688
Anal. Chem. 1983, 55, 688-692
Atomic Fluorescence Spectrometry with Inductively Coupled Plasma as Excitation Source and Atomization Cell M. A. Kosinski, Hlroshi Uchida,' and J. D. Winefordner" Depatfment of Chemistty, Universiw of Florida, Gainesviiie, Florida 326 1 1
A novel approach is presented In whlch two Inductively coupied plasmas (ICPs) are used In an atomlc fluorescence system. One ICP is used as the excltatlon source and the second ICP Is used as the atomlzatlon cell. Emlssion, excltatlon, and fluorescence anaiytlcal curves of growth, and verllcal dlstributlons for zinc atornlc and calcium Ionic fluorescence intenshies are obtained. Llmhs of detection for 16 elements are compared to fluorescence detection llmits using other radiatlon sources. Interelement effects, spectral interferences, and noise sources are also discussed.
The inductively coupled plasma (ICP) has long been used as an effective vaporization, atomization, excitation, and ionization (VAEI) cell for emission spectrometry (1). Many physical parameters of the ICP have been investigated and a review by Barnes (2)contains references covering the various operating principles of the ICP. However, in emission, inherently complex spectra are produced, resulting in the need of a high-resolution monochromator to isolate the analytical lines of interest. Xenon arc lamps (3),electrodeless discharge lamps (EDLs) (4), boosted hollow cathode lamps (HCLs) (5), ICP emission (6),and dye lasers (7)have all been previously used as excitation sources for flame atomic fluorescence spectrometry (AFS). However, in flames, chemical and ionization interferences along with quenching effects of fluorescence by flame gas molecules are a major concern. The use of the ICP as an atomization cell for AFS should be beneficial due to its high volatilization and quantum efficiency (8). The first use of the ICP as an excitation source for flame AFS was reported by Hussein and Nickless (9). Relatively poor detection limits were obtained, though. Since then, improvements in the sample introduction and the plasma stability have resulted in the increasing use of the ICP as an excitation source for flame AFS. Epstein et al. (6) have shown the advantage of the ICP as compared to other AFS sources is its flexibility with respect to the available intense atomic and ionic line radiation. Changing from one element to the next is simply done by aspirating a different solution into the source ICP. Omenetto et al. (10)have examined ICP emission profiles by using the resonance monochromator method. In this method, a low concentration of analyte is aspirated into a flame and increasing concentrations of the analyte are aspirated into the ICP. The resulting fluorescence from the flame is monitored, allowing certain limiting characteristics of the ICP emission profile to be inferred. The properties of high intensity, long-term stability, narrow line width, and freedom from self-reversal all contribute to the ICP being an excellent excitation source for AFS. Also, due to the narrow line width of spectral lines in the ICP, spectral interferences are insignificant. On leave from Industrial Research Institute of Kanagawa Prefecture, Yokohama 236, Japan.
