. In fact the mean particle size diameters were 4 and 2 pm after 0.5 and 3 h of grinding with ZrOl beads respectively. Particle size distri- bution analysis showed that more than 90% of the particles were below 22 pm after 0.5 h and below 5 pm after 3 h of grinding with Zr02 beads. After the particle size reduction studies the relation between the particle size distribution in the slurry and the analyte emission intensity observed in the ICP was investi- gated slag slurries (0.1 % m/v) were prepared from each of the three sieved fractions.They were as before 180-63 63-37 and t 3 7 pm and were subsequently analysed for Ca Fe Ti and Na by ICP-AES. The results obtained (as illustrated for Ca Fe Ti and Na in Fig. 4) show that the Ca A1 1.2 C Particle size fraction Fig. 2 Effect of different particle size fractions on the emission intensity of Si Ca and Al A 180-63 ,urn; B 63-37 pm; and C (37 pm 100 5-10 10-15 15-20 20-25 2g30 Particle size/pm Fig. 3 Effect of grinding time upon the particle size distribution of slag slumes using the wet Zr02 bead method unshaded area 0.5 h; and shaded area 3 h Ti 1.8 Ca Fe 1.6 3 1.4 s 1.2 CI .- C $ g J 1.0 .- 3 0.8 . 0.6 0 2 0.4 E w 0.2 Q) c. .- .- n v Particle size fraction Fig. 4 Effect of slag particle size upon the emission intensity of Ca Fe Ti and Na A before sieving (1 80-63 ,urn); B after sieving (63-37 pm); C after sieving and wet grinding with ZrO beads (<37 ,urn) simple sieving of the sample can double the signal (e.g.the emissions from fractions A and B for Ca). All of the analyte signals were increased further by grinding with Zr02 beads but the magnitude of the increase was not equal for each of the different elements under study for instance the Ti and Na emissions show a higher percentage increase with wet- grinding than the Fe emission signal (compare B and C fractions in the oberved signals in Fig. 4). The relationship observed between the Zr02 grinding time and the analyte intensities for all of the elements under study is illustrated in Fig. 5. The signal increases with grinding time (as smaller particles are produced with extended grinding which in turn affects the solid transport 2 f 2 1.0 2 1.5 .- - .- fn Q) U .- 0.5 0 fn .- .- E n - - w Time Fig.5 Effect of grinding time using the wet ZrO bead method upon analyte emission intensity from slag slurries unshaded area 0.5 h; and shaded area 3 h400 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1991 VOL. 6 efficiency). However the emission intensity did not in- crease to the same extent for the different elements. This indicates that the slag is a multicomponent material and that the different degrees of hardness of each of the components has resulted in different particle size distribu- tions for each mineral. Selection of Dispersant The use of a dispersant helps to maintain the small particles of a slurry sample in suspension.In order to achieve a stable suspension of ground slag the dispersants tested were Triton X- 100 sodium pyrophosphate (Na4P207) and am- monia solution. The effect of the dispersants was investi- gated using slurries (0.1 Oh d v prepared with each of the dispersants) which were aspirated into the ICP. The stability of the observed analyte signals was determined over a period of 6 min. The results showed that in suspensions prepared in 0.02 0.1 1 or 2% m/v Triton X- 100 a depression of the signal was evident and the stability of the signals was poor. A slurry prepared in 0.1Oh m/v Na4P20 showed good stability when monitoring the emis- sion lines of Ca and Mg. The ammonia solution (0.35% v/v) as found by Ebdon and Collier for kaolin slurries,13 proved to be the most effective dispersant for the fine slag particles as it produced the most stable emission signals for all of the elements under examination.Effect of Solids Content in the Suspension In order to evaluate the influence of the percentage