JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 809 Simultaneous Multi-element Determination Using Helium or Argon Plasma for Graphite Furnace Capacitively Coupled Plasma Atomic Emission Spectrometry* Glen F. R. Gilchrist Peter M. Celliers Huacheng Yang Changbin Yu and Dong C. Liang Aurora Instruments Ltd. 191 West 6th Avenue Vancouver British Columbia Canada V5Y 1K3 Furnace atomization plasma excitation spectrometry (FAPES) has been heralded as a promising new technique for simultaneous multi-element determinations at sub-ppb concentrations. Graphite furnace capacitively coupled plasma atomic emission spectrometry (GFCCP-AES) is a specific configuration of FAPES technology. Reported here is the first application of GFCCP-AES to simultaneous multi-element ultratrace determinations.Comparison of analytical sensitivity and signal characteristics are made for Ar and He plasmas in GFCCP-AES. The emission pulses for Cd Zn TI Pb Ag Sb Au and Sn were higher and broader in the Ar plasma than in the He plasma. The noise levels were approximately equal for both plasmas hence the signal-to-noise ratio was superior for the Ar plasma. Mechanisms of vaporization atomization and excitation are discussed and compared with those proposed by other workers. Keywords Plasma furnace emission; furnace atomization plasma excitation spectrometry; graphite furnace capacitively coupled plasma; inductively coupled plasma atomic emission spectrometry; graphite furnace atomic absorption spectrometry Low pressure gas discharges in hollow cathodel and hollow anode2*' sources have been used for atomic emission spectrometry (AES). It has been ~hownl-~ that electrother- mal atomization coupled with plasma excitation spectro- scopy can provide detection limits comparable to graphite furnace atomic absorption spectrometry (GFAAS).The requirement for low pressure discharge hence evacuation of the furnace work head after loading limits the applica- tion of furnace atomization non-thermal excitation spectro- scopy (FANES) in practical analytical chemistry. Liang and co-workers4~s and Sturgeon and co-worker~.~.~ have reported the development and merits of furnace atomiza- tion plasma excitation spectrometry (FAPES) which is capable of operating at atmospheric pressure. Graphite furnace capacitively coupled plasma atomic emission spec- trometry (GFCCP-AES) is a configuration of FAPES technology for practical analytical chemistry.All of the workers cited here have recognized that the plasma furnace is a promising source for simultaneous multi-element analysis with detection limits in the picogram range for most metals and non-metals. The greater sensitivity of GFCCP-AES over inductively coupled plasma atomic emission spectrometry (ICP-AES) is due to the high atomization efficiency of GFCCP-AES the inefficiency of nebulization for sample introduction in ICP-AES and the longer residence time of excited species within the analysis volume 250-1000 ms for GFCCP versus < 1 ms for ICP- AES. An advantage of GFCCP-AES over GFAAS is that the high excitation efficiency of the r.f.plasma allows simulta- neous determination of elements at the detection limits of GFAAS.l-lO Another advantage is that non-metals some of which are difficult to analyse using GFAAS can be analysed using GFCCP-AES. Sturgeon et al.' and Banks et a[.* have reported absolute detection limits of 94 and 29 pg respectively for P. Huang and blade^,^ using an atmospheric pressure parallel plate CCP as a gas chroma- tography detector have determined that the limits of detection for F C1 Br I C N and 0 range from 10 to 78 pg SKI. The emission spectrum created within the furnace removes the need to provide a hollow cathode lamp or electrodeless discharge lamp as the source for absorption spectrometry. * Presented at the 1993 European Winter Conference on Plasma Spectrochemistry Granada Spain January 10-1 5 1993.Examination of the fundamental characteristics of the GFCCP discharge along with pragmatic studies of vapori- zation atomization and excitation mechanisms matrix effects chemical modification and practical applications are essential to realize the full potential of this technique. One