JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 53 Matrix Effects of Potassium Chloride and Phosphoric Acid in Argon Inductively Coupled Plasma Atomic Emission Spectrometry Bojan BudiC and Vida Hudnik National Institute of Chemistry Hajdrihova 79 POB 30 67 7 75 Ljubljana Slovenia Changes in analyte emission intensity in the presence of low concentrations of potassium chloride and phosphoric acid were studied in inductively coupled plasma atomic emission spectrometry. It has been shown that both matrices can cause significant depressant effects on the relative emission intensities of Sr Mg Mn Fe Cu and Zn under normal analytical operating conditions and that the lowering of intensity was more pronounced for KCI than for H,PO at the same analyte to matrix concentration ratio.The matrix effects can be corrected by increasing the power supplied to the plasma which was found to correlate with the excitation energy of the emission line. Keywords Inductively coupled plasma; atomic emission spectrometry; matrix interference; radiofrequency forward power; matrix effects correction Although chemical matrix effects are less severe in inductively coupled plasma atomic emission spectrometry (ICP-AES) than in other spectrometric techniques they can influence the intensity of the emission signal and thus introduce errors into the analytical procedure. It has been shown that such effects occur not only for samples with relatively high concentrations but even for samples with relatively low concentrations of concomitant elements or mineral acids.'-7 From the extensive studies in this field it has been found that generally in the normal analytical zone of the plasma matrix effects cause depression of analyte emission signals for both easily ionizable elements (EIEs) or non-EIEs as concomitant species8 To understand the mechanisms by which these matrix effects occur investigations concerning transport processesg-" and those in the plasma it~elfl"'~ were carried out.Since emission from the ICP source is heterogeneous and the processes that produce emitting species are still not completely understood no one model is able to predict exactly which conditions are optimum for analyte excitation. To compensate for salt/acid effects different methods can be used including matrix match- ing standard additions internal reference or the method based on the measurement of the intensity of the hydrogen (HP) emission line at 486.13 nm.14 However these methods are either time consuming or do not eliminate matrix effects completely because the analyte emission signal seems to be dependent both on changes in transport processes and changes in the plasma which are not a linear function of the matrix concentrat ion.' Matrices containing phosphate or alkaline elements are frequently encountered in the analysis of different minerals such as silica after dissolution with phosphoric acid or after fusion with alkaline salts or hydroxides. In the present study the effects of potassium chloride and phosphoric acid concen- tration on emission signals of elements with different ionization potentials and excitation energies (Sr Mg Mn Fe Cu Zn) were investigated.Power supplied to the plasma was varied from 1.0 to 1.6 kW while other experimental conditions were kept constant and the possibility of correcting for salt/acid effects by increasing forward power is presented. Experimental Apparatus An Applied Research Laboratories ARL 3520 OES sequential vacuum spectrometer equipped with an SAS 11 automation system for instrument control data acquisition and data manipulation was used. The instrumentation and operating conditions used which remained virtually constant with the exception of the forward power are listed in Table 1. Samples Table 1 Instrumental and operating conditions Spectrometer Grating R.f. generator Plasma torch Nebulizer Argon flow rate Observation height Solution uptake rate ~~ Monochromator with 1 m radius concave grating in Paschen-rounge mounting Linear dispersion 0.926 nm mm-' Quartz-controlled 27.12 MHz and automatic network.Operating power between 1.0 and 1.6 kW Fassel type Glass Meinhard type TR-30-3A Inner 1.0; intermediate 0.8; outer 12 15 mm above the load coil 2.2 ml min - (unforced) 1 min-' were nebulized without the use of a peristaltic pump and six replicate measurements were made. The spectral characteristics of the analyte lines studied are presented in Table 2. Reagents Analytical-reagent grade salts were used to prepare all solu- tions. The analyte concentration was lOpgml-' and the matrix concentrations were 0.0313 0.0625 0.125 0.25 