Anal. Chem. 1983, 55, 1045-1050
1045
Three-phase Argon Plasma Arc for Atomic Emission Spect rornet ry Terence R. Mattoon' and Edward H. Plepmeier" Department of Chomistty, Oregon State University, Corvallis, Oregon 9733 1-4003
A new three-phase argon plasma arc was developed for analytical emlsslion spectrometry. The power supplled to the plasma arc utilizes a simple clrcult wlth three-phase ac llne A three voltage stepped (downvla a transformer to 104 .V, Concentric quartz:tubing assembly Is used to supply argon and sample aerosol 110 the plasma. A Bablngton-type nebullrer wlth a spllt sample aerosol flow stream was constructed and used to supply sample aerosol to the three-phase plasma arc. The sample stre,am is entralned In the center of the plasma and appears slniilar to the sample stream entralned In an Induction coupled plasma. Desolvatlon of the sample aerosol prlor to reachlng the plasma enhanced the emission signal by 30%. Deteciloni llmlts between 0.1 and 2 pg/mL were obtalned for seven elemcents uslng a 1/4-m monochromator. Calibration curvles yielded a llnear dynamic concentratlon range of approxiimately 3 orders of magnltude.
A new electrical three-phase argon plasma arc has been invented that entrains a sample stream in the center of hot plasma. The analytical advantages of the high temperatures of electrical plasmas t o atomize and excite samples have been recognized for decades. However, one of the well-known difficulties of most plasma sources is their inability t o obtain enough contact between an aerosol sample stream and hot plasma t o provide for efficient sample atomization ( I ) . Induction coupled plasma (ICP) sources suffered from this problem until the early 1960'9, when t h e sample stream was successfully entrained in the center of a toroidal shaped plasma (2). T h e ICP inow provides relative freedom from matrix effects, excellent detection limits, and a wide dynamic concentration ramge for multielement analysis, due in large part t o the relatively long contact time of the sample stream with hot plasma as it passes through the center of the plasma (3-6). During our work with argon plasmas that are generated we discovered that a plasma having between two elecitrodes a triangularly shaped horizontal base could be maintained with three electrodes by connecting each electrode to one phase of a three-phase, 60-Hz electrical power source. We then found that a sample aerosol stream could be introduced u p through the center of the plasma. This paper reports the first use of the new three-phase plasma source for elemental analysis and the initial studies of the electrical and spectral characteristics of the plasma source.
(a,
EXPERIMENTAL SECTION Electrodes. The plasma arc burns between three 0.040-in. (1.0 mm) diameter 2 % thoiriated tungsten electrodes (Tungsten Electrodes, Teledjme Wah Chang, Huntsville, AL). The electrodes are in a horizontal plane and point toward the center of the triangle determined by the tips of the electrodes, Figure 1. The tips of the electrodes are 5 mm apart. The electrodes are held in place and connected electrically to electrode holders (Spectra-Metrics Inc., Andover, MA), Figure 2. The electrode holdeirs are cooled with a water flow rate of 100 mL/min. The electrodes are cooled with argon that surrounds them, flowing at 0.5 L/min, from the electrode holder into the plasma. A flow rate of 0.5 Present address: Hewlett-Packard, Sunnyvale, CA. 0003-2700/83/0355-1045$0 1.50/0
Table I. Quartz Tubing Assembly Sizes configuration
outer tube
1 0.d. (mm) i.d. (mm) 2 0.d. (mm) i.d. (mm) 3 0.d. (mm) i.d. (mm) 4 0.d. (mm) i.d. (rnm) 5 0.d. (rnm) i.d. (mm)
26.0 24.0 25.0 21.5 21 19
18.5 15.5 16.0 13.5
middle tu be
inner tube
4 17.0 14.5 13.5 12.5 12.0 10.0 12.0 10.0
3 4 3 8
