444
Anal. Chem. 1901, 53, 444-450
(15) Yen Bower, E. L.; Winefordner, J. D. Anal. Chim. Acta 1978, 702,1. (16) Kalyanasundaram, K.; G-ieser, F.; Thomas, J. K. Chem. Phys. Len. 1977, 57, 501-505. (17) Habarta, J. G.; Cline Love, L. J. 30th Pittsburgh Conference on Analyilcal Chemistry and Applied Sectroscopy, Cleveland, OH, 1979; Abstract No. 601; ABC Press: Monroevllle, PA. (18) Khuanga, U.; McDonald, R.; Sellnger, B. K. 2. Phys. Chem. (Wesbaden) 1976, 101, 209-223; Chem. Abstr. 1976, 85, 1987582. (19) McGlynn, S. P.; Daigre, J.; Smith, F. J. J . Chem. Phys. 1963, 39, 675-679. (20) Vo-Dinh, T.; Yen, E. L.; Winefordner, J. D. Talenta 1977, 24, 146. (21) Vo-Dinh, T.; Hooyman, J. R. Anal. Chem. 1979, 57, 1915. (22) Kalyanasundaram, K.; Thomas, J. K. J . Am. Chem. Soc. 1977, 99, 2039-2044.
(23) Dorrance, R. C.; Hunter, T. F. J. Chem. Soc.,Faradiy Trans. 7 1977, 73, 1891-1899. (24) Thomas, J. K. Acc. Chem. Res. 1977, 70, 133-138. (25) Nakajima, A. Bull. Chem. Soc. Jpn. 1974, 4 4 , 3272. (26) Ham, J. S. J . Chem. Phys. 1953, 21, 756. (27) Cardinal. J. R.; Mukerjee, P. J . Phys. Chem. 1978, 82, 1614-1620. (28) Cardinal, J. R.; Mukerjee. P. J . Phys. Chem. 1978, 82, 1620-1627.
RECEIVED for review May 27, 1980. Accepted December
1,
1980. This work was presented in p& at the 31st Pithburgh Conference On Chemistry and Spectroscopy, Atlantic City, NJ, March 14, 1980; Abstract No. 820.
Matrix Interferences in Graphite Furnace Atomic Absorption Spectrometry by Capacitive Discharge Heating C. L. Chakrabartl," C. C. Wan, H. A. Hamed, and P. C. Bertels Department of Chemistry, Carleton University, Ottawa, Ontario, Canada, K1S 586
An analytical technique has been developed that employs an anisotropic pyrolytic graphite tube atomizer which is heated at very high heating rates (up to 100 K ms-') by capacitive discharge to produce high temperature (up to 3300 K) and an isothermal conditlon. Synthetic samples of saline water were analyzed by capacitive discharge technique and also by the conventional graphite furnace atomic absorption spectrometry (AAS), using for the latter the Perkin-Elmer HGA 76B. Recoveries by the capacitive discharge technique and the conventional graphite furnace AAS were typically about 100 % with the former technique and 12-75 % with the latter technique. There is also another very significant diff e r e n c e l h e background corrector was not required nor used with capacltlve discharge technique, whereas the background corrector was required and used with the conventional graphite furnace AAS. Since the sensitivities of capacitlve discharge technlques are independent of the matrix, calibration curves are neither necessary nor used for analysis. The sensitivity constant is evaluated for a given analysis line and for given experlmentai condltions using a single-point calibration with a standard of the anaiyte salt in ultrapure water. The mass of analyte M in unknown samples is evaluated from its AWakand the sensitivity constant.
Numerous workers (1-13) have reported matrix interferences in graphite furnace atomic absorption spectrometry (GFAAS). Slavin et al. (14-16) have reported various ways of reducing interferences especially with lead. Commercial electrothermal atomizers which are used in graphite furnace atomic absorption spectrometry have two serious limitations-their low heating rates limit them to much lower sensitivity than is achievable at much higher heating rates and they are nonisothermal (spatially and temporally). The result of the latter is that samples which are vaporized from hot central parts condense at the cold ends of the graphite tube resulting in much lower residence times, hence, lower sensitivities; also, the condensed analytes are revaporized in the next analysis producing memory effects and erroneous results (17). Also, their slow heating rates, nonisothermal conditions, and the much lower vapor temperatures into which the samples are vaporized cause severe matrix interferences 0003-2700/81/0353-0444%01.OO/O
of various kinds: spectral, chemical, and physical interferences. The effects of heating rates in graphite furnace atomic absorption spectrometry have been reported in earlier publications (18, 19).
