!
If the calorimeter is to be calibrated by one experiment and used in others, care must be taken to be sure that the same heat flux distribution over the surface of the cavity results during both experiments. Thus approximately the same type reaction cell should be used in the cavity during both experiments. This difficulty may be circumvented somewhat by placing thermocouples such that they totally surround the cavity, thus removing angular dependencies. These difficulties plus others arising from measurement techniques apparently have been overcome because these calorimeters give good results.
I n this case 0 _ _L ) means that the additional term neglected is of order l/(r0 - 2 ) . From this equation, it can be seen that if r0 is large in comparison with the time interval of the heat input P(z), then Equation 53 can be approximated by (54)
and again the integral of the temperature deviation a t any radius E is approximately proportional to the heat flux into the sink. This equation also indicates that the constant of , time interval of proportionality is dependent upon T ~ the measurement. Thus, when a calibration experiment is followed by a n actual measurement, the time intervals in both experiments must be approximately the same, and this time interval must be larger than that of both the heat pulses. This problem should be considered when a quick reaction is used as a calibration for a slow reaction. I n this case, the tendency is to stop the calibration experiment before the proper T~ is reached. Calorimeter Usage. In the theory developed previously, the heat sink was assumed to be infinite; in actual usage this sink is large but finite. For the infinite approximation to be true, two conditions must be satisfied. The total heat expected during an experiment must not make a significant change in the steady state temperature of the heat sink; the outside edge of the heat sink must be far away from the cavity such that the temperature change at this edge is practically zero. T o determine if these conditions are satisfied, the analysis should be carried out for each particular geometry and heat sink materials.
SUMMARY
The mathematical description of the principle of heatburst calorimetry was developed in general above. This theory can be used to design calorimeters and improve their operation by applying it to each particular calorimeter. Sometimes computer solutions may be required because of geometry or materials of construction. The specific examples above suggest that a three-dimensional geometry probably will give better performance since no approximation is involved in the application of theory to the spherical cavity. We suspect that this has something to do with the difference in the Green’s function for three-dimensions as compared to that in twodimensions. In operation, when a calibration is followed by an actual experiment, the time of both experiments should be approximately the same and the type container in the cavity should also be the same. RECEIVED for review August 13, 1970. Accepted November 11,1970. -
Flame Emission Spectrometric Determination of Aluminum, Cobalt, Chromium, Copper, Manganese, Niobium, and Vanadium in Low and High Alloy Steels Velmer A. Fassel, Robert W. Slack, and Richard N. Kniseley Instifute for Atomic Research and Department of Chemistry, Iowa State Unicersitj , Ames, Iowa 50010
T h e flame emission spectrometric determination of AI, Co, Cr, Cu, Mn, Ni, and V in low and high alloy steels i s presented. No prior chemical separations are required. A nitrous oxide-acetylene flame, formed on a commercial slot burner normally employed for atomic absorption spectrometry, is used to excite the atomic spectra. The spectra a r e dispersed by a 0.5-meter, table-model spectrometer, and intensities are measured with a digital integrator. NBS certified low and high alloy standard samples were employed for calibration purposes. “Chemical” or ‘Lphysical” interference effects were not evident since congruent analytical curves were obtained for a wide range of steel compositions. Coefficients of variation for quantitative determinations were equivalent or superior to those published for atomic absorption measurements.
DURINGTHE PAST two decades, approximately 20 papers (1-3) were published on the flame emission spectrometric Mavrodineanu, “Bibliography on Flame Spectroscopy. Analytical Applications: 1800-1966.” Nut. Bur. Sfand. (US.) Misc. Pub(. 281 (1967). (2) M. Margoshes and B. F. Scribner, ANAL.CHEM.,40, 223R (1968). (3) J. D. Winefordner and T. J. Vickers, ibid., 42, 207R (1970). (1) R.