of the challenges facing plasma furnace spectroscopists is to explain the observed discrepancies in the operation of He and Ar plasmas. Difficulty in igniting and maintaining Ar plasmas have been reported,1° yet Ar has a lower first ionization energy than He (15.75 ev and 24.59 eV respectively) and therefore it should be easier to ignite an r.f. plasma in Ar than in He. Hettipathirana and Blades" have calculated that the electron ionization rate (the number of ionizing collisions per cm per Torr of an electron) is 120 times higher in He than in Ar which might be the reason why He plasmas are easier to ignite.Other differences in the operation of Ar and He plasmas have been noted. The emission of C I CN and C2 observed in Ar but not in He is thought to arise from sputtering of the centre electode by Ar. l o Sturgeon et allo have reported that their Ar plasma was not sufficiently energetic to excite CO+ (19.39 and 19.66 eV) as was observed in their He plasma but Ar gave better sensitivity for Ag Cu Mn Pb Ni and Fe. Hettipathirana and Blades12 have used emission and absorption to examine the temporal characteristics of Ag Pb and Mn and to elucidate some of the vaporization and atomization mechanisms in the plasma furnace.They have shown that the centre electrode if it is not heated by the plasma can act as a second surface for condensation and re-vaporization. They have also shown that the plasma shifts the appearance temperature and the peak tempera- ture of Pb to lower values. Determination of elements in real samples may require improved sensitivity and reduced matrix interferences which can be obtained by using matrix modification platform atomization and isothermal atomization. Stur- geon et al.13 have obtained 3-10-fold improvement in detection limit and 6-1 7-fold improvement in sensitivity over wall atomization for Cd Pb and Ag by using a platform and Pd modification. They have also reported 100% recovery of Pb with up to 24 pg of NaCl added to the plasma furnace.For wall atomization and no modifier Gilchrist and Liang14 have reported 93 and 55% recovery of T1 with 0.08 and 1.2 pg of NaCl respectively added to the furnace. Hettipathirana and BladesI5 have compared810 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 matrix interferences from NaCl and NaN03 on the emission signal of Pb and Ag. Sturgeon et aLL6 have used plasma furnace emission to determine Cd and Pb in sediment and biota. The objectives of this work are to show that GFCCP-AES can be used for the simultaneous ultratrace determination of Cd Zn T1 Pb Ag Sb Au and Sn to demonstrate that stable plasmas can be ignited and maintained in Ar and to compare the signals obtained from Ar and He plasmas during multi-element analysis.Experimental Instrumentation An A1 7000 (Aurora Instruments Vancouver Canada) simultaneous multi-element plasma furnace emission spec- trometer was used for simultaneous determinations. An A1 2000 (Aurora Instruments) spectrometer was used for atomic absorption and single element atomic emission measurements. The A1 7000 and A1 2000 are unique new instruments so discussion of some of their noteworthy features is warranted here. Both spectrometers use a state- of-the art transversely heated pyrolytic graphite coated integrated contact graphite furnace that provides temporal and spatial heating profiles that are nearly isothermal. The furnace is 19 mm long with a 5.8 mm i.d. The furnace acts as a counter electrode to a 1 mm diameter pyrolytic graphite coated graphite rod positioned along the furnace central axis.An Ar or He atmospheric pressure plasma is generated by applying high voltage r.f. power to the central electrode while maintaining the furnace at ground poten- tial. The entrance optics of the A1 7000 image a region of the plasma that is 2 mm high and 0.025 mm wide. Normally the furnace is aligned so that the bright region of the plasma adjacent to the centre electrode is focused onto the entrance slit. Using this imaging system no emission signal from the heated furnace centre electrode or plasma species is observed. Blank firings and background signals plotted with the vertical scales used in Figs. 1-4 