In this work, the utilization of the ICP as an excitation source for atomic and ionic fluorescence in a second ICP was investigated (ICP-ICP-AFS). Emission, excitation and fluorescence analytical curves of growth were obtained and, from these, further information about the line profile of the ICP was revealed. For the conventional emission curve of growth, the emission from the source ICP was plotted vs. concentration. For the excitation curve of growth, a fixed, low concentration of the analyte was aspirated into the atomization ICP, while increasing concentrations were introduced to the source ICP. Thirdly, the fluorescence analytical curve of growth was obtained by aspirating a fiied high concentration (20 mg mL-') into the source ICP while increasing concentrations of the analyte were introduced to the atomization ICP. Vertical distributions of atomic and ionic species densities were obtained and detection limits for 16 elements are compared to other ICP techniques. Further, a few examples of chemical and spectral interferences were examined. Finally, the noise sources limiting the measurements were investigated and suggested improvements are discussed. EXPERIMENTAL SECTION Instrumentation. A block diagram of the experimental system is shown in Figure 1. The source ICP (Model 2500K, Plasma-Therm, Kresson, NJ) utilized a conventional short torch (Model T1.0, Plasma-Therm). The normal operating conditions were 2.0 kW RF power, 15 L m i d plasma support argon flow rate, and 20 psig nebulizing gas pressure. A 20 mg mL-' and@ excitation solution was used. Two exceptions, however, were aluminum and sodium. Aspiration of 20 mg mL-l solutions of aluminum and sodium clogged the pneumatic nebulizer (Model PN5601, Plasma-Therm) after only a few minutes of operation. For these elements, a 10 mg mL-' excitation solution was used and a nebulizer gas pressure of 40 psig was used to avoid clogging. The atomization ICP (Model 1500D,Plasma-Therm, Kresson, NJ) utilized an extended-sleevetorch (Baird Corp., Bedford, MA) and was operated at 0.7 kW for the atomic lines and 1.0 kW for the ionic lines of nonrefractory elements, as well as the atomic and ionic lines of refractory and rare earth species. The only exception in this case was vanadium for which 1.2 kW was used. The observation height in the atomization ICP for the atomic lines of nonrefractory elements was 40 mm above the top of the extended-sleeve torch. This position corresponded to a height of 80 mm above the load coil since the outer sleeve of the torch extends for 40 mm beyond the top of the load coil. The observations height for all other lines was 20 mm above the top of the torch (60 mm above the load coil). The plasma support argon flow rate was 15 L mi&, with no auxiliary argon flow being used, and the nebulizer gas pressure was 35 psig (Model T-230-Ae, J. E. Meinhard Assoc., Santa Ana, CA). The source radiation obtained (10-30 mm above the load coil) by aspirating 20 mg mL-l aqueous solutions of the analyte into the source ICP was chopped at a frequency of 550 Hz (Model 7500, Rofin-Math ASSOC.,Great Neck, NY) and focused onto the atomization ICP through the use of spherical quartz S1-UV grade lenses (50 diameter, 75 mm focal length). The resultant fluorescence was observed at a 90° angle through the use of a 350 mm focal length monochromator (Model EU-700, Heath Co., Benton Harbor, MI; now GCA Corp., McPherson Instruments,
0003-2700/83/0355-0688$0 1.50/0 Q 1983 Amerlcan Chemlcal Society
ANALYTICAL CHEMISTRY, VOL. 55, NO. 4, APRIL 1983
j
Lens
Lens CEpper
/'J
Lens
FH&#-HT Reference
Amplifier
889
Amplifier
\i chromator
Supply
Figure 1. Block diagram of the ICP-ICP-AFS system. z
5 50-
I o3 3xl# Io4 concentration of Zn in source I C P pg mL-1
e .-
a 40-
z
$ 30-
=
Figure 8. Excitation and emission curves of growth for zinc atom at 213.9 nm (RF powers: 2.0 kW source ICP, 0.7 kW atomization ICP): (A) excitation curve of growth, 40 mm observation height above the top of the extended-sleeve torch (80 mm above the load coil), 1 pg mL-' zinc aspirated into atomization ICP (B)emission curve of growth, 20 mm observation height above the load coil (conventional short torch).
20-
-
0
10 20 30 40 50 Helght cbove top of extended sleeve t o r c h , mm
Figure 2. Vertical distributions of fluorescence intensities for (A) &(I) at 213.9 nm, 1 p g mL-l atomization, 20 mg mL source ICP, and (B) Ca(I1) at 393.4 nm, 1 p g mL-' atomization, 20 mg mL source ICP. RF powers: 2.0 kW source ICP, 0.7 kW (zn), and 1.0 kW (Ca) atomization ICP.