186
0
(FES) determination of various alloying constituents in low and high alloy steels. Of these papers, only two described procedures which were based o n the direct nebulization of the dissolved sample into the flame; the remaining publications described methods based on prior chemical separation of the iron matrix. In contrast, of the approximately 50 papers published or presented on the flame atomic absorption spectrometric (AAS) analysis of steels during the past ten years (4-IO), about 10% of the papers described a procedure requiring the prior chemical separation of the iron; the remaining papers described simple procedures based on the direct ______ -
(4) P. H. Sholes, A~ialyst.93, 197 (1968). (5) W. Slavin, “Atomic Absorption Spectroscopy,” Interscience, New York, N. Y., 1968. (6) “Bibliography on Atomic Absorption Spectroscopy,” Varian
Techtron. Monrovia, Calif., 1968. (7) F. J. Feldman, J. A. Blasi, and S. B. Smith, Jr., ANAL.CHEM., 41, 1095 (1969). (8) Y . Endo, H. Hata, and Y . Nakahara, Tetsu To Hugurze, 55, 216 (1969). (9) P. Konig, K. H. Schmitz, and E. Thiemann, Z. Anal. Chem., 244, 232 (1969). (10) R. W. Taylor, Mid-America Symposium on Spectroscopy, Chicago, Ill., June 1970.
ANALYTICAL CHEMISTRY, VOL. 43, NO. 2 , FEBRUARY 1971
nebulization of the dissolved steel samples into the flame. These historical developments strongly imply that prior chemical separation of the iron matrix is required for the FES determination of most alloying constituents, but that these separations are not generally required if flame AAS techniques are employed for the analyses. In fact, in some of the papers on the AAS analysis of iron and steels, this requirement is strongly implied o r specifically stated, without documentary scientific evidence to support the conclusion. Since both AAS and FES are based on observing the free atoms released from the nebulized sample in the flame, it is, a t first glance, difficult to understand or interpret the rather divergent historical development of AAS and FES techniques for the analysis of steels. To properly appreciate these developments, it is important to recall that the AAS literature contains the often repeated but disputable claims or generalizations that AAS techniques should be less susceptible t o chemical (11-14, spectral (15-17), and physical (11, 18, 19) interference effects, and that their powers of detection (12, 20-23) should be considerably superior to FES techniques. That these claims were not generally based o n scientific data, fact, or logic has been adequately documented (24-26). The focal point of our introductory discussion has been to suggest that there has been an unjustified neglect in applying FES to the analysis of steels, and other metals as well. This neglect has been particularly incongruous when it is realized that FES possesses several distinctive advantages over AAS observations (24-26) that have not been generally recognized. In view of the several attractive features which FES provides for the sequential, multielement analysis of steels, a critical appraisal of its actual capabilities appeared t o be in order. The results of this evaluation are summarized in this paper.
EXPERIMENTAL FACILITIES AND PROCEDURES Selection of Flame. Three features should be provided by the burner and flame. These are: a nebulizer-spray chamber burner for the premixing of the oxidant, fuel, and aerosol; a chemical environment conducive to the efficient atomization of the elements; and a temperature high enough t o provide optimal signal to noise ratios. The premixing of gases in the spray chamber burner allows a natural separation of (11) T. S. West, in “Trace Characterizations: Chemical and Physical,” Nut. Bur. Stand. (US.)Monogr. 100, p 281 (1967). (12) T. S. West, Analyst, 91, 69 (1966). (13) Y. Chan and P. Wong, Talanta, 15, 867 (1968). (14) G. W. Ewing, “Instrumental Methods of Chemical Analysis,” McGraw-Hill, New York, N. Y., 1969. (15) W. Slavin, Appl. Spectrosc., 20, 2816 (1966). (16) H. L. Kahn, J. Clrem. Edirc., 43, A7 (1966). (17) H. L. Kahn and W. Slavin, Inr. Sci. Techno/., 60-65, (Nov. 1962). (18) A. Walsh and J. B. Willis, in “Standard Methods of Instrumental Analysis,” Part A, F. J. Welcher. Ed., Van Nostrand, Princeton, N. J., 1966, pp 106-117. (19) A. Walsh, in “Advances in Spectroscopy,” Vol. 11, H. W. Thompson, Ed., Interscience. New York, N. Y., 1961, pp 1-22. (20) R. Mavrodineanu and H. Boiteux, “Flame Spectroscopy,” Wiley, New York, N. Y., 1965, p 193. (21) J. Robinson, “Atomic Absorption Spectroscopy,” Marcel Dekker, New York, N. Y., 1966, p 6. (22) W. Slavin, At. Absorption Newslett., No. 24, p 15, Sept. 1964. 41, (l), 25A (1969). (23) G. D. Christian, ANAL.CHEM., (24) V. A. Fassel, Plenary Lectures, Chemical Institute of Canada Conf., Ottawa, Canada, May 1969; and International Atomic Absorption Symposium, Sheffield, England, July 1969. (25) E. E. Pickett and S. R. Koirtyohann, ANAL.CHEM.,41, (14), 28A (1969). (26) J. D. Winefordner, V. Svoboda and L. Cline, Critical Rev. Anal. Chem., 1, 233 (1970).
