Table II. Recovery Data For Sulfate
Table I. Beer’s Law Data for Sulfate Low dilutionn
a
s042- present, ppm S 0 4 z - found, ppma
High dilutionb
so4*-,PPm
A, 587 nmc
S 0 4 2 - . ppm
A, 625 nma
1.2 2.4 3.0 4.2
0.31 0.82 1.10 1.53
4.1 8.1 11 14
0.40 0.84 1.02 1.40
24 ml diluted to 100 ml. 5 m l diluted to 100 ml. Effective molar absorptivity of 31,000, A = 0.15 for the blank. Effective molar absorptivity of 9,400, A = 0.10 for the blank.
80-90% complete. The molar absorptivity of the chloranilate ion is about 1000 in the visible region and close to 25,000 in the ultraviolet region a t the optium p H value ( 3 ) . The sensitivity for sulfate using barium iodate and measuring visible absorption should be as high as the chloranilate method measuring ultraviolet absorption. The major difference is that a dilution is required using barium iodate because of the solubility of the compound and the high molar absorptivity of the starch-iodine complex. Probably the most important consideration in comparing barium chloranilate and barium iodate for measuring sulfate would be the relative solubilities. This factor would be important as far as interferences are concerned since other anions which form insoluble barium salts would release chloranilate or iodate ions. The extent of the interference would depend on the relative solubilities of the barium compounds involved. The solubility of barimoles per um chloranilate is reported to be 5.2 X liter in 50% ethyl alcohol ( I ) . From this work, it is estimated that the solubility of barium iodate in 75% ethyl moles per liter. The alcohol is approximately 1-2 x solubility could be lowered by employing other solvents miscible with water, such as acetonitrile (7). It was found in this work that although the blank is reasonably low in 1:l ethanol-water the Beer’s law plot is linear only above
2.1 4.2 6.2 9.3 17 a
2.0 4.1 6.4 9.6 17
Error,
YO
-3.8 -1.5 +2.6 4-3.0 -0.0
High dilution procedure. A measured at 625 nm.
5 ppm sulfate and precision is poor. Results using 95% ethanol or methanol were satisfactory but sensitivities were lower than in 3: 1 ethanol-water. A rather extensive study of the interference of chloride, fluoride, and phosohate ions on the determination of sulfate with barium chloranilate has been reported ( 3 ) . Interferences using barium iodate might be expected to be somewhat similar although differences would arise because of differences in the solubility of barium chloranilate and barium iodate. A very limited study of interferences in this work showed that bicarbonate, borate, hydroxide, and sulfite ions could not be tolerated while bromide, chloride, and nitrate ions had no effect. The barium iodate procedure should be useful for the determination of sulfate using visible absorption a t a much lower level than is possible with barium chloranilate. The sulfate concentration range with barium iodate using the lower dilution is about the same as can be measured using the ultraviolet absorption of the chloranilate ion. Some flexibility exists as to the concentration range covered using barium iodate since different dilution factors can be used. Received for review September 25, 1972. Accepted November 30, 1972. This work was part of the M.A. thesis of Willie L. Hinze, Sam Houston State University. May 1972. The authors express their appreciation to the Robert A. Welch Foundation of Houston, Texas, for partial support of this research.
I CORRESPONDENCE lnterelement Effects in the Flame: Spectroscopic Determination of Aluminum, Molybdenum, and Vanadium in a Nitrous Oxide-Acetylene Flame Formed on a Circular Slot Burner Sir: In recent years renewed attention has been focused on interelement effects in flame spectroscopy as a result of repeated observations that a variety of concomitants enhanced the atomic absorption or emission signals of a number of elements in nitrous oxide-acetylene flames. These enhancements have been difficult to explain because they occur for systems that heretofore would have been expected to show depressions of the analyte signal. Among the most perplexing of these observations were those of Dagnall et al. ( I ) , who studied the effect of con-
(1) R M . Dagnall, G. F Kirkbright, T S West, and R Wood,
Chem., 42, 1029 (1970).
Anal.