L/min argon for electrode cooling was the minimum effective flow rate, while flow rates of argon above 2 L/min distorted the plasma shape. As new electrodes heat, they form a molten globule on the end, which reaches a maximum diameter of approximately 2 mm. This globule has no adverse effect on the plasma arc as far as arc wander. During operation, the lengths of electrodes decrease at a rate of less than 1 mm/h. Electrical Power. The power used to sustain the argon plasma arc was derived from three-phase ac line voltage. Three-phase ac line voltage consists of three sine waves 120' out of phase from each other, Figure 3. Three-phase ac line voltage is transmitted through four wires with one wire corresponding to each phase and the fourth wire serving as a common (neutral). The voltage between any two of the three phases was 208 V,,. This voltage was reduced with a three-phase wye-connected transformer to 104 VmB, Series power resistors (nominally 1 Q, each consisting of two 2 4 , 200-W resistors in parallel) were used to drop the voltage to the plasma and limit the current to 24 Aave.The actual resistances of the power resistors varied up to 2 Q during operation because of heating, even with fan cooling. Therefore, it might be better to use 2-Q or 3 4 resistors with a higher power rating, e.g., 800 W. Standard 30-A circuit breakers were placed in each leg. The power Impply, including the transformer was enclosed in a cabinet 56 cm wide, 43 cm deep, and 63 cm high. Because of the phase angle relationship of the three currents, the current in each leg of the delta-connected plasma was 14 A (8). The voltage wave forms across the resistors were sinusoidal, indicating that sinusoidal current was being supplied to each electrode. The voltage wave form between any pair of electrodes was approximately sinusoidal, with a root mean square value of 11V ., Therefore, the power dissipated by each leg of the plasma was 170 W, for a total plasma dissipation of 510 W. Quartz Tubing Assembly. Quartz tubing was used to supply argon to the plasma and to introduce the sample aerosol into the center of the plasma. The original quartz tubing assembly consisted of two concentric quartz tubes. The argon to support the plasma flowed through the outer quartz tube and the sample aerosol flowed through the inner quartz tube. However, by adding a third middle concentric quartz tube, a more stable plasma was formed. This three concentric quartz tubing assembly, Figure 4, is similar to the quartz tubing assembly in the ICP. The middle quartz tube stabilizes the plasma by preventing the argon flowing through the outer tube from pushing the plasma away from the electrodes. Various flow rates of argon were attempted through the region between the middle and inner quartz tubes, but no benefit was apparent. Several diameters of each quartz tube were tested in various combinations with each other, Table I. In the following para@ 1983 American Chemical Society
1046
ANALYTICAL CHEMISTRY, VOL. 55, NO. 7, JUNE 1983
TRANSFORMER WYE-CONNECTED
DELTA L O A D
C I R C U I T BREAKERS
B '!
\
\
ALUMINUM S P A C E R
Figure 1. Schematic of circuit to control current to the plasma. ARGON
-ALUMINUM
SPACER
'I PLE
Flgure 4. Quartz tubing assembly. Scale is approximate only.
Figure 2. Electrode holders: (A) electrode, (e) ceramic cylinder, (C) metal body, (D) water out, (E) Plexiglas, mounted on 1.5 in. diameter adjustable posts on an optical table, (F)bok connecting electrode holder to Plexiglas, (G) electrical connection, (H) water in, (I) argon in, (J) bolt to hold N in place, (K) bolt to hold electrode in place, (L) argon out, (M) thumb screw to center electrode, (N) removable sheath.
I'
I ,'
.
Figure 3. Three-phase ac line voltage wave forms. graphs the considerations for determining the finally adopted diameters of the three quartz tubes will be discussed. The inner quartz tube required a small 0.3-mm exit orifice so the sample aerosol would have a highly velocity to overcome aerodynamic barriers of the plasma. A small exit orifice also produces a small diameter sample aerosol stream, which has less effect on the current flow between the electrodes than a larger diameter. The aerosol tended to condense a t the entrance of the 3 mm i.d. inner quartz tube, but no condensation occurred for the 7 mm i.d. quartz tube. The inner quartz tube was placed 8 mm below the plasma, so the tip of the tube would not melt. A larger distance from the inner quartz tube to the plasma caused the sample aerosol to spread out too much before entering the plasma, resulting in a large diameter sample introduction stream, which interfered with the current flow between the electrodes and caused a noisy plasma. The choice of the outer quartz tube diameter was influenced by two factors. A larger diameter required more argon to support the plasma. Too small a diameter caused the corresponding electrode gaps to be too small, which resulted in difficult sample introduction for the inner tube that was used. The best inside diameter of the outer quartz tube was 19 mm.