of solid present in the suspension on the ICP emission signal intensity slurry calibration graphs were obtained by vary- ing the concentration of solids in the slag slurry (from 0.05 to 0.5Oh d v ) . The results obtained are shown in Fig. 6. The intensity increases linearly with the amount of slag in the slurry in all instances except for Mg where the observed response was not linear above 0.2% d v . This can be attributed to self-absorption phenomena which were also detected in aqueous solutions of Mg2+ with concentrations similar to those of the slag slurry (at 0.2% m/v).Analytical Application When the slag slurry was analysed by ICP using aqueous standards and using the slurry preparation as detailed (k 3 h grinding with Zr02 beads) the recoveries (as calculated from the slurry to aqueous solution signal ratio) ranged from 60 to 8O% depending on the analyte. It was thought 0.1 0.2 0.3 0.4 0.5 C Concentration of solids (% m/v) Fig. 6 The effect of solid concentration upon the emission intensity of A; Si Ca Al Fe Ti Na and K; and B Mg in a slag slurry Table 3 Comparison of elemental content of a slag as determined by the slurry technique and by acid digestion. Each value is the mean of 3 determinations with a 1 s integration time Acid digestion Slurry atomization Oxide Content (%) RSD (Oh) Content (%) RSD (%) Si02 CaO MgO Fe203 A1203 Mn02 TiOz Na20 KZ0 40.99 8.66 4.10 3.16 0.43 0.91 2.52 0.22 0.56 0.5 0.6 0.8 0.9 1.3 1.5 1.2 1.3 4.8 41.15 8.54 4.40 3.14 0.38 1.02 2.54 0.16 0.50 1.7 2.5 1.2 2.1 5.2 1.8 1.4 3.7 5.3 that these low recoveries might be due to transport losses of the slag slurry aerosol in the commercial spray chamber and possibly incomplete atomization andor excitation in the As shown previously by other w o r k e r ~ ~ * ~ ~ ~ J ~ poor reco- veries are related to particle size.A particle size distribution of < 5 pm (see Figs. 3 and 5 ) should overcome this problem but at the expense of unacceptably long grinding times. Therefore a more practical approach was finally selected the sieved sample was ground for 30 min with Zr02 beads (a more reasonable period than 3 h for routine determina- tions); and the loss in sensitivity was compensated for by using a reference slag (in slurry form) in order to calibrate the ICP. As described above this method of calibration for slurry samples has been used by several workers notably Sugimae and Mizoguchi for airborne particles,16 and Halicz and Brenner for various geological material^.^ These work- ers found that close matching of sample and standard suspensions was vital for correct calibration.Although Sugimae and Mizoguchi concluded that differences in the response to the iron present in air particles between samples and reference standards was mainly due to the particle size composition Halicz and Brenner in addition stressed that the textural mineralogical and chemical compositions must also be similar.For the present investi- gation this was achieved by the use of the slag reference material Slagg SI-Ss (Institutet fur Metallforskning Ger- many) which underwent the same treatment as the samples so as to satisfy the conditions mentioned above. The total concentration of Si Ca Al Mg Fe Ti Mn Na and K obtained by the direct analysis of slag suspension by ICP are given in Table 3. These results are comparable in terms of accuracy and precision with those obtained by the acid digestion method described above. Good agreement is also observed between the results obtained by both methods (direct analysis by slurry atomization and the acid digestion of the sample). However the precision is worse when the slag slurry is introduced into the ICP.For routine analysis the poorer precisions observed with the slurry technique are more than compensated for by simplicity and speed of operation which reduces the analysis time the risks of contamination and loss and the use of