JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 465 Investigations on the Determination of Chloride and Bromide by Furnace Atomic Non-thermal Excitation Spectrometry and Furnace Ionic Non-thermal Excitation Spectrometry* Klaus DittrichJ Bernard Radziuk and Bernhard WelzS Department of Applied Research Bodenseewerk Perkin-Elmer GmbH 0- W-7770 Uberlingen Germany The determination of chloride and bromide by non-thermal excitation spectrometry in a graphite furnace using both atomic and ionic spectral lines was investigated. The most sensitive determinations could be made at the ionic lines CI II 479.545 nm and Br II 470.486 nm. The addition of an ionization buffer provided constant plasma conditions resulting in improved linearity of the calibration function. Under optimum conditions and in the presence of appropriate buffers detection limits of 0.6 ng for chloride and 2 ng for bromide were obtained.Keywords Furnace atomic non-thermal excitation spectrometry; halide determination; atomic and ionic lines; helium plasma; ionization buffer Non-metals are much less commonly determined using atomic spectrometric techniques than are the metallic elements. This is because the most sensitive resonance lines of the non-metals are found in the vacuum ultraviolet (VUV) spectral range and can therefore be monitored only at the cost of increased instrumental complexity. Also of significance is the fact that the less sensitive lines which result from transitions between higher excited states of atoms or ions and can be measured in the UV or visible range have as those in the VUV excitation potentials greater than 9 eV thus requiring more energy than is available in most of the commonly used electrically gener- ated plasmas such as inductively coupled plasmas (ICPs) and sparks.In addition to the requirement for high excitation energy considerable energy is needed for mole- cular dissociation because non-metals often form very stable compounds with metals or other non-metals. On the other hand the energies needed for vaporization are often not very large since halides are among the more volatile compounds. Considering the above it is not surprising that the microwave-induced plasma (MIP) with high electronic temperature and low gas temperatare has in most instances been used for the determination of non-metals by atomic spectrometric methods.Furnace atomic non-thermal excitation spectrometry (FANES) which was developed by Falk and co-~orkers,~-~ uses a non-thermal excitation source based on a hollow cathode discharge. The possibilities of this technique for the determination of fl~oride,~ chl~ride,~ sulphate,6 phos- hate,^^^ ammonium and nitrate9 using atomic emission have been described previously. For chloride molecular non-thermal excitation spectrometry (MONES) a particu- lar variation on FANES which made use of the non- thermal excitation of the stable diatomic molecule MgCl was developed and a detection limit of 0.24 ng of C1 was ~btained.~ This was a considerable improvement over FANES for which the best detection limit that could be measured was 8 ng.5 On the other hand it has previously been reported that 80 pg of chloride could be detected using FANES.4 Other techniques for the determination of halides based on molecular-emission cavity analysis (MECA) and electrothermal evaporation molecular absorption spectro- metry (ETE-MAS) have been described by Dittrich.l0 * Presented in part at the 1990 Winter Conference on Plasma -f Present address Institute for Analytical Chemistry University 8 To whom correspondence should be addressed.Spectrochemistry St. Petersburg FL January 8th-l3th 1990. of Leipzig D-0-70 10 Leipzig Germany. Detailed investigations of the determination of chloride and bromide by FANES and furnace ionic non-thermal excitation spectrometry (FINES) were undertaken in the present work.A helium plasma was used with the aim of developing a sensitive analytical technique for chloride and bromide in aqueous solution and of studying the factors influencing the determination with a view to resolving the discrepancy mentioned above. Experimental Instrumentation A 1 m off-axis Ebert monochromator (Perkin-Elmer Nonvalk CT USA) with an 1800 lines mm-I grating a reciprocal linear dispersion of 0.72 nm mm-I and a wavelength positioning resolution of 3.5-4 pm obtained by means of a stepping motor under microcomputer control was used. The central portion of a tube cross-section located midway between the sampling hole