The outer diameter of the middle quartz tube helped determine the velocity of the support argon flow to the plasma a t a given outer quartz flow rate. A larger diameter caused a higher argon flow velocity which reduced plasma flicker. However, for a small distance between the outer quartz tube and the middle quartz tube, any misalignment of the two quartz tubes resulted in severe, uneven flows of argon to the plasma from one side to another. By choosing an outside diameter of 12.4 mm, a misalignment between the middle and outer quartz tubes resulted in only a small argon flow unevenness. The middle quartz tube was placed approximately 5 mm below the plasma. The three holes in the outer quartz tube for inserting the three electrodes were 7 mm in diameter. The three holes were 120° apart and 7 mm from the top of the quartz tube, which allowed for viewing the sample emission 10 mm above the electrodes. This viewing area is above the high current density regions and high background regions of the plasma and yet still provides good excitation of the analyte atoms. Operating the Arc. Prior to igniting the plasma arc, the coolant water, power supply fans, power supply, heating coil, and argon were turned on in that order. A 15 cm, 12 mm i.d. Pyrex tube with a 3.5 cm long, 12 mm diameter piece of graphite in the end was then touched to all three electrodes, igniting the plasma. The plasma would ignite with an outer tube argon flow between 3 L/min and 10 L/min. Igniting the plasma with an argon flow through the sample injection tube was difficult, so was avoided. The outer tube argon flow rate was adjusted to 9 L/min for most emission spectrometry. The sample aerosol was injected into the plasma a t a flow rate of about 1 L/min. If the sample injection was not in the center of the plasma, viewed through arc welding mask filters, the electrodes were adjusted by moving the posts on which the electrode holders were mounted. Usually alignment prior to igniting the plasma arc was sufficient to obtain proper sample injection into the center of the plasma. Nebulizer. The Babington-type nebulizer (9, IO) was constructed out of Plexiglas and standard fittings, Figure 5 . The stainless steel sphere had an orifice of 0.2 mm. This was the smallest orifice that could be drilled in stainless steel with the equipment available. Since a nebulizer based on the Babington principle requires argon flow through this orifice at high pressure for efficient aerosol production, the resulting argon flow into the nebulizer was 2.5 L/min a t 60 psi. Since the flow rate was too high for the plasma arc, a split stream system was developed. The split stream system was developed so that the sample aerosol would travel two similar paths, Figure 5. The exit tip diameter of a Pyrex sample auxiliary tube was varied to obtain the desired argon flow rate to the plasma. The resulting sample aerosol flow was about 1 L/min. Heating coils made of nichrome wire were wrapped around both the sample injection tube and the sample auxiliary tube. The
ANALYTICAL CHEMISTRY, VOL.
Flgure 5. Babington-type nebulizer: (A) quartz sample injection tube, (6)Pyrex auxiliary sample tube, (C) plastic SzYagelok (all other Swagelok is stainless steel), (P) O-ring, (E) Plexiglas, (F) stainless steel thread rod (3/8-32), (G) glass rod, (H) plug, (I) stainless steel sphere.