CAPACITIVE DISCHARGE TECHNIQUE Cresser and Mullins (20) have suggested the use of a large electrolytic capacitor for rapid heating of metal filament atomizers. A practical device for capacitive discharge heating has been patented (21). L'vov (22) has reported the use of a capacitor bank as a source of electrothermal energy for heating graphite atomizers. Capacitive discharge technique has enhanced the sensitivity of GFAAS (23) and has made it relatively free from matrix interferences (24). This relative freedom from matrix interferences has been accomplished with an anisotropic pyrolytic graphite tube atomizer which is heated at very high heating rates (up to 100 K ms-') by capacitive discharge producing isothermal conditions and high temperatures (24). Conyentional instrumental analytical techniques require careful calibration of the instrument with chemically analyzed standards or synthetic standards of known composition. When analyses of miscellaneous materials are required, the task of providing the required range of standards becomes insurmountable and instrumental techniques then lose their accuracy, since accurate analyses generally necessitate the use of standards which are closely similar in composition to the sample for analysis. The reason for such close matching of standards and samples in their composition is that the sensitivity of conventional instrumental techniques depends on the sample composition, often in a complicted way. The necessity of preparing standards of the same composition as that of the unknown samples involves prior knowledge of the sample composition and chemical treatment or modification of the standards and/or samples-the latter exposes the samples to the risk of contamination and/or loss; the consequence is questionable results, especially in trace and ultratrace analysis. The new technique described in this paper employs electrothermal heating of an anisotropic pyrolytic tube atomizer in atomic absorption spectrometry producing very fast rates of heating (up to 100 K ms-') and an isothermal condition, both spatial and temporal (23). This technique dispenses with the calibration curve and yields analytical results directly from the absorbance of the unknown sample 0 1981 American Chemical Soclety
ANALYTICAL CHEMISTRY, VOL. 53, NO. 3, MARCH 1981
Table I. Experimental Conditions and the Precision of the Proportionality Constant for Equation 1 atomizaanalysis tion proportionality day line, temp, constant of % no. element nm K eq 1 RSD" 4.3 2670 5.85 x 10" 1 Pb 283.3 2670 7.43 x 10" 6.6 2 Pb 283.3 7.6 2820 7.90 x lo1* 1 Mn 279.5 2820 8.89 x 10'' 4.0 2 Mn 403.1 2.5 2820 1.21 x 10" 3 Mn 403.1 2970 1.65 x 10" 3.0 1 Ni 232.0 3.5 2300 2.30 X 10'' 1 Cd 228.8 2300 9.90 X l O I 3 7.6 2 Cd 228.8 " Three different concentrations-each concentration was determined in triplicate, making a total of nine determinations. The value represents percent relative standard deviation of the mean of three determinations. and the use of a simple equation relating the absorbance to the mass of the analyte. By simplifying and shortening the analytical procedure, this technique often eliminates the sample preparation stage with its attendant risk of contamination and/or loss of analytes and gives accurate and precise results at a fraction of the time and cost of the conventional graphite furnace atomic absorption spectrometry. Equation 1 is a modified version of that given by L'vov (W),
where AWakis the peak absorbance, C1is a coefficient determined by atomic and spectroscopic constants and by experimental conditions, T i s the absolute temperature, S is the cross-sectional area of the graphite tube, P is the pressure inside the graphite tube, and C1' C,('P,7/SP).Equation 1 is used for analysis as follows. The proportionality constant C1'is evaluated for a given analysis line and for given experimental conditions using a standard made of the analyte (salt) dissolved in ultrapure water. The mass of analyte M in an unknown sample is evaluated from its AW&value and the proportionality constant C1'of eq 1.
EXPERIMENTAL SECTION Apparatus. The details of the apparatus and accessories and their operation have been reported in an earlier paper (23). Reagents. Stuck solutions of Pb, Cd, and Mn were prepared separately by dissolving the appropriate mass of pure metals (199.9% pure) in pure nitric acid and the solutions made up to 1000 pg/mL with ultrapure water of resistivity 18.3 MR cm obtained directly from a Milli-Q2 water purification system (Millipore Corp, Mississauga, Ontario, Canada). Stock solutions of NaCl, MgCl,, and CaCl, were prepared separately by dissolving the appropriate mass of the pure salts (ACS reagent grade purity) in ultrapure water containing 0.1 % (v/v) HCl. These solutions were then diluted with ultrapure water to make the solutions contain 5% (w/v) of the salts. All test solutions were prepared by serial dilution of the above stock solutions with ultrapure water immediately prior to determination. Solution Sampling Procedure. Sample solutions (5.0 X lo4 dm3) are injected into the atomizer by means of an Eppendorf syringe fitted with disposable plastic tips.
Solid Sampling Procedure. A solid sample weighing from 0.2 to 2 mg is delivered to the central part of the graphite tube
and the loaded graphite tube is then reweighed to determine the weight of the solid sample. The sequence of steps making up the heating program is as follows: The sample is dried at 370 K for 10 s. During this time, the capacitor bank is charged to the desired voltage. The capacitor bank and the atomization stage of the temperature control unit are fired sequentially, with a minimum resolution of 1 ps. The atomizer temperature is set by adjusting the voltage of the temperature control unit. The electrical energy from the capacitor bank is used for extremely fast heating of the atomizer to its maximum temperature; the atomization stage of the temperature control unit is used to maintain the temperature of the atomizer constant at some preselected value. Absorption pulses are recorded with a Model 549 storage oscilloscope (Techtronix,Inc. Portland, OR) fitted with a Type. 1A7A high-gain differential plug-in. The signal trace is photographed with a Polaroid camera (Tektronix, Inc.). In the case of elements which were present in the materials in relatively high concentrations, their less sensitive analysis lines were employed. By use of analysis line of the desired sensitivity and optimization of the operating conditions,the proportionality constant of eq 1 was evaluated as follows. Three standard solutions of the analyte of different concentrations were employed. Each of the standard solutions was analyzed in triplicate, and the arithmetic mean of these triplicate determinations for each of the three solutions was averaged to provide the value for the proportionality constant. The concentrations of the unknown samples were determined from their absorbance values obtained with the above conditions and suitable mass of solid samples or 5.0 X lo4 dm3 volume of the dissolved (aqueous solution) samples as the cases might be and solving eq 1 for the unknown mass of the analyte element.