and 0.5 moll-l. For temperature measurements iron was added at a concentration of 50 pg ml-' to KCl or &PO,.Table 2 Wavelengths excitation potentials (Ed and ionization poten- tials (Ei) of the spectral lines considered Spectral line A1 I1 Sr I Sr I1 Mg I1 Mn I Mn I1 Fe I Fe I Fe I1 Fe I1 c u I c u I1 Zn I Zn I1 Mg 1 Wavelength/nm 260.92 460.73 407.77 285.2 1 280.27 279.48 257.61 385.99 360.67 256.69 275.33 324.75 224.70 213.86 202.55 EJeV 4.62 2.69 3.04 4.35 4.42 4.44 4.81 3.21 6.13 5.91 7.77 3.82 8.24 5.80 6.12 EJeV 5.98 5.69 7.64 7.43 7.90 - - - - - - 7.72 9.39 - -54 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 Excitation Temperature and Electron Number Density Measurements In order to determine the excitation temperature a set of iron lines reported by Blades and Caughlin16 was used by applying the Boltzmann plot method.For the determination of the effect of matrix concentration and power supplied to the plasma on the excitation temperature the method described by Houk et which does not require transition probabilities was used T,dE*ij AE* - kT, 1nR T, = where Ii/Ij =ratio of line intensities at matrix concentration or forward power other than at the reference point (Ii/Ij)ref= ratio of line intensities at the reference point xef = reference value of the excitation temperature E*i(i) =excitation energy of state i( j) above the ground state and I =net line intensity of state i or j respectively. Electron number density was measured using the HB 486.13 nm line." A correction for instrumental and Doppler broadening was applied to the H/J profile before calculation of the electron number density.Results and Discussion Influence of Matrix Concentration on Analyte Emission Intensities Measured emission intensities in different matrices are depen- dent on the concentration of matrix element. It was found that matrix effects are strongly spatially dependent and therefore enhancement or depressant effects can occur at different spatial positions of the plasma.' Ramsey et d3.' observed a depression of emission intensities in the presence of matrices and found that the magnitude of the depression was dependent on both the matrix and the analyte emission line. A decrease in excitation temperature in the presence of a matrix was found to be relationed to the lowering of the analyte emission sensitivity.' Similar conclusions were reached by Yoshimura et aL2 for the presence of mineral acids.Olesik and Williamsen7 found by measuring both fluorescence and emission intensities in the presence of different matrices that the number of ground-state ions decreased in the presence of matrix elements but more analyte ions were excited. Therefore matrix effects can counteract each other. The influence of KCl and H3P04 concentration on the relative intensities of Sr Mg Mn Cu and Zn atom and ion emission lines was measured. The analyte concentration was constant while the matrix concentration was varied between 0 and 0.5 moll-'. For these measurements forward power was kept constant at 1.3 kW. The matrix effect M is defined as the percentage difference in the net line signals between solu- tions of the analyte with and without matrix.For the matrices studied a decreasing effect is observed (Figs. 1 and 2). At the same molar ratio of KC1 and H3P04 to the analyte the matrix effect in the presence of KC1 [Fig. l(a) and (b)] was greater than when H3P04 [Fig. 2(a) and (b)] was added. Also the difference in matrix effect between analytes studied was larger for KC1 than for H3PO4 under the same experimental con- ditions. From Figs. 1 and 2 it is also obvious that the depressing effect on signals for atomic lines of Mn Mg Cu and Zn is only slightly greater than that on the ionic lines of the same elements. The matrix effect (M,%) in the presence of 0.5 mol I-' H,P04 is in the range 7-10% while at the same concentration of KCl this effect is more pronounced (12-15%) with the exception of Sr atomic and ionic emission lines [Fig.l(a) and (b)]. Yoshimura et aL2 obtained similar depressant effects when 0 -5 - 10 - 15 - 20 I I I I I -25 I I I I I I 0 0.10 0.20 0.30 0.40 0.50 [KCll/mol I-' Fig. 1 Matrix effect (M%) on (a) atomic and (b) ionic emission lines as a function of KCl for A Mn; B Cu; C Mg; D Zn; and E Sr. Concentration M(%) is defined as [(Zm-1,,)Zn] x 100 where I and I, refer to the net line emission intensity in water solutions and in the presence of matrix respectively Zn Ca and Mg emission intensities were measured in the presence of HNO and H2S04. To find a correlation between a change in the physical state of the plasma and observed