primary function of the heating coils was t o prevent sample condensation in the sample tubes. Condensation in the sample tubes results in uneven flows and a noisy plasma. An added advantage of heating the sample aerosol was that less power was required for the plasma 1x1heat and desolvate the sample droplets. The power supplied through the heating coils was 25 W for the sample injection tube and 20 W for the sample auxiliary tube. The sample solution delivery to the nebulizer was controlled by a peristaltic pump (Cole-Parmer). The nebulization efficiency was less than 2 %, Nebulization efficiency is the ratio of the amount of solution that leaves the nebulizer via the sample tubes to the amount of the sample solution entering the nebulizer. It has been reported by several workers, (11) that one drop of 10% Triton X 1100 ([ (tetramethylbuty1)phenyll hydroxypoly(oxyethanediyl))added to 100 mL of sample solution improves signal stability. Triton X 100 is a surfactant which reduces surface tension and tends to reduce the formation of large drops in the nebulizer chamber. Large drops can cause turbulence in the flow of fine particles 1 o the plasma. The addition of the Triton X 100 was also reported to aid in overall nebulizer drainage. One drop of 10% Triton IC 100 was added to every 100 mL of all sample solutions, with the exception of the sodium solutions, because sodium was present in the Triton X 100. Spectrometric Observation System. A 0.25-m Aminco grating monochromator was used with a 600 grooves/mm grating blazed for 500 nm first order. The reciprocal linear dispersion is 6.6 nm/mm first order. The monochromator has a limited scanning range of 15 nm per scan at a scan rate of 10 nm/min. The observed spectral Iband-pass was 0.20 nm. A current-to-voltage converter for a 1P28’4photomultiplier used MeKee-Pedersen components in a MP-1001 console. The photomultiplier tuble, in an Aminco (Catalog No. C4-6155) housing, was operated at -725 V from a regulated dc power supply (MP1002). The anodic current from the photomultiplier tube was converted to a voltage by means of a chopper-stabilized operational amplifier (MP-1031) with a high impedance resistance selector (MP-1009) set at lo7 a. For most studies, a 1.0-s time constant was used along with a Fluke 8020A multimeter to measure the voltages. Typical voltages were between 10 mV and 1V. For the background emission scans, a 0.1-s time constant was used along with a Heath Model SR-204 strip chart recorder. Emission from the plasma was directed t o the monochromator via two 350-mm focal length aluminized mirrors with a MgF, coating for high reflectance in the UV wavelength region. The mirrors were mounted in the over-and-under “2” configuration to reduce astigmatism produced by the side-by-side mirror configuration in the monochromator (12). The mirror mounts enabled independent control of the pitch and yaw and provided calibrated adjustments of the emission viewing region both vertically and horizontally.
RESULTS AND DISCUSSION Nebulizer Characteristics. T o determine the optimum solution flow rate to the nebulizer, a 100-mg/L barium solution
55, NO. 7, JUNE 1983
1047
SAMPLE FLOW RATE, rnl/min
Relative emission intensity of Ba solution (0)and S I N vs. sample solution flow rate to t h e nebulizer.
(X)
and a blank solution were pumped, via a peristaltic pump, to the nebulizer a t various flow rates, and the resulting emission signals a t 455.4 nm were measured with a 1-s time constant. A plot of the net relative emission vs. the sample flow rate is shown in Figure 6. In Figure 6 it can be seen that the relative emission signal increases with sample solution flow rate, but so does the noise, as indicated by the standard deviation bars at each point. Individual points showing the net signal to noise ratio for each are also shown (right-hand ordinate). The uncertainty or noise, N , was calculated from the standard deviation of the lOO-mg/L barium solution total signal and the standard deviation of the blank signal. The S I N improves with flow rate except for the last point. At this high flow rate the sample solution probably no longer produces a thin film over the Babington sphere; rather, a thick film is probably produced which results in large droplet formation during the nebulization process. The optimum flow rate was 18 mL/min, which was used for the rest of the experiments. The respom;e time of the nebulizer, the time required to reach 63% of the new signal, was found to be 10 s using a 500 mg/L Ba solution. Heating Coil. T o measure the effect of the heating coil on the intensity of an emission signal, a 100 mg/L barium solution was nebulized into the plasma, with the heating coil off. The emission signal was measured a t 455.5 mm and recorded on a strip chart recorder with a 0.1-s time constant. After a stable signal was obtained, the heating coil was turned on. The emission signal increased, and when the emission signal reached a new stable level, the heating coil was turned off. The emission signal decayed down to approximately the original level before the heating coil had been turned on. The net signal increase due to having the heating coil on was 30%. Background Emission. A spectrum of the background emission was obtained over the wavelength region 600-250 nm, Figure 7. All of the 70-nm scans shown in Figure 7 are not at the same scale sensitivity due to the fact that the optical efficiency of the spectrometer varied over the 600-250 nm wavelength region. The peaks in the background emission spectrum correspond to tungsten emission lines. The tungsten emission results from the use of 98% tungsten electrodes. The other 2% is thorium. No peaks corresponding to thorium emission were seen above the background in the background emission spectrum. No peaks corresponding to argon emission above the background were seen in the background emission spectrum either. The background signal increased as wavelength decreased from 600 t o 460 nm. In addition to an actual change in emission intensity, this signal change may be due in part to the fact that the grating of the monochromator was blazed
1048
ANALYTICAL CHEMISTRY, VOL. 55, NO. 7, JUNE 1983 Ni (I) 352.5 nm EMISSION
Figure 9. Profile of Ni emission in the plasma source. BACKGROUND EMISSION FOR Ni 352.5 nm
C WAVELENGTH (nm)
Figure 7. Background emission spectra.