and the tube end adjacent to the anode was imaged 1:l on the entrance slit (50 pm x 2.5 mm) using f/l 1 optics in order to match the aperture of the monochromator.For measurements above 400 and 700 nm cut-off filters WG 320 and GG 475 (Schott Mainz Germany) respectively were used. The FANES source with gas control unit was manufac- tured in the Central Institute for Optics and Spectroscopy of the Academy of Sciences of the German Democratic Republic Berlin Germany (see also references 2-10 and 0 Fig. 1 Schematic diagram of the FANES source 1 and 2 inlets for internal gas flow (gas entering at inlet 1 flows through the tube in all instances); 3 and 4 inlets for external gas flow which reaches the outer surface of the tube through channels in the graphite cones; 5 connection for the evacuation of the furnace (3 and 4 may also be used for evacuation but at a slower flow rate than through 5); and 6 sample introduction port466 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1991 VOL.6 Table 1 Temperature-pressure-discharge programme for CI and Br FANES and C1+ and Br+ FINES measurements Time/s Temperature/"C 0-1 1 20- 152 (12 "C s-1) 11-14 152 14-18 152-352 (50 "C s-') 18-3 1 31-32 32-35 35-45 45-60 60-63 63-65 65-70 70-7 1 352 352 352 352 352 15001 Maximum power 2200 Maximum power 20 20 Phase Pressure/flow* 1 +2+3+4 inlet 6 outlet 5 closed Atmospheric/ Drying At mospherid 1 +2+3+4 inlet 6 outlet 5 closed Vacuum/ 1 +2 closed 5+6 closed Evacuation through 3+4 1 +2 closed 6 closed Evacuation through 3 + 4 + 5 Reduced pressure/ 1 +2 inlet 3+4+6 closed Evacuation through 5 Reduced pressure/ 1 +2 open 3+4+6 closed Evacuation through 5 Reduced pressure/ 1 + 2 open 3 + 4 + 6 closed Evacuation through 5 1 +2+3+4 inlet 5 + 6 closed Atmospheric/ 1 +2+3+4 inlet 6 outlet 5 closed Vacuum/ Pyrolysis Close lid Evacuation? Evacuation3 Pressure stabilization§ Atomization Discharge on )I Purge Discharge on (1 Filling Open lid * The designations 1-6 refer to Fig.1. t During this phase water-solvent residues are removed by means of the sample introduction port in order to keep the anode $ During this phase the maximum rate of evacuation is applied for the removal of residual gases. 9 The optimum pressure and flow conditions are set during this phase. fl Atomization temperatures between 1300 and 2000 "C were investigated and the optimum was selected. I/ Discharge current was also investigated and optimized (in the range 25-1 50 mA).The measurement took place for 8 s between 58 and compartment clean. 66 s. Fig. 1). An HGA-600 power supply (Perkin-Elmer) was used for the electrical heating of the graphite tube. The power supply was controlled from an Epson PCe microcom- puter (Seiko Epson Corporation Nagano Japan) by means of a National Instruments (Austin TX USA) PC2A GBIB interface corresponding to the IEEE-488 standard. Pyro- lytic graphite coated graphite tubes (Perkin-Elmer Part No. B089 1 504) were used exclusively. Samples were introduced using a Model AS-60 furnace autosampler (Perkin-Elmer). The hollow cathode discharge was generated using a current stabilized high voltage power supply (NCL 1200-1 50 pos Heinzinger Rosenheim Germany) also under microcomputer control. The photomultiplier current was converted into voltage using an operational amplifier circuit with a time constant of 60 p s and digitized at a rate of 100 Hz using a 12 bit analogue to digital ( N D ) converter.For data acquisition and manipulation an Epson AX microcomputer (Seiko Epson) and custom software were used. Evaluation was based exclusively on peak height. Procedure Samples 10 p1 in volume were pipetted onto the graphite tube of the FANES source dried and pyrolysed (see Table 1). At this point the FANES head was closed and evacuated. A flow rate of helium gas was selected such that a pressure of 3.5 x lo3 Pa was established (for a discussion of the influence of gas flows and directions see under Optimiza- tion of Pressure and Flow Conditions and of the Tempera- ture Programme).The ignition of the discharge and the electrothermal vaporization/atomization should occur as nearly as possible at the same time. In addition to electrothermal atomiza- tion sputtering contributes to the transport of sample intoJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. SEPTEMBER 199 1 VOL. 6 46 7 the gas phase. However detailed investigations into this phenomenon were not carried out. Stock Solutions Stock solutions with concentrations of 1 and 10 mg ml-I of halide