RESULTS AND DISCUSSION Table I presents the experimental conditions used and, also, the precision of the proportionality constant for eq 1. It can be seen from Table I that precision of the mean varied from 2.5 to 7.6% RSD. These relatively low values of the percent relative standard deviation are a measure of the validity of the proportionality expressed by eq 1. Since Table I has established the constancy (within i 7 . 6 % RSD) of the proportionality constant of eq 1, considerable savings in time can be achieved by using a single standard in place of three standards. Further savings in time can be achieved by running a single standard and each sample in duplicate in place of a larger number of replicates. In act& analysis of samples, such economy of measures will mean much faster turn-around of samples without any appreciable loss of accuracy. Tables 11-VI present the results of analysis of various materials. Table V shows a comparison of the analytical results obtained by employing the new technique with the results of the conventional GFAAS, using the Perkin-Elmer heated graphite atomizer HGA 76B. Optimized experimental conditions were used in both cases. Tables V and VI show that recoveries of 100 f 12% are generally obtainable. Table V also shows that the results of the conventional GFAAS suffer from extremely severe depression by the matrix, whereas the matrix interferences have been almost completely eliminated by the new technique. The significance of these results is that matrix interferences are mostly eliminated by this technique and the
Table 11. Recoveries, Solution Sampling, United States Geological Survey Marine Mud, MAG-1 element Pb
445
analysis line, nm
most probable value, kdkg
recovered value," kg/kg
recoveries,b
283.3 232.0 403.1
(2.04 k 0.13) x lo-' (7.02 t 0.66) x lo-' (1.02 k 0.04) x 10-3
(2.01 k 0.13) x 10-5 (6.97 f 0.23) x lo-' (0.98 k 0.21) x 10-3
98.5 99.3 95.9
Ni Mn " The * values represent one standard deviation of five successive replicate determinations. successive replicate determinations.
%
Arithmetic means of five
446
ANALYTICAL CHEMISTRY, VOL. 53, NO. 3, MARCH 1981
Table 111. Recoveries, Solid Sampling, National Bureau of Standards Oyster Tissue, SRM 1566 analysis line, certified value, recovered value: element nm kglkg kglkg
recoveries,b %
(4.8 i 0.3) x (4.7 i 0.10)x 10-7 98.5 (3.2 f 0.4) x (3.4 f 0.22) x 10-6 105 (1.75 f 0.06) x lo-' (1.72 f 0.06) X 98.3 " The c values represent one standard deviation of five successive replicate determinations. bArithmetic means of five successive replicate determinations. Pb Cd Mn
283.3 326.1 403.1
Table IV. Recoveries, Solid Sampling, National Bureau of Standards Bovine Liver, SRM 1577 analysis line, certified value, recovered value: element nm kglkg kglkg Pb Mn Fe
283.3 403.1 305.9
recoveries,b %
(3.5 * 0.15) x 10-7 (1.10i 0.04) x 10-5 (2.73 f 0.09) x 10-4
(3.4 f 0.8) x 10-7 (1.03 + 0.10) x 10-5 (2.68 i 0.08) x 10-4
a The i values represent one standard deviation of five successive replicate determinations. successive replicate determinations.
102 107 102 Arithmetic means of five
Table V. Recoveries, Solution Sampling, Synthetic Aqueous Samples
element Cd
analysis line, nm 228.8
mass of analyte, kg 1.0x 1 0 4
Pb
217.0
5.0 x 10-13
Mn
279.5
2.0 x 10-13
matrix in excess over the analyte
recoveries expressed as % of the amount of analyte taken conventional technique with new HGA 76Ba techniqueb
NaCl 700000-fold + MgCl, 84000-fold + CaCl, 28000-fold NaCl 30000-fold + MgCl, 3600-fold + CaC1, 1200-fold NaCl 25000-fold + MgC1, 3000-fold t CaCl, 1000-fold
37
100
12
100
75
100
a HGA 76B is heated graphite atomizer 76B used with an atomic absorption spectrophotometer, Model 603 (The PerkinElmer Corp., Norwalk, CT). The relative standard deviation of the mean of five replicate determinations by the conventional graphite furnace atomic absorption spectrometry using HGA 76B = f 8%. The relative standard deviation of the mean of five replicate determinations by the capacitive discharge technique = i 12%.
