62
ANALYTICAL CHEMISTRY, VOL. 51, NO. 1, JANUARY 1979
Table I. Results for Determination of lron in Absorbance Range 0.13 t o 1.4 on Single Disc concentration, ppm taken found error, ppm 1.00
1.01
4.00
4.00 7.97
8.00 15.00 1.20 2.80
5.50 11.00
13.50 9.50
+ 0.01 0.00
- 0.03
+ 0.01
15.01
1.20 2.77 5.50 10.96 13.46 9.50
0.00
- 0.03 0.00 - 0.04 - 0.04 0.00
input the data for the least-square fit calculation. Input the number of t h e standard solutions as well as their concentrations in the same order as they have been loaded in the disc. Check the presented correlation coefficient as an indication of any scattering on the “stored” working curve. In most cases t h e correlation coefficient should be better than 0.9998. A t least three standard solutions are required. Once the working curve has been established and stored, the results may be presented in concentration form simply by calling the “concentration output“ routine.
RESULTS AND DISCUSSION T h e percent relative standard deviations of the measured absorbance values as a function of the absorbance a t various numbers of averaged rotations from R = 1 t o R = 1000 are shown in Figure 7 . T h e theoretical curves which are based on photon statistics are shown as the solid lines and illustrate t h a t for every 10-fold increase in the number of rotations at 7. constant speed there is a t 10 increase in precision. I t can be seen in Figure 7 that the experimental points are quite close to t h e theoretical solid lines. This indicates t h a t the rotating-disc system follows photon statistical considerations, a t least down to % RSDs of about 0.02% for 1000 revolutions (about 30 s total elapsed time) but only 1-2 s measurement time for each cuvette). When the number of revolutions is increased t o 10 000, the precision improves to about 0.01 70
RSD which indicates that other factors are causing variation in the absorbance measurements. I t can be observed in Figure 7 t h a t the 70 RSD for any given number of revolutions does not change much from about 0.3 t o 2 absorbance units, which is a point previously emphasized ( 1 3 , 1 4 )but often neglected. The quantitative analytical performance of the system was evaluated by using the classical method for the photometric determination of iron with 1,lO-phenanthroline. Typically obtained analytical errors over a very wide absorbance range of about 0.13 to 1.4 A are shown in Table I. The measurement was continued for 1000 revolutions a t about 1800 rpm. The errors shown are typical, with errors generally small b u t occasionally about 170because of sample handling problems. Although the performance of the Rotating-Disc Module is well within our initial expectations, there are obvious improvements that could be made to the hardware and software of the system. Work is now underway to make these improvements. A major application of the module is in the high-precision recording multisample spectrophotometer previously described ( 4 ) .
LITERATURE CITED (1) B. W. Renoe, R. P. Gregory I V , J. Avery, and H. V. Malrnstadt, Clin. C h m . ( Winston-Salem, N . C . ) , 20, 955 (1974). (2) S. D. Brunk, T. P. Hadjiioannou, S. I . Hadjiioannou, and H. V. Malrnstadt, Clin. Chem. (Winston-Salem, N.C.), 22, 905 (1976). (3) S. D. Brunk and H. V. Malrnstadt, Clln. Chem. (Winston-Salem, N.C.), 23, 1054 (1977). (4) J. P. Avery and H. V. Mairnstadt, Anal. Chem., 48, 1308 (1976). (5) N. G. Anderson, A m . J . Clin. Pathol., 53, 778 (1970). (6) . . N. G. Anderson, C. A. Burtis, J. C. Mailen, C. D. Scott, and D. D. Willis, Anal. Lett., 5, 153 (1972). (7) C. A. Burtis, J. C. Mailen, W. F. Johnson, C. D. Scott, T. 0. Tiffany, and N. G. Anderson, Clin. Chem. (Winston-Salem, N . C . ) ,18, 753 (1972). 18) C. D. Scott and C. A. Burtis. Anal. Chem.. 45. 327 (1973). C. A. Burtis, W. F. Johnson, J. C. Mallen, J. B. Overton, T.’O. Tiffany, and M. B. Watsky, Clin. Chem., ( Winston-Salem, N.C.), 19, 895 (1973). (IO) T. 0. Tiffany, Crit. Rev. Clin. Lab. Sci., 5 (3), 129 (1974). (11) J. Avery and H. V. Malrnstadt, manuscript in preparation. (12) T. A. Woodruff and H. V . Malrnstadt, Anal. Chem., 46 1162 (1974). (13) H. V. Malrnstadt, M. L. Franklin, and G. Horlick, Anal. Chem., 44, (a), 63A (1972). (14) J. Avery, R. P. ’&gory IV, B. W. Renoe, T. Woodruff, and H. V. Mairnstadt, Clin. Chem., ( Winston-Salem, N.C.), 20, 942 (1974).