were prepared from KC1 KBr and NaCl (Suprapur Merck Darmstadt Germany). The analyte and ionization buffer solutions were obtained from the stock solutions by dilution. Results and Discussion Wavelength Selection Intensities of spectral lines measured in a helium MIP have been reported by Tanabe et aLl1 Since it could be assumed that excitation mechanisms in the FANES hollow cathode glow discharge would be similar to those in the MIP the most intense of these lines were investigated.From Table 2 it can be seen that the relative intensities measured for different sources vary significantly. It is noteworthy that under the conditions described atomic lines were found to be the most intense for both chlorine and bromine. The large discrepancy between the results of this study and those of Tanabe et was due to differences in the excitation conditions. Tanabe et al. introduced gaseous samples into the MIP under constant conditions; thus a particular electron pressure and degree of ionization was established resulting in the reported intensities. In the present study the electron pressure was determined not only by the discharge parameters but also particularly by the presence of the easily ionized cationic component of the analyte salt.Thus the electron pressure was higher than that in an MIP so that the ionization equilibrium was shifted in the direction of neutral atom formation resulting in a partial inversion of the relative intensities. The values obtained within the same level of ionization are similar for FANES/FINES and the MIP. The following lines were chosen as given in Table 2 for further investigations C1 I (C1 FANES) 725.662 nm; C1 11 (C1+ FINES) 479.545 nm; Br I (Br FANES) 635.074 nm; and Br I1 (Br+ FINES) 470.486 nm. Whereas the pair of ion lines chosen arise from analogous transitions the C1 I line at 725.662 nm does not correspond to the Br I line at 635.074 nm but rather to a Br I line at 655.981 nm.The Br I line at 635.074 nm is analogous to the C1 I line at 741.412 nm. On the basis of intensities for the chlorine lines it would be expected that the Br I line at 655.98 1 nm would be the most intense for the determination of bromide. This line however is overlapped by the very strong Ha spectral line at 656.279 nm so that it was not possible to confirm this theoretical assumption experimentally. Wavelength positions were found using the stepping motor with the aid of lines emitted by a barium hollow cathode lamp Ba I1 493.409 nm and Ba I1 455.403 nm for the spectral range between 400 and 500 nm. For the spectral range 650-800 nm the lines of a potassium hollow cathode lamp K I 769.898 nm and K I 766.491 nm or the Li I line at 670.784 nm were used.Optimization of Pressure and Flow Conditions and of the Temperature Programme As shown in Fig. 1 there are a variety of possible flow conditions. Firstly the conditions required to maintain a stable discharge were investigated. This was possible when helium was introduced through inlets 1 and 2 and pumped out through outlet 5. Inlets 3 and 4 were closed. The flow rates of the gas through the two inlets were measured. Optimum discharge conditions were obtained with flows of 27 and 8 ml s-l through inlets 1 and 2 respectively. These values refer to atmospheric pressure. The actual flow rates at the discharge pressure of 3.5 x lo3 Pa are significantly higher.It is surprising that the discharge is stabilized by such a strong gas flow through the tube. Further optimization by for example reducing the cross-section of the vacuum outlet would be useful since the residence time of the atoms would thus be increased. This however could not be carried out for practical reasons. Based on the optimization described in detail below the temperature-pressure-dis- charge programme given in Table I was used for all further measurements. Table 2 Relative intensities for selected spectral lines of chlorine and bromine as measured in the FANES source compared with the MIT wavelength tables and with the MIP Intensities MIPT FANES/FINES$ Element Wavelength/ Energy levels/ MIT* and line nm Transition cm-' ( 4 9 (a) @Ill (a) (b) c1 I 741.412 4s 4P=j/2-4p 'Pg/2 71 954-85 438 150 3 21 18 18 c1 I 725.662 4s 4P5,2-4p 4S9/2 7 1 954-8 1 73 1 200 14 100 100 100 Cl I 754.709 4s 4P3/2-4p 4Sg12 72 484-85 73 1 25 6 42 44 44 c1 I1 479.545 4s 5Sq-4p 'P3 107879-128 730 250 100 100 47 100 c1 11 48 1.006 4s 'Sq-4~ 'P2 107 879-128 663 200 76 76 37 18 c1 I1 48 1.946 4s 5Sq-4p 5P1 107 879-128 622 200 