JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 199 I VOL. 6 19 Application of Platform and Palladium Modification Techniques With Furnace Atomization Plasma Emission Spectrometry Ralph E. Sturgeon Scott N. Willie Van T. Luong and Shier S. Berman Division of Chemistry National Research Council of Canada Ottawa Ontario K1A OR9 Canada Figures of merit (limit of detection sensitivity and precision) for eight elements (Ag Cd Pb Mn Sn TI As and Se) were obtained using furnace atomization plasma emission spectrometry (FAPES) in combination with atomization of samples from a L’vov platform and the use of a Pd modifier (0.5 pg). A 50 W He plasma was utilized. Precision of replicate integrated signals averaged 2.8% (range 1.743%) for injection by hand of samples at concentrations 13-70-fold above the detection limit.Estimated limits of detection and sensitivities for integrated intensities improved 3-1 0- and 6-1 7-fold respectively for volatile elements (Cd Pb and Ag) compared with sample atomization from the tube wall. Added NaCI up to 24 pg had no effect on the recovery of the integrated signal from Pb when a 100 W plasma was used. Keywords Helium plasma; atomic emission spectrometry; graphite furnace; platform; palladium modifier Furnace atomization plasma emission spectrometry (FAPES) is a novel analytical atomic emission technique based on a combined source wherein the inherent advantages associated with a graphite furnace atomizer (for sample vaporization and atomization) are intimately coupled with those of an atmo- spheric pressure radiofrequency (r.f.) He plasma (for element excitation) in a single convenient unit.’-s Efficient multi- element detection thus becomes feasible using the graphite furnace even for elements having lines of high excitation energy which are normally not accessible through reliance on graphite furnace thermal techniques alone.6 Analytical figures of merit have previously been reported for several elements (Ag Cd Pb Ni Fe Be Bi Cu and P) in a broad effort to characterize the FAPES technique.’ De- tection power in the low pg range is available although preci- sion of replicate signals with sample atomization from the furnace wall has an average relative standard deviation (RSD) of 5%.Additionally analyte emission transients for volatile elements (Cd Pb and Ag) occur early during the initial period of rapid heating and gas expansion from the analytical volume.Smith er al.’ reported on the interference from an easily ionized element (NaCl) on the response for Ag using a FAPES system. A depressive effect was observed when more than 350 ng of Na were present (a 40 000-fold excess over Ag) presumably due to a modification of the excitation characteris- tics of the plasma in the presence of Na. We have also noted interference effects on the determination of Cd and Pb in digested samples of marine sediment and biological tissues,“ although their origin was not investigated. Many matrix effects have been successfully eliminated from graphite furnace atomic absorption spectrometry (GFAAS) when samples are atomized from a L’vov platform7 and chemical modifiers such as reduced Pd are present.x Similar benefits should accrue with FAPES; thus it was the purpose of this study to assess the feasibility of transferring the above GFAAS technology to the FAPES system.Report- ed here are the improved limits of detection (LODs) sensitiv- ity and precision of replicate signal measurement for a number of elements when atomized from a L’vov platform in the presence of reduced Pd as compared with similar figures of merit arising from sample atomization from the tube wall. It is also shown that interference from NaCl on the response from Pb is reduced when these techniques are implemented although they are not as efficiently eliminated as they are in GFAAS. NRCC No. 325 17. Experimental Instrumentation The FAPES system has been fully described previo~sly.*~~~~ Briefly a 13.56 MHz r.f.He plasma was supported inside a Perkin-Elmer Model HGA-2200 graphite furnace. An Ar ex- ternal sheath gas was maintained at a flow-rate of 1 1 min-’ and an He internal plasma gas was set at 200 ml min-I. All signals were digitized with 12 bit resolution and acquired at the rate of 8OOO points per transient (irrespective of the sampling period). In this study grooved pyrolytic graphite coated tubes and solid pyrolytic graphite L’vov platforms (Perkin-Elmer) were used. Reagents Stock solutions (loo0 mg 1-I) were prepared by dissolution of the high-purity metals (Cd Pb Ag Mn Sn and T1) or their salts (As,O and Na,SeO,). Working standards were obtained by dilution of the stock solution with high-purity de-ionized distilled water (DDW) acidified to 1% v/v with sub-boiling quartz distilled HNO (for Cd Pb Ag TI and Mn) or HCl (for Se As and Sn).A 3% m/v solution of high-purity NaCl (Ventron) was prepared in DDW and further purified of trace metals by passage through a column of 8-hydroxyquinoline immobilized on silicasv