RECEIVED for review April 3, 1978. Accepted September 25, 1978. Work was partially supported by N I H research grant
HEW PHS GM 21984-03.
Filament in Furnace Atomization Atomic Absorption Spectrometry V. P. Garnys“ and L. E. Smythe Department of Analytical Chemistry, University of New South Wales, Kensington 2033, Australia
A new approach to furnace atomic absorption spectrometry uses multiple sample desolvation and ashing of samples on a filament before introduction into a graphite furnace for atomization. Filament in furnace atomization (FIFA) can more than triple the analysis rates over conventional sequentially operated furnaces. The versatility and scope of the new technique is illustrated by the determination of Pb in large numbers of blood and hair samples and also by the determination of Ag, Au, Cd, Fe, Mn, and Zn in acidlfied aqueous solutions. Under typical routine conditions, the RSD was 5 to 10% for 1-pL aliquots and the best RSD was 1.8%.
With AAS a t the ultratrace level, processing of microsamples in flame heated crucibles, such as the Delves Cup ( I ) , 0003-2700/79/0351-0062$01 O O / O
tantalum boat (2), and the newer porous graphite capsule (3) is more rapid than electrothermal atomizers, since several crucibles may be dried and ashed simultaneously prior t o atomization. Electrothermal, carbon, or metallic filaments or strips have received extensive study because of their low power requirements, rapid heating rates, low cost, and ease of sample application ( 4 ) . However, open filaments do suffer from very short atom residence times, varied distribution of the atom cloud, and very large thermal gradients as the atom cloud leaves the filament. Consequently, some of the disadvantages of open filaments are reduced sensitivity, critical optical beam alignment, and increased matrix interferences ( 5 ) . Graphite furnaces offer advantages over flame heated crucibles and electrothermal filaments in that they partially 1978 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 51, NO. 1, JANUARY 1979
63
4 L
Figure 1. Plan schematic of FIFA showing: (1) CRA63 furnace workhead (heavy dashed outline), electrodes, and furnace (section): (2) insulated assembly carriage; (3) filament anchorage screw: (4) base plate; (5) filament and coils; (6) ratchet transport mechanism; (7) slide rod: (8) filament power supply terminals: (9) conducting plate and filament support rod; (10) external thermocouple for monitoring furnace temperature; ( 1 1) electrode end stops
enclose and provide a more even distribution of the atom cloud at temperatures up to 3000 "C, increase atom residence times, and provide a metal reducing environment for the condensed and vapor phase (5,6). Consequently, furnaces offer generally better detection limits, relatively simple alignment, a n d lowered interferences ( 4 ) . However, graphite furnaces do have some disadvantages. Manual microsample addition is exacting and tedious ( 7 , 8 )and the behavior of the sample during drying a n d ashing cannot be directly observed. Drying and ashing in graphite furnaces are normally carried out sequentially with inert gas environments which promote pyrolysis rather than oxidation. After atomization, the furnace must cool t o below t h e analyte solvent boiling point, before the next sample solution can be injected. Automation of microsample addition has assisted with pipetting precision b u t not the rate of analysis. Also the equipment is expensive and requires some care for successful operation. Recently, aerosol deposition in furnace atomization developed by Matousek (9),has led t o the introduction of the "Fastac" system (10) which is claimed t o improve speed a n d versatility. However, detailed performance d a t a on this latter system are not yet available. With t h e most popular furnaces, although graphite has excellent electrical and thermal properties, it cannot be used regularly with strong oxidizing acids such as HC104 and HzSO,. Consequently graphite furnace atomizers have reduced versatility for mineral a n d biological sample digests, which use these acids extensively t o solubilize t h e trace metals. In this publication we report t h e development of a