46 46 26 55 Br I 635.074 5s 4P5/2-5p 'P9,2 63 430-79 172 200 19 100 100 100 Br I1 470.486 5s 5sp-5p 5P3 93927-115176 250 100 100 86 100 Br I1 487.550 5s 5pp-5p 5P3 93 927-1 14 818 400 60 60 38 44 Br I 734.856 5 s 4p3/2-5P 2D$2 64 90 1-78 805 500 15 75 76 76 Br I 700.521 5 s 4P3/2-5p 'P9/2 64901-79 121 200 14 73 83 83 * See reference 12.j. See reference 11. # Results from the present study. Discharge current 100 mA no ionization buffer amount of material 100-200 ng of chloride or fj Measured relative intensities.1 Intensities normalized relative to the highest intensity in the respective ionization state. bromide. These were not optimized conditions (see text).468 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 CI- concentrationhg per 10 pl Fig. 2 Dependence of net line intensity (IN) for C1 FANES at 725.662 nm on the C1 concentration (as NaC1). Discharge current 100 mA; atomization temperature 1500 "C Optimization of Excitation Conditions for the Determination of Chloride General investigations of Cl FANES The spectral line 725.662 nm was used in the investigation of C1 FANES. The dependence of the measured intensity value on concentration is shown in Fig.2. The concave shape of the curve can be interpreted only in terms of a relative increase in the concentration of chlorine atoms with the amount of sodium chloride. This is a result of the suppression of ionization due to the increasing concentra- tion of electrons arising from the increased sodium content. The conditions in the plasma are described by the following equations Na+Na++e- E1=5.14 eV C1 *C1+ +e- E,=13.01 eV NaCle + II (1) where E is the ionization energy. Both ionizations may be caused by interaction with helium but the degree of ionization is higher for sodium. Thus an increase in the over-all concentration of sodium chloride in the plasma shifts the equilibria in the direction shown by the thicker arrows. The effect of potassium bromide on chlorine emission intensities is presented in Fig.3. It is evident that two competing processes determine the effect of KBr on chlorine emission. These are signal enhancing effects resulting from the increase in electron pressure and thus in the concentration of chlorine atoms NaeNa+ +e- El= 5.14 eV + I1 K e K + +e- Ei=4.34eV II NaCl- + KBr .- CleCl+ +e- Ei= 13.01 eV + II BreBr+ +e- Ei= 11.84 eV and a signal depressing effect resulting from the presence of the salt in increasing amounts. This affects the volatiliza- tion process and can also result in further changes in the structure of the plasma. It was concluded based on the results presented in Figs. 2 and 3 that an ionization buffer may be necessary for the application of FANES to the determination of chloride.Optimization of current and atomization temperature for Cl FANES with and without KBr as the ionization buffer In Fig. 4(a) and (6) the effects of current and atomization temperature (tube temperature) respectively on net inten- sity are shown. The results confirm the effect of ionization 0 500 1000 K- concentration/ng per 10 pi Fig. 3 Dependence of net line intensity (IN) for C1 FANES at 725.662 nm on the K+ concentration of the solution (as KBr). Discharge current 100 mA; atomization temperature 1500 "C; analyte 50 ng of C1 as NaCl 25 50 100 150 ilm A m W B W I 1400 1600 1800 2000 TPC Fig. 4 Effects of (a) discharge current and (b) atomization temperature on the net line intensity (IN) for Cl FANES in the presence (A) and absence (B) of the ionization buffer 200 ng of K+ (as KBr).(a) Atomization temperature 1500 "C; and (b) discharge current A 125 and B 45 mA. Analyte 100 ng of C1 (as NaCI) on C1 FANES. Without an ionization buffer there is an initial increase in net intensity with the discharge current followed by a rapid decrease due to increasing ionization. The atomization temperature has virtually no effect on Cl FANES in the absence of a buffer because it has no influence on the extent of ionization of the chlorine atoms. In the presence of the ionization buffer KBr the ionizing effect of the discharge current is suppressed and the signal enhancing effect dominates. The results indicate that for currents greater than 150 mA which were not achievable using the present apparatus the same ionization effect as in the absence of the buffer might be observed.Also the maximum net intensity for C1 FANES was not obtained in the presence of the buffer. This may be a result of the change in plasma characteristics due to an increase in the concentration