PLUME
Flgure 10. Profile of background emlssion in the plasma source.
b-4
SAMPLE MAIN PLASMA
I
,QUARTZ
Y
TUBE
Figure 8. Schematic diagram of visible plasma regions.
at 500 nm, and as the scan moves away from the wavelength at which the grating was blazed, the stray light increases (13). Noise in Background Signal. The noise in the background signal usually limits the detection limit in analytical emission spectrometry. A calculation of the background noise components of the emission signal was obtained for a blank solution using propagation of error techniques as described by Ingle (14). The background signal intensity was determined by making repetitive voltage readings of a blank solution a t 324.7 nm. The time constant was 1.0 s, so readings were taken every 5 s. The standard deviation of the blank signal was 1.3 X V, and the dark current noise was 0.3 X 10" V. The background shot noise calculated according to the methods in ref 14 was 0.8 X V. It is apparent that the major source of noise in the background signal is background flicker, which is usually the case when there is a high background signal like that seen with this three-phase plasma. Plasma Regions. T o make the various regions of the plasma more visible, a 100 mg/L sodium solution was nebulized in the plasma. By visually viewing the plasma through a UV rejection, visible attenuation filter, several distinct regions of the plasma could be seen, and these are schematically shown in Figure 8. A very intense orange color was seen in the central sample region which was due to the sodium emission. The plume region was also orange but not as intense as the center. The plasma region appeared greenish through
the UV rejection, visible attenuation filter. Emission Profile of the Plasma. The profile of the net emission intensity from a 500 mg/L nickel solution, measured at 352.4 nm, vs. the horizontal distance from the center of the plasma and the vertical distance above the electrodes is plotted in Figure 9. The area viewed by the spectrometer was calibrated by backlighting the monochromator and calibrating the micrometers on the mirror mounts, which independently controlled the vertical and horizontal position of the viewing region, as previously described in the Experimental Section. The outer quartz tube limited viewing any lower than 12 mm above the electrodes. As can be seen from the plot of relative emission signal in Fiture 9, the net emission signal is largest in a 2-mm central region of the plasma and closest to the electrodes. This shows that the sample stream is entrained in the center of the plasma (rather than escaping around the outside). A profile of the emission intensity from a blank solution was also plotted, Figure 10. The relative emission intensity scales for Figures 9 and 10 are different. The background emission profile is also largest when viewed through the central region of the plasma but does not decrease as rapidly with height as does the nickel emission. Analytical Curves. Analytical curves were obtained for copper, sodium, and aluminum a t 324.7 nm, 588.9 nm, and 309.2 nm, respectively. These three elements were chosen because they represent three distinct types of elements. Copper is a transition metal and usually works well for analytical emission spectrometry. Sodium is an alkali metal with a low ionization potential. In high-temperature plasmas, sodium has a tendency to ionize, whereas in lower temperature flames sodium does not ionize as much. Aluminum forms a refractory oxide which requires higher temperatures to atomize. The argon and sample flow conditions for obtaining the copper and sodium analytical curves were essentially the same:
ANALYTICAL CHEMISTRY, VOL. 55,
Table 11. Anallytical (Curves
concn range, pointls mg/L
no.
