V O L U M E 26, NO. 5, M A Y 1 9 5 4 development of comparatively small deposits or concentrations of niobium-tantalum ores. It has been estimated that a single accurate chemical analysis of this type costs about $100 to perform. Unless this cost is a relatively small fraction of the total value of a deposit, such deposits have not been considered economical to exploit. T h e n many complete analyses can be made rapidly and inexpensively, the cost of an initial analysis should 110 longer be a detrimental factor to the development of small deposits of niobium-tantalum ores. ACKNO W LEDGMElrr T
The cooperation of Allan H. hfacmillan and Thomas E. Green, who performed the chemical analyses, and of Maurice J. Peterson, who made the spectrographic analyses, is appreciated. George E. .4shby, now with the J. J. Maguire Co., did much of the preliminary work in developing the equipment and the general technique of quantitative mineral analysis. This investigation was performed under the direct supervision of John E. Conley, with the general supervision of Paul M. iimbrose. Their continued interest in this problem encouraged the authors to prepare this paper for presentation and for final publication.
805 LITERATCRE CITED
(1) Adler, L., and Axelrod, J. AI., J . Opt. SOC.Amer., 43, 769 (1953). ( 2 ) ;Itkinson. R. H.. Steipman. J.. and Hiskes. C. F.. ANAL.CHEY., 24,477 (1952). (3) Behr. F. -4.. and Zinparo. P. W., Xorelco Reporter, 1, 3 (1953). (4) Birks, L. S., Rev. Sci.>nstr., 22, 891 (1951). (5) Birks, L. S., and Brooks, E. J., . ~ N A L . CHEY.,22, 1017 (1950). ( 6 ) Birks, L. S., Brooks, E. J., and Friedman, H., IM., 25, 692 (1953). (7) Briswy, R. M.,Ibid., 24, 1034 (1952). (8) Despujols. J., J . phys. radium, 13, Suppl. to No. 2, 31A (1952). (9) Friedman, H., and Birks, L. S.,Rea. Sci.Instr., 19, 323 (194s). (10) Gillam, E., and Heal. H. T., Brit. J . A p p l . Phys., 3, 353 (1952). (11) Hevesy, G. von, “Chemical Analysis b y X-Rays and Its rlpplications,” Kew Tork, RIcGraw-Hill Book Co., 1932. (12) Rlortimore, D. hl., and Romans, P. A,, J. Opt. Soc. Amer., 42, 673 (1952). (13) Schoeller. W. R.. and Powell. -4.R., Analust, 50, 485 (1925); 53, 264 (1928). (14) Woods, G. A., Atomic Energy Research Establishment, Harwell, England, CRLiAE-62 (1950). RECEIVED for review September 24, 1953. Accepted January 22, 1954. Presented in p a r t a t the Pittsburgh Conference on Analytical Chemistry and -4pplied Spectroscopy, 1953.
Some Interferences in Flame Photometry ROY D. CATON, JR.,
and
RAYMOND W. BREMNER
Department o f Chemistry, Fresno State College, Fresno, Calif.
Interference considerations have played a very prominent role in the measurement of metal ions in solution by some workers, but certain aspects of interference effects caused by the so-called inert materials have been given little or no consideration by others. Viscosity obviously is an important factor in the flame spectrophotometric determination of metal constituents in solution. Because the rate of flow of the solution through an aspirator or orifice into the flame is a function of its viscosity, several authors have added various substances to standard solutions to give viscosities approximating those of the solutions being analyzed. The present study w-as undertaken in the hope of finding viscosity correction factors or otherwise correlating the effects of viscosities. This was accomplished in part, but other effects, dependent upon the particular viscosity-regulating additive used, play an important role in changing the flame intensity. Particle size, which is one of these effects, is studied by means of photomicrographs.