Table VI. Recoveries by the Capacitive Discharge Technique, Solution Sampling, Synthetic Aqueous Samples recoveries expressed as % of the amount matrix in excess over analysis line, nm mass of analyte taken the analyte of analyte taken" Cd 228.8
Mn 279.5
Pb 283.3
a
Cd 2.0 x kg taken as nitrate kg Cd 1.0 x taken as nitrate Mn 1.5 x kg taken as nitrate kg Mn 1.5 X taken as nitrate Pb 2.0 x l O - I 3 kg taken as nitrate Pb 2.0 x kg taken as nitrate
ultrapure water
104
ultrapure water + NaCl 600000-fold + MgCl, 72000-fold + CaCl, 24000-fold ultrapure water
105
ultrapure water + NaCl 100000-fold + MgC1, 12000-fold + CaC1, 4000-fold ultrapure water
ultrapure water + NaCl 25000-fold + MgC1, 3000-fold + CaC1, 1000-fold The relative standard deviation of the mean of five replicate determinations = i 12%.
sensitivity is independent of the matrix. This is of great practical and theoretical importance since analysis by GFAAS has been all along characterized by matrix interferences. There is also another very significant differencebackground correction was not required nor employed with the new
98.2 98.5
105 100
technique, whereas background correction was required and was employed with the conventional GFAAS. Since analysb of complex materials (e.g., seawater) by the conventional GFAAS usually requires the removal of interfering matrices (including those that give extremely intense background ab-
ANALYTICAL CHEMISTRY, VOL. 53. NO. 3, MARCH 1981 447
sorption) from the sample prior to the determination of the analy?es. background correction represents a serious limiting factor in the speed and accuracy of analysis by the conventional GFAAS. Even removal of the bulk of the interfering matrices in the charring (pyrolysis) stage of the heating cycle of the conventional GFAAS by employing the time-resolved selective volatilization and atomization technique (26, 27) requires precise determination of the heating program that is neceseary for removing the bulk of the matrix without any loss of the analytea,and correction for the residual background absorption by the background correction technique. Since background absorption and other matrix interferences are mostly eliminated by this new technique, large economy of operation results from greatly simplified and shortened analytical procedures of this technique permitted by the absence of matrix interferences including background absorption. However, this technique does not eliminate interferences from atomic lines which may overlap the analysis l i n e t h i s is also a spectral interference but of infrequent occurrence. Elimination of the background absorption by the new technique results in better accuracy. With use of solid-samplingtechnique in conjunction with the new technique, -100% recoveries were obtained. Solid-sampling technique also eliminates any potential problem of soaking of the sample solutions into the graphite tube and the effect of such soaking on the peak absorbance values. The conventional GFAAS with solid-sampling technique is even more subject to interferences than solution-sampling technique. Solid-sampling technique, applied in conjunction with the new technique, eliminates the enormous dilution factor involved in solution-samplingtechnique and thereby enhancw greatly the relative analytical sensitivity. This is in addition to the large enhancements in the absolute sensitivity of this technique; e.g., for Cu, Ni, and AI, the enhancements are 27-fold, 24-fold. and 20-fold, respectively (23). In Table V the chloride interference in the conventional graphite furnace atomic absorption spectrometry can be explained by using Pb as an example (similar explanation is also valid for Cd and Mn since these metals and their chlorides have volatilities similar to those of Pb). Loss of volatile chlorides of Pb apparently occurn a t the moment of vaporization of the bulk of NaCI, MgCI2, and CaC12, i.e., when the surface temperature of the graphite hrbe is 9o(tloOO OC. This effect can be explained by the “carrier effect” or “carrier distillation” as known in emission spectroscopy (%), which is also called covolatilization by some other workers. I t is reasonable to expect that evaporation of the salts would cause simultaneous introduction of lead into the vapor phase as chlorides. Sturgeon et al. (17) have reported that the vapor-phase temperature lags considerably behind the surface temperature in the conventional graphite furnace atomic absorption spectrometry. At the above temperature the lead chlorides are only partly dissociated and are partly lost by explusion (as explained later) and to a lesser extent by convection and by diffusion. The earlier the lead chlorides are vaporized the longer is the time they have for the loss proeess to deplete the analyte-containing compounds. However, in the above case, the primary mechanism for the loss of anal@-containing compounds is by explusion with the rapidly expanding inert gas a t the initial phase of heating in the atomization cycle when the graphite surface temperature is rising very rapidly; the loss is also partly due to convection and diffusion. It would be instructive to investigate the specific roles of the capacitive discharge technique in removing matrix interfereces which have plagued the conventional graphite furnace atomic absorption spectrometry. It is unquestionable that the capacitive discharge technique allows isothermal
kg 01 Pb (as niirate) Flpue 1. Oscllbscoplc traces IW 5.0 X h an aqueous solution atomized at a heating rate of 1.3 K ms-’ and a final temperature 01 2200 K. Pb A 283.3 nm: curve A (bfltrace). Pb (as nitrate) in 0.5% (w/v) NaCl aqueous soMiMI; curve B (rhJht trace). Pb (as nitrate) in an aqueous solution. c
, ,
, .
