320
Anal. Chem.
1981, 53, 320-324
ionization process. Equation 20 will then give the atomic concentraton and eq 19 the ion concentration of the element of interest.
LITERATURE CITED (1) (2) (3) (4) (5)
(6) (7) (8) (9)
Li, K. P. Anal. Chem. 1976, 48, 2050. Li, K. P. Anal. Chem. 1977, 49, 2086. Li, K. P. Anal. Chem. 1978, 50, 628. Boss, C. B.; Hieftje, G. M. Anal. Chem. 1979, 57, 895. L’vov, B. V.; Kruglikova, L. P.; Polzik, L. K.; Katskov, D. A. J . Anal. Chem. USSR(€ngl. Transl.) 1975, 30,551. L’vov, 8. V.; Krugiikova, L. P.; Polzik, L. K.; Katskov, D. A. J. Anal. Chem. USSR (Engl. Transl.) 1975, 30, 545. Boss, C. B.; Hieftje, G. M. Anal. Chem. 1974, 49, 2112. Bastiaans, G. J.; Hleftje, G. M. Anal. Chem. 1974, 46, 901. Aikemade, C. Th. J. In “Analytical Flame Spectroscopy”; Mavrodineanu, Ed. McMillan: London, 1970; Chapter 1, p 1.
(10) Kirkbright, G. F.; Sargent, M. In “Comprehensive Analytical Chemistry”; Svehla, Ed.; Elsevier: Amsterdam, 1975; Vol. IV, Chapter 2, p 201. (1 1) L’vov, 8. V.; Katskov, D. A.; Krugiikova, L. P.; Polzik, L. K. Spectrochim. Acta, Part B 1976, 378, 49. (12) Hails, D. J. Spectomchim. Acta, Part B 1977, 328, 221. (13) Jacob, P. W.; RusselJones, A. J . Phys. Chem. 1988, 72,202. (14) Weast, R. C., Ed., “Handbook of Chemistry and Physics”, 56th ed.; CRC Press: Cleveland, OH, 1975-1976. (15) Gurvich, L. V.; Veitz, I.V. Dokl. Akad. Nauk SSSR 1956, 708, 659. (16) Gurvich, L. V.; Veitz, I. V. Izv. Akad. Nauk SSSR, Ser. f l z . 1958, 22, 873. (17) Bulewicz, E. M.; Sugden, T. M. Trans. Faraday Soc. 1959, 55, 720.
RECEIVED for review June 9,1980. Resubmitted September 25, 1980. Accepted November 12, 1980.
Acid Effects in Laser-Enhanced Ionization Spectrometry Terry 0. Trask and Robert
B. Green”
Depattment of Chemistry, Chemistry Building, University of Arkansas, Fayettevllle, Arkansas 7270 1
Mineral acids are often added to samples prior to flame spectrometric determinations. Mineral acids produce enhancement and suppression of analyte slgnals In laser-enhanced loniratlon (LEI) spectrometry. Thls paper characterizes these effects, examines them in detall, and provides explanations for the observed slgnal behavior. The addition of nitric acid to standard and sample solutions is suggested as a means of improving LEI signals while preventing adsorption on container walls durlng storage.
In laser-enhanced ionization (LEI), thermal ionization of an analyte atom in a flame is enhanced by a pulsed dye laser tuned to an absorption transition. The laser-related current pulse is detected with electrodes and monitored with conventional electronics. LEI, a special case of the optogalvanic effect ( I ) , has been directed toward analytical flame spectrometry (2-11). Detection limits which are superior to those obtained with existing flame spectrometric methods have been reported for many metals (3-5). Since the LEI signal depends on the analyte transition probability and ionization potential, many transitions which are not suitable for purely optical techniques have been used successfully (3-5). Two photon (5) and stepwise excitation (6, 7) schemes have produced increased selectivity and sensitivity in some cases. Excited states of lanthanum, yttrium, and scandium oxides have been observed in seeded hydrogen/air flames (8). A four-level model of the atom has been used to describe the LEI signal production mechanism by a combination of optical and collisional processes (9). Recently, atomic resonance line lasers have been applied to LEI spectrometry (10). The potential of LEI spectrometry for trace metal determinations has prompted the investigation of interferences in this laboratory. Previously electrical interferences due to low ionization potential matrices have been examined for rod (11) and plate (12) electrodes. The present work is concerned with the effects of mineral acids on LEI signals. These effects must be accounted for since acids are often added to samples prior to flame determinations. Acid dissolution is integral to the preparation of many real samples. Acids are also routinely
added to standards and samples to prevent adsorption during storage. Often chromatographic pretreatment of samples requires addition of acid solutions for elution of the desired constituents. In addition to the practical aspects, characterization of interferences will contribute to a better understanding of the signal collection process and therefore better techniques for interference reduction.