T
HERE are many analytical techniques for determining small quantities or low concentrations of most of the elements which exhibit flame spectra, but the flame method offers the analyst two considerable advantages: It provides a precision which cannot be matched by any other spectrochemical method and a complete analysis can be made in a few minutes. Water, biological fluids, and liquid food products are easily analyzed for the alkali and alkaline earth metals with high specificity. Minerals, ceraniicF, glass, soils, alloys, metals, biological tissues, and food products are easily analyzed after an appropriate solution of the substance is made. However, the presence of foreign ions and molecules in the unknown substance is often the most troublesome factor in flame photometry and causes the majority of errors and a loss of ana-
lytical accuracy. Berry et a / . ( 1 ) and Parks et al. (10) denionstrated the depressing effects of inorganic acids and inorganic salts upon the flame intensities of sodium and potassium. Other workers have shown similar results (3, 8). Several workers have also shown how the presence of certain organic solutes may either enhance or depress flame intensity, depending on the substance used (1-3, 6, 9, 10). Several workers have briefly discussed the effects of viscosity of the solution and attributed a portion of the interference of foreign substances to their alteration of the rate of atomization of the sample into the flame (1,2,4, 5,9). Berry et al. ( 1 ) used sucrose ap an example of a compound which caused large errorq because of viscosity. Bills et al. ( 2 ) showed that the presence of 5000 p p.m. of glycerol with 5 p p.m. of sodium depressed the flame intensity 24% while it depressed the rate of atomization into the flame by only 6%. Mosher et al. (9) showed the depressing effects of varying amounts of gelatin upon the flame intensity of sodium and potassium and attributed these effects to viscosity. Conrad and Johnson (4)also attributed errors in the analysis of petroleum oils to viscosity. Some authors (1, 2 ) advocate correcting for interferences by dilution of unknowns to minimize them. Empirical correction cui vee have also been wed with success (4,‘7,8,11),and a method eliminating the interfering effects of diverse ions in water analyG i G by the addition of “radiation buffers” has been employed by Keet et al. ( I f ) In many cases, standards are compounded to approximate the composition of the sample being analyzed (9,10). The references mentioned are not an exhaustive list, but serve to indicate the present status of the problem of interference. The present study was made to determine the magnitude of interference attributable to viscosity only, in some solutions containing interfering substances. The question of whether or not the particle size of the spray might have some bearing upon the interference effects of foreign substances also presented itself and a study of this aspect v.-as made. It was decided to ex-
ANALYTICAL CHEMISTRY
806 amine the effects of sucrose, glucose, urea, and gelatin on the determination of sodium, potassium, and calcium. APPARATUS AND STANDARD SOLUTIONS
Beckman Model DU Quartz Spectrophotometer with Flame Attachment. This apparatus, employing an oxygen-natural gas flame, was used during all the experimental work. The spectrophotometer and the flame attachment are described in detail by Gilbert el al. (6). Compressed air for the flame attachment was taken from a pipeline at 65 6 pounds per square inch. A water trap followed by a diaphragm-type pressure regulator was installed in the air line just before it reached the instrument. Natural gas from the city lines was passed through a diaphragm-type pressure regulator before entering the instrument. The additional control valves within the instrument held pressure fluctuations within very narrow limits at the burner. Zeiss Photomicrographic Apparatus. This instrument was loaned by Twining Laboratories of Fresno, Calif. Reagents. All reagents were of analytical reagent grade and conformed to A.C.S. specifications. The weights used were calibrated against a set calibrated by the Xational Bureau of Standards All volumetric glassware used was of Sormax quality, which conforms to NBS specifications. Solutions were stored in glass-stoppered borosilicate glass bottles which had been previously seasoned by steaming out for 20 minutes and storing full of distilled water for 30 days or more. Distilled water stored in such bottles for several months was found to remain uncontaminated; no potassium, sodium, or calcium was shown to be present, using the instrument settings used in the experimental work. Calibration Solutions. Stock solutions of 1000 p.p.m. of sodium and 1000p.p.m. of potassium were separately prepared from chlorides which had been previously dried a t 110" C. for 2 hours. A stock solution containing 1000 p.p.m. of calcium was prepared by dissolving the required amount of calcium carbonate in the minimum amount of dilute hydrochloric acid and diluting with distilled water in a volumetric flask, Aliquot portions of these solutions were diluted to prepare a series of calibration solutions for each metal, the concentration range employed being 0 to 100 p.p.m. Solutions Containing Interfering Materials. Stock solutions of 1M sucrose and 2M glucose were prepared for the subsequent preparation of solutions of lesser concentration. Considerable difficulty in keeping the solutions from spoiling because of bacterial action was first encountered but the addition of a few drops of carbon disulfide to the solutions was found to prevent such decomposition. Presence of the small amount of carbon disulfide that dissolved was shown to produce no observable effert upon flame intensity, rate of aspiration, or particle size. A 4-M stork solution of urea was found to be stable for the entire period of about 1 month during its use. A special calfskin gelatin, free from sodium, potassium, and calcium, was obtained from the Eastman Kodak Co. The gelatin was in sheet form and of very high purity. The sheets were quickly washed in distilled water to remove possible surface contamination, then with ethyl alcohol to remove the water, dried in a vacuum oven at 70" C. overnight, and stored in a desiccator. Ten grams of the dried gelatin mere dissolved in a small portion of hot water and diluted to 1 liter to obtain a stock solution containing 10 mg. of gelatin per ml. of solution. The stock solution was found to be stable for only about 2 days, after which a cloudiness and scum rendered it unusable. For this reason all solutions containing gelatin were measured immediately after they were prepared. Aliquot parts of the foregoing stock solutions and the metal stock solutions were separately placed in 100-ml. volumetric flasks and diluted to prepare solutions containing 100 p.p.m. of the desired metal ion and varying amounts of the contaminants.