Acton, MA). The entrance and exit slit widths were 1mm and the slit height was 10 mm. The photomultiplier tube (Model R928, Hamamatau, Inc., Middlesex, NJ) waa operated at -loo0 V (Pacific Precision Instruments H.V. Supply, Concord, CA). The output was preamplified (Model 427, Keithley Instruments, Inc., Cleveland, OH) and fed to a lock-in amplifier (Dynatrac 391, Ithaco Inc., Ithaca, NY). The resulting signal was fed to a laboratory constructed integrator. The 10-s integrations were displayed on a digital voltmeter (Model 8000A, John Fluke Mfg. co., Inc., Seattle, WA). For the spectral inteference studies, the output of the lock-in was fed to a chart recorder (Model 5000, Fisher Scientific Co., Pittsburgh, PA). Neutral density filters allowed verification of the linearity of the fluorescence and emission photomultiplier response. Chemicals. Stock solutions of 20 mg mL-l were prepared from the pure metal when possible or from reagent grade chemicals, dissolved in the minimum amount of acid, and successively diluted with distilledJdeionized water.
RESULTS AND DISCUSSION Vertical Distributions of Atomic and Ionic Fluoresence Intensities. In Figure 2, vertical distributions are given for zinc atomic and calcium ionic fluorescence intensities. The fluorescence intensity at 5 mm above the top of the extended-sleeve torch was taken to be the same for both elements. However, the signal-to-noise ratio was found to be poor at this height. Consequently, an observation height of 40 mm above the top of the extended-sleeve torch (80 mm above the load coil) was adopted for zinc and was also found to be optimal for the other atomic lines of nonrefractory elements investigated. I t can be seen from Figure 2 that beyond 40 mm above the top of the extended-sleeve torch, the zinc fluorescence intensity drops off rapidly due to quenching by air entrainment in the plasma (8). For ionic, all other lines, a 20 mm observation height above the top of the extended-sleeve torch (60 mm above the load coil) was adopted. Above this height, the fluorescence in-
lo3 3x10~ io4 Concentration of Cc In source I C P p g rnL-' Figure 4. Excitation and emission curves of growth for calcium ion at 393.4 nm (RF powers: 2.0 kW source ICP, 1.0 kW atomization ICP): (A) excitation curve of growth, 20 mm observation height above the top of the extended-sleeve torch (60 mm above the load coil) 1 pg mL-' calcium aspirated into atomization ICP; (B)emission curve of growth, 20 mm observation height above the load coil (conventional short torch).
tensity for these lines became very weak due to the decrease in temperature of the plasma (11) and air entrainment (8). Curves of Growth. In Figures 3 and 4, the emission and excitation curves of growth for zinc and calcium are shown. As previously shown for both elements in ICP-excited flame-AFS (IO),the emission curves of growth for both elements yield a limiting slope of unity at low atom densities in the source ICP and a slope of 0.5 at high atom densities. Also, the excitation curves of growth yield a limiting slope of unity a t low atom densities in the source ICP and a slope approaching zero a t high atom densities. The limiting slopes of Figures 3 and 4 agree well with theory (IO)and show that self-reversal is absent in the ICP under the conditions used in these analysis. In Figure 5, fluorescence analytical curves of growth are given for calcium and zinc; these curves show that at low atom densities in the atomization ICP, a limiting slope of unity is obtained while at high atom densities a slope
690
ANALYTICAL CHEMISTRY, VOL. 55, NO. 4, APRIL 1983
Table I. Limits of Detection
species A1 I Ca I Ca I1 co I c o I1 Cr I Cr I1 cu I Fe I Fe I1 Mn I Mn I1 Mo I Mo I1 Na I Nd I1 Ni I Ni I1 Sm I1 Tb I1 VI v I1 Yb I Yb I1 Zn I Zn I1
limits of detection, ng mL-' laser-excited ICP-flame-AFSC ICP-AFS HCL-ICP-AFSe
ex x / f l x , nm ~
ICP-ICP-AFS
309.278 394.4011396.152 422.673 393.366 240.725 228.616 357.869 205.552 324.754 248.327 259.940 279.482 279.827 257.610 313.259 386.411 202.030 588.995 430.358 232.003 231.604 363.429 350.917 411.178 309.311 390.3261290.882 398.799 369.419 213.856 206.200
8000
1000
60 2 40 300 900 2000 30 1300 100
4
150 9
5 100 1
0.08
11
5
2
10
2 6
1 10 0.3
2 30
1500 12000
100
loob
0.5
N.D.