EXPERIMENTAL SECTION Apparatus. The experimental apparatus has been described previously (11). The excitation source, a CMX-4 linear flashlamp pumped dye laser with frequency doubling capability (Chromatix, Inc., Sunnyvale, CAI, was operated at 10 Hz with rhodamine 6G laser dye (Exciton Chemical Co., Dayton, OH). The flame cell was a commercial atomic absorption premix burner (Model 370, Perkin-Elmer Corp., Norwalk, CT) with a 10-cm slot burner head. Ionization current measuremenb utilized two horizontally parallel cathodes (1mm X 70 mm tungsten rods or 1mm X 10 mm X 70 mm molybdenum plates), 12.5 mm apart and centered 10 mm above the burner head. The laser beam was directed down the center of the flame and between the electrodes. The dual cathodes were maintained at -1000 V (Model 415B HV Power Supply, Fluke Manufacturing Co., Inc., Seattle, WA) with respect to the burner head which was used as the anode. The burner head was insulated from the burner body with Teflon tape so that the current drawn through the flame could be monitored on the low voltage side of the flame. Current pulses corresponding to the enhanced ionization signal were filtered, amplified, displayed on an oscilloscope (Model 1741A, Hewlett-Packard, Go., Palo Alto, CA), and processed with a boxcar signal averager (Model 162 with Model 164 or 165 gated integrator, Princeton Applied Research, Inc., Princeton, NJ) with a 0.5-/.mgate and 1-5 time constant. The oscilloscope and boxcar averager were triggered by a photodiode which monitored the laser pulse. The output of the boxcar averager was read out on a strip chart recorder (Sargent-Welch, Skokie, IL). Reagents. Aqueous standards were prepared from 99.97% indium metal (Matheson Coleman and Bell, Norwood, OH) according to the procedure described in ref 13. Reagent grade hydrochloric, nitric, sulfuric, and phosphoric acids (Fisher Scientific Co., Fairlawn, NJ) as well as high-purity hydrochloric, nitric, and sulfuric acids (Ultrex, J. T. Baker Chemical CO., Phillipsburg, NJ) were used to prepare the acid solutions. Procedure. The procedure is outlined in ref 11. Much of the data are reported in terms of “relative signal”, Le., the LEI signal of the indium standard under experimental conditions compared
0003-2700/81/0353-0320$01.00/00 1981 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 53, NO. 2, FEBRUARY 1981 321 ~~
~~
Table I. Comparison of LEI Signal Enhancements for 100 n g / d Indium in Reagent-Grade and Ultrex Acids % enhancement (in arbitrary units) acid (0.1 M ) reagent grade Ultrex HNO, H2SO4 HCI
48 69 58
48 66 51
log A C I D C O N C E N T R A T I O N
(moles l i t e r )
Effect of acid concentration on the LEI sgnal for 100 ng/mL indium: (A) phosphoric acid, (B) nitric acid, (C) sulfuric acid, (D) hydrochloric acid. Figure 1.