*
CALIBRATION OF FLAME SPECTROPHOTOMETER
Gas, Air, and Oxygen Pressures. Flame emission of a metal differs markedly with fluctuations in the gas, oxygen, and air pressures. For each metal there will be certain pressures for which a maximum emission is obtained. These conditions may even differ with changes in concentration of the element activated in the flame. To find the correct pressures for maximum emission for each metal, the following procedure was used: A standard solution of 100 p.p.m. of the metal was atomized into the flame.
Table I. Air Pressure, Lb./Sq. Metal Inch Sodiurp 25 25 Potassium 20 Calcium
Instrument Settings Used
Oxygen Pressure, Inches of Water 30 30 32
Gas Pressure, Cm. of Isopropyl Alcohol 2.7 2.7 2.5
Slit Width, Mm. 0.05 0.05 0.20
Wave Length, Mr 589 767 556
Table 11. Calibration Data for Sodium, Potassium, and Calcium Standard Solutions Solution
Concn., P.P.I\I.
Potassium
100
80 60 50 40 30 20
10 5
Flame Intensity Atomizer I Atomizer I1
100.0 81.6 61.3 51.6 40.6 27.0 16.5 6.3 2.9
100.0
85.6 68.0 59.0 48.7 24 0 10.7 4.6 100.0 81.5 61.9 52.1 41.8 21.5 11.1 1.8
The gas was turned up to give a vigorous flame, the oxygen pressure adjusted for maximum flame intensity, and then the air pressure adjusted for maximum flame intensity. After a good air pressure was selected, the oxygen pressure was readjusted for maximum reading. As a final check, the gas was turned a little higher or lower and the oxygen finally adjusted to see if the new maximum was above the preceding one. Slit Width. This was set to a value which allowed a minimum ratio of background radiation to net total radiation, with the instrument sensitivity set near its maximum limit. Wave Length. The wave-length selector dial was first set a t the required reading and then carefully adjusted until a maximum response was obtained on the galvanometer needle while the sample was atomized into the flame. Phototube. h red-sensitive phototube was used for wave lengths above 625 mp and a blue-sensitive phototube for wave lengths below 625 mG. Before flame measurements were made, it was necessary to activate the phototube for 5 minutes with light from a standard solution to obviate erratic reading obtained when the first flame measurements were made. Heater Voltage. The spray chamber is surrounded by a heating jacket which was connected to a variable transformer to control the temperature of the heater. The heater temperature must be high enough to prevent condensation of the spray in the spray chamber. Hon ever, too high temperatures caused apparent emission by the distilled water blank, probably from the hot glass surface. The transformer setting used was 80 volts for solutions atomized with an old-type atomizer and 100 volts for solutions atomized with a newer atomizer which delivered more sample to the flame. The instrument settings used in the evperimental work are listed in Table I. Calibration Curves. The 100 p.p.m. solutions were used as reference solutions with instrument dial set at a flame intensity reading of 100. Before and after each reading the dark current was checked for constancy and adjusted if necessary. When the reference solution produced a reading of 100 f 0.5 both before and after running a sample, it was assumed that the instrument
V O L U M E 26, NO. 5, M A Y 1 9 5 4 had not varied enough to produce significant errors during the analysis. Five or more readings were taken for each solution and the results were averaged. The average error in precision ' of the reported flame intensity. Wlieii was never more than 1% not atomizing samples into the flame, distilled water was introduced to prevent overheating the spray chamber because such overheating caused a very appreciable drift of readings as thc chamber cooled. Calibration data were obtained when two types of atomizers were in operation. The two atomizers were identical with thc exception of the bore of the capillary tubing with which the). were constructed: atomizer I and atomizer 11, the latter having a capillary diameter about three times that of the former. Both atomizers were supplied by the Beckman Co., atomizer I1 being of a more recent design.