new patented ( 1 1 ) technique which combines the advantages of a metallic filament with those of a graphite furnace. Of greater significance is the much increased sample analysis rate, since simultaneous multiple sample drying and ashing are possible outside t h e furnace. Results are presented which illustrate t h e versatility a n d applicability of t h e technique t o t h e determination of lead in large numbers of blood and hair samples collected during t h e course of a survey and investigation of 1200 Sydney school children (12). T h e scope of the technique is also illustrated by determination of Ag, Au, Cd, Fe, Mn, a n d Zn in acidified aqueous solutions. EXPERIMENTAL Apparatus. Atomic absorption measurements were made with a Varian-Techtron Mode! AA-5 atomic absorption spectrometer fitted with a model CRA 63 graphite furnace atomizer (Varian Techtron, Springvale, Victoria) to which the filament in furnace
Figure 2. Filament in furnace atomization (FIFA) apparatus attached to a Varian Techtron model CRA-63 atomizer. The location of the thermocouple and removed portion of the pyrolytic graphite coat are shown. Legend, the same as Figure 1
atomization (FIFA) apparatus was attached as shown schematically in Figure 1. Furnace Modifications and Temperature Monitoring. Sample solutions were dispensed with a 5-&LHamilton syringe with Chaney adaptor (Pfizer Diagnostics, New York, N.Y. 10917) set to deliver either 1or 2 bL. Deterioration of the contact between the electrodes and the resistance heated furnace, due to electrode slippage, was prevented by constructing two screw-in end stops. A central portion of the hard, outer pyrolytic graphite coat of the furnace was stripped off on a lathe to improve the contact between the furnace and the electrodes. This modification eliminated localized arcing and uneven heating of the furnace. The graphite furnace temperature during the dry, ash, and atomize cycles was monitored by locating a small junction chromel-alumel thermocouple 1.5 mm above the furnace. The atomic absorption signals were measured on a fast recorder (Mace FBQ-100, N.I.C. Instruments, Sydney, Australia). The thermocouple output was either recorded on the chart recorder or during routine operation on a digital millivoltmeter (Digital Panelmeter, ADPOlOE. Analog Devices. Norwood, Mass. 02062). Calibration of the external thermocouple output with respect to the internal furnace temperature, was carried out with chromel-alumel and tungsten/26% rhenium-tungsten/5% rhenium thermocouples placed inside the furnace and at higher temperatures with an optical pyrometer (Leeds and Northrup Co., Nort,h Wales, Pa.). Filament in Furnace Atomization. An apparatus was constructed as shown in Figures 1 and 2 whereby an electrically heated metallic or carbon filament or strip could be passed through a small groove cut in the lowest portion of the graphite furnace of the Varian-Techtron model CR4-63 atomizer. Several samples, ranging in volume from 0.5 to 20 pL, could be loaded onto the specified parts of the filament as shown in Figure 3. All of the samples could then be simultaneously dried and ashed by passage of a low current (2 to 5 A; 3 to 15 W) through the filament. A 9-A Variac voltage regulator was used in series with a 240 V/6 V transformer to power the filament. A digital voltmeter with a 10-A facility (John Fluke Mfg. Co., Mountlake Terrace, Wash. 98043, model 8000A-05) was used to monitor the wire current. Filament Construction. Precise location of liquid samples is difficult on a bare, hot wire. It was found that a small coil, or spring of tungsten wire wound about a typical tungsten filament. would contain a larger aliquot of the sample, assist in location of the sample, prevent its movement along the wire, and also act as an efficient high surface area heat radiator which allowed samples to be rapidly dried, ashed, and efficiently atomized. Rapid and precise placement of 1- to 20-pL samples on the small coils was achieved after short experience. A 90 mm X 0.25 mm tungsten wire filament with five to ten sample locating coils made from 0.125-mm tungsten wire comprised the filament which was passed through the graphite furnace.