element copper sodium aluminum
4 6 7
1-500 1-500 5-1000
1049
Table 111. Detection Limits (pg/L) % deviation of
of
NO. 7, JUNE 1983
log-log slope 0.98 1.07 0.89
f f f
0.08 0.04 0.06
1000 mg/L intens -15 -40
element copper sodium nickel
h,
nm
strontium aluminum
324.7 589.0 341.4 352.4 257.6 403.1 407.8 309.2
barium
455.4
manganese
TPP 100
5
17
10
17
300 200 300 400 2000
20 5
396.1
a 1L/min sample aerosol flow rate to the plasma and an outer tube argon flow rate of 9 L/min. However, under these same operating conditions, aluminum yielded a very low intensity emission signal even at high concentration levels. By lowering the sample aerosol flow rate to the plasma to 0.5 L/min and lowering the outer tube argon flow rate to 6 L/min, the aluminum emission increased by over an order of magnitude. The lower sample aerosol flow probably enabled a longer residence time for the sample in the pllasma, and the lower argon flow rate through the outer tube probably resulted in a higher plasma temperature, which would increase the atomization efficiency of the aluminum. The lower outer tube argon flow resulted in a less aerodynamically stable plasma, which increased the noise. The emission signals from sodium and copper were also evaluated at these lower argon and sample aerosol flow rates. Copper gave a slightly lower emission ;signal a t these lower flow rates than was obtained a t the higher flow rates. A possible explanation for this is the fact that less sample per unit time is entering the plasma so the total copper population was lower. The sodium (emission signal a t these lower flow rates was much less than at the higher flow rates. This may have been due to increased ionization of the sodium, and the wavelength monitored, 589.0 nm, is not an ion line. An attempt was made to view the sodium ion line at 288.1 nm, but due to the fact it is 3 orders of magnitude less intense than the un-ionized sodium line 589.0 nm in the ICP (1.9, arid the fact that there is an intense tungsten line at 287.9 nm, the sodium ion line a t 288.1 nm could not be seen in the three-phase plasma arc above the bac,kground emission with our low-resolution spectrometer. The slopes of the three log-log analytical curves were calculated by using linear regression from an H P 32E calculator. The error in the slope was calculated by using a standard formula for calculating the error in the slope (16). The results are shown in Table 11. For Cu and Na the 1000 mg/L point was not included in the fit since its intensity deviated significantly from the straight line. A slope of 1.00 indicates that the relationship between the emission signal and concentration is linear and passes through the origin. Within the error of the slope, copper has a slope of one and sodium has a slope close to one. Aluminum has a slope slightly less than one, which could result from selfabsorption or from the fact that the standards were made from salts and at high salt concentration the larger salt particles are inefficiently atomized under these conditions of flow rate, plasma size, and power dissipation. At 1OMmg/L the sodium and copper analytical curves showed negative deviations due to either self-ablsorption or the high salt concentration. The linear dynamic concentration ranges for the copper, sodium, and aluminum are approximately 3 orders of magnitude, if the asisumption is made that the analytical curves are linear down to the detection limits, which will be presented in the next sectim. The linear dynamic concentration range for the three-phase plasma compares favorably with the DCP linear dynamic concentration ranges of 2 to 3 orders of magnitude (17). Concentrations for the three-phase plasma
DCP lit.
ICP
lit.
1 0.2
19 21
6 0.04
20 19
0.3
20
2 0.4
20 19
17 17
500
are 1to 2 orders of magnitude larger than the concentrations used for the DCP linear dynamic concentration range. Consequently, if the detection limits can be improved with the methods indicated below, the linear dynamic range could be pushed toward the 5 to 6 orders of magnitude (18) dynamic range of the ICP. Detection limits for seven elements are given in Table 111, as well as the wavelength used and detection limits reported for the ICP and DCP. The detection limit for our results is given as the concentration that gives a signal equal to twice the standard deviation in the background signal (19). The ICP detection limits chosen for comparison with the three-phase plasma detection limits were the detection limits obtained with an ICP using a Babington-type nebulizer (20). For those elements for which a detection limit was not determined by an ICP with a Babington-type nebulizer, the best available detection limits for an ICP using other pneumatic nebulizers ( I 7,21)were used to compare to the detection limits obtained with the three-phase plasma. The detection limits for the three-phase plasma arc are between a factor of 15 and 400 higher than the detection limits reported for the DCP, and a factor of 33-1200 higher than the detection limits reported for the ICP. There are several possible ways in which the detection limits in the three-phase plasma arc could be improved. First, a spectrometer with improved spectral resolution and stray light rejection would significantly improve the line-to-background ratio. The use of carbon electrodes would reduce spectral interferences and reduce the amount of light entering the monochromator that can become stray light. Flicker noise could be reduced by a more rigid, optimally shaped, and carefully aligned quartz tubing assembly. Other possibilities include stabilizing the three-phase power supply and improving the nebulizer system. The interference of concomitants was not studied. It seems likely that such interferences will be present, just as they were for early ICP’s, until the design has been optimized to avoid them. Advantages that the new three-phase plasma source has over the DCP and ICP include lower initial cost and simplicity of the power supply. No radio frequency noise is generated. The electrodes provide control over the shape of the plasma, and, in particular, the diameter of the central core region in which the sample stream is entrained. The new electrical three-phase argon plasma source (patent applied for) is the only argon plasma, beside the ICP, to have an area in the center of the plasma for effective sample aerosol contact with hot plasma. Registry No. Argon, 7440-37-1.