807 To correct for viscosity, rates of atomization were determined by use of the apparatus shown in Figure 6. The atomizer was inverted as shown, the small bulb attached and filled with sample using a small dropper, the air pressure turned up to 25 pounds per square inch, and the time of flow of the liquid from the first to the second mark on the bulb measured with a stop watch. The time of atomization of the same volume of water was also deterIO0
I
I
A. \
8.
I
I
I
I
I
I
1
CORRECTED FOR VISCOSITY UNCORRECTED FOR VISCOSITY
-
-
EFFECTS OF VISCOSITY VARIATIONS CAUSED BY SOME ORGANIC SOLUTES ON FLAME INTENSITY
Experimental Procedure. The effects of glucose, urea, sucrose, and gelatin upon the emission of sodium, potassium, and calcium were determined by measuring the flame intensities of 100 p.p.ni. solutions of the metals, which separately contained varying amounts of one of the organic substances, against the corresponding pure 100 p.p.m. solutions of metal salts as reference standards. Each flame intensity was converted to met,al concentration espressed as parts per million by niwns of t,he appropriate calibration curve, Figure 1. The interference effects of each of the organic substances on the determination of each of the metals a w shown i n columns three of Tables 111, IV, and V and by curveB i n Figures 2 to 5 , in which all the data are plotted but only the, curve for potassium is drawn. of the sodium and potassium data were obtained n-hen atoniizer I n-as in use, with the esccption of the gelatin interference data. rill calcium measurements n-ere obtained when atomizer I1 n-as in use. In no case did all!. of these organic substances contribute to the background emksion of the flame at the wave length and slit Pettings w e d f o r thcx metals, as shown tiy running I)lanlis. Rates of Atomization. .It this phaw of the work, it was thought that a large part of the interference caused by each of the organic. solutes was due to increased viscosity and consequent, decrease 'n rate of atomization of solution into the flame.
Figure 1. Calibration Curves for Sodium, Potassium, and Calcium Chloride
CALCIUM 0 0
0.4
I
I
1.2
0.8
MOLARITY
I
1.6
GLUCOSE
2 .o
Figure 2. Effect of Glucose on Metal Found in a Standard Solution Containing 100 P.P.M. of Metal
Table 111. Corrections for 1-iscosity Interferences Caused by Contaminants in Solutions Containing 100 P.P.M. Sodium Concn. of Interfering Flame Intensity Solute Sucrose, molarity 0.0000 100.0 0.0293 94.2 85.0 0.0882 78.4 0.149 62.8 0.303 51.1 0.464 40.9 0.632 32.8 0.800 27.5 0.914 25.2 0.988 Glucose, molarity 0.0000 100.0 0.0500 95.1 n io00 89.4 82.5 0.2000 66.0 0.5000 55.6 0.8000 49.3 1,000 36.3 1,500 26.3 2.000 Urea, molarity 0.0000 100.0 0.0500 97.3 95.6 0.1000 92.6 0,2000 87.4 0.4000 82.3 0.6000 0.8000 78.8 1.000 74.8 1.600 67.2 2,000 61.0 3.000 51.5 4.000 46.1 Gelatin, ing./ml. 0.0 100.0 1.0 94.0 91.2 2.0 4.. 0. 87.6 86.2 6.0 84.1 8.0 81.4 10.0
Sodium Found, P. P. 11.
Corrected Correction Sodium Ratio Found, P.P.31.
100.0 89.3 73.8 64.2 44.5 32.4 23.6 17.5 14.2 12.7
1 000 1,020 1.065 1 122 1.276 1,482 1.768 2.135 2.483 2.725
100.0 91.1 78.6 72.0 56.8 48.0 41.7 37.4 35.3 34.6
100.0 90.7 81.0 70.0 48.0 36.6 30.6 20.0 13.4
1.000 1,020 1.040 1.081 1.230 1.407 1,551 2.030 2.780
100.0 92.5 84.2 75.7 59.0 51.5 47.5 40.6 37.2
100.0 94.6 91.6 86.3 78.5 69.7 64.6 59.0 49.2 42.3 32.8 27.8
1.000 1.002
1.013 1.016 1.020 1,025 1.050 1.056 1,099 1.149
100.0 94.8 92.0 87.0 79.5 70.8 66.9 60.5 51.7 44.7 36.0 32.0
100.0 88.8 83.4 77.1 74.5 71.3 67.2
1,000 1,050 1.092 1,200 1.278 1.433 1.568
100.0 93.2 91.1 92.5 95.2 102.2 105 4
1.004 1 ,008
ANALYTICAL CHEMISTRY
808 mined between each two measurements for the solutions to check consistency. Five or more measurements of each solution were made and the results were averaged. Results were obtained with an average error in precision of not more than 1 % of the reported result. Viscosity Correction Ratios. The time of atomization of each solution was divided by the time of atomization of water measured a t the same temperature. The resulting ratio was a measure of how many times slower than water the solution was atoniIO0
I
I '
1
4
0
1 I I I I I CORRECTED FOR VISCOSITY 8. UNCORRECTED FOR VISCOSITY I
A.