yield straight horizontal lines. The A1 7000 has a 0.75 m Rowland Circle configuration with 0.02 nm spectral band- pass.Background correction is carried out by simultane- ously monitoring the background emission 0.08 nm higher and lower than the analyte line. The data acquisition system operating at 300 Hz is fast enough to provide 100-400 signal and background measurements for each element depending on the emission pulse width of that element. The emission lines used in this work in nm were Cd 228.80 Zn 2 13.86 T1276.79 Pb 283.3 1 Ag 328.07 Sb 206.83 Au 242.80 and Sn 286.33. An r.f. oscillator and power supply have been developed to provide reliable ignition and stable running of the Ar plasma. Optimization of the oscillator for the He plasma performed by adjusting the impedance matching between the oscillator and the plasma furnace causes enhancement of the emission signals in He however sensitivity and detection limits are still inferior to those in the Ar plasma reported here.The Ar plasma could not be made to ignite and run reliably in the He-optimized oscillator. A 27.12 MHz free running oscillator was optimized to ignite and maintain an Ar plasma and used throughout this work. Temperature measurements were made using a built-in IR pyrometer focused onto the outside of the furnace directly below the dosing hole. The built-in pyrometer was calibrated against an Ircon Model 1100 IR optical pyro- meter (Ircon Niles IL USA) focused through the dosing hole onto the inside of the furnace directly below the dosing hole. Procedures Drying pyrolysis and atomization temperatures were tested to obtain conditions that gave a reasonable response for all eight elements.All elements had peak heights not more than 1 5% lower than for their individually optimized conditions except Sn which had a peak height about 50% below its optimum value. All samples were dried by heating at 90 "C for 5 s then raising the temperature at a rate of 1 K s-l for 25 s. The charring temperature was 400 "C and the furnace heating rate was 3500 K s-l. The atomization temperature was 2100 "C unless stated otherwise. Furnace cut-off experiments to investigate analyte vaporization processes were conducted as follows He or Ar plasma gas was chosen and the furnace was loaded with 20 pl of 25 ppb of Cd Zn Ti Pb Sb Au and Sn and 0.05 ppb of Ag multi-element test solution. The drying and pyrolysis steps were run in the normal manner but during the atomization cycle the furnace power was cut off after 400 ms by which time the furnace wall was heated to 1800 "C.When the furnace reached room temperature the centre electrode was removed and the furnace was cleaned at 2400 "C for 2 s. The centre electrode was then returned to its place and a normal atomization cycle was run with the dry furnace. Reagents and Materials Stock solutions containing 10 000 pg ml-I were prepared separately by dissolving the appropriate mass of pure metal (Cd Zn Pb Sb Au and Sn 99.999%) or pure metal oxide (T1 99.99%) in 50% aqua regia (HC1-HNO3 3+ 1). These solutions were transferred into calibrated flasks and made up to volume so that the final solution contained 5% aqua regia. The Ag stock solution was prepared in a similar way except that nitric acid was used instead of aqua regia.Multi-element test solutions containing Cd Zn T1 Pb Sb Ag Au and Sn were prepared daily by serial dilution of single-element stock solutions. All test solutions were made up to contain 1% v/v nitric acid and stored in plastic containers. Test solutions of 20 pl in volume were manually injected into the furnace. Results and Discussion Analytical Performance In GFCCP-AES like GFAAS discrete samples are placed into the atomizer then dried and thermally treated prior to atomization. Thermal pre-treatment is used to remove much of the solvent and matrix and in some cases to decompose the sample. During the atomization step the analyte is vaporized atomized and excited in the r.f. plasma then the analyte vapour is removed from the analytical volume by diffusion.Convolution of the supply function and the removal function produces an emission signal pulse. Figs. 1-4 show the transient emission signals