A 2500 mg I-’ stock solution of Pd chemical modifier was prepared by dissolution of the high- purity metal (Spex Industries) in HNO,. A 250 mg I-’ working solution of Pd was prepared by dilution of the stock with DDW. Procedure All wavelengths (A) were set using the appropriate hollow cathode (Ag Cd Pb Ti Mn and Sn) or electrodeless- discharge lamps (As and Se). The lines selected are summar- ized in Table 1. A nominal spectral bandwidth of 0.08 nm was used for all measurements.In order to compare the results with those from previous studies a 50 W (forward power) He plasma was used. The effect of forward plasma power on the recovery of Pb response in the presence of NaCl was investi- gated for forward powers of up to 100 W. Analytical figures of merit were obtained for each element. Sample volumes of 5 yl were introduced into the furnace by hand using a syringe (Hamilton) equipped with a short section of Teflon tubing. The palladium modifier solution (250 mg 1-I) was co-injected with the sample in the amount of 0.5 pg (2 pl). Atomization conditions were not extensively studied and cannot be considered as truly optimum for any given element.20 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL.6 Table 1 using atomization from the tube wall and no modifier (see reference 5 ) Absolute detection limit. Values in parentheses are those obtained LOD*/pg Excitation Element h/nm energy/eV Peak height Peak area Ag 328.1 As 200.3 Cd 228.8 Mn 279.5 Pb 283.3 Se 196.0 Sn 284.0 TI 377.6 3.78 7.54 5.4 1 4.43 4.37 6.32 4.79 3.28 0.76 ( 1.2) 0.92 (2.0) 0.90 7.2 (21) I03 1700 10 17 0.50 (4.8) 2.0 (6.8) 5.0 7.4 (46) 370 3600 29 37 * LOD based on 30 criterion where o is the estimate of the standard deviation of repetitive measurements of the blank. Samples were charred at 500 "C during which time forward power was applied and the He plasma spontaneously estab- lished. The furnace was allowed to cool to room temperature") and the sample subsequently atomized using the maximum power heating mode of the HGA-2200 controller.Final atomi- zation temperatures (in "C as measured on the furnace power supply meter) selected were 2100 (Cd Pb Ag and Tl) 2500 (Mn and Sn) and 2700 (As and Se). Blank signals recorded for each analyte were obtained by atomizing 5 p1 volumes of DDW containing 1% v/v HNO or HCI along with 2 pl of 250 mg 1-I Pd. Background correction with this system was limited to se- quential measurement of a blank solution and/or an unloaded furnace and plasma cycle. The high time resolution of the de- tection system permitted unambiguous background correction to be implemented although it must be noted that with the simple samples run here no structured background was en- countered. The effect of various masses of NaCl on the response from 5 ng of Pb was investigated.The sample was atomized under various conditions including directly from the tube wall from the wall following the addition of 5 pg of Pd from the plat- form and from the platform following the addition of 5 pg of Pd . Measurements of the time-dependent temperature of the surface of the graphite tube were made using a calibrated Ircon Series 1100 automatic optical pyrometer (Niles) which was focused through the sample introduction hole onto the back wall of the tube. Measurements of the plasma reflected power levels were obtained from the appropriate 1/0 port on the rear of the RFX-600 r.f. power supply (Advanced Energy Indus- tries). This 0-5 V signal was directly compatible with the data acquisition system.Results and Discussion No analytical difficulties were experienced when using the platform with the FAPES system. In general the plasma was easier to establish in this configuration as the centre electrode (r.f. antenna) is in closer proximity to the ground established through the platform (thus a higher field strength resulted). Additionally plasma stability was enhanced at lower input powers. Continuum background intensity however increased 5-1 5-fold (depending on the wavelength). This increase was noted at room temperature and was not due to the image of the incandescent platform overlapping the entrance slits of the spectrometer but rather to the increased plasma power density in the tube as a result of the volume restriction caused by the platform.At high temperatures generally at times beyond the duration of the analyte signal the background intensity was 0 1 2 3 4 5 fh Fig. 1 Time-resolved analyte emission signals for atomization from a platform. Continuous line 0.5 kg Pd modifier; broken line no modifier. (a) As 10 ng; 200.3 nm. (h) Cd 50 pg; 228.8 nm. (c) Pb 500 pg; 283.3 nm. ( d ) Sn 1.25 ng; 284.0 nm higher from the tubes equipped with a platform due to detec- tion of the incandescent image of the platform. Typical emission signals for several of the analytes are shown in Fig. 1. The effect of the Pd modifier is particularly evident for the most volatile elements. In the absence of Pd signals from As and Se were poorly resolved from the back- ground. Signals for Pb Cd and Ag were clearly delayed in time (temperature) relative to those obtained without Pd whereas this time shifting of the signal was less evident for elements such as Mn and Sn. These observations are consis- tent with those reported in the literature dealing with GFAAS.Peaks displayed for As and Cd in Fig. 1 (and that for Se. not shown) exhibit poor signal to noise characteristics in the absence of Pd because at the 500 "C char temperature used here significant amounts of these elements are lost prior to atomization.