depressant effects in the presence of matrix the electron number density ne and excitation temperature T,, were measured.Electron number density was established to be 2.15 x lo1' cm-3 under the experimental conditions applied. No difference was observed when either KCl or H3P04 were added at concentrations of up to 0.5 mol I-'. Because of the lack of local thermodynamic equi- librium in the ICP,I9 various species and states can show a considerable difference in T,, and emission lines having different excitation potentials could be considered to have their own excitation temperature. However for analyte excitation studies in the ICP this parameter is often used." In the present experiments regardless of the addition of KCl or H3P04 T, was found to be fairly constant ie. 6490 K. The uncertainty in the temperature measurement was estimated to be about +300 K.In addition the method described by Houk et which does not require transition probabilities was used. For a comparison of T, for atomic and ionic spectral lines two sets of iron line pairs were chosen Fe I 360.67/Fe I 385.99 nm and Fe I1 275.33/Fe I1 256.69 nm. As a reference value an excitation temperature of 6490 K for the blank solution was chosen. The changes in excitation temperature determined in this way did not exceed a value of & 90 K. These findings disagree with the observations of Yoshimura et a!.,' which could be owing to somewhat different operating conditions. Since no change in n and T, in the plasma was observed when KC1 or H3P04 were added it seems more probable that in this case transportation and vaporization effects predomi- nate.For a detailed study of this phenomenon it will beJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 55 0 -5 - 10 2 5500' % 7000 r I I 1 I !!? 1 -I5 t -20 t -25 I I I 1 I 0 0.10 0.20 0.30 0.40 0.50 IH,PO,l/mol I - ' Fig.2 Matrix effect (M%) on (a) atomic and (b) ionic lines as a function of H,PO concentration for A Mn; B Cu; C Mg; D Zn; and E Sr. M(%) as in Fig. 1 necessary to measure the spatial distributions of the emission and fluorescence intensity of analyte lines n and T,,,. Effect of Forward Power on Relative Emission Intensities and Excitation Conditions Experimental conditions such as carrier gas flow rate or power supplied to the plasma can significantly effect the plasma conditions and thus the relative analyte emission intensit- ies.16*20 Forward power is one of the experimental parameters that can be used for ICP optimization and the effects of forward power on emission intensities and signal-to-back- ground ratios have been widely described in the literature.21 In order to assess the possibility of correcting matrix effects by changing the power supplied to the plasma in the presence of KCI and H3P04 forward power was varied between 1.0 and 1.6 kW in increments of 0.15 kW while other operating conditions and matrix concentration remained constant.The elements Al Sr Fe Cu Mg Mn and Zn were considered. 3.5 3.0 2.5 2.0 1.5 1 .o 0.5 I I I 0.95 1.10 1.25 1.40 1.55 Forward power/kW Fig. 3 Influence of increasing forward power on the electron number density (n,) Fig.4 Relationship between forward power and T, calculated by using a reference temperature of 6490K for (a) atomic and (b) ionic Fe emission lines Electron number density and T, were measured in the same increments as analyte emission intensities and are pre- sented in Figs. 3 and 4. Since the main objective was to find a relationship between changes in T, on one hand and forward power on the other the method on the relative basis without transition probabilities mentioned above was applied. From Fig. 3 it is evident that n increases almost linearly when forward power increases and a similar observation was made for T, (Fig. 4). As a reference point a temperature value of 6490K determined by the Boltzmann plot method with a forward power of 1.3 kW was chosen.The difference between T, calculated for Fe I and Fe I1 lines (intensities were measured for six replicates) is evident indicating the lack of local thermodynamic equilibrium conditions and the tendency for T, for atomic lines to be more affected by changes in forward power. The effect of the forward power on relative emission intensity for Sr Mg and Zn is illustrated in Figs. 5 and 6. From Fig. 5 it can be seen that relative atomic emission intensities (each value was normalized to that of the highest value) behave quite differently depending mainly on the excitation energy (Ex) of the analyte emission line. It is obvious that at lower values of Ex relative intensity decreases [Sr 460.73 nm line with