with the LEI signal under standard conditions. The relative signal curves are recorded as percent signal increase or decrease compared to a neutral, aqueous indium standard (100% relative signal). A blank was used for each solution combination to determine the analyte signal. LEI pulse broadening (11) occurred for phosphoric acid solutions only. Thus the position of the boxcar averager gate remained constant, i.e., "single-point analysis", for all experiments except those involved with phosphoric acid solutions. RESULTS AND DISCUSSION Indium signal behavior in mineral acid solutions has been characterized for LEI spectrometry. Figure 1illustrates that the presence of common mineral acids results in significant and varied changes in the LEI signal for 100 ng/mL indium. Although some of the observed interferences have been reported for other flame spectrometric techniques (14,15),other interferences are unique to LEI spectrometry. Practical and theoretical understanding of signal behavior in acid solutions is essential to the exploitation of LEI spectrometry for a variety of atomic determinations. Variations in experimental conditions have significant effects on the LEI signal. Modifying the electrode geometry, applied voltage, and/or position will alter LEI signals from indium in neutral, aqueous solutions due to changes in field strength in the flame (11,12). The signal for indium in acid solutions is equally sensitive to these changes but only relative differences in signal magnitude were observed, Le., additive interferences. Changes in sample matrix and, to a lesser extent, flame composition contribute to an increased ion concentration in the flame. This leads to modified signal collection characteristics due to variations in the effective electrical potential at the excitation site resulting in multiplicative interferences (11, 12). The formation of an ion sheath about the cathodes at high ion concentrations enhances and then suppresses the indium signal. The signal suppression observed with reagent-grade phosphoric acid was due to alkali contamination. This was not a problem for reagent-grade hydrochloric, nitric, or sulfuric acids (see Table I). When these reagent grade acids were
J
1.5
20
25
30
A C E T Y L E N E FLOWRATE I Mer$ m,""lel
Effect of acetylene flow rate on the LEI signal: (A) 100 ng/mL indium, (B) 100 ng/mL indium in 0.1 M hydrochloric acid, (C) 100 ng/mL indium in 0.1 M nitric acid, (D) 100 ng/mL indium in 0.1 M sulfuric acM. Figure 2.
compared with high-purity (Ultrex) acid solutions, negligible differences in LEI signal were recorded. High-purity phosphoric acid was not available for testing but on the basis of the commercial assay, 1.0 M H3P04contains at least 16 g / m L sodium and 1 g/mL potassium. Comparable levels of sodium and potassium have been shown to suppress indium signals in aqueous solution (1I ) . Since reagent-grade hydrochloric, nitric, and sulfuric acids were of acceptable purity, the remaining experiments were confined to these acids. Although similar in appearance, the signal variations illustrated in Figure l were only secondarily related to sheath formation for hydrochloric acid solutions and not a t all for nitric and sulfuric acid solutions. For example, Figure 2 illustrates the effect of changing the acetylene flow on the indium signal for aqueous and acid solutions. Indium signals in hydrochloric, sulfuric, and nitric acid solutions peaked a t the same acetylene flow rate (air flow was maintained a t 15 L/min for all cases). Increasing concentrations of acid also did not shift the maxima and only the relative signal level was changed. Similar behavior was observed in hydrogen (14 L/min)/air (15L/min) and hydrogen (16.4L/min)/nitrous oxide (10.7L/min) flames. Figure 3 illustrates the signal threshold for indium in various acid solutions. At greater than approximately 750 V applied, the curves resembled threshold data for indium in sodium matrices ( I I ) , but the minimum applied voltage for observation of a LEI signal was approximately the same for all three acid solutions. Solutions of higher acid concentration also produced the same applied voltage threshold. Previous work has shown that low ionization potential matrices produce an ion sheath at the cathodes which increases voltage thresholds with increasing con-
322
ANALYTICAL CHEMISTRY, VOL. 53, NO. 2, FEBRUARY 1981
8
iyl
D
t
6
2oo
I
/
C
2
a
z -
-
”4 w 2
W I
0
d
,
-
2
I
1000
500
1500
APPLIED VOLTAGE I-VOLTSI
C 1
Figure 3. Effect of applied voltage on the LEI signal: (A) 100 ng/mL
indium, (B) 100 ng/mL indium in 0.1 M hydrochloric acid, (C)100 ng/ml indium in 0.1 M nitric acid, (D) 100 ng/mL indium in 0.1 M sulfuric acid.