:
W e 2. oscikmupa traces fa2.5 X 10-l~ kg of Pb (as niirate)in an aqueovs solution atmized at a heating rate of 40 K ms-I and temperatwe 01 2200 K. Pb A 283.3 nm: culve A. Pb (as nitrate) In 0.5% (wlv) NaCl aqueous solution: curve B. Pb (as nitrate) In an aqueous solution; cuve C. 0.5% (wlv) NaCl aqueous solution.
(constant-temperature) atomization and very high heating rates, whereas the conventional graphite furnace atomic absorption spectrometry does not (17.24). The effect of very high heating rates on isothermal atomization is seen in Figures 1and 2 Figure 1 Bhows millompic traces of 5 X lW3kg of P b atomized a t a heating rate of 1.3 K m-’under nonisothermal conditions a t a fmal temperature of 2200 K, using the conventional GFAAS but with an anisotropic pyrolytic graphite tube. The left trace is for P h (taken as nitrate in an aqueous solution) in a 0.5% (w/v) NaCl aqueous solution and the right trace is for lead nitrate alone in aqueous solution. I t can be seen from Figure 1that in the presence of the chloride matrix the lead appearance time is shorter and, hence, the lead appearance temperature is lower; also, the pulse amplitude is lower than that of the lead nitrate alone. In this case, the peak height mode of measurement would give a lower result (relative to the lead standard) and a decrease in the sensitivity of lead. However, since the arean under both the absorption pulses are about the same, the integrated absorbance mode of measurement would give nearly correct results, whereas the peak height absorbance mode would not. Figure 2 shows oscilloscopic traces for the above systemgcurve A one trace, curve B two traces (for duplicate runs)-using the capacitive discharge technique, a heating rate of 40 K md,and isothermal atomktion at a f d (isothermal) temperature of 2200 K (same as that of Figure 1);m e C (the bottom trace) ia for the blank containing an aqueous solution
448
ANALYTICAL CHEMISTRY, VOL. 53, NO. 3, MARCH 1981
of 0.5% (w/v) NaCl and shows that the background absorption due to the NaCl matrix is completely absent, which is highly significant. It is seen from Figure 2 that the lead absorption pulses are identical in both amplitudes and areas, and the matrix effects have been completely eliminated. It can also be seen from Figure 2 that the lead signal appears much later in time (compared to Figure 1) and, hence, the appearance temperature for lead is considerably higher than that in Figure 1. In Figure 2 the considerably higher appearance temperature for Pb and atomization under constant temperature are crucial for eliminating chloride matrix interferences, as will be seen from later discussion. In Figure 2 both the peak height absorbance mode and the integrated absorbance mode of measurement would give correct results. Since the same anisotropic pyrolytic graphite tube was used in both Figures 1 and 2, equilibration of the surface temperature throughout the graphite tube was very fast in both cases. However, in Figure 1 atomization of lead occurred under nonisothermal conditions at a temperature which was lower than that in Figure 2 and at a time when the temperature of the atomizer increased at a very rapid rate from 600 to 2200 K; hence, the pressure inside the graphite tube changed from 1to 3.7 atm, which resulted in partial expulsion of the lead chlorides vapor; also the extent of dissociation of lead chlorides to lead atoms was less at the lower temperature. The conclusion is that the elimination of the matrix effect in Figure 2 was mostly due to atomization at a constant, higher temperature and a much faster rate of heating with the capacitive discharge technique. It has been reported in the literature ( 2 S 3 1 ) that the chloride interference is a vapor-phase type of interference, caused by inadequate heat delivery to the analyte element as it moves out of the atomizer surface. This kind of interference includes either a recombination of analyte atoms with interferent species or a lack of dissociation of volatile analytecontaining compounds before atomization temperature is achieved. This kind of interference is generally a suppression, since it involves a depletion of the observable analyte atomic population. Also, the amount of chloride interference is dependent upon the cation with which the chloride is associated; those metal chlorides which vaporize and dissociate to provide chlorine at the temperature a t which the analyte element is normally atomized show the greatest interference. Using time-resolved studies Czobik and Matousek (29) have shown that the degree of lead signal suppression depends upon the availability of atomic chlorine for recombination in the vapor phase at the time of lead atomization. L'vov (22)has theoretically and experimentally illustrated the effects of chlorine as a vapor-phase interferent and its suppression by the addition of excess LiN03, thereby binding excess C1 through formation of thermally stable LiC1. Frech and Cedergren (31) also attribute chloride interferences in lead determination to the thermodynamic stability of gaseous PbClz and PbCl species at the low vapor-phase temperature at which they are vaporized. From thermodynamic calculations these authors (31) have found that the amount of lead chlorides decreases with increasing temperature in the ashing step. In the atomization step, the temperature of the furnace has to be raised from a relatively low temperature to a high preset atomization temperature, thereby scanning part of the temperature range in which lead chlorides are vaporized as mentioned above. The time required to reach the final temperature is important in determining the distribution of lead among various lead compounds. If the temperature is raised slowly, the formation of volatile lead chlorides will be more probable than when rapid heating is employed. The above findings of Frech and Cedergren (31)fully agree with those of the present study, which emphasizes the crucial role of very
100
I HEATING RATE,Kms-I
Flgure 3. Recoveries of lead (added as nitrate) from an aqueous solution containing magnesium chlwMe as the matrix as a function of the heating rate: Pb A 217.0 nm; 1.0 X lo-', kg of Pb; 2.5 X lo-' kg of MgCi,; atomization temperature = 2570 K. 'OOT
2o 0
t 8
16
48
32
I
64
HEATING RATE / K ms-'
Figwe 4. Recoveries of cadmium (added as nitrate) from an aqueous solution containing 1.0 X lo-*kg of NaCl 1.2 X lo-' kg of MgCI, 4.0 X lo-'' kg of CaCi, as a function of the heating rate: Cd A 228.8nm; 4.0X lo-'' kg of Cd; (0)capacith/edischarge, temperatwe = 1900 K; (A)HGA 768,temperature = 1900 K and heating rate = 2.28 K ms-'.