comitants on Al, Mo, V, Ti, and Zr emission and on Al, Mo, and V fluorescence in an inert gas separated nitrous oxide-acetylene flame formed on a circular slot burner. These authors reported that enhancements were, in general, significantly larger in thermal emission than in fluorescence, and that a few concomitants produced large depressions of the analyte signal. No hypotheses were proposed to account for either the effects observed or the differences between the emission and fluorescence results. We have recently reinvestigated a number of the more prominent enhancement effects observed in emission by Dagnall et al. Although our experimental conditions were similar to those used by Dagnall et al., we were unable to confirm the surprisingly large enhancements they reported. This brief communication summarizes our results. ANALYTICAL CHEMISTRY. VOL. 45, NO. 4, APRIL 1973
815
Table I. Effect of Concomitants on the Flame Atomic Absorption and Emmission of AI, Mo, and V Analyte Concomitant This work Dagnali et a / Mo
V AI
Ni VW) AI cu
0
- 1O%Q.b OQ
+ 10%
K
-70%
Zn WWI) Na
- 10% - 20% - 70%
co
-50% 00
AI Na W(VI) Mo(VI)
+
0 0" 1O%Q
(1)
+2900%
+ 485% + 100% + 58% + 55% + 52% + 23% -
40% 16%
-I- 30% 12%
+ + +
22% 11%
Data corrected for emission by concomitant at wavelength used. Data corrected for absorption by concomitant at wavelength used. a
EXPERIMENTAL The circular slot burner and shielding head were constructed according to the design of Hingle et al. (2). The sample was introduced with a pneumatic nebulizer through a Perkin-Elmer spray chamber, both with and without a flow spoiler. Gas flow rates were 10-12 I./min nitrous oxide, 5-6 l./min acetylene, and about 20 l./min argon. Horizontal and vertical positioning of the burner head to 20.05 mm was achieved with the racking mechanism previously described (3). The external optical system was similar to that used by Fassel et al. (4). A mask over the lens nearest the spectrometer with a , 3-mm diameter hole was used when the spatial distribution of analyte in the flame was studied. Interference studies were made both with and without the mask; it was not used for the detection limit measurements. Slit heights were 10 mm, and the slit widths were 0.1 mm. The spectrometer, detector, and electronics have been previously described f5), with the following exceptions: a Keithley Model 417 picoammeter was used for emission signal amplification: data acquisition utilized an Infotronics CRS 80 Digital Readout System operated in the linear mode with 8-second signal integration. All metal salt solutions were prepared from reagent-grade chlorides or the metal dissolved in a minimum of HCI except Mo, V, and W solutions, which were prepared from ammonium molybdate, metavanadate, and tungstate, respectively. A small amount of HC1 was added, where necessary, to repress hydrolysis. Analyte concentrations were 25 pg/ml; concomitant concentrations were 2500 pg/ml. The wavelengths used were: AI, 3961.5 A; Mo, 3132.6 A (absorption), 3903.0 A (emission): V, 4379.2 A. These wavelengths are normal ones for the elements studied. However, it is not entirely clear exactly which wavelengths were used by Dagnall et al. for their studies of Mo and V; consequently, we are not certain that we have used the same ones.
(2) D. N. Hingle, G . F. Kirkbright, M. Sargent, and T. S. West, Lab. Pract. 18, 1069 (1968). (3) J. A. Fiorino, R. h. Kniseley, and V. A. Fassel, Spectrochim. Acta. 23B. 413 (1968). (4) V. A. Fassel, J. 0. Rasmuson, R. N. Kniseley, and T . G. Cowley, Spectrochim. Acta, 25B, 559 (1970). (5) J. B. Willis, Spectrochim. Acta, 25B, 487 (1970).
816
0
ANALYTICAL CHEMISTRY, VOL. 45, NO. 4, APRIL 1973
RESULTS AND DISCUSSION Our first goal was an attempt to duplicate some of the unusually large enhancements observed in emission by Dagnall et al., and to check the same systems in absorption. The large enhancements were not observed in emission, nor were any significant differences found between absorption and emission measurements. The results are summarized in Table I, tabulated as percentage changes produced by the concomitants indicated a t the positions in the flame where analyte absorption or emission was a maximum. The data for A1 and Mo represent both absorption and emission data; V was studied in emission only. The comparison of results shown in Table I follows a consistent pattern, i.e., we observe: (a) depressions or no effect in those instances where Dagnall et al. reported 100 to 2900% enhancements; (b) either depressions or smaller enhancement& when the Dagnall et al. percentage enhancements are below 100%; and (c) a greater percentage depression when Dagnall et al. reported depressions. It is of interest to note that in the two instances where Dagnall et al. reported depressions (Na and Co on Mo) we found depressions as well. We cannot account for the discrepancies between the two sets of data shown in Table I. To be sure, the gas flow rates used in the present work were about twice those employed by Dagnall et al. However, the flame stoichiometry was almost exactly the same for the two experimental systems. Furthermore, analyte signals were maximized at the gas flow rates used in our work and these signals decreased as the flow rates decreased. In a further effort to compare the two experimental systems, we determined detection limits in emission for A1 and Mo, using the argon-shielded flame, and found values that almost exactly duplicated those reported by Dagnall et al. Our inability to confirm the unusually large enhancement effects reported by Dagnall et al. suggests that their data have limited transferability, even if the experimental conditions are similar. In addition, our observations cast doubt on the general validity of the Dagnall et al. conclusion that: (a) the virtue of the atomic fluorescence technique lies in its relatively high freedom from these interference effects; and (b) the observation of free atoms of Al, V, and Mo in this flame in thermal emission is subject to severe chemical or physical interference whereas, in marked contrast, observation of the free atoms in fluorescence is not. A. C. West1 R. N. Kniseley V. A. Fassel
Ames Laboratory-USAEC and Department of Chemistry Iowa State University Ames, Iowa 50010
1 0 n leave from Lawrence University, Appleton, Wis. 54911.
Received for review July 24, 1972. Accepted December 11, 1972.