of charge carriers. The atomization tempera- ture has an effect in the presence of the ionization buffer. At first the net intensity increases because the volatilization of the large amounts of material is improved. At higher temperatures the intensity decreases since the extent of ionization of the easily atomized K atoms is also increased resulting in a change in plasma characteristics. Optimization of discharge current and atomization tempera- ture for CI+ FINES with and without ionization buffer These experiments were carried out at the 479.545 nm spectral line.Increasing concentrations of the ionization buffer gave the anticipated signal depression resulting fromJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 - ( b ) - B - PA I 1 I 469 1500 h c v) C .- = 1000 e ,Z 500 c ; .- v 0 100 300 500 700 900 K- concentrationhg per 10 pI Fig. 5 Dependence of the net line intensity (I,) for C1+ FINES at 479.545 nm on the K+ concentration of the solution (as KBr). Discharge current 100 mA atomization temperature 1500 "C; analyte 50 ng of C1 as NaCl (A) or KCI (B) 2000 - c v) c .- 1500 c s .- $ 1000 v ,z 500 I I 2500 3000k 0 50 100 150 1400 1600 1800 2000 ilmA TPC Fig. 6 Effects of (a) discharge current and ( b ) atomization temperature on the net line intensity (I,) for C1+ FINES at 479.545 nm in the presence (A) and absence (B) of the ionization buffer 200 ng of K+ (as KBr).(a) Atomization temperature 1500 "C; and ( b ) discharge current 110 mA an increase in electron pressure (Fig. 5). There is no significant difference between the results for NaCl and KCI. As for FANES (Fig. 3) it is apparent that the volatilization impeding effect dominates for larger amounts of KBr. The effects of discharge current and atomization temper- ature on the net C1+ FINES line intensity are shown in Fig. 6(a) and (b). The results are as expected. The very strong increase of signal with current results from increased excitation and ionization. The ionization suppressing effect of the K+ ions in the presence of the ionization buffer affects only the magnitude of the increase.The atomization temperature has no significant effect [see also Fig. 4 (b)]. A discharge current of 1 10 mA was used because the stability of the discharge was reduced at higher currents. A comparison of Figs. 4 and 6 confirms that ionization is the major factor affecting the relationships presented. It was not possible to measure the effect of atomization temperatures greater than about 2000 "C in the presence of the ionization buffer as the discharge became unstable. Optimization of Excitation Conditions for the Determination of Bromide For the determination of bromide the spectral lines Br I 635.074 nm and Br I1 470.486 nm were used. By analogy with the studies on C1 FANES and C1+ FINES the effect of the ionization buffer KC1 on Br FANES and Br+ FINES was investigated.The results are given in Fig. 7. The same 1000 r 0 500 1000 K- concentration/ng per 10 pl Fig. 7 Dependence of the net intensities (Z,) for Br FANES (A) and Br' FINES (B) on the additional K+ concentration in the solution (as KC1). Discharge current 100 mA; atomization temper- ature 1500 "C; and analyte 100 ng of Br+ (as KBr) I 1 - 100' 50 100 150 1400 1600 1800 2000 ilmA TPC Fig. 8 Effects of (a) discharge current and (b) atomization temperature on the net intensity (Z,) for Br FANES at 635.074 nm in the presence (A) and absence (B) of the ionization buffer 224 ng of K+ (as KCl). (a) Atomization temperature 1500 "C; (b) discharge current B 30 and A 1 10 mA. Analyte 100 ng of Br- (as KBr) I I 50 100 150 ilm A 2000 1000 Fig.9 Effects of (a) discharge current and (b) atomization temperature on the net intensity (ZN) for Br+ FINES at 470.486 nm in the presence (A) and absence (B) of the ionization buffer 224 ng of K+ (as KCl). (a) Atomization temperature 1500 "C; (b) discharge current B 30 and A 110 mA. Analyte 100 ng of Br- (as =r) relationships as for chlorine (see Figs. 3 and 5) are valid for bromine i.e. the depression of the Br+ FINES signal is much stronger than that of the FANES signal. This can be explained in terms of the ionization equilibria KeK+ +e- E,=4.34eV Br*Br+ +e- KBre + II (3) E = 1 I .84 eV and the increasing volatilization interference. On the other hand the increase in the Br+ FINES signal at low concentrations of K+ is similar to that for Br FANES and not as strong as for C1 FANES.This is because the analyte with 100 ng of Br- as KBr already contains 50 ng of K+,470 