hazardous chemi- cals when compared with classical acid dissolution. ICp.7.12,14 Conclusions Direct solids analysis for the total concentrations of Si Ca Mg Al Fe Mn Ti Na and K in slags by ICP-AES is possible by nebulizing suspensions of each. As previously s ~ o w ~ ~ * ~ ~ ~ J ~ the most critical factor affect- ing the ICP signal has proved to be the particle size of the ground slag generally the ICP analytical signals increase with a reduction in the particle size distributions of theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1991 VOL.6 40 1 slurries. The longer the grinding time the closer the emission signals of slurries come to those of aqueous solutions containing an equivalent concentration of the analyte being studied. By using the proposed procedure the ratios between the two emission signals as stated earlier were always below 1.0 (normally between 0.6 and 0.8 depending on the analyte). This variability demonstrates the complexity of the multicomponent nature of the slag material and points to preferential particle size reduction depending on the hardness of each mineral component when the longer wet-grinding times are not used. On the other hand when the solid content in the slurry was increased from 0.05 to 0.5% m/v no viscosity effects were observed.Therefore the reduced recoveries observed are perhaps due to less effective transport of the slag slurry from the nebulizer to the ICP as compared with aqueous solutions in addition to the possibility of incomplete volatilization and/or atomization (the extent of which would be mineral or element dependent).I4 As found by other workersgJ6 these phenomena necessitate the use of slag reference materials in the form of slurries instead of aqueous solutions for the calibration of the instrument. The results obtained show that the slurry technique holds promise for the direct and accurate analysis of solid slags by ICP-AES and with the calibration limitation detailed above offers a straightforward non-hazardous technique for monitoring both major and minor elements in routine slag analysis.References Fuller C. W. Hutton R. C. and Preston B. Analyst 1981 106 913. Wilkinson J. R. Ebdon L. and Jackson K. W. Anal. Proc. 1982 19 305. Ebdon L. and Wilkinson J. R. J. Anal. A t . Spectrorn. 1987 2 39. Ebdon L. and Wilkinson J. R. J. Anal. A t . Spectrom. 1987 2 325. 5 McCurdy D. L. and Fry R. C. Anal. Chem. 1986,58,3126. 6 Ebdon L. Foulkes M. E. Parry H. G. M. and Tye C. T. J. Anal. At. Spectrom. 1988 3 753. 7 Dick W. A. Page J. R. and Jewell K. E. Soil Sci. 1985,139 21 1. 8 Williams J. G. Gray A. L. Norman P. and Ebdon L. J. Anal. At. Spectrom. 1987 2 469. 9 Halicz L. and Brenner I. B. Spectrochim. Acta Part B 1987 42 207. 10 Verbeek A. A. and Brenner I. B. J. Anal. At. Spectrom. 1989 4 23. 11 Speirs G. A. Dudas M. J. and Hodgins I. W. Clays Clay Miner. 1983 31 397. 12 Ebdon L. and Collier A. R. Spectrochim. Acta Part B 1988 43 355. 13 Ebdon L. and Collier A. R. J. Anal. At. Spectrom. 1988 3 557. 14 Ebdon L. Foulkes M. E. and Hill S. J. Anal. At. Spectrom. 1990 5 67. 15 Watson A. E. and Moore G. L. S. A). J. Chem. 1984,37,81. 16 Sugirnae A. and Mizoguchi T. Anal. Chim. Acta 1982,144 205. 17 Watson A. E. S. Afr. J. Chem. 1986 39 147. 18 Ambrose A. J. Ebdon L. Foulkes M. E. and Jones P. J. Anal. At. Spectrom. 1989 4 219. 19 Ebdon L. and Cave M. R. Analyst 1982 107 172. 20 Boumans P. W. J. M. Line Coincidence Tables for Inductively Coupled Plasma Emission Spectrometry Pergamon Press Oxford 1980. 21 Fernindez Sdnchez M. L. Palacio Suarez J. Ferndndez Molina E. and Sanz-Medel A. J. Anal. At. Spectrom. 1987,2 491. 22 Fernindez SAnchez M. L. PhD Thesis University of Oviedo 1989. 23 Garcia Alonso J. I. PhD Thesis University of Oviedo 1985. 24 Ebdon L. Cave M. R. and Mowthorpe D. J. Anal. Chem. 1980 115 179. Paper 0/0262 1 B Received June 12th 1990 Accepted March I9th I991