+
+
high heating rates in determining the extent of chloride matrix interferences with lead determination. Cadmium and manganese and their chlorides are similar to lead and lead chlorides in their volatilities, and a similar situation has been demonstrated for chloride interference with manganese determination (32). The above explanation is valid for the chloride interferences with determination of Cd and Mn. The effect of heating rates on chloride interferences is presented in Figure 3 which shows the recoveries of lead from a MgClz matrix as a function of the heating rate. It is seen from Figure 3 that a severe depression in the signal is caused by the matrix at lower heating rates. As has been explained later, this interference is due to the covolatilization of lead as lead chlorides occurring at a relatively low temperature at a time when the temperature of the atomizer is increasing rapidly-the final temperature is 2570 K. The depression in the signal becomes progressively smaller with increasing heating rates till the heating rate of about 14 K ms-' is reached or exceeded, when the analytical recovery becomes and remains 100%. Figure 4 shows the recoveries of cadmium from an aqueous solution of NaCl + MgC12 + CaC12matrices as a function of heating rates at the isothermal atomization temperature of 1900 K. This figure shows that even at the highest heating rate used (63 K ms-'), the recovery curve is still rising (although very slowly), indicating a need for still higher heating rates for greater recoveries-this is in view of the fact that the atomization temperature of 1900 K used is adequate for complete atomization. The importance of heating rates and atomization a t a constant temperature on the matrix inter-
ANALYTICAL CHEMISTRY, VOL. 53, NO. 3, MARCH 1981
440
t
2o 10 0 I2
OD4
20 0 24
30
0.36 0 00 0 I2 M A S S OF MATRIX a I O O / k a
4.0 0 40 0 16
0.50 U q C k 0 2ococ12
Flgure 5. Recoveries of lead (added as nitrate) from an aqueous solution containing the matrix: sodium chloride + magnesium chloride + calcium chloride as a function of the mass of the matrix (WX 217.0 nm; 5.0 X kg of Pb): (0)capacitive discharge, temperature = 2550 K and heating rate = 40 K ms-'; (A)HGA 768, temperature = 2550 K and heating rate = 1.25 K ms-'.
2o
t
01
20 024
40 0 40
0 08
0 16
80 0 96
60 072 0 24
M A S S OF M A T R I X
0 32
d
10 NoCI
1 IOONoCl I 2 Yqc12
0 4 C0Cl2
1o91kO
Flgure 8. Recoveries of cadmium as a function of the mass of the matrix: sodium chloride magnesium c h k r i i 4- calcium chkride (Cd X 228.8 nm; 1.0 X lo-'' kg of Cd): (0)capacitive discharge, temperature = 2000 K and heating rate = 40 K ms-'; (A)HGA 768, temperature = 2000 K and heating rate = 2.28 K ms-'.
+
ferences in the conventional graphite furnace atomic absorption spectrometry and the relative freedom from matrix interference which is provided by the new technique is illustrated by Figures 5-7. It is clear from Figures 5-7 that the conventional graphite furnace atomic absorption spectrometry with its much lower heating rates suffers from Severe matrix interferences even at relatively low concentration of the matrix, whereas the new technique with its much higher heating rates is relatively free from chloride matrix interferences and probably other interferences as shown in Tables 11-VI. The relative freedom from matrix interferences provided by this new technique is primarily due to its high rate of heating and atomization at a constant temperature-the vapor-phase temperature is also higher than that of the conventional GFAAS. The above-mentioned higher temperature of the vapor phase is due to the much faster attainment of a constant temperature of the graphite surface and, also, to the smaller difference between the vapor-phase temperature (lower) and the graphite surface temperature (higher) in the capacitive discharge technique. The smaller difference in the temperature is due to the smaller internal diameter (3 mm) of the graphite tube used in the new technique (compared to 6 mm in the conventional graphite furnace atomic absorption technique) and to the faster attainment of a constant temperature of the graphite surface because of high anisotropy of the graphite used in the construction of the graphite tube for the new technique. Because of the slow heating rates of commercial graphite furnaces these furnaces are unable to provide atomization a t the constant, high temperature required for eliminating matrix interferences. Figures 5-7 show that there is a slight decrease in the analyte signals given by the new technique a t high concentrations of the matrix. The matrices (NaCl+ MgClz + CaC12)
0.1 0 06
1.0 0.12 a04
0.02
1.5 0.10 0.06
2D
2.5 NoU
0.24
0.3 yPU2
0 00 MASS OF MATRIX a I O ' l k p
0.1 CoCl2
Flgure 7. Recoveries of manganese as a function of the mass of the matrix: sodium chlorkle magnesium chloride calclum chbride (Mn kg of Mn): (0)capacitive discharge, temX 279.5 nm; 2.0 X perature = 2820 K and heating rate = 40 K ms-'; (A)HGA 768, temperature = 2820 K and heating rate = 1.25 K ms-'.