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL. 6 21 ~~~ Table 2 Absolute sensitivity for platform atomization and Pd modifier. Values in parentheses are those obtained using atomization from the tube wall and no modifier (see reference 5 ) Peak height/ Peak area/ Element nA ng-' nA s ng-' Ag As Cd Mn Pb Se Sn TI 3260 (370) 2.3 320 (51) 850 64 (4.7) 0.2 1 110 120 720 ( 120) 0.67 110 (7.4) 250 19 (1.1) 0.13 25 60 Table 3 Precision of replicate measurements.Precision expressed as % RSD for n > 8 at analyte concentrations (+fold in excess of estimated LODs RSD (%) Element Ag As Cd Mn Pb Se Sn T1 Peak height 1.1 (30) 5.2 (100) 2.2 (50) 4.7 (130) 2.1 (70) 5.4 (25) 3.9 (125) 2.8 (30) Peak area 2.0 (50) 4.3 (25) 2.6 (25) 4.2 (25) 2.0 (70) 2.4 (16) 1.7 (40) 3.1 (13) K 75 1 '+\ I 1 I 1 0 20 40 60 80 100 Mass of NaCVpg Fig. 2 Effect of adding NaCl on the recovery of an integrated signal in- tensity from 5 ng of Pb in A the presence and B the absence of 5 pg of Pd modifier. (a) Atomization from the tube wall; (h) atomization from the platform Figures of Merit Table 1 presents the estimated LODs obtained with the present system.For comparative purposes those for Cd Pb and Ag re- ported in a previous studys using only sample atomization from the tube wall are also given. It is clear that there is an en- hancement (3-10-fold) in performance with the platfonn- modifier system for these elements. Peak height LODs are generally superior to those characterizing the signal area because of the rapid evolution of analyte atoms at high temper- ature coupled with their rapid diffusive loss from the analyti- cal volume. Detection limit data for the FAPES system compare favour- ably with GFAAS data (within a factor of 2) with the excep- tion of As and Se." The wavelengths of these two elements are in a region of poor response for the Hamamatsu R446 photomultiplier tube (PMT) used in this study and the emis- sion lines have higher excitation energies than those used for the other elements.At 200nm the radiant sensitivity of the PMT is less than 25% of its maximum. Accounting for this factor alone should make the LOD for As obtained with FAPES comparable to that with GFAAS." The LOD for Se would be similarly enhanced but performance of FAPES would still be more than ten times worse than GFAAS. The lower excitation energy for the Se transition certainly cannot account for its poorer performance relative to As. These two factors however severely limit analytical performance. Line selection in the present study was made on the basis of sensi- tivity rather than any rigorous comparison of signal to noise or detection limit criteria.Non-resonance lines of As Sn and T1 were utilized because sensitivity was 25-fold greater for As at the 200.3 than at the 193.7 nm line 15-fold greater for Sn at the 284.0 than at the 224.6 nm line and 4-fold greater for T1 at the 377.5 than at the 276.8 nm line. Table 2 summarizes the sensitivities of the elements in the terms of absolute response per ng for both peak height (nA) and area (nA s); Table 3 gives the precision of replicate meas- urements. Precision is largely determined by the repeatability of injection of samples into the furnace by hand. As expected over-all precision is slightly better for area measurements. The presence of the platform has on average served to increase the precision of replicate measurements by a factor of 2 over that obtained for a similar suite of elements when atomization occurred from the tube wall.' This effect is most noticeable for the volatile elements with atomization from the wall since these are released during the early period of rapid heating of the furnace and expulsion of the internal gas.Additionally any instability in the plasma which might occur during this period of rapid heating will have less influence on signals for those analytes released at higher temperatures. Effect of NaCl on Response From Lead The effect of increasing the mass of NaCl added on the re- sponse from Pb when atomized from both the platform and the wall in the presence and absence of Pd modifier was investi- gated.It was noted that greater signal recoveries were obtained as the forward power to the plasma was increased. In the pres- ence of 18 Fg of NaCl the peak height signal recovery for Pb was 70%. This increased to 80% at a forward power of 75 W and remained unchanged at 100 W power (it should be noted that quoted forward