of 50 ppb of Cd T1 Pb and Au respectively collected simultaneously with Zn Ag Sb and Sn (not shown here). Cadmium T1 Pb and Au were chosen for presentation because they cover a broad range of volatilities and are representative of the emission pulse shapes observed for the eight elements studied. The emission pulses of the eight elements studied were between 3.8 and 8.7-fold higher in the Ar plasma than in the He plasma. The relative performance of the Ar and He plasmas changes with a change in impedance matching between the oscillator and the plasma furnace but a 2-4-fold higher sensitivity was always observed with the Ar plasma.Table 1 gives the mean peak height values of the emission signal obtained for replicate injections of the eight-element test solution under the compromise conditions detailed earlier.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993. VOL. 8 A Time/ms Fig. 1 Emission pulses for 50 ppb of Cd test solution collected during eight simultaneous determinations in A Ar and B He plasmas 600 c. .- C 2 500 2 e c. .- 3 400 > v) c .- 9 300 .- C .- $ 200 E .- W 100 0 r I 500 1000 1500 2 Time/ms Fig. 2 Emission pulses for 50 ppb of T1 test solution collected during eight simultaneous determinations in A Ar and B He plasmas Ti me/ms Fig. 3 Emission pulses for 50 ppb of Pb test solution collected during eight simultaneous determinations in A Ar and B He p 1 as m a s 2500 I A fn J .g 2000 - ip 2 c >.cn e .- 0 1000 2000 3000 Time/ms 81 1 1 Fig. 4 Emission pulses for 50 ppb of Au test solution collected during eight simultaneous determinations in A Ar and B He plasmas Table 1 Peak height values'J for 20 pl of multi-element test solution containing 50 ppb of Cd Zn T1 Pb Sb Au and Sn and 0.1 ppb of Ag. Values are the mean k the standard deviation of at least five replicate determinations and are given for peak two of emission pulses having two peaks Element Ar plasma He plasma Cd 1100k40 289k 1 1 Zn 813k 11 1402 19 T1 630 k 32 125k 12 84k 12 Pb 507 k 44 Ag 660k 55 102k7 Sb 324 k 32 53+ 13 Au 2260k 110 530 k 45 Sn 156+ 12 18k4 Figs. 1-4 and Table 1 show that for the conditions reported here the Ar plasma provides emission signals for Cd Zn T1 Pb Sb Ag Au and Sn that are both higher and broader than those in the He plasma.The complete mechanism that causes this difference in emission signal is not yet fully understood. The enhanced signal in the Ar plasma is probably due in part to slower diffusion of the analyte atoms through Ar. However attenuation of analyte diffusion by the massive Ar atoms and ions is not sufficient to account for the whole extent of the signal enhancement. Differences in the nature of the Ar and He plasmas especially differences in electron mobilities and the number of collisions per electron probably cause some of the observed differences in sensitivities for the analyte atoms in the different plasmas.Hettipathirana and Blades'' have determined that the electron ionization rate is 120 times greater in He than in Ar which probably accounts for the easier ignition and maintenance of He plasmas. Ionizing collisions are how- ever much more energetic than the collisions that excite emission from the analyte atoms studied here. Excitation energies for these elements varied from 3.27 eV for the Ag 328.07 nm line to 5.97 eV for the Sb 206.83 nm line. A possible reason for the improved sensitivity in Ar compared with He might be found in differences in the elastic electron-neutral collision cross-sections for lower energy electrons especially in the energy range of 3-6 eV which is the most important energy range for excitation. The bulk of the plasma electrons have energies in the thermal range812 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL.8 around 0-3.0 eV. It is interesting to note that the collision cross-sections for these low energy electrons are aHe=6 x cm2 for He and Ar re~pectively.1~ This Ar cross-section is remarkably low about an order of magnitude less than in He. In the energy range of 0-5 eV the He cross-section decreases monotoni- cally while the Ar cross-section increases to a maximum value of around 4 eV which is nearly an order of magnitude larger than