+
+
used in Figures 5-7 are major components of seawater, but the highest concentrations of the above components used represent 1.2-5-fold dilution of their concentrations in seawater. Even the heating rate of 63 K ms-' used in Figure 4 was unable to yield 100% recoveries of the analytes when the above three c o m p o n e n ~of the matrix were present together in high enough concentrations. It is possible that a combination of higher heating rates and optimum atomization temperatures together with an optimum atomizer configuration which will allow higher heating rates and higher temperatures to be used may give higher recoveries. Optimum atomizer configuration will include the shape, the tube wall thickness, and the internal diameter of the graphite tube, which' determine the speed with which the temperature is equilibrated throughout the graphite tube and the vapor phase. It may be stated that direct determination of trace elements in seawater presents a challenge in handling a difficult matrix (inorganic salts) present in an overwhelming excess compared to most other matrices (in the usual solution sampling technique) encountered in real samples. The reason for selecting a matrix similar to saline water for testing this new technique was to determine its ability to handle a difficult matrix; but, as it turned out, the effort met with rather limited success at the present stage of development of this technique. This work is consistent with the following vapor-phase mechanism as primary mechanism for chloride interferences in the determination of Pb, Cd, and Mn in chloride matrices as proposed by Frech and Cedergren (31). 1220 K
PbClz(s) __* PbCl,(g)
-
PbCl(g)
+ Cl(g)
CONCLUSIONS Although this method does not need nor use calibration curves, it uses a single-point calibration and is not an absolute method of analysis but is close to an absolute method of instrumental analysis. With continuing improvement in the capacitive discharge technique and in the design of the tube atomizer, it is expected that even closer approach to an absolute method of analysis by the capacitive discharge tech-. nique will be possible. This method of analysis gives accurate and precise results at a fraction of the time and cost required by the conventional methods of analysis by GFAAS. LITERATURE CITED Krasowski, J. A.; Copeland, T. R. Anal. Chem. 1979, 51, 1843. Kdrtyohann, S. R.; Pickett, E. E. Anal. Chem. 1988, 38, 1087. Koirtyohann. S. R.; Pickett, E. E. Anal. Chem. 1985, 37, 801. Jackson, K. W.; West, T. D. Anal. Chlm. Acta 1972. 59, 187. Alger, D.; Anderson. R. 0.; Mines, I. S.; West, T. S. Anal. Chlm. Acta 1971, 57, 271. (6) Anderson, R. 0.;Johnson, H. N.; West, T. S. Anal. Chlm. Acta 1971, 57, 355. (7) Matousek, J. P. Abstracts of 3rd Annual Meeting, Federation of Analyticai Chemistry and Spectroscopy Societies, 1978 Meting, Phlladelphia, PA, Nov 1978; Paper 308.
(1) (2) (3) (4) (5)
450
Anal. Chem. 1981, 5 3 , 450-455
(8) Czobik, E. J.; Matousek, J. P. Anal. Chem. 1078, 50,2. (9) Smeyers-Verbeke, J.; Michotte, Y.; Van den Winkel, P.; Massant, D. L. Anal. Chem. 1978, 48, 125. (10) Cruz, R. B.; Van Loon, J. C. Anal. Chim. Acta 1074, 72, 231. (11) Churella, D. J.; Copeland, T. R. Anal. Chem. 1078, 50,309. (12) Hutton, R. C.; Ottaway, J. M.; Platt, T. Abstracts of 3rd Annual Meeting, Federation of Analytical Chemistry and Spectroscopy Societies, 1976 Meeting, Philadelphia, PA, Nov 1976; Paper 307. (13) Eklund, R . H.; Holcombe, J. A. Anal. Chim. Acta 1970, 709,97. (14) Manning, D. C.; Slavin, W. Anal. Chem. 1078, 50, 1234. (15) Slavin, W.; Manning, D. C. Anal. Chem. 1070, 57, 261. (16) Manning, D. C.; Slavin, W.; Myers, S. Anal. Chem. 1079, 51, 2375. (17) Sturgeon, R. E.; Chakrabarti, C. L. Prog. Anal. At. Spectrosc. 1078, 7 , 1. (18) Gregoire, D. C.; Chakrabarti, C. L.; Bertels, P. C. Anal. Chem. 1978, 50. 1730. (19) Gregolre, D. C. Ph.D. Thesis, Carleton University, Ottawa, Ontario, Canada, 1978. (20) Cresser, M. S.; Mullins, C. D. Anal. Chim. Acta 1074, 68, 377. (21) Manthey, K. German Patent 2245 610, 1974. The Perkin-Elmer Corp. Bodenseewerk Geratetechnik GmbH. 7770, Uberlingen, West Ger-
“la”,
(22) L vov, B. V. Spectrochim Acta, Part 8 1078, 338, 153. (23) Chakrabarti, C. L.; Hamed, H. A.; Wan, C. C.; Li, W. C.; Bertels, P. C.;