64
ANALYTICAL CHEMISTRY, VOL. 51, NO. 1 , J A N U A R Y 1979
0
(a) Detail of tungsten filament and sample locating coils. (b) Five- and 20-pL samples on coils. (c) Positioning of seven 20-pL samples on filament during drying
‘
1
O 1
Figure 3.
02
0 3
04
0 5
0 6 uq/mL
Figure 4. Typical standard additional calibration curves. 0 = normal 0.09 p g Pb mL-’ blood-HNO, hydrolysate; 0 = elevated 0.40 pg Pb
mL-’ blood-HNO, hydrolysate; A = “Hyland Control Serum”, 0.04 pg
This filament was then used for most of the analyses of the more volatile elements such as Pb, Cd, Ag, Au, Zn, Mn, and Cu. The locating coils were also stretched from 3 to 4 mm before location, t o improve their matrix residue capacity and to improve their ability to release vapor during atomization. They were prevented from sliding on the m a n filament by carefully crimping their ends with blunted side cutter pliers. The strong capillary action of the locating coils ensured quantitative and reproducible sample transfer from the microsyringe. Figure 3 shows the construction of the filament and the location of sample aliquots. Filament Travel Mechanism. In this study, a manually operated filament travel mechanism was devised consisting of a 5-mm pitch sawtooth notched brass rod and spring loaded ratchet. By adjustment of the throw of the transport lever, samples could be advanced into the furnace aC 5- or 10-mm steps at selected speeds. The filament always rested, during atomization, in a 0.25-mm deep groove cut longitudinally into the floor of the graphite furnace. After the samples were simultaneously dried and ashed on the filament outside the furnace, it was operated either in a continuous mode, where the filament was advanced into tne furnace already a t atomization temperature, or in a pulse atomization mode. In this latter mode, the filament was advanced into the furnace below atomization temperature and then pulse heated by using step atomization with zero dry and ash time settings. Reagents. All of the reagents used were either redistilled or of commercially available high purity grade. Nitric acid was either B.D.H. Aristar grade or vapor phase distilled from an all-Teflon, infrared heated two-bottle type still. For the determination of zinc. triply distilled water was required to reduce the blank readings. For ultramicro in-situ digestions of blood, a mixture of 7 0 7 ~ perchloric acid (Fredrick Smith, AR) containing 2000 ppm of concentrated orthophosphoric acid (Ajax Chemicals, AnalaR) was used. This will be referred to in Lhe text as HC104 (0.270 HaPo,).
RESULTS A N D DISCUSSION Lead i n Blood and Hair. Filament in furnace atomization has been extensively tested in a survey (12) which required t h e determination of lead in approximately 3000 capillary blood samples ranging in volume from 10 t o 70 pL and approximately 4000, 1-cm hair segments, ranging in weight from 1 t o 10 mg. Figure 4 shows typical standard addition calibration curves for blood which has been oxidatively hydrolyzed with nitric acid at 80 “ C in polyethylene vials (13,1 4 ) . Even with FIFA. freshly digested 1:l blood:HN03 hydrolysates did leave a residue after ashing in air, which caused a gradual sensitivity
Pb mL-’ serum-”0, hydrolysate. Conditions: Pb line 283.3 nm, 1430°/2.5 s, 1 pL of direct hydrolysate without in-situ microdigestion
5
30r
\\ .o
CUECAW” 2
5
pol
OPERAT3P i
Typical continuous trace for direct determination of Pb in hair-HNO, hydrolysates. Conditions: Pb line 283.3 nm, 1430°/2.5 s, 1 ILL hydrolysate, overall analysis rate 19 s/sample. Good interaperator reproducibility is demonstrated for 1-cm hair segments 1.1, 1.2 and 2.1, 2.2, 2.3. Standard additions to albumin were used for calibration Figure 5.