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 thus shifting the origin of the curves. Moreover small amounts of the additives appear to improve the volatilization behaviour of the analyte. Optimization exper- iments were accordingly carried out with and without the ionization buffer containing 224 ng of K+ as KC1. The results of a study on the optimization of the discharge current and atomization temperature for Br FANES and Br+ FINES with and without an ionization buffer are presented in Figs. 8 and 9. The effects of current and temperature are evidently very similar to those for C1 FANES (Fig. 4) and C1+ FINES (Fig. 6). For Br FANES in the absence of an ionization buffer the signal decreases with increasing current [Fig.8(a)]. In the presence of a buffer the enhancement of excitation due to the current is dominant. The net intensity for Br+ FINES increases in all instances with current. Also by analogy with C1+ FINES the signal for Br+ FINES decreases somewhat in intensity in the presence of the buffer. The maximum achievable intensity is limited by the stability of the plasma at high currents. The effect of temperature on Br FANES corresponds exactly to that on C1 FANES. When Br+ FINES measurements were made in the presence of the ionization buffer the rapid introduction of relatively large amounts of easily ionized material resulted in discharge instabilities showing that the reported values were not comparable to those obtained for C1+ FINES.Results A typical series of FINES measurements is depicted in Fig. 10. It can be seen that the half-widths of the emission peaks are of the order of 0.1 s. Thus high time resolution and exact selection of integration times are required for the evaluation of peak areas. Since both C1 and Br can be brought into the gas phase by sputtering heating of the tube and ignition of the discharge must be synchronized to within a few tenths of a second. The background emission intensity varies with the tube temperature. Prior to the analytical signal tube tempera- tures are between 300 and 500 "C. After the signal the temperatures reach 1600 "C. The following improvements in instrumentation would be desirable based on the above considerations (i) sub- millisecond time resolution; (ii) optimized integration intervals (between appearance time and end time); (iii) background correction by means of wavelength modula- 0 8 t / s Fig.10 A typical series of C1- determinations using C1+ FINES for samples containing A 10; B 20; C 50; and D 100 ng of C1-. Discharge current 1 10 mA; atomization temperature 1600 "C; ionization buffer 200 ng of K+ as KBr; integration time as marked 0.7 s immediately after the start of atomization 4000 - c v) .- 5 3000 .= 2000 ; 5 v 1000 0 50 100 0 50 100 CI- concentration/ng per 10 pI Fig. 11 Calibration graphs for (a) C1 FANES at 725.662 nm; and (b) C1+ FINES at 479.454 nm in the presence (A) and absence (B) of the ionization buffer (KBr) ; .= 2000 e ,z v 1000 20 100 200 20 100 200 Br concentratiodng per 10 pl Fig.12 Calibration graphs for (a) Br FANES at 635.074 nm; and (b) Br+ FINES at 470.086 nm in the presence (A) and absence (B) of the ionization buffer (KCl) tion; and (iv) provision for gas stop and reduced flow in order to increase the residence time of atoms in the tube. Growth curves were generated for chlorine and bromine using atomic and ionic lines under the conditions described in Tables 1 and 3. The results for both chloride [Fig. ll(a) and (b)] and bromide [Fig. 12(a) and (b)] show an improvement in linearity in the presence of an ionization buffer. There is not much difference in sensitivity for atomic emission despite the fact that considerably higher currents were found to be optimum in the presence of the buffer.This is apparently balanced by the degradation in volatilization conditions caused by the buffer. Since NaCl was used for chlorine measurements at higher concentrations of analyte the sodium acts as an ionization buffer and the addition of KBr results only in poorer volatilization. Chloride is detected with higher sensitivity using FINES when no buffer is used whereas the opposite behaviour was recorded for bromide. This is partly because a higher current could be maintained for all bromide concentrations measured with the buffer whereas for chloride the currents were identical both with and without the buffer and because ion line intensities vary with the square of the discharge current. Although the use of an ionization buffer did not always result in higher sensitivity the plasma conditions became less dependent on the amount of analyte added as evidenced by the improved linearity. Thus the use of an ionization buffer for practical applications is in general advantageous.Conclusions Chlorine and bromine atoms generated from the respective halides can be determined successfully using FANES and the ions using FINES.