powers are simply the output levels of the RFX 600 r.f. supply; the amount of power dissipated in the tuning network is unknown even when reflected power levels are near zero). Integrated signal recovery was 70 70 and loo% respectively at these same power levels. As a conse- quence the interference study was conducted with a forward power of 100 W. It is possible that it is the initial temperature of the centre r.f. electrode (before atomization) that deter- mines to some extent the degree of interference encountered.As the forward power is increased there is a concomitant in- crease in the electrode temperature. This might influence the degree of interaction between the sample vapour and the elec- trode surface as discussed in more detail below. Fig. 2 shows the recovery of the integrated signal intensity from 5 ng of Pb in the presence of various masses of added NaCl (corrected for an NaCl blank). The effect of 5 pg of Pd modifier is evident. A depressive interference effect is almost22 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL. 6 0 125 100 p 75 2 .- $ 5 0 C - 2E 1 2 3 4 5 ds 2400 2000 1600 1200 g 800 400 0 0 1 2 3 4 5 us Fig. 3 Characteristics of the Pb-NaCI interference system in a 100 W He plasma.(a) A Temperature-time characteristics of the tube wall; 8 reflected power-time characteristics. (h) Lead emission transient. Broken line 5 ng of Pb with 5 pg of Pd; continuous line 5 ng of Pb with 5 pg of Pd and 60 pg of NaCl immediately established in the absence of modifier irrespec- tive of whether or not atomization occurs from the tube wall or the platform. It is not clear from these data whether atomiza- tion from the tube wall or the platform is to be preferred in the presence of small amounts of salt; as the mass of NaCl in- creases beyond 30pg the benefits of atomization from the platform become more obvious. It is equally obvious however that the interference-free range was not increased by the use of a platform although the interference observed at higher masses of NaCl is clearly less with the platform compared with the in- formation obtained from atomization from the wall.Significant depressive interference effects on the signals from several analytes by NaCl have also been reported by Falk et a1.12 using the hollow cathode FANES system. The extent of interference was apparently due to the degree of temporal overlap of the Na C1 and analyte vapour populations in the an- alytical volume. In the present situation the volatilities of NaCl and Pb are very similar and it is expected that there is significant element population overlap with the result that this system can be considered one of the most severe situations likely to arise. Fig. 3 shows the emission signals arising from the atomiza- tion of 5 ng of Pb in 1 % HNO and in the presence of 60 pg of NaCI. In both instances 5 pg of Pd were present as a modifier.Integrated signal recovery is only about 50% in the presence of this amount of NaCl. It is interesting to note the early de- crease in background signal intensity accompanying the rapid release of sample from the platform. This apparent ‘quench- ing’ of the plasma might be due to the momentary formation of a more conductive vapour plume within the analytical volume leading to the formation of a ‘mini-arc’ and the tran- sient shrinkage of the plasma volume actually imaged onto the entrance slit. Alternatively the centre r.f. electrode might be acting as a cooler second surface onto which a portion of the volatilized matrix may condense thereby altering the plasma excitation processes.A short time later this surface layer is re- volatilized as the electrode heats to the temperature of the hotter tube wall by radiation. Such plasma ‘quenching’ was observed only in the presence of large amounts of NaC1. It is clear from Fig. 1 that such effects are absent when only Pd and analyte are atomized. Reflected power losses which occur during the atomization of this sample are also shown in Fig. 3 along with the tempera- ture of the tube wall. The change in reflected power level with temperature is not influenced by the presence of NaCl in the sample but is apparently determined only by the temperature. It is also evident from this figure however that it is not the temperature of the tube wall that accounts for the time depen- dence of the reflected power level.This is clear from the pres- ence of an ‘induction’ period which occurs between establishment of maximum reflected power and steady state tube wall temperature. Thus it appears that it is the tempera- ture of the centre r.f. electrode that influences the magnitude of the reflected power level the latter being heated by radiation from the tube wall. The reason for a rise in reflected power and hence a decrease in the efficiency of the coupling of the r.f. energy into the system is not yet clear. At high tempera- tures however themionic emission of electrons from the graphite surfaces occurst3 which may alter the impedance matching