LITERATURE CITED (1) Kelrs. C. D.; Vlckers, T. J. Appl. Spectrosc. 1977, 37,273. (2) Greenfield, S.;Jones, I. LI.; Berry, C. T. Analyst (London) 1964, 8 9 , 713. Also U S . Patent 3467471, Sept 16, 1969; Brlt. Patent 1109602, 1968; flrst application, Oct 21, 1963. (3) Fassel, V. A. Science 1978, 202, 183. (4) Fassel, V. A. Anal. Chern. 1979, 5 7 , 1290A.
1050
Anal. Chem. lQ83,55, 1050-1054
(5) Barnes, R. M.; Schlelcher, R. G. Spectrochlm. Acta, Part 8 1975, 306, 109. (6) Barnes, R. M., CRC Crlf. Rev. Anal. Chem. 1978, 7, 203. (7) Murdick, V. A., Jr.; Plepmeier, E. H. A"/. Chern. 1974, 4 6 , 678. (8) Willis, C. H.; Chandler, H. M., Jr., Introduction to Electrical Engineering"; D. Van Nostrand: Princeton, NJ, 1957. (9) Babington, R. S. Pop. Scl. 1973, May, 102. U S . Patents 3421 692; 3 421 699;3 429 058;3 425 059;and 3 504 849. (IO) Fry, R. C.; Denton, M. 8. Appl. Spectrosc. 1979, 33, 393. (11) Plasma Line 1980, 1 , 2. (12) Klueppei, R. J.; Coleman, D. M.; Eaton, W. S.;Goldstein, S.A,; Sacks, R. D.; Walters, J. P. Spectrochlm. Acta, Part B 1978, 338, 1. (13) Skogerboe, R. K.; Urasa, I. T.; Coleman, G. N. Anal. Chem. 1976, 30, 500. (14) Ingle, J. D., Jr. "Notes on Basic Spectrometric Measurements", Oregon State University: Cowallis, OR, 1978;p 67.
(15) Winge, R. K.; Peterson, V. J.; Fassel, V. A. Appl. Spectrosc. 1979, 33, 206. (16) Dixon, W. J.; Massey, F. J. "Introduction to Statistical Analysis"; McGraw-Hill: San Francisco, CA, 1969;p 195. (17) Skogerboe, R. K.; Urasa, I.T. Appl. Spectrosc. 1978, 32, 527. (18) Kirkbright, G. F.; Ward, A. F. Talanta 1974, 2 1 , 1145. (19) Vickers, T. J.; Winefordner, J. D. I n "Analytical Emission Spectroscopy"; Grove, E. L., Ed.; Dekker: New York, 1972;Part 11, p 333. (20) Garbarino, J. R.; Taylor, H. E. Appl. Spectrosc. 1981, 35, 153. (21) Fassei, V. A.; Kniseley, R. N. Anal. Chern. 1974, 4 6 , I l l O A .
RECEIVED for review September 7 , 1982. Accepted February 4,1983.