380 100 30O0Ob N.D. 8000 1000
10
400
90 10000
150 30 6 600
0.5
0.5
a Excitation Xlfluorescence A , if different from excitation wavelength. Not detectable. (6). Reported by Kosinski et al. (11). e Reported by Demers and Allemand ( 1 3 ) .
of -0.5 is obtained. According to theory (IO),the source ICP is acting as a "line" excitation source compared to the absorption profile in the atomization ICP; this occurs even though the higher temperature of the source ICP (2.0 kW RF power, short torch) would seem to cause the Doppler halfwidth to be larger than that of the cooler atomization ICP (1.0 L min-') at 15 mm above the load coil with a conventional short torch in the ICP (12). The signal suppression a t high aluminum concentrations is due to the increase in sample solution viscosity. In combustion flames, the presence of phosphorus in the analyte matrix or flame gases results in the formation of refractory compounds (17). This formation results in a consequent decrease in the fluorescence signal. In Figure 6, the effect of the presence of phosphorus on calcium ionic fluorescence is shown. The intensity was slightly suppressed by the presence of phosphorus, possibly due to the increase in sample solution viscosity (18, 19). Spectral Interferences. One of the advantages in ICPICP-AFS should be relative freedom from spectral interferences due to the narrow bandwidth of spectral lines in the source and atomization ICPs. Here, we investigated spectral overlap in two cases: (i) Zn (213.856 nm) and Cu (213.853 nm); (ii) Co (231.160 nm) and Ni (231,096,231.234nm). The results obtained are summarized in Table 11. Difficulty is encountered in the measurement of zinc in a copper matrix by atomic absorption of ICP atomic emission spectrometry (ICP-AES) since the zinc resonance line a t 213.856 nm is subject to direct interference from the copper nonresonance line at 213.853 nm. This interference has been reported (20) for flame atomic absorption analysis and requires elaborate procedures to remove the copper prior to analysis. Even with an echelle spectrometer (spectral band-pass = 0.003 nm), this line pair has been shown to exhibit an overlap (21). The measurement of a 10 yg mL-l zinc solution in a 5000 pg mL-' copper matrix using a 20 mg mL-' zinc solution for excitation resulted in no significant excitation of copper fluorescence using a monochromator spectral band-pass of 2
Flgure 7. Fluorescence spectra for a 1000 yg mL-' Co-1000 pg mL-' NI solution: RF powers, 2.0 kW source ICP, 0.7 kW atomization I C P observation height, 40 mm above the top of the extended-sleeve torch (80 mm above the load coli); (A) 20 mg mL-' cobalt excitation, (B) 20 mg mL-l nickel excitation. Spectral band-pass was 1 nm.