Gregolre, D. C.; Lee, S. Anal. Chem. 1080, 52, 167. (24) Chakrabartl, C. L.; Wan, C. C.; k m e d , H. A.; Bertels, P. C. Nature (London) 1080, 288, 246. (25) L’vov, B. V. “Atomic Absorption Spectrochemical Analysls”; Adam H i b r Ltd.: London, 1970; p 221. (26) Nakahara, T.; Chakrabarti, C. L. Anal. Chim. Acfa 1079, 704,99. (27) Chakrabartl, C. L.; Wan, C. C.; Li, W. C. Spectrochlm. Acfa. Part 8 1080, 358, 93. (28) Schroll, E. 2 . Anal. Chem. 1083, 798, 40. (29) Czobik, E. J.; Matousek, J. P. Anal. Chem. 1078, 50,2. (30) Hageman, L. R.; Nichols, J. A.; Niswanadham, P.; Woodriff, R . Anal. Chem. 1079, 57, 1406. (31) Frech, W.; Cedergren, A. Anal. Chim. Acfa 1078, 82, 83. (32) Hageman, L.; Mubarak, A.; Woodrlff, R. Appl. Spectrosc. 1070, 33, 3.
RECEIVED for review March 3, 1980. Accepted October 28, 1980. H.A.H. is grateful to the Government of Iraq for a postgraduate scholarship. This work was supported by research grants from the Natural Sciences and Engineering Research Council of Canada.
Slurry Atomization Direct Atomic Spectrochemical Analysis of Animal Tissue Norita Mohamed and Robert C. Fry* Department of Chemistry, Kansas State Universtty, Manhattan, Kansas 66506
A method Is described which albws r a m , dkecl flame atomic absorption analyds of animal tlssue. The procedure Involves a 2mIn tissue homogenization followed by dlrecl atomlzatlon of the homogenate with an updated version of the clog-free BaMngton nebulizer. A conventlonal air-acetylene slot burner and the method of transient nebullzer sampling are used to avoid burner slot clogglng. Excellent agreement Is found between the resuits of the overall 5mln slurry atomlzatlon procedure and those of lengthy conventional dry/wet ashing methods for the determlnatbn of Cu, Mn, and Zn In fresh beef llver and beefsteak. The recovery achleved for Cu, Mn, and Zn was 100 i 3 % using slurry atomization. Thls Indicates that the homogenized tissue particles are sufficiently small to be nebullzed, transported through the tapered cone Jarrell Ash spray chamber, and atomized by the tlme they reach the observation zone located 4 mm above the premixed l O t m air-acetylene burner slot.
Sample preparation is one of the most time consuming, unreliable, and hazardous aspects of the determination of trace metals in animal tissue. The Food and Drug Administration, Kansas City district, now invests up to 3 days’ time in the wet ashing of 25-g meat composites. The extended time of wet ashing results from the large size of the sample, which in turn, is needed to make sure that the sample is representative. The practice of wet ashing and analyzing a single homogenized meat composite made up of many individual samples arises from the large number of samples in the monitoring program, the extended time of the wet ashing treatment, and manpower considerations. The use of perchloric acid can accelerate the wet ashing of smaller samples; however, the safety factor with this reagent becomes especially critical when the samples are as large as 25 g. Individual sample identity and overall analytical sensitivity are sacrificed when a composite of samples is analyzed (since an individual sample is now “diluted“ with other items in the 0003-2700/81/0353-0450$01.00/0
composite). In addition, wet or dry ashing treatment and extraction procedures may lead to contamination by reagents, volatilization loss, or loss related to retention by insoluble residues and surfaces of the ashing container (1). In an attempt to eliminate wet and dry ashing, several authors (2-4) have studied the atomization of ground and sieved powder suspensions in older, large-bore total consumption burners having excessive flicker noise and extremely poor sensitivity. Dilute suspensions of extensively ground and sieved rock samples (5) and coal powder have been atomized (6,7) by using a conventional premixed flame. The coal slurry response was only -20% of normal. A correction factor (5X) was then determined by wet ashing comparisons. None of these earlier nebulizer-based systems proved sufficiently “clog free” to handle foods. Fry and Denton recently introduced the first clog-free nebulizer for atomic spectroscopy capable of handling high solids food slurries (8,9). This was based on the revolutionary new Babington principle of aerosol generation which involves violent interruption of a solution film flowing across a small orifice from which compressed gases emerge at supersonic velocity (IO, 11). The resultant high-density, finely dispersed aerosol (consisting of small droplets in the diameter range