Table I. Experimental Facilities and Operating Conditions Perkin-Elmer No. 290-0019 Burner burner with No. 303-0195 NzO-CzHzburner head. Slot width, 0.38 mrn The p(N20/C2H2)value was Flame stoichiometry maintained at 2.0 except for vanadium when p was adjusted to 1 . 6 Spherical quartz lens; 3.5-cm External optics diameter and 11-cm focal length Jarrell-Ash, Model 82000, 0.5Spectrometer meter Ebert mounting, scanning spectrometer with 1180 grooves/mm grating, blazed for 2500 A ; effective aperturef/8,6; reciprocal linear diseersion at the exit slit is 16 A/mm in first order Fixed bayonet ; 16 X 0,025 mm Slits entrance and exit slits were used, corresponding to a spectral slit width of 0 . 4 A. Automatic scan speeds of 500, Scanning action 250, 125, 50, 20, 10, 5, and 2 A/min are available. Wavelength counter accyracy is 1 2 A with 1 0 . 5 A reproducibility EM1 6256B S-13 response mulDetector tiplier phototube New Jersey Electronics CorporaDetector power supply tion, Model S-325-RM (500-2500 V, 0-10 rnA) Keithley Instruments preamplifier, Preamplifier Model 4170 Keithley Instruments amplifier, Amplifier Model 417. Scale expansion achieved through appropriate zero suppression and gain controls Hewlett-Packard Model 7001A Recorder x-y Recorder Superior Electric Company 1 Voltage regulator KVA Stabiline automatic voltage regulator, Model 1E5101 (for detector power supply, amplifier and recorder) Infotronics Corp., Model CRSReadout system 80 Digital Readout System, Series No. 3564. Measured Interval (sec), 8 ; mode, linear
larger aerosol droplets and some preevaporation of solvent before the particles reach the flame. Both of these factors contribute to minimization o r elimination of “chemical” (or “solute vaporization”) interferences (27). A flame formed from the premixed gases also develops a desirable spatial separation of the various reaction zones. This zonal separation enables the analyst to sample radiation only from the flame zone which produces maximal signal-to-noise ratios. With reference to the utilization of “hot” flames, it is appropriate to note that the population of excited states, and hence line intensities, increases exponentially with temperature (20), and that “chemical” interferences decrease with increasing temperatures (27). However, one must recognize that the increased ionization and the increased complexity of the observed spectra, resulting from higher flame temperatures, can cause difficulties. The essential features discussed above are conveniently provided by commercially available nitrous oxide-acetylene slot burners designed especially for AAS. (27) V. A. Fassel and D. A. Becker, ANAL.CHEM.,41, 1522 (1969).
ANALYTICAL CHEMISTRY, VOL. 43, NO. 2, FEBRUARY 1971
187
Mn 403076
0 190 %
TIME (MIN )
Figure 1. Emission of Mn line as a function of time Mn
403076 403307 403449
a
d
N B S 153 0.219 w t % Mn
Figure 2. Relative intensity measurements of Mn lines The fact that line emission intensities are enhanced in the same way as line absorbances by increasing the flame depth has been clearly documented (28). The commercial burner used in the present study is described in Table 1. Spectrometer and Accessories. The features which are desirable in a modern FES facility were recently summarized by Pickett and Koirtyohann (25); the facilities described in Table I provide the performance characteristics detailed by these authors. It is worth noting that the spectrometer is a small, table-model instrument that forms the nucleus of a popular, commercially available atomic absorption instrument. Thus, the spectrometer requirements for performing emission or absorption analyses are, in essence, very similar. To facilitate the wavelength location of spectral lines, two micropotentiometers (MP) were individually coupled directly to the wavelength drive of the spectrometer. The voltage drop developed across the micropotentiometer during a wavelength scan provided a reference signal for the horizontal drive of the x-y recorder. Intensity Measurements. Since line emission intensities were stable over reasonable periods of time (a typical example (28) S . R. Koirtyohann and E. E. Pickett, Appl. Spectrosc., 23, 597 (1969).