Ex=2.69 eV Fig. 5(a)] a similar effect was observed at the Fe 385.99 nm line (Ex= 3.21 eV) whereas emission intensity for atomic lines of Mg Mn and Zn having Ex values between 4.35 and 4.44 eV increased when forward power was increased from 1.0 to 1.6 kW.Plots for Mg and Zn are presented in Fig. 5(b) and (c). The relative emission intensity of the Cu atomic line at 324.75 nm (Ex = 3.82 eV) remains quite unaffec- ted when forward power is increased from 1.0 to 1.6 kW. It is most likely that both processes i.e. a decrease in the analyte atom population and an increase in the analyte excitation contribute to the changes in relative emission intensities when forward power is increased. For analyte ion emission lines (illustrated in Fig. 6 for Sr Mg and Zn) increasing the forward power causes a significant increase in signal for all the lines considered with the exception of the Sr I1 407.77 nm line [Fig 6(a)].In the latter case similar behaviour for Sr as for Ba ionic emission can be assumed as in the investigation of Olesik22 who found that the increase in the Ba ionic emission signal was due to increased excitation56 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 0.6 0.4 0.2 A ( b ) 1.1 1.0 - 0' 1 I I I 1.1 1 1 .o 0.9 0.8 0.7 0.6 0.5 0.4 1.00 1.15 1.30 1.45 1.60 Forward power/kW Fig. 5 Relative atomic emission intensities as a function of the forward power A without matrix; and in the presence of B KCl and C H,PO for (a) Sr (b) Mg and (c) Zn rather than to a change in the Ba ion number density which decreased when forward power was increased from 0.5 to 1.0 kW.From Figs. 5 and 6 it is obvious that in the presence of KC1 or H3P04 the depressant effect occurs to a similar extent for both atomic and ionic emission lines between 1.0 and 1.6 kW forward power and that lowering of relative emission intensit- ies is slightly greater in the presence of a KCl than in a H3P04 matrix. For correction of the matrix effects for ionic emission lines an increase of the forward power is needed. In Fig. 7 the correlation between Etot (which is the sum of excitation and ionization potential) and AP the power necessary to correct for the matrix effect of 0.5mo11-1 H3P04 using an initial forward power of 1.1 kW is presented. From Fig. 7 it is evident that in the range of E, between 12 and 16 eV (where most of the analytically useful lines are found) an increase in AP to about 25 W to correct for the H3P04 matrix effect is necessary.At lower values of E, much higher forward power for the correction of matrix effects is needed however for emission lines (such as the Sr 407.77 nm ionic line) which are subject to a large matrix effect (M>20%) and have low E,, matrix effects cannot be eliminated completely in this way. It is also shown (Fig. 7) that the behaviour of the Cu ionic line (E,,,= 15.96eV) differs from other analytes. It is interesting to note U 1.00 1.15 1.30 1.45 1.60 Forward power/kW Fig. 6 Relative ionic emission intensities as a function of the forward power A without matrix; and in the presence of B KC1; and C H,PO for (a) Sr (b) Mg and (c) Zn 17 cu Fe 14 < 1 3 - Q2 12 - 11 10 - - 9 - 8 ' I 1 I I 0 50.00 100.00 150.00 200.00 250.00 AP/W Fig.7 Relationship between E, of Sr 430.55; Sr 407.77; A1 260.92; Fe 256.69; Cu 224.70; Mn 257.61; Mg 280.27; and Zn 202.55 nm ionic emission lines and forward power increase needed to compensate for the H,PO (0.5 moll-') acid effect. A value of 1.1 kW was chosen for the initial forward powerJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 57 Table3 Results of the determination of Zn Fe Mn Mg and Sr in the presence of H,PO (0.4 mol I-'). Results given in pg ml-' Element Wavelength/nm A* Bt c D4 Zn 202.55 130 120 130 131 Fe 256.69 38 35 38 38 Mn 257.61 115 110 116 115 Mg 280.27 155 147 157 155 Sr 407.77 265 232 234 262 Sr 430.55 240 247 267 - * A In solution. 7 B Without matrix effect correction.1 C Matrix effect correction by forward power compensation. 