centration (11,12). Therefore, both Figures 2 and 3 suggest that the addition of acids produces enhancement of the LEI signal by other means. As Figure 1indicates, all acids enhanced the indium signal. This signal increase occurred with no increase in noise except for phosphoric acid solutions. The signal enhancement in nitric acid remained constant over a relatively large concentration range. Only a t concentrations less than M did the LEI signal begin to diminish linearly to the signal level for a neutral, aqueous solution of indium. At very high concentrations of nitric acid the signal decreased due to viscosity effects which will be discussed later. The increased signal observed for indium in nitric acid at concentrations up to approximately 3 M was similar to signal enhancements reported in atomic absorption (16, 17) and atomic emission spectrometry (18). The source of the enhancement is burner “memory”. The memory is attributed to condensation of part of the neutral analyte solution on the interior of the burner head. At trace levels, a ieduced steady-state concentration of analyte will still be maintained in the flame even with condensation. When an acid solution not containing the analyte is aspirated into the burner, there is a temporary increase in the analyte concentration in the flame due to leaching of analyte from the burner head. In LEI spectrometry, this temporary increase in the indium concentration was reflected by a transient current (Le., voltage) surge as shown in Figure 4. The voltage spike eventually decayed to a signal level determined by the indium concentration and the acid and its concentration. Figure 5 shows that the indium condensation on the burner head increased with aspiration time. Each data point was obtained by aspirating neutral, aqueous indium for the specified time (shown on the abscissa) and then aspirating 20 mL of a 0.1 M nitric acid solution. The amplitude of the resulting voltage spike was measured on a storage oscilloscope. After the neutral, aqueous indium solution was aspirated for approximately 10 min, the voltage transient due to subsequent acid aspiration began to saturate the current preamplifier. Since the detection of the ionized analyte in LEI spectrometry is presumably due to reduction of cations at the cathode surface, leaching of residual indium from the cathodes
2
3
4
TIME (microseconds)
Figure 4. Comparison of LEI signals: (A) signal for neutral, aqueous
100 ng/mL indium, (B) transient signal increase when 0.1 M nitric acid was aspirated following the indium solution. This is a transposition of photographs of two separate oscilloscope traces.
C
Ol
_ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ _ _ _-_- - - - - - - - - 2.6
7.6 ASPIRATION TIME
12.5
(MINUTES)
Flgure 5. Effect of aspiration time on the LEI signal for 100 ng/mL
indium: (A) neutral, aqueous solution signal; (B) transient signal, each data point represents a single experiment in which the indium solution was aspirated for the indicated time followed by a 0.1 M nitric acid aliquot.
and reexcitation was also a potential source of the observed voltage spikes. Solutions (10 wg/mL) of indium or nickel were aspirated for up to 10 min with previously cleaned LEI
ANALYTICAL CHEMISTRY, VOL. 53, NO. 2, FEBRUARY 1981
'"&le 11. 0.1 M Nitric Acid Enhancement as a Function of Indium Concentration % signal [indium], Mg/mL enhancement 1.00 0.75 0.50 0.25 0.10 0.05
10 9 30 53 187 496
cathodes held at -1000 V. Voltammetry of these cathodes showed no oxidation waves for indium or nickel, respectively, which would knave indicated their deposition from the flame. The same experiment was repeated for several sets of cathodes. The cathodes were then rinsed with 0.05 M hydrochloric acid, and the resulting solution was evaporated to reduce its volume. Atomic absorption spectrometry of these solutions produced no indium or nickel signals; LEI spectrometry of the same solutions also yielded no signals. There are several possible explanations for the absence of the analyte plated on the cathodes. The concentration may be too low for determination by these techniques. The analyte may be a minority charge carrier or any electrodeposited metal may flake off prior to determination. In experiments where the electrodes were removed from proximity with the flame during aspiration of a neutral, indium solution and then replaced for aspiration of an acid solution, identical voltage transients were observed discounting the importance