decrease in t h e lead absorbance over about 100 samples. Hair:HN03 (1:10 w/v) hydrolysates prepared in a fashion similar t o the blood method did not exhibit such sensitivity decreases, possibly because of t h e higher HNO, strength, reduced matrix components, and greater lead concentration. Figure 5 shows a typical lead in hair hydrolysate calibration and. on t h e same continuous trace, the good inter-operator reproducibility is illustrated for samples 1.1,1.2, 2.1, 2.2, and 2.3 representing 1-cm hair segment hydrolysates. In-situ U l t r a m i c r o Digestion with O x i d i z i n g Acids. Unlike graphite, tungsten and tantalum are almost inert to hot, concentrated HC104 and H2S04. In-situ ultramicro digestions with concentrated HC104, H2S04, H3P04, and HNO, were used t o minimize or remove organic ash deposits and to chemically modify the blood matrix to be more suitable for routine furnace analysis. T h e increased thermal stability of lead phosphate as compared to lead nitrate after in-situ addition of concentrated HC104 with and without H3PO4, required higher furnace temperatures t o maintain peak height sensitivity when
ANALYTICAL CHEMISTRY, VOL. 51, NO. 1 , J A N U A R Y 1979
r
C 8 2 x-
I
70 58
65
-
D;"
6 0 -
Figure 6. Typical continuous trace for analysis of blood-"0, hydrolysates (0.09, 0.28 pg Pb mL-') after in-situ digestion with HCIO, (0.2% H,PO,); using Pb line 283.3 nm; 1600°/2.5 s; 1-pL sample aliquot plus 3 pL HCIO, (0.2% H,PO,); overall analysis rate 24 ?./sample. Calibration is by standard additions to the Internationally standardized blood-"0, hydrolysate.
phosphate was added. Figure 6 shows continuous calibration and reproducibility traces for lead in blood when 3 pL of HC104 (0.2% H,PO,) was used in situ on the tungsten filament t o remove residual organic material from 1 pL of blood:HN03 hydrolysate. For the analyses of lead by FIFA in 0.1 M saline solution, nitric acid was used in situ to convert NaCl to NaN03 so t h a t NaCl interference a t the lead peak position could be eliminated. This saline solution was used in our International Interlaboratory Survey of Lead in Freeze-dried and Oxidatively Hydrolyzed Blood, Saline and Water ( 1 5 ) in which 27 laboratories from 9 countries participated. Atomization Surface and Filament Lifetime. Graphite furnaces are severely limited in their use for continuous analysis of large numbers of samples because of the deterioration of the graphite surface after repeated atomizations, probably partly due to oxidation and sublimation of the graphite but mainly to residue buildup in the furnace surface (14). In contrast, the tungsten filament surface did not appear to deteriorate and the sample locating springs remained bright and without residue even after 300 to 500 determinations of blood hydrolysate. A further advantage was t h a t for 500 determinations on a 5-sample filament, each sample position was heated to atomization temperatures only 100 times. For practical purposes, t h e calibration curve for a new' filament is similar to the curve for a filament which has performed 400 determinations. Similar retention of sensitivity between filaments was observed for lead and other elements when either low residue solutions or in-situ HC104 digestions were used. Figures 7 and 8 indicate consistent atomization efficiency for the tungsten filament under routine operating conditions for hair and blood hydrolysates. To obtain these error curves, within-run means and standard deviations for replicated standards and samples were calculated and plotted. Clearly, replicates were less precise a t lower concentrations and more uniform precisions were obtained a t higher concentrations. T h e more regular error curve of Figure 7 for 277 hair: concentrated H N 0 3 hydrolysate determinations, may result from the higher lead level in hair hydrolysate solution. Also the inorganically more simple hair matrix, which left very little residue on the filament, possibly allows more consistent vaporization during atomization. For 1-pL aliquots, 5 to 10% RSD was typical while the best RSD of 1.8% is similar to typical precisions obtainable with automated dispensers.