nm. This is both because the radiation emitted by the source ICP is in the form of sharp lines and because the copper nonresonance transition at 213.853 nm (11203-57949 cm-l) is difficult to excite due to the excited lower level of this transition. Nickel and cobalt are also subject to spectral interference problems in the wavelength region of 231 nm. The radiation emitted from the source ICP while aspirating a 20 mg mL-l cobalt solution was used to excite a solution containing 1000 pg mL-l nickel and 1000 pg mL-' cobalt. On the other hand, the radiation emitted from a 20 mg mL-' nickel solution was also used to excite this same solution. As can be seen from Table 11,no significant nickel fluorescence results when cobalt excitation is used and, conversely, no significant cobalt fluorescence is seen with nickel excitation. Also, Figure 7 shows the simplified fluorescence spectra obtained with a monochromator spectral band-pass of 1 nm. Noise Considerations and Scatter Interferences. In order to determine the principle source of noise in this technique and to see if scattered radiation from the source ICP was present, a noise study was carried out for zinc at 213.9 nm and sodium at 589.0 nm. This study enabled an estimate of noise in the ultraviolet and visible regions of the spectrum to be obtained. The analytical precision (4 s time constant, 10 s integration time, with 16 consecutive readings) for the measurement of high concentration solutions was on the order of 2% for zinc and 10% for sodium and was limited to some extent by the source ICP stability. Other researchers (22) have reported
892
Anal. Chem. 1963, 55, 692-697
ICP emission precision to be limited by the fluctuations in sample transport and nebulization to about 1 % . It is reasonable, therefore, that the precision is slightly worse with two ICPs. The long-term stability of the ICP provides an advantage over other sources used for AFS such as electrodeless discharge lamps, which must be carefully thermostated (23),and the Eimac short-arc xenon lamp (24),which has a lower intensity in the ultraviolet. Also, there was no significant difference in the precision when the measurements for zinc and sodium were carried out in a matrix containing 10 mg mL-l calcium. However, there was an increase in scattered radiation. The precision for the measurement of low concentration solutions was about 3-4 times worse than that for high concentration solutions. The scatter signal due to the calcium matrix at the zinc and sodium lines resulted in a 2-5 times increase in signal over the blank level, respectively. Even though scatter was observed due to the calcium matrix, it did not act as a significant noise source. The precision for the low concentration analysis was about the same with and without the 10 mg mL-’ calcium matrix. Consequently, if present, scatter can be subtracted out by using the two-line technique (201, assuming that the scatter level does not change appreciably in the wavelength vicinity of the atomic fluorescence line. While the fluorescence detection system is well optimized for a background shot-noise limited system, the optical transfer of radiation from the source ICP to the atomization ICP could be improved by at least an order of magnitude through the use of an ellipsoidal reflector (25,261placed behind the plasma to collect a much larger solid angle of emission. Indeed, based upon solid angle considerations, we are presently collecting less than 1 % of the source radiation using 50 mm diameter lenses. Also, a mirror placed behind the atomization ICP in the direction of the fluorescence monochromator might improve detection powers by two times and aspiration of higher concentration excitation solutions could also be used to increase the source intensity in some cases. However, clogging of the nebulizer and aerosol tube of the torch may result as was noticed in this work for aluminum and sodium. The use of demountable torches with large diameter aerosol tubes might improve the situation.