of analyte leaching from the cathodes. The observation of voltage surges and enhancements were not unique for indium and have been recorded for sodium and nickel as well. This is expected since acid enhancement of the LEI signal, similar to that for atomic absorption spectrometry should not be analyte specific. Table I1 shows that the relative importance of the signal enhancement diminished as the indium concentration was increased. Therefore, a more sensitive flame technique will be more susceptible to this interference. The signal behavior for indium in nitric acid solution can almost certainly be explained by condensation of the analyte in the burner, but other factors contributed to the behavior observed for hydrochloric and sulfuric acid solutions. Barring contamination or anion effects, all LEI signal vs. acid concentration curves should resemble those obtained for nitric acid solutions. With increasing acid concentration, there should be a linear rise in signal until the maximum enhancement is reached, a flat response over a concentration range determined by the viscosity of the acid, and a gradual decrease in signal when viscosity effects become important. The behavior of sulfuric acid solutions can be characterized similarly to nitric acid except that the concentration range which gave a constant signal was very small. In previous studies of sulfuric acid effects on indium absorption, signal depression was observed by Popham and Schrenk (14). No signal enhancement was reported but the indium concentrations were 500 times greater than the concentration used in the present study. This again emphasizes the greater relative importance of acid effects at trace concentrations of analyte. The suppression of the signal may be attributed to lowering of the solution aspiration rate due to viscosity effects as the sulfuric acid concentration was increased. The onset of viscosity effects occurred a t a much lower sulfuric acid concentration because of its much higher viscosity. The aspiration rate was held constant for these experiments. The signal could be improved by increasing the aspiration rate, but since more indium was leached from the burner, the concentration range for a constant signal did not increase.
(In,
323
Table 111. Comparison of 100 ng/mL Indium Signals in Several Flames re' for neutral
flame HJair H,/N,O C,H,/air
% enhancement
solutions HNO,
H,SO,
HCI
197 197 172
249 272 227
177 185 163
100 270 300
The hydrochloric acid interference was similar in appearance to sulfuric acid, but since its viscosity is approximately the same as nitric acid, other factors must be considered. In atomic absorption spectrometry, decreases in indium absorption in the presence of hydrochloric acid have been attributed to the formation of diatomic indium chloride in the flame (19,20). Only small reductions in the atomic indium population were observed for higher indium and acid concentrations in previous studies. At the lower indium and acid concentrations in the present work, such small changes would be obscured by the acid enhancement due to the removal of condensed analyte in the burner. The onset of signal suppression at lower hydrochloric acid concentrations than for nitric acid solutions is most likely due to alteration of the flame's electrical environment. The presence of hydrochloric acid provides a large electron sink through a two-body dissociative attachment reaction. HC1+ e- Cl- + H (21). The LEI signal behavior observed for hydrochloric acid solutions (Figure 1)was, in fact, similar to the change in flame conductivity with the addition of chlorine to alkali-laden flames reported by Padley et al. (22). In dc measurements in this laboratory, the ion current in the flame initially increased with increasing hydrochloric acid concentration and then decreased apparently due to electron attachment. When the analyte is in a low ionization potential matrix, both enhancement and suppression of the signal have been observed (11, 12). In acid solutions of indium, the relative degree of enhancement due to a sodium matrix was lowered but the limiting sodium concentration for observation of an LEI signal remained essentially unchanged. The relative enhancement was lower for acid solutions while the absolute enhancement increased because a higher concentration of both indium and sodium should be leached from the burner into the flame. When the laser was tuned to the sodium line at 589.0 nm, no difference was observed in the LEI signal