0 D Matrix effect correction by sample matching. that the intensity of the same Cu emission line deviated from some other elements when the relationship between relative emission intensity for ionic lines was plotted against excitation potential for different concentrations of mineral acids.2 It is possible since the Cu 224.7 nm line lies near the ionization limit of Ar (16 eV) that this line is not excited by the same process as other lines. The relationship between relative atomic emission lines and forward power (Fig. 5 ) shows that for the correction of matrix effects for these spectral lines much larger changes (e.g. a decrease for the Sr 460.73 nm line) of power supplied to the plasma are needed than for the ionic lines.Applications The procedure described above was also used in the practical analysis of H,P04 samples containing Mg Mn Fe and Zn at lower concentration. In Table 3 a comparison of the results of the analysis are presented when no power correction was applied (using calibration curves for water solutions) when there was an increase in AP of 25 W for all the elements analysed and when calibration curves obtained with sample matching were used. The forward power was set up initially at 1.1 kW. The analyte and synthetic solutions contained 0.4 moll-' H3P04. A relative standard deviation of less than 3% (n = 6) was obtained. Good agreement between results obtained with power correction and matrix matching except for the Sr 407.77 nm line was achieved.Power correction is probably preferable because it is easy controlled and the preparation of synthetic solutions for matrix matching is time consuming. Thus for practical analytical problems when ana- lysing elements in the presence of matrix elements at higher concentrations the ionic emission lines with higher values of Etot could be corrected more easily for matrix effects by changing the power supplied to the plasma than could ionic emission lines having low values of Etot or atomic emission lines. However correction could be effectively performed for the emission lines whose intensities are influenced to a similar extent by the matrix. Conclusions Both KC1 and H3P04 matrices have a significant depressant effect on analyte emission intensities in ICP-AES.The matrix effects were enhanced when the concentration of KC1 and H3PO4 was increased up to 0.5 moll- Atomic lines appear to be more subject to changes in matrix concentration and the depressant effect was slightly greater when KC1 was added to the analytes in the same molar ratio as H3P04. No change in n and T, in the ICP was observed in the presence of matrix. Under the operating conditions applied the magnitude of the depressions of the analyte emission signals does not appear to correlate with the ionization potential or the excitation energy. It seems more probable that concomitant elements have a complex influence on analyte emission intensities which is a function of a greater number of parameters. Radially resolved emission and fluorescence data should be useful for the detailed study of matrix effects.By changing the power supplied to the plasma from 1.0 to 1.6 kW it was shown that the matrix effect does not change significantly. A correlation between the total excitation energy in a range of about 12-16eV for ionic emission lines and forward power required to correct for matrix effects was found. In this range of excitation energies for ionic emission lines a relatively small increase in forward power (about 20-30 W) was found to be needed to achieve similar performance to that found without a KCl or H3PO4 matrix. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 References Shen X.-e. and Chen Q.4 Spectrochim. Acta Part B 1983 38 115. Yoshimura E. Suzuki H. Yamazaki S.and Toda S. Analyst 1990 115 167. Ramsey M. H. Thompson M. and Walton S. J. J. Anal. At. Spectrom. 1987 2 33. Thompson M. and Ramsey M. H. Analyst 1985 110 1413. Ramsey M. and Thompson M. J. Anal. At. Spectrom. 1986 1 185. Ramsey M. H. and Thompson M. J . Anal. At. Spectrom. 1987 2 497. Olesik J. W. and Williamsen E. J. Appl. Spectrosc. 1989,43 1223. Sun D.-h. Zhang Z.-x. Qian H.-w. and Cai M.-x. Spectrochim. Acta Part B 1988 43 391. Farino J. Miller J. R. Smith D. D. and Browner R. F. Anal. Chem. 1987,59 2303. Bates L. C. and Olesik J. W. J. Anal. At. Spectrom. 1990,5 239. de Loos-Vollebregt M. T. C. Peng R. and Tiggelman J. J. J. Anal. At. Spectrom. 1991 6 165. Blades M. W. and Horlick G. Spectrochim. Acta Part B 1981 36 881. Rybarczyk J. P. Yester C. P. Yates D. A. and Koirtyohann S. R. Anal. Chem. 1982 54 2162. Botto R. I. Spectrochim. Acta Part B 1985 40 397. Olesik J. W. and Den S.-J. Spectrochim. Acta Part B 1990 45 731. Blades M. W. and Caughlin B. L. Spectrochim. Acta Part B 1985 40 579. Houk R. S. Shoer J. K. and Crain J. K. Spectrochim. Acta Part B 1987 42 841. Griem H. R. Spectral Line Broadening by Plasmas Academic Press New York 1974. Caughlin B. L. and Blades M. W. Spectrochim. Acta Part B 1984,39 1583. Olesik J. W. and Bradley K. R. Spectrochim. Acta Part B 1987 42 377. Boumans P. W. J. M. and de Boer F. J. Spectrochim. Acta Part B 1977 32 366. Olesik J. W. Spectrochim. Acta Part B 1990 45 975. Paper 3/04528E Received July 28 1993 Accepted September 21 1993