ACKNOWLEDGMENT The authors wish to thank Nicolo Omenetto of the University of Pavia, Pavia, Italy, and Edward Voigtman, Benjamin Smith, and Gary Long (University of Florida) for their helpful discussions and comments during this work. Registry No. SOz, 7446-09-5; BaS04, 7727-43-7;Vz06,131462-1; SOz, 7631-86-9;copper, 7440-50-8. LITERATURE CITED Fassel, V. A.; Knlseley, R. N. Anal. Chem. 1974, 4 6 , 1110A. Barnes, R. M. Crlt. Rev. Anal. Chem. 1978, 7 , 203. Demers, D. R. Appl. Spectrosc. 1968, 22, 797. Mansfield,J. M., Jr.; Bratzel, M. P., Jr.; Norgordon, H. 0.; Knapp, D. 0.;Zacha, K. E.; Wlnefordner, J. D. Spectrochlm. Acta, Part B 1968, 2 3 8 , 389. Lowe, R. M. Spectrochlm. Actci, Part B 1971, 268. 201. Epsteln, M. S.; Nlkdel, S.; Omenetto, N.; Reeves, R.; Bradshaw, J.; Wlnefordner,J. D. Anal. Chem. 1979, 51, 2071. Fraser, L. M.; Wlnefordner, J. D. Anal. Chem. 1971, 4 3 , 1693. Uchlda, H.; Koslnskl, M. A.; Omenetto, N.; Wlnefordner, J. D. Spectrochlm. Acta, In press. Hussein, Ch. A. M.; Nickless, G. Paper presented at the 2nd ICAS, Sheffield, England, 1969. Omenetto, N.; Nlkdel, S.; Bradshaw, J.; Epsteln, J. S.; Reeves, R. D.; Wlnefordner, J. D. Anal. Chem. 1979, 51, 1521. Kosinskl, M. A.; Uchlda, H.; Wlnefordner, J. D. Talants, in press. Uchlda, H. Spectrosc. Lett. 1981, 14, 665. Demers, D. R.; Allemand, C. D. Anal. Chem. 1981, 53, 1915. West, A. C.; Fassel, V. A.; Knlseley, R. N. Anal. Chem. 1973, 45, 1587. Murayama, S. Spectrochlm. Acta, Part B 1970, 2 5 8 , 191. Borowlec, J. A.; Boorn, A. W.; Dlllard, J. N.; Cresser, M. A.; Browner, R. F.; Matteson, M. J. Anal. Chem. 1980, 52,1054. Alkemade, C. Th. J.; Voorhuls, M. H. Fresenlus’ 2. Anal. Chem. 1958, 163, 91. Greenfield, S.; McGeachin, H. M.; Smith, P. B. Anal. Chim. Acta 1987, 8 4 , 76. Uchlda, H.; Matsui, H. Bunko Kenkyu 1978, 2 7 , 110. Larklns, P. L.; Wlllls, J. 8. Spechochlm. Acta, Part B 1974, 2 9 8 , 319. Zander, A. T.; OHaver, T. C.; Kellher, P. H. Anal. Chem. 1977, 4 9 , 638. Boumans, P. W. J. M.; DeBoer, F. J. Spectrochim. Acta, Part 8 1977, 3 2 8 , 365. Browner, R. F.; Batel, B. M.; Glenn, T. H.; Rletta, M. E.; Winefordner, J. D. Spectrosc. Lett. 1972, 5 , 311. Cochran, R. L.; Hleftje, G. M. Anal. Chem. 1977, 4 9 , 2040. Shull, M.; Wlnefordner, J. D. Anal. Chem. 1971, 4 3 , 799. Benettl, P.; Omenetto, N. 0.; Rossl, G. Appl. Spectrosc. 1971, 25, 57.
RECEIVED for review November 5,1982. Accepted December 20,1982. This work was supported by AFOSR-F-49620-80C-0005.
Furnace Atomic Absorption Spectrometry Atomizer with Independent Control of Volatilization and Atomization Conditions Darryl D. Siemer Exxon Nuclear Idaho Co., CPP 602, Idaho Falls, Idaho 83402
A graphite furnace atomizer featurlng Independent control of the temperatures of both an atomization zone and a spatlally separate volatlllzatlon zone was constructed and characterized. Separate optically sensed temperature feedback controlled power supplies were employed to heat both zones. A number of previously documented matrix Interference problems observed In analyses performed with furnaces of conventional construction were found to be greatly reduced or ellmlnated with thls system.
Recent work in this laboratory has been directed toward
the improvement of the routine reliability (“ruggedness”) of trace element determinations in complex samples by graphite furnace atomic absorption spectrometry (GFAAS). These projects have included the modification of furnace power supplies to incorporate temperature feedback control (I);a study of the biases that the “slow” electronic signal processing circuitry commonly found in AAS spectrometers primarily designed for flame AAS impose onto GFAAS signals (2);and, finally, the modification of the atomizers themselves in order to increase the effective gas-phase temperatures experienced by volatile analyte elements “atomized” in several versions of Varian Techtron’s carbon rod atomizer (CRA).
0003-2700/83/0355-0692$01.50/00 1983 American Chemlcal Society