that for He. At 4 eV electron energy aH,=2.3x 10-{6 cm2 and aA,=6Ox 10-16~m2.17Thus in Ar the collision rate of electrons in the 4-6 eV range is nearly two orders of magnitude higher than in the thermal range of 0.0-3.0 eV.It is possible that this strong energy dependence of the elastic collision cross-section in Ar can lead to an enhancement of the electron energy distribution function in this energy range. Electrons in the thermal energy range (0.0-3.0 eV) will tend to absorb energy from the r.f. field through collisional absorption and thereby diffuse out to higher energies. However in the higher energy ranges (4-6 eV) the much stronger collision cross-section will tend to trap electrons in this energy range until they lose energy through inelastic processes (e.g. atomic excitation). Further study of this hypothesis requires detailed kinetic studies of the electron velocity distribution function including the effects of the r.f. field and the energy dependence of the collision cross-sections.cm2 and aA,=0.6 x Features of the Emission Pulses Fig. 1 shows that for the Ar plasma the Cd emission pulse has a shoulder on the trailing edge and for the He plasma the emission pulse is smaller and has only one peak. In both Ar and He the appearance time of the Cd pulse is virtually zero due to the 400 "C pyrolysis temperature and the high heating rate of the furnace. Fig. 2 shows that for the Ar plasma the T1 emission pulse has two peaks the first peak appearing at the onset of atomization while the second peak appears at about 500 ms. In the He plasma the T1 emission pulse has only one peak but it might be two peaks that are not resolved in time. Fig. 3 shows that in Ar Pb like T1 has two peaks the first peak appearing at the onset of atomization while the second peak appears at about 600 ms.In He the Pb emission pulse has a shoulder on the rising edge indicating that two peaks are present but not resolved in time. Fig. 4 shows that in Ar Au like T1 and Pb has two peaks the first occurring at the onset of the atomization step and the second appearing 1200 ms later. In He two peaks are also seen for Au the first appearing almost immediately and the second appearing at 700 ms. In the Ar plasma Zn TI Pb Ag Sb and Au have two resolved peaks and Cd has two unresolved peaks as shown by the shoulder on the trailing edge of the pulse. In the He plasma two peaks are observed only for Pb Sb and Au. The single peaks observed in the He plasma for Zn T1 Ag and Sn appear at least 100 ms after the start of the atomization cycle (Zn) and the appearance time increases in the same manner as does the appearance time of peak 2 for these elements in the Ar plasma.The above observations suggest that analyte elements have two vaporization events one from the furnace wall and one from the centre electrode. For Zn T1 Ag and Sn in the He plasma no emission pulse is observed during the initial vaporization from the tube wall because these elements are not sufficiently atomized or excited or the peaks are not resolved as with Cd in the Ar plasma. Fig. 5 shows the appearance times for the emission pulses of the eight elements studied in the Ar plasma (circles) and in the He plasma (squares). For the reasons given above the single peak observed for Cd Zn T1 Ag and Sn in the He plasma has been treated as peak 2.When double peaks are observed the first peak (open symbols) appears almost instantaneously (< 150 ms) while the 1500 0 0 0 . m * 500 m p;::. 0 Cd Zn TI Pb Ag Sb Au Sn Fig. 5 Appearance time of the emission pulses collected during eight simultaneous determinations in the Ar plasma (0 peak 1 ; 0 peak 2) and the He plasma (0 peak 1; H peak 2); for emission pulses with one peak that peak is given as peak 2 second peaks (filled symbols) appear at later times (up to 1400 ms). The appearance time of the second peaks for both Ar and He plasmas increases with a decrease in the volatility of the elements. The second peaks in the He plasma (filled squares) have appearance times 150-700 ms earlier than those in Ar plasma (filled circles).Hettipathirana and BladesI2 have used atomic absorption to determine that gas-phase analyte species initially