for aqueous or acid solutions. This indicated that the sodium concentration in the flame at this level (approximately lox greater than the indium concentration) was essentially unaffected by additional sodium leached from the burner head. This was also suggested by the nearly identical sodium matrix concentration ranges in aqueous, nitric, and hydrochloric acid solutions. Voltage thresholds were the same for all solutions at the same sodium concentration indicating little or no modification of the cathode sheath. Table I11 compares indium signals in 0.1 M acid solutions in three different flames. The lower temperature hydrogen/& flame produced indium signals in neutral solutions approximately 60% lower than either the acetylene/air or hydrogen/nitrous oxide flames. This large difference may be attributed to a lower atom fraction in the hydrogen/air flame. The acetylene/air flame yielded a slightly higher signal than the hydrogen/nitrous oxide flame not only because of a higher atom fraction and a higher atomization efficiency (23)but also because the latter flame required a greater cathode separation (12). The increased cathode separation is necessary due to the increased width of the flame. Table I11 also reveals that hydrogen-based flames yielded larger signal enhancements for acidic solutions than the acetylene flame. Apparently natural flame ions produced by hydrocarbon combustion re-
-
324
Anal. Chem. 1981, 53, 324-330
duced the augmentation of the LEI signal by acidic solutions due to sheath formation around the cathodes. Both cylindrical rods (11)and plates (12)have been used for cathodes in LEI spectrometry with the latter preferred for their higher resistance to electrical interferences. Acid concentration curves for both rods and plates were similar as would be expected due to the nature of the acid effects. The analytical calibration curves for acidic solutions of indium were linear and parallel to the curve for indium in a neutral aqueous solution. Enhancement for a particular acid concentration results in the extrapolated limit of detection being lower than for a neutral, aqueous solution. There will be a deviation from linearity for hydrochloric and sulfuric acid solutions if the acid concentration does not remain constant among a group of samples and/or standards. The maintenance of the nitric acid concentration within a relatively wide concentration range in a series of samples permits the generation of linear analytical curves. Since nitric acid solutions yield a constant signal over 4 orders of magnitude, precise measurement of concentration (or volume) is not crucial. Nitric acid should be added to all solutions for LEI determinations to improve the signal, with no increase in noise, reduce measurement time, and prevent possible damage to electronics by high-voltage transients. Other acids may also be used for LEI spectrometry with the recognition that their concentration is more critical within a group of samples and/or standards. With the exception of phosphoric acid, reagentgrade acids which are much less expensive than ultrapure acids are adequate for analytical determinations by LEI spectrometry. This work has also confirmed the necessity of acidifying solutions stored in glass or plastic containers to pH -1 to prevent variations in analyte concentration due to adsorption on container walls (24).
LITERATURE CITED (1) Green, R. B.; Kelier, R. A,; Schenck, P. K.; Travis, J. C.; Luther, G. C. Appl. Phys. Lett. 1978, 29, 727-729.
(2) Green, R. B.; Keller, R. A.; Schenck, P. K.: Travis, J. C.; Luther, G. C. J. Am. Chem. Soc. 1976, 98, 8517-8518. (3) Turk, G. C.;Travis, J. C.; DeVoe, J. R.; O'Haver, T. C. Anal. Chem. 1978, 50, 817-820. (4) Travis, J. C.;Turk, 0. C.; Green, R. B. ACS Symp. Ser. 1978, No. 85,91-101. (5) Turk, 0. C.;Travis, J. C.:DeVoe. J. R.; O'Haver, T. C. Anal. Chem. 1979,57, 1890-1896. (6) Turk, G.C.;Mallard, W. G.; Schenck, P. K.; Smylh, K. C. Anal. Chem. 1979 .- . - , 51 - . , 2408-2410 - . - - - . . -. (7) Gonchakov, A. S.;Zorov, N. B.; Uuzyakov, Yu. Ya.; Maveev, 0. I. Anal. Left. 1979, 12, 1037-1048. (8) Schenck, P. K.; Mallard, W. G.; Travis, J. C.; Smyth, K. C. J. Chem. Phys. 1978, 69, 5147-5150. (9) . . Travis. J. C.:Schenck. P. K.: Turk, G. C.; Mallard. W. C. Anal. Chem. 1979,51, 1516-1520. (10) Ehrlich, D. J.; Osgood,R. M. Jr.; Turk, G. C.; Travis, J. C. Anal. Chem. 1980,52, 1354-1356. (11) Green, R. B.; Havrilla, G. J.; Trask, T. 0. Appl. Spectrosc. 1980, 3 4 , 561-569 ... ...