188
is shown in Figure l), three relatively simple intensity measurement techniques may be used. Two of these alternatives are illustrated in Figure 2 for the determination of Mn. In this ipstance, the weakest of the three components (Mn 4030.76 A) is least subject to spectral interferences from other constituents that may occur in steels and is measured for quantitative determinations. The first alternative simply involves a scanning of the spectrum in the immediate vicinity of the line of interest followed by measurement of the recorded peak height. Since the measurement time for this alternative is short (essentially the “residence time” of the recorder pen at the peak), random disturbances in emission or electronic noise may exert too large a n influence o n the readings. A considerable improvement in precision of observations can be effected by one of two related integration techniques. One of these, illustrated in the lower portion of Figure 2, involved “parking” the spectrometer o n the line, and recording alternately the background signal (pure Fe solution nebulizing) and the line signal (sample nebulizing) for periods of -10 sec. A longer time constant in the electronic system may be used to advantage for this measurement technique. The improvement in precision is readily evident in the figure. Rather than using the recorder as an integration device, the amplified photocurrent may be integrated directly by commercially available integrators and printed as backgroundcorrected, relative spectral-line intensities. The latter integration technique was employed in collecting the data summarized in this paper. The integration time periods were 8 sec for both background and line plus background readings. Quantitative Calibrations. The NBS Standard Reference Materials (29) includes an extensive collection of well characterized steels whose nominal cornpositions cover the range of interest to the steel industry. These reference samples are ideally suited for calibration purposes. First, the general applicability of the dissolution procedure may be evaluated o n actual compositions commonly used industrially. Second, if a congruent analytical curve is obtained from a set of reference samples of wide composition, with no experimental points which depart significantly from the curve, then the analyst may conclude that spectral, chemical, or physical interferences for the range of compositions studied either do not exist or d o not make a significant contribution to the measured relative intensities. F o r the calibration experiments, representative sets of low and high alloy compositions were selected from the NBS collection. Dissolution Procedure. One-gram samples of the steels were dissolved in a mixture of 30 ml of 1 :1 HC1 and 5 ml of “Os. The resultant solution was evaporated to near dryness and baked at 200 “C for 5 min. The dried residue was redissolved in 10 ml of concd HCl, and diluted to a final volume of 200 ml, yielding a solution containing 5.0 mg of sample per milliliter. For several of the low alloy (high carbon) and high silicon steel samples, visible insoluble residues remained. Optical emission spectrographic examination of these residues showed insignificant or nondetectable levels ofthe seven impurity elements studied in this investigation.
RESULTS AND DISCUSSION Wavelength scan recordings of the analysis lines are shown in Figure 3. The background recordings were made while solutions containing 2.5 mg of pure Fe per milliliter were being nebulized. These solutions were equivalent in iron content to an actual sample containing 50 wt of Fe. Since spectral interferences from Fe lines were not evident, the arbitrary assumption was made that a reference blank solution whose F e content was equivalent to a sample containing 50 (29) Standard Reference Materials, Nut. Bur. Stand. Publ., 260 (1968).
ANALYTICAL CHEMISTRY, VOL. 43, NO. 2, FEBRUARY 1971
(US.)Misc.
AI
Cr
Mn
3961.58, 0 Cl43 %
3570 7 8, 007%
4030.8a 0.69 %
V
a
4379.2
a
005%
I
1
I
;I
I 4WtW&
SAMPLE
Ni
3515.1 0.053%
--JL
BACKGROUND +whbww
.--%+AM
Figure 3. Wavelength scans of analytical lines and background
If:
E
E ;;
I
I ~ I ; (;,ti)
NBS 33 c ( 3 N i l
4
i
N B S 199 ( L o w A l l o y )
W
a
1
0
L
.
1 I I Ill
l
0 01
7
1
LUL4
01
10
CHROMIUM C O N C E N T R A T I O N ( w t % )
00 2 0 05 01 02 ALUMINUM CONCENTRATION ( w t % 1
Figure 6. Analytical curve for the determination of chromium
Figure 4. Analytical curve for the determination of aluminum
11
164 N I . I I C ~ ! N B S I ~ ~
cu
NBS 169177N1.20Cr1
/
NBS 160119Cr. 9".