Determination of Trialkyllead Compounds in Water by Extraction and Graphite Furnace Atomic Absorption Spectrometry Walter R. A. De Jonghe, Wllly E. Van Mol, and Fred C. Adams" Department of Chemistry, University of Antwerp (U.I.A.), Universiteitsplein 1, 6-2610 Wilrijk, Belgium
A slmple extraction procedure Is descrlbed for the sensltlve determlnatlon of traces of trlalkyllead compounds in water. After enrlchment of the sample by a fast vacuum dlstlllatlon technique and saturatlon of the resldual volume with sodium chlorlde, the analytes are extracted In chloroform. By Incorporation of specific purlflcatlon steps, Interference from other forms of organic and lnorganlc lead Is completely ellmhated. The flnal chloroform extract is treated wlth a sulfurlc acld solution in order lo transfer the trlalkyllead compounds present back into an aqueous solution. The analysis Is completed by graphlte furnace atomic absorption spectrometry. Under normal laboratory conditions, a detection limlt of 0.02 pg can be achieved wlth I - L samples. The method was developed to investlgate the occurrence of trlalkyllead compounds In envlronmental water samples. Experiments wlth splked lake water have establlshed that there Is a conslderable loss of lnltlally added trlethyllead chlorlde upon exposure to sunllght, while trimethyllead remains falrly stable.
I t is well established now that environmental samples may contain substantial quantities of tetraalkyllead compounds (TAL) originating from their use as antiknock agents in gasoline and possibly from natural alkylation of inorganic lead (1-5). Numerous methods have been developed to detect the occurrence of these contaminants in air, water, sediment, and biological matrices. On the other hand, only a few methods are available for the environmental determination of trialkyllead ions (TriAL), the highly toxic photochemical and metabolic dealkylation products of TAL. Those described include the extraction and spectrophotometric measurement of colored alkyllead complexes (6-10) and direct polarographic techniques (11). Individual TriAL species allegedly have been determined by thin-layer chromatography (12) or by gas chromatography with an element-specific detector (13-15). Other procedures are reviewed in detail elsewhere (16). Although proven valuable in laboratory studies dealing with only one or two species, a number of the techniques adopted
so far are difficult to apply with real samples, in which theoretically four different TriAL compounds can be encountered: trimethyllead (TriML), dimethylethyllead (DMEL), methyldiethyllead (MDEL), and triethyllead (TriEL). Moreover, for a complete analysis, they require pure standards of the intermediate species, DMEL and MDEL, the preparation of which is troublesome (17). The present study describes an improved extraction procedure based on the well-developed technique of salting out the TriAL ions as neutral species into an organic solvent (12, 14,18,19),combined with graphite furnace atomic absorption spectrometry (GFAAS). In contrast to earlier approaches where other forms of organic and inorganic lead may give rise to serious interferences, the proposed determination is highly specific for TriAL even in the presence of up to 100 bg L-l of inorganic lead salts. Stability tests of TriML and TriEL in water are discussed to demonstrate the practical applicability of the method. EXPERIMENTAL S E C T I O N Apparatus. A Perkin-Elmer 503 atomic absorption spectrometer is used, in combination with a PE HGA-74 graphite furnace atomizer. Absorbances are read as peak height from a Hitachi Perkin-Elmer56 strip-chart recorder and from the built-in peak-read device of the spectrometer. The furnace is flushed with argon at a flow rate of 300 mL m i d internally and 900 mL min-' externally. The gas flow can be changed during the atomization stage to increase sensitivity, by means of the stopped (gas-stop), reduced (miniflow),and continuous flow settings. Solutions are injected with Eppendorf micropipets with disposable polypropylene tips. The light source is a lead hollow-cathode lamp, operated at 10 mA. All analyses are done at the 283.3-nm line with a spectral bandwidth of 0.7 nm. The miniflow mode is used in the atomization step, as this provides a suitable compromise between the sensitivity and linearity pursued. For a 2 0 - ~ Lsample injection the optimal temperature/time program is as follows: drying at 100 OC for 30 s, charring at 500 OC for 20 s, atomizing at 2300 "C for 10 s and glowing-out at 2700 OC for 10 s. For the optimization of the extraction procedure, a number of measurements are also performed by means of a Varian 3700 gas chromatograph with flame ionization detection (GC-FID). It
0003-2700/83/0355-1050$01.50/00 1983 American Chemical Society