deposi- ted on the furnace wall condense on the centre electrode and are re-vaporized and atomized when the electrode is heated by radiation from the furnace wall. This phenome- non was observed here using an A1 2000 atomic spectro- meter which has the same furnace and r.f. power supply as the A1 7000 but uses a Czerny-Turner monochromator and can be configured for absorption or emission. Hettipathi- rana and BladesI2 have reported that the plasma heated the centre electrode preventing condensation of analyte so no second peak was observed in the emission mode. For this work neither plasma heated the centre electrode and no analyte loss was observed when the plasma was on for 30 s at a furnace temperature of 400 "C.Adjusting the impedance matching between the oscillator and the plasma furnace can lead to heating of the centre electrode even when the furnace is held at room temperature. When this is done only one peak is observed. With no heating by the plasma the centre electrode may have enhanced heating by conduction in He compared with Ar since the thermal conductivity of He is 9-10 times greater than that of Ar. Thus in He earlier vaporization and atomization of the analyte condensed on the centre electrode is observed. The utility of the centre electrode as a second surface for condensation and re-vaporization is demonstrated by a two step pyrolysis programme which can be used to eliminate the early small peak from emission pulses with two peaks.The temperature for the first pyrolysis step is chosen so that all of the analyte stays in the condensed phase on the graphite tube wall. The temperature of the second step is chosen so that the analyte is vaporized and condenses on the centre electrode but is not lost from the tube. For example Pb can be pyrolysed at 550 "C in the first step then at 900 "C for 5 s in the second step without loss from the furnace. A two step pyrolysis programme was not used in this work because Ag Au and Sn could not be transferred to the centre electrode without causing Cd to be lost from the furnace prior to atomization. The observed features of the emission pulses are consis- tent with a mechanism in which the analyte is vaporized from the tube wall and then condenses on the centre electrode.When the centre electrode is heated by radiationJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 813 800 600 - UJ U .- 5 400 t E $ 200 (D >. - U .- s o A 0 . 0 . - *-& 0 500 1000 1500 2000 Time/ms Fig. 6 Results of temperature cut-off experiments for Pb in (a) He plasma and (b) Ar plasma A first firing; and B second firing m C .; 600 v) E 400 W 200 .- L I 0 500 1000 1500 2000 Time/ms Fig. 7 Results of temperature cut-off experiments for Au in (a) the He plasma and (b) the Air plasma A first firing; and B second firing from the tube wall and by conduction from the gas phase18 the analyte is re-vaporized atomized and excited by the plasma. This distillation of solvent matrix and analyte may prove to be beneficial in removing matrix interferences.Gilchrist and LiangI4 obtained 93% recovery of T1 in a 4 ppm NaCl matrix by pyrolysing at 600 "C and 55% recovery of T1 in a 60 ppm NaCl matrix by pyrolysing at 500 "C; in GFAAS without platform or modifier all the T1 would be lost from the furnace under the conditions described above.I9 Direct proof of analyte condensation on the centre electrode is presented below. Cut-off Experiments As detailed under Experimental partial heating of the furnace removal of the centre electrode purging the furnace and then replacing the electrode and re-firing allows one to determine if any analyte has condensed on the centre electrode. Figs. 6 and 7 show the results for Pb and Au respectively obtained during multi-element cut-off experi- ments run using Ar and He plasmas.Fig. 6(a) shows that for the He plasma Pb is completely vaporized and removed from the furnace during the first firing (solid line) because no emission pulse is observed during the second firing (broken line). In Fig. 6(a) a small ragged peak can be seen at the beginning of the atomization step. This peak is probably due to Pb vaporizing from the wall as molecular species (PbO