(12) Havrllla, G. J.; Green, R. B. Anal. Chem. 1980, 52, 2378-2383. (13) Smith, 8. W.; Parsons, M. L. J. Chem. Educ. 1973, 50, 679-681. (14) Popham, R. E.; Schrenk, W. G. Spectrochim. Acta, Part 6 1969, 248, 223-233. (15) Nakahara, T.; Musha, S. Anal. Chim. Acta 1975, 80, 47-59. (16) Dong, A. E. Appl. Spectrosc. 1973, 2 7 , 124-128. (17) Brachaczek, W. W.; Butler; Pierson, W. R. Appl. Spectrosc. 1974, 28, 585-587. (18) Koirtyohann, S.R.; Pickett, E. E. Anal. Chem. 1988, 40, 2068-2070. (le) Haraguchi, H.; Fuwa, K. Spectrochlm. Acta, Part 6 1975, 308, 535-545. (20) Haraguchi, H.; Furuta, N.; Yoshimura, E.; Fuwa, K. Anal. Chem. 1978, 46, 2066-2069. (21) Miller, W. J.; Gould, R. K. J . Chem. Phys. 1978, 68, 3542-3547. (22) Padlev, P. J.; Page, F. M.: Suden, T. M. Trans. Faraday SOC.1981, 57, 1552-1562.(23) DeGalan, L.; Samaey, G. F. Spectrochim. Acta, Part 6 1970, 256, 245-259. .. -. .. (24) Robertson, D. E. Anal. Chem. 1988, 40, 1067-1072.
RECEIVED for review October 3, 1980. Accepted November 17,1980. This research was supported by the National Science Foundation under Grant No. CHE 79-18626. This work was presented in part at the 179th National Meeting of the American Chemical Society, Houston, TX, March 1980, and the 22nd Rocky Mountain Conference on Analytical Chemistry, Denver, CO, Aug 1980.
Time Shifts and Double Peaks for Lead Caused by Chemisorbed Oxygen in Electrothermally Heated Graphite Atomizers S. G. Salmon, R. H. Davis, Jr., and J. A. Hoicombe' Department of Chemistry, The University of Texas at Austin, Austin, Texas 787 12
A mechanism for Pb vaporization from a graphite atomizer surface, that has been aitered by chemisorbed 02,is presented. Two major types of active sites on graphite provide different mechanisms for PbO reduction and Pb vaporizatlon. Deactivation of these sites by the chemisorption of O2 causes a shift to the secondary release mechanism. The two release mechanisms account for the double peaks and appearance temperature shifts which are observed for pb. Volatile metals show this effect because their vaporization temperatures lie between optimum O2 adsorption at 500 OC and total desorp tion at 950 'C. The proposed mechanism may also explain the function of some matrix modifiers and metal carbide coated graphite atomizers.
Environmental and toxicological concerns have made P b one of the more widely determined elements by using atomic 0003-2700/81/0353-0324$01 .OO/O
absorption spectroscopy. Flameless atomic absorption with an electrothermally heated graphite atomizer gives a much higher sensitivity for Pb but often suffers from the appearance of double peaks and time-shifted signals. Double peaks for P b and other relatively volatile metals, such as Cd and Zn, have been reported for a variety of sample types and atomizer designs (1-7). The nonreproducible appearance of a double peak can be the source of a large analytical error if the height of the atomic absorption peak is measured. Integration of the atomic signal may reduce the analytical impact of changing peak shape but can also result in the loss of sensitivity in some atomizers. This is due to the fact that integration is better suited to an isothermal environment and long residence times, Le., >1s. These conditions are not met by some commercial atomizers, especially for the volatile metals (49).In addition, timc-shifted signals resulting from analyte release at different temperatures still may produce a measurable analytical error. Elimination of the double peak or time-shifted signal may be 0 1981 American Chemlcal Society