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 47 1 Table 3 Conditions used for the measurement of analytical growth curves Discharge current/ Atomization temperature/ mA "C Wavelengthl With Without With Without Ionization buffer Method nm buffer buffer buffer buffer per 10 pl C1 FANES 725.662 125 45 1600 1500 200 ng of K+ (as KBr) C1+ FINES 479.545 110 110 1600 1600 200 ng of K+ (as KBr) Br FANES 635.074 110 30 1600 1100 224 ng of K+ (as KCl) Br+ FINES 470.486 125 100 1600 1500 224 ng of K+ (as KCl) Table 4 Results for FANES and FINES Method C1 FANES without IBt C1 FANES with IB C1+ FINES without IB C1+ FINES with IB Br FANES without IB Br FANES with IB Br+ FINES without IB Br+ FINES with IB Wavelengthl nm 725.662 725.662 479.545 479.545 635.074 6 3 5.074 470.486 470.486 Working range of calibration graph Z,=37.2 x ng of C1- 50-100 ng of C1- 1,=29.3 xpg of c1- 0-100 ng of C1- Ih=65.7xpg of C1- 0-20 ng of C1- I,=36.1 x p g of C1- 0- 100 ng of C1- I =4.9 x ng of Br- 100-200 ng of Br- I,= 5.2 x ng of Br- 0-200 ng of Br- I,=9.7 x ng of Br- 40-200 ng of Br- I,=19xng of Br- 0-200 ng of Br- Blank value (arbitrary units) 194 239 289 284 215 295 452 46 3 * I,/Z,=Ratio of intensity for 100 mg of halide to intensity for blank solution.'f IB = Ionization buffer. Relative standard deviation of blank (%) 10.6 12.7 10 7.4 8.5 10 15 13 lS1IB* 19 12 23 13 2.3 1.8 2.1 4.1 Limit of detection/ ng 0.85 1.3 0.46 0.6 5.1 5.8 4.6 2 The ionization buffer containing potassium e.g. as KBr for C1 FANES should be used in the practical analysis of unknown samples in order to keep the electron pressure in the plasma constant. As a result and in particular at low concentrations of analyte the maximum sensitivity for atomic lines is increased and that for ionic lines decreased. Thus the linearity of the calibration function is good and a first order polynomial fit can be made. Under optimum conditions i.e.in the presence of the ionization buffer and using the required parameters atomization temperature 1600 "C and discharge current 110 mA the ion lines for both elements are the most sensitive. The ratios measured between atom and ion line intensities in the presence of the ionization buffer (calculated from the sensitivities Table 4) are 8 1 100 for Cl I:C1 I1 and 27 100 for Br 1:Br 11 which are nevertheless shifted in the direction of atomic line intensity as compared with the MIP (see Table 2). There is an improvement of an order of magnitude in detection limit compared with our previous measurements on C1 FANESS The results for C1+ FINES are similar to those for MgCl MONES which were however obtained in an argon p l a ~ m a . ~ In general in the evaluation of FANES FINES and MONES it must be remembered that depending on the composition of the plasmas and owing to the relatively low gas and high electron temperatures it is essential to study and to account for the dissociation and ionization equilibria MeM++e- X e X + +e- M X s + II (4) in order to obtain accurate results. In the present work this was achieved using potassium (K+K++e-) as the ioniza- tion buffer. References 1 2 3 4 5 6 7 8 9 10 11 12 Falk H. Hoffmann E. and Ludke Ch. Fresenius Z. Anal. Chem. 1981 307 362. Falk H. Hoffmann E. and Liidke Ch. Spectrochim. Acta Part B 1981 36 767. Falk H. Prog. Anal. '42. Spectrosc. 1980 3 181. Falk H. Hoffmann E. and Liidke Ch. Prog. Anal. At. Spectrosc. 1988 11 417. Dittrich K. and Fuchs H. J. Anal. At. Spectrom. 1987 2 533. Dittrich K. Fuchs H. Berndt H. Broekaert J. A. C. and Schaldach G. Fresenius J. Anal. Chem. 1990 336 303. Dittrich K. and Fuchs H. J. Anal. At. Spectrom. 1989 4 705. Dittrich K. and Fuchs H. in preparation. Dittrich K. Fuchs H. Mermet J. M. and Riviere B. J. Anal. At. Spectrom. 1991 6 313. Dittrich K. CRC Crit. Rev. Anal. Chem. 1986 16 223. Tanabe K. Harauchi H. and Fuwa K. Spectrochim. Acta Part B 1981 36 119. Wavelength Tables with Intensities in Arc Spark or Discharge Tubes of the Massachusetts Institute of Technology (MIT) ed. Harrison G. E. John Wiley New York 1939. Paper 0/0395 7H Received August 30th 1990 Accepted May 3rd. 1991