1 I
-_
3MoI
005
NBS 160 119 Cr. 9Ni. 3 M o I
1
I
,
1
01
C O E A L T CONCENTRAT1ON(wt
NES 161 164 Ni, 17CrI NES 3 3 c ( 3 N i l
z
NBEI26D136NI
Ili I 1 I 1
002
1
t
I-
I I I I I , 05
IO
0.5L 001
%I
I
I
I
1
I I I I I
005 0.1 COPPER CONCENTRATION ( w t , % )
05
Figure 5. Analytical curve for the determination of cobalt
Figure 7. Analytical curve for the determination of copper
wt % of Fe represented a satisfactory background simulation of the actual samples. The Fe content of the latter ranged from 0.5 up to 99 wt %. I t is of interest to note that a close match in the Fe content of the reference and sample solutions is required for several of the AAS procedures which have been described (4,30>.
Analytical Curves. Representative sets of low and high alloy compositions were selected for each element studied from the available NBS Standard Reference Materials. It is worth noting that one of the samples (NBS-169) was actually a Ni-Cr alloy containing only 0.5 wt Fe. The identification of the samples selected are shown in Figures 4 to 10. Each plotted point represents the average of five observations on different days. Normalization of the data from day to day was assured by adjusting the photomultiplier
+
-
(30) G. F. Kirkbright, A. M. Smith, andT. S. West, Analyst, 91,700 (1966).
ANALYTICAL CHEMISTRY, VOL. 43, NO. 2, FEBRUARY 1971
189
400
o q
I
I
I
I I I I I I
I
200
N05 122d I L O r A l l o y !
60
NBSI26bi36N81
NBS 153 i a C o , a m , z
20
c. I
w. 4 c r . 2 v I
I
0.5 MANGANESE CONCENTRATION ( w t
%I
iJ
Figure 8. Analytical curve for the determination of manganese 25
001
I I I I I
0.05 01 NICKEL CONCENTRATION Iw t % )
05
Figure 9. Analytical curve for the determination of nickel voltage and picoammeter range setting until a predetermined photocurrent was obtained while a reference solution of the element was nebulized under the prescribed flame conditions. The horizontal bars attached to the calibration points indicate the range of concentrations reported by the analysts who participated in the NBS certification programs. The points themselves represent NBS Cerlified Values ; the latter signify the best concentration values obtainable at the time the certificates were issued and do not necessarily certify true compositions. Of all of the points plotted in the figures, only the C r data o n NBS 1296 and 122d show significant departure from the analytical curve. Since low alloy compositions are involved here, it is reasonable to assume that the various constitutionally related interference effects are not responsible for these outlying points. The general disposition of the remaining calibration points along single congruent analytical curves for each of the elements provides strong experimental support for the following general conclusions: that the technique of alternately recording the background signal (pure Fe solution nebulizing) and the line signal (sample nebulizing) adequately corrects for residual spectral background; and that if a premixed nitrous oxide-acetylene flame is employed for the emission spectrometric determination of alloying constituents and residual impurities steels, chemical separation of the iron matrix is not generally required and chemical o r physical interferences either do not exist or d o not exert a 190
NBS 169(77Ni,20Crl
4 NBS 7 3 c I I 3 C r l 4
NBS 122 d (Low Alloyl NBS 1551Low A l l o y l
NBS 341 i20N1.2Cr:
3L I
2 5
0 01
I
I
~
Ill11
005
I
L
01
10
VANADIUM CONCENTRATION ( w t % )
Figure 10. Analytical curve for the determination of vanadium significant effect on the emission intensities. It is particularly significant that the calibration points for the Ni-Cr alloy (NBS-169) show no deviation from the analytical curves. Thus, the often repeated generalizations found in a sizable fraction of the atomic absorption spectrometry literature that flame emission techniques are highly susceptible to these interferences are contradicted by the experimental results discussed above. It is beyond the scope of this paper to discuss the determination of each of the constitutents in detail, but it is appropriate to draw attention t o several important points. The direct nebulization of dissolved steel samples has been previously employed for the determination of Cr and M n as major constitutents ( > O S wt %) in steels and slags (31, 32). For the AAS determination of Al, Konig el al. ( 9 ) have reported that reference “blank” solutions should have approximately the same Fe content as the test solution. Figure 5 shows that calibration point for the Ni-Cr alloy (NBS169), which contains only 0.54 wt Fe, does not depart from the analytical curve derived from low alloy steel calibration standards. This observation suggests that less stringent requirements on the similarity in total composition between reference standards and samples apply when a nitrous oxide-acetylene flame is used in either absorption o r emission. The early literature (4, 30, 33-35) on the AAS determination of Cr repeatedly refers to the suppression of Cr absorption by the presence of Fe in the solutions. Beyer’s (36) data, however, shows that the simple expedient of employing reference standard solutions whose Fe content approximated the concentrations found in samples accommodated the suppression effects in the calibrations. The analytical curve for C r (Figure 7) suggests that the total composition of reference standards and samples need not be matched closely for FES measurements. With reference to the Co, Cu, Mn, Ni, and V data shown in Figures 6 and 8 t o 11, it is worth noting that the calibration points for the Ni-Cr alloy, which contained Fe, fell either directly on the analytical curve? only 0.54 wt (31) W. A. Dippel and C. A. Bricker, ANAL.CHEM.,27,1484 (1955). (32) S. Ikeda, Sci. Rep. Pes. Inst. Tohok~iUnic., Ser. A . , 8, 463 (1956). (33) K. Kinson, R. J. Hodges, and C. B. Belcher, Anal. Chim. Acta, 29, 134 (1963). (34) W. J. Price and P. A. Cooke, Specfrocision, 18, 2 (1967). (35) A. Glammorise, At. Absorption Newsier?., 5, 113 (1966). (36) M. Beyer, ibid., 4, 212 (1965).