With the r.f. tuner presently used (Heathkit Model 2060A) reflected power levels could not be decreased below 30 W (for 100 W forward power) when the system was at a temperature of 2100 K.Some finite value of resistance is es- tablished in the gas phase which currently cannot be matched to the 50 i2 source output with this tuner. A similar effect has been noted with both the hollow cathode and anode versions of the FANES technique whereby the voltage drop establish- ing the low pressure glow discharge in these devices is significantly decreased owing to the release of thermionic elec- trons from the tube wall as it is heated above 1800-2000 K.I2.l4 At atmospheric pressure the mean free path for electrons in He lies in the pm region.” Thus thermionic electrons released from the centre electrode surface might have a more pro- nounced effect on plasma excitation than those released earlier from the hotter but more distant tube wall. Electron mobility and mean free path are probably fieid dependent and with the present centre electrode configuration will vary with radial po- sition.As a non-uniform radial field distribution is involved and is most intense at the centre electrode electrons generated in this region will contribute the most to the observed analyte emission intensity (assuming excitation by electron impact). Thermionic electrons originating from the surface of the centre electrode are likely to have a greater effect on plasma process- es than those generated some distance away at the tube wall where the field is less intense. It is generally accepted that there are two populations of electrons in plasmas each with a separate temperature.1 2 ~ 1 6 A high temperature group promotes excitation and ionization whereas a lower temperature higher density fraction is involved in collisional de-excitation pro- cesses. Thermionic emission from the tube wall may serve to flood the plasma with a large number of low-energy electrons thereby reducing the analyte emission because of the increased collisional de-excitation rate. The intensity of the He I plasma gas line at 587.6 nm in- creased approximately 5-fold with temperature over the initial ramp heating stage (to 2100 K) of the atomization cycle. This presumably arises because the decrease in plasma particle density accompanying the increased temperature produced a longer mean free path of electron excitation.The intensity of this line subsequently decreased to a level of only about 3- fold greater than the room temperature intensity. In an attempt to re-tune the plasma continuously by hand as the temperature was ramped it was almost possible to eliminateJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1991 VOL. 6 23 this subsequent decrease. Obviously a complete understand- ing of these observations will require fundamental informa- tion on the mechanistics of level populations because in addition to the above there is a substantial increase in the number density of thermionic electrons with increased tem- peratureI3 and these may exert considerable influence on the over-all excitation processes. Conclusions It would appear that all of the advances made in the use of the stabilized temperature platform furnace concept’ currently so important to the field of GFAAS can be conveniently utilized with the FAPES technique.As expected detection limit sensi- tivity and precision significantly improved for all of the vola- tile elements when atomized from a platform in the presence of Pd. It also appears that this approach will be useful in helping to control some matrix interferences which are present when using this technique. The authors thank B. Hutsch Ringsdorff Werke Germany for supplying the centre electrodes. References I Liang D. C. and Blades M. W.. Spectrochim. Acfa. Part B 1989 44,1059. 2 3 4 S 6 7 8 9 10 11 12 13 14 IS 16 Sturgeon R. E. Willie S. N. Luong V. T. and Berman S. S. and Dunn J. G. J. Anal. At. Specworn. 1989,4,669. Smith D. L. Liang D. C. Steel D. and Blades M. W. Specfro- chim. Acta Part B 1990,45,493. Sturgeon R. E. Willie S. N. Luong V. T. and Berman S. S. J. Anal. At. Specrrom. 1990,5,635. Sturgeon R. E. Willie S. N. Luong V. T. and Berman S. S. Anal. Chem. 1990,62,2370. Littlejohn D. Anal. Proc. 1989 26,92. Slavin W. Camick G. R. Manning D. C. and Pruszkowska E. At. Speca-osc. 1983,469. Beach L. M.. Spectroscopy 1987,2,2 1. Sturgeon R. E. Berman S. S. Willie S. N. and Desaulniers J. A. H. Anal. Chem. 1981,53,2337. Manning D. C. and Slavin. W. Specrrochim. Ada Part B . 198540 461. Slavin W. in Graphite Furnace AAS a Source Book Perkin-Elmer Norwalk 1984. Falk H. Hoffman E. and Ludke Ch. Prog. Anal. Spectrosc. 1988 11,417. Sturgeon R. E. Berman S. S. and Kashyap S. Anal. Chem. 1980 52 1049. Harnly J. M. Styris D. L. and Ballou N. E. J . Anal. At. Spectrorn. 1990,5 139. Westwood W. D. Prog. S w - Sci. 1976 7 71. de Galan L. Spectrochim. Acfa Part B 1984,39,537. Paper 0i02625E Received June 6th 1990 Accepted October 15th 1990