-.. 0 1 IO
20
3c
40
50
p g n ~ ~ E A 3
Figure 7. Quality control error curve for 277 determinations (including 101 standards) of Pb in hair-HNO, hydrolysates. Each point ( 0 )
represents standard deviation and mean of at least 5 calibration standards or sample replicates at t h e indicated concentration
20
I t
I
Figure 8. Quality control error curve for batches of Pb in blood-HNO, hydrolysates with in-situ "20, (0.2YO H,PO,) ultramicro digestion. The points represent means of calibration standards and sample replicates for: 0 = 118 determinations, including 51 standards; 0 = 193 determinations, including 68 standards; 0 = 614 determinations, including 224 standards.
For blood:HNO, hydrolysates after in-situ digestion with the HCIOI (0.2% H3P0,), a s l o ~deterioration of the atomization efficiency occurred during 600 determinations. Without in-situ digestion, blood residue buildup resulted in sensitivity decreases similar to those with normal graphite furnace analysis as reported by us previously ( 2 4 ) . This lends further support to our view t h a t ash buildup causes either physical obstruction or occlusion which results in decreased atomization efficiencies. Figure 8 shows three typical error curves for blood analysis runs of 118, 193, and 614 determinations. The 5 to 10% RSD for 118 determinations deteriorated to 15 to 20% RSD for 614 determinations. However, in normal use for routine calibration, the 614 determinations would be subdivided typically into groups of approximately 100 determinations. In this case, the precision of the standards would be maintained within 5 to 10% RSD. T h e conditions of the graphite furnace were not found to be important provided that end stops were used to control electrode slip and imperfect electrode contact was improved
66
ANALYTICAL CHEMISTRY, VOL. 51, NO. 1, JANUARY 1979
Table I. Analysis Conditions and Calibration Equations for Cd, Ag, Au, Mn, Zn, and Fe in Aqueous Solutions Using Filament in Furnace Atomization conditions Cd Ag Au Mn Zn Fe wavelength, 228.8 328.1 242.8 279.5 213.8 372.0 nM
concentration range aliquot, W L atomization rate
0-100 pg
0-100 pg
0-100 pg
1.0
1.0 1430" 12 s
1430"/2 s
1390"/2 s
1.0
0-300 pg
0-600 pg
1.0 1710" / 2 s
1200"/2 s
0-10 ng
1.0
1.0 2800" 12 s
calibration curve equation coefficients a h r2
3.44 5.59
1.00
2.50 1.51 1.00
0.00 3.00 1.00
0.00 1.50 1.00
- 3.27 0.47 0.996
0.021 0.019 0.983
Equation for Cd, Ag, Au, Mn: A = a + bC (pg). Equation for Zn: In A = b In C (pg) + a . Equation for Fe: A = h In C (ng) + a , where A = absorbance and C = concentration in either picograms (pg) or nanograms (ng). by removing (on a lathe) the central portion of the hard outer pyrolytic graphite coat of the furnace. T h e external chromel-alumel thermocouple was used to monitor both the furnace temperature a t which the next sample was advanced and the maximum furnace temperature during atomization. The thermocouple output did not appear t o deteriorate over several thousand atomizations. Maximum filament life was obtained when the tension of the supporting filament arms was adjusted to a minimum t o compensate for filament expansion during heating. Maximum Temperature and Scope of FIFA. Tungsten wire was mostly used in the preliminary development of FIFA because of its ready availabilit,y, general suitability, and low cost. Tantalum, rhenium, molybdenum, and graphite strips were also constructed and evaluated and their detailed evaluation will be presented in further publications. The major disadvantages of tungsten arise from its brittleness due t o macrocrystallinity (19)but more importantly, the formation of tungsten carbide eutectic mixtures which reduce the melting point of tungsten from 3400 "C to 2500-2800 "C (26). These eutectic mixtures can form on the surface of a filament during high atomization temperatures while it is in contact with the floor of a graphite furnace. Because of their low ash mass, relative to the filament, samples leaving carbon residues after ashing were not expected to contribute significantly to filament deterioration by carbide formation. For tungsten filaments, carbide formation does limit the scope of elements to those that may be atomized completely below 2500 "C. However, many of the elements of environmental, clinical, and geochemical interest volatilize below 2500 "C and may still be determined. Table I demonstrates the applicability, linearity, and similar sensitivity of FIFA as compared with furnaces for t h e determination of cadmium, gold, silver, zinc, and manganese and iron. Some preliminary experiments indicate success in producing pyrolytic graphite strips of sufficient conductivity and strength to replace metal filaments for the determination of less volatile elements. The results in Table I also show an extended calibration curve for iron atomized from a graphite strip, demonstrating t h a t the FIFA technique has a scope which is comparable to t h a t of conventional graphite furnaces. The furnace temperature settings used for t h e elements studied were similar to those published for conventional furnaces (6, 27,28). High purity graphite yarns and cloths ("Thornel", Union Carbide Cor-
poration) gave poor precisions because of excessive sample creep and uneven heating along the yarns. CONCLUSIONS Filament in furnace atomization (FIFA) is a new method using multiple sample desolvation and ashing on a filament, with subsequent introduction of the filament into a graphite furnace for atomization. The simultaneous multiple drying and ashing of samples can more than triple the analysis rates over sequentially operated conventional furnaces. The external tungsten filament described is also more accessible t o rapid sample application and will withstand strongly oxidizing agents such as HC104, H2S04,and HN03. This allows in-situ ultramicro digestions of biological samples to remove organic material or t o chemically convert compounds t o minimize spectral interferences or improve atomization characteristics. However, the major advantages of FIFA should flow from a reduced dependence on tube quality and prolonged calibration stability. LITERATURE CITED H. T. Delves, Analyst (London),9 5 , 431 (1970). H. L. Kahn, G. E. Peterson, and J. E. Schallis, At. Absorpt. Newsi., 7 , 35 (1968). D. A. Katsov, L. P. Kruglikova, and B. V. L'Vov, Zh. Anal. Khim., 3 0 , 238 (1975). A. Syty, Crit. Rev. Anal. Chem., 155-228 (Oct. 1974). G. F. Kirkbright, Anayst (London), 9 6 , 609 (1971). W . C. Campbell and J. M. Ottaway, Talanta, 21, 837 (1974). F. J. M. J. Maessen, F. D. Posma, and J. Balke, Anal. Chem., 46, 1445 ( 1974). C. J. Pickford and G. Rossi, Analyst(London),9 7 , 647 (1972). J. P. Matousek, Talanta, 24, 315 (1977). H. L. Kahn, R. G. Schleicher and S. B. Smith Jr., Ind. Res., 101 (Feb. 1978). Australian Patent Application No. PC415176, "Filament in Furnace Atomic Absorption Spectrometry". V. P. Garnys, R. Freeman, and L. E. Smythe, Proc. Conf. on Ambient Lead and Heaith, 31-49, 1974, Health Commission of New South Wales, Sydney, 1976. V. P. Garnys and J. P. Matousek, Clin. Chem., ( Winston-Salem, N.C.), 21, 891 (1975). V. P. Garnys and L. E. Smythe. Talanta, 22, 881 (1975). V. P. Garnys and L. E. Smythe. presented at R.A.C.I. SymposiumWorkshop on Furnace A.A.S. Techniques, Sydney, 7 November 1975. F. Hall and N. F. Spooner, ISA Trans., 4 , 355 (1965). M. J. Maessen and F. D. Posrna, And. Chem., 46, 1439 (1974). E. J. Czobik and J. P. Matousek, Talanta, 24, 573 (1977). R . H. Forster and A. Gilbert, J . Less-Common Met., 20, 315 (1970).
RECEIVED for review April 24,1978. Accepted September 21, 1978.