-
ized. \-iscosity ratios were also measured but they exceeded time ratios, up to 13'% for the more concentrated solutions. Since each ratio of the time of flow through an Ostwald viecometer was found to approximate closely the atomization time ratio for the same solution when atomizer I was used and since atomization rates were comparatively difficult to obtain, viecometer time ratios determined at the same temperature were used for this atomizer. The hydrostatic pressure of the solution in the bulb, Figure 6, caused negligible errors when atomizer I was used with 25 pounds per square inch of air, which was the only air pressure used with this atomizer. The contribution of the small amounts of the metal chloride present in the solutions to the rates of atomization was negligible in all cases, as shonn by running blanks. With the installation of atomizer 11, the atomization time ratios were no longer equal to the visconieter time ratios, showing the equality of the ratios for atomizer I to be coincidental. Therefore, for solutions atomized with the newer assembly, it was necessary to determine directly the rates of atomization. It was found that the pressure chusing suction of sample into atomizer dropped
CALCIUM
Table IV. Corrections for Viscosity Interferences Caused by Contaminants in Solutions Containing 100 P.P.M. Potassium Concn. ,of Interfering Solute
Flame Intensity
Potassium Found, P.P.M.
Correction Ratio
Sucrose, molarity 0.0000 0 0300 0.0600 0 1000 0.3000 0 4500 0 6000 0 8000 1 000 Glucose, molarity
100.0 90.2 80.4 71.5 45.4 30.4 20.8 11.6 7.0
100.0 89.0 78.8 69.9 44.5 32.4 24.5 15.6 10.5
1.000 1,020 1.040 1.070 1.276 1.455 1.715 2.160 2.750
100.0 90.8 82.0 74.8 56.8 47.1 42.0 33.7 28.9
100.0 90.7 82.6 69.2 44.0 32.2 26.0 15.3 8.9
100.0 89.8 81.4 67.7 43.2 33.8 29.0 19.3 13.9
1.000 1.020 1.040 1.081 1.230 1.407 1.551 2.030 2.780
100.0 91.6 84.6 73.2 53.1 47.6 45.0 39.2 38.6
100.0 97.1 94.0 86.4 79.0 72.4 65.1 60.7 48.5 43.1 31.1 22.3
100.0 96.6 93.4 85.3 77.5 70.8 63.5 59.3 47.3 42.2 33.0 25.8
1,000 1,002 1,008 1.013 1.016 1,020 1.025 1.050 1.056 1,099 1.149
100.0 96.8 93.8 86.0 78.5 71.9 64.8 60.8 49.7 44.6 36.3 29.6
100.0 92.1 88.0 83.5 80.3 75.4
100.0 88.4 82.7 77.6 73.6 68.0
1.000 1.045 1.092 1,200 1.320 1.470
100.0 92.4 90.3 93.1 97.1 100.0
n onoo
0.0500 0 1000 0 2000 0 5000 0 8000 1 000 1.500 2,000
Urea, molarity 0 0000 0 0500 0 1000 0 2000 0 4000 0 6000 0 8000 1 000 1 600 2 000 3 000 4 000 Gelatin, mg./ml. 0.0 1.0 2.0 4.0 6.0 8 0
1 ,004
Corrected Potassium Found, P.P.M.
CALCIUM
m
I
0.8
I
I
2.4
1.8
I
3.2
MOLARITY UREA Figure 4. Effect of Urea on Metal Found in a Standard Solution Containing 100 P.P.M. of Metal
8. UNCORRECTED 0
1
2
3
FOR 5
VISCOSITY 8
7
8
9
1
M G ~ M L . . GELATIN Figure 5. Effect of Gelatin on Metal Found in a Standard Solution Containing 100 P.P.M. of Metal
0
V O L U M E 26, NO. 5, M A Y 1 9 5 4
809
from 130 to 90 cm. of water when atomizer I was replaced by atomizer 11, because of the lower air press u r e ( T a b l e I ) required for atomizer 11. T h u s f o r solutions atomized at 20 pounds