ANALYTICAL CHEMISTRY, VOL. 43, NO. 2, FEBRUARY 1971
A 396153 ,i pv nJ(Z- -C- n
1%
h
3&50 COO5%
3gk69
cu 324754
Mr 403076
COC2 YC
OCl%
002%
NI
3515 OF,
h
4374 24
3 OCj 46
'3Cl %
-A .
Figure 11. Recording of line emission at lowest determinable concentration
Cu 3247.54 8
0.01 010 5 010
c v, = t
or within the range of concentrations reported by the analysts who participated in the NBS certification program. Lowest Determinable Concentrations. The lowest concentration points plotted on the analytical curves should not be considered the lowest concentrations determinable in a quantitative sense. If the reasonable assumption is made that reliable quantitative measurements require a single/noise,,, of 10, then the concentration level that would provide this signal level can be readily computed from experimental data. For the integrating readout system employed in the present study, the lowest concentration determinable is then simply the concentration, in wt Z, required to give a n integrated line signal which is ten times greater than the rnis deviation of integrated background readings. rhese concentrations are shown in Table 11. The practical utility of these values is illustrated in Figure 11, which shows actual recordings made with solutions of iron containing the seven impurities at the computed lowest determinable concentration. The reproducibility of measurements a t these concentration levels is accurately reflected by the repeated runs for Cu shown at the bottom of the figure. The coefficient of variation figure of 5 % is based on measurements of the integrated net line signal.
Table 11. Lowest Determinable Concentrations Lowest determinable Wavelengtli, A concentration, wt Element 3961.53 0,0025 AI 3453,jO 0,005 co 3578,69 0.002 Cr 3247,54 0.01 cu 4030,76 0.02 Mn Ni 3515.05 0.005 4379.24 0.01 V Table 111. Precision Re1 Element A1
co Cr
Wavelength, A 3961,53 3453,50
Concentration, wt ?z 0.043 0.14 0.055
stand
:za
dev, 0.1
3578,69 3247.54 cu 0.081 Mi i 0.219 4030.76 3515.05 Ni 0.046 V 0,053 4319,24 Since only 10 determinations were r u n , only one figure is given for the relative standard deviation.
2 1
I 0.8
2 4
significant
PRECISION
A realistic measure of precision for the analytical procedure described in this paper was obtained by determining the reproducibility of results obtained over a period of time. Three reference standards were chosen per element; those of highest and lowest concentrations were elected as calibration points, and the third designated as the unknown. Three signal integrations were made for each line and its associated background. The average values of the net relative line intensities were employed in establishing the concentration of the unknown. The procedure was repeated on 10 consecutive days. The results are summarized in Table 111. It is seldom possible t o draw definitive comparisons between these data and most of the AAS precision data reported in the literature,
because the experimental conditions under which the AAS results were obtained were either not adequately specified o r they differed greatly from those used in the present study. In those instances where good comparisons may be drawn, the FES precision data summarized in Table I11 were comparable to or superior to published AAS values.
RECEIVED for review September 21, 1970. Accepted November 19, 1970. Work was performed in the Ames Laboratory of the U. S. Atomic Energy Commission, Iowa State University, Ames, Iowa.
ANALYTICAL CHEMISTRY, VOL. 43, NO. 2, FEBRUARY 1971
191