PbC1220.21) partially dissociating in the gas phase then condensing on the centre electrode. Fig. 6(b) shows that for the Ar plasma some of the Pb vaporized during the first firing condenses on the centre electrode and produces an emission pulse during the second firing (broken line). Fig. 7(a) shows that for the He plasma some Au is atomized and excited during the first firing condenses on the centre electrode and produces an emission pulse appearing at 1000 ms during the second firing.Fig. 7(b) shows that the same phenomenon occurs in the Ar plasma except that the emission pulses are larger due to the greater sensitivity of Au in Ar. Signals for the remaining six elements were also collected during the cut-off experiments. For the Ar plasma Cd and T1 were completely vaporized during the first step. Emission signals for Zn Pb Sb Ag and Au were observed during both steps indicating that they were vaporized during the first firing condensed on the centre electrode and were then vaporized atomized and excited during the second firing. No emission signal was observed for Sn during the first firing but an Sn emission pulse was observed during the second firing.These observa- tions are consistent with Sn being vaporized as molecular species (SnO SnC1,20,21) during the first firing but hardly dissociating so that excited atoms are not observed. The SnO or SnCl condensed on the centre electrode might be vaporized atomized and excited during the second firing. For the He plasma Cd Zn T1 Pb and Sb were completely vaporized in the first step as shown by their emission signals in the first step and no emission signal in the second step. This is consistent with greater heating of the centre electrode by conduction in He than in Ar thereby reducing or eliminating condensation of analyte on the centre electrode. Emission signals for Ag and Au were observed during both steps indicating that Ag and Au were atomized excited and condensed on the centre electrode during the first firing.The Ag and Au were then re- vaporized from the centre electrode atomized and excited during the second firing. The absence of an Sn emission signal during either step indicates that Sn was vaporized as a molcular species (SnO SnC1220,21) and removed from the furnace without being atomized and excited sufficiently to be observed. Atomization Curves Fig. 5 shows that vaporization and atomization from the centre electrode occur earlier in a He plasma than in an Ar plasma. Fig. 8 shows the effect of atomization temperature on the emission pulses of Pb and Ag in Ar (solid symbols) and in He (open symbols). The optimum atomization temperatures of Pb and Ag in the Ar plasma are 400 and 500 "C higher respectively than those in the He plasma.Fig. 8 also shows that Pb and Ag emission pulses can be observed at I 100 "C in the He plasma while no signal can be observed at this temperature in the Ar plasma. Table 2 shows that the optimum atomization temperatures for Cd T1 Pb Ag Sb and Au are 300-500 K lower in the He plasma than in the Ar plasma. The optimum atomization temperatures for longitudinally heated GFAAS in the absence of chemical modifiers are given for comparison. One possible explanation for the earlier appearance of the analyte emission pulse for the He plasma is that He having thermal conductivity about nine times greater than Ar heats the centre electrode by conduction to a greater extent than does Ar.Heating of the electrode by the plasma was not observed prior to atomization for either plasma gas in814 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 2000 [ 1 ioo - 1200 1600 2000 Atomization temperaturePC Fig. 8 Atomization temperature versus emission intensity for Pb (A He; B Ar) and Ag (C He; D Ar); emission intensity for the largest peak (peak 2) is given when two peaks occur Table 2 Optimum atomization temperatures for GFCCP-AES in Ar and He and for GFAAS Element Ar Plasma He Plasma GFAAS* Cd 1600 1100 1500 Zn 1800 1800 2000 Ti 1900 1600 2100 Pb 1800 1400 2000 Ag 2000 1500 2400 Sb 2150 1700 2200 Au 21 50 1700 2200 Sn 2100 2 100 2 500 *Longitudinally heated graphite furnace. this work so any plasma-induced heating would have to occur during the atomization cycle when direct measure- ment of the centre electrode temperature is very difficult.The above notwithstanding the A1 7000 oscillator can be configured to heat the centre electrode if desired. Sturgeon et a[.’ have reported a decrease of 400-500 K in the appearance temperature of P in their FAPES source relative to the appearance temperature in GFAAS. Hettipa- thirana and Blades1* have reported a decrease in the appearance temperature of Pb in GFCCP-AES relative to GFAAS and no change in the appearance temperature of Ag and Mn. Both workers have suggested that the observed shifts to lower appearance temperatures are due to plasma induced dissociation of volatile analyte containing mole- cules. This mechanism does not explain why elements that do not form volatile molecular species have different appearance temperatures in Ar compared with He.Silver and Au certainly do not form volatile molecular species but have appearance temperatures in He lower than those in Ar. Investigation of heating and sputtering of the centre electrode by He and Ar and bombardment of gas-phase molecules by the plasma gases is required to elucidate the different vaporization atomization and excitation mecha- nisms in these two plasma gases. AES at concentrations that have previously been the domain of GFAAS and ICP mass spectrometry. An r.f. oscillator can be built that reliably produces a stable capacitively coupled 27 MHz Ar plasma. Four to nine times greater sensitivity is observed for the Ar plasma relative to the He plasma for the conditions reported here.Volatile metal species that vaporize prior to atomization are likely to condense on the centre electrode. Analyte atoms and molecules that have condensed on the centre electrode are re-vaporized and atomized when the electrode is heated by radiation and conduction from the tube wall. Second surface condensation and re-vaporization of solvent matrix and analyte might be advantageous for reducing matrix interferences. Sputtering or heating by the He plasma can contribute to analyte vaporization from the centre elec- trode. Financial support from the National Research Council of Canada (IRAP-M Grant No. 4 I 138W) the Science Council of British Columbia [Technology B.C. Grant No. 17(T-3)] and the Western Economic Diversification Program (B-90- WD-0592) is gratefully acknowledged.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 References Falk H. and Ludke Ch. Prog. Anal. Spectrosc. 1988,11,4 17. Ballou N. E. Styris D. L. and Harnly J. M. J. Anal. At. Spectrom. 1988 3 1141. Frech W. Baxter D. C. and Hutsch B. Anal. Chem. 1986 58 1973. Liang D. C. and Blades M. W. Spectrochim. Acta Part B. 1989 44 1059. Smith D. L. Liang D. C. Steel D. and Blades M. W. Spectrochim. Acta Part B. 1990 45 439. Sturgeon R. E. Willie S. N. Luong V. Berman S. S. and Dunn J. G. J. Anal. At. Spectrom. 1989 4 669. Sturgeon R. E. Willie S. N. Luong V. and Berman S. S. Anal. Chem. 1990 62 2370. Banks P. R. Liang D. C. and Blades M. W. Spectroscopy 1992 7 36. Huang D. and Blades M. W. Appl. Spectrosc. 1991,45 1468. Sturgeon R. E. Willie S. N. Luong V. T. and Dunn J. G. Appl. Spectrosc. 199 1 45 14 13. Hettipathirana T. D. and Blades M. W. Spectrochim. Acta Part B. 1992 47 493. Hettipathirana T. D. and Blades M. W. J. Anal. At. Spectrom. 1992 7 1039. Sturgeon R. E. Willie S. N. Luong V. and Berman S. S. J. Anal. At. Spectrom. 1991 6 19. Gilchrist G. F. R. and Liang D. C. Am. Lab. 1993 March 25 34U. Hettipathirana T. D. and Blades M. W. J. Anal. At. Spectrom. in the press. Sturgeon R. E. Willie S. N. Luong V. and Berman S. S. J. Anal. At. Spectrom. 1990 5 635. Brown S. C. Basic Data of Plasma Physics MIT-Wiley New York 1959 p. 16. Wu S. Chakrabarti C. L. and Rogers J. T. Prog. Anal. Spectrosc. 1987 10 1 11. Fuller C. W. Anal. Chim. Acta 1976 81 199. Frech W. Lundberg E. and Cedergren A. Prog. Anal. At. Spectrosc. 1985 8 257. Gilchrist G. F. R. Chakrabarti C. L. Hughes D. H. and Ashley J. F. Anal. Chem. 1992 64 1144. Conclusions Simultaneous multi-element determination of Cd Zn T1 Pb Ag Sb Au and Sn can be carried out using GFCCP- Paper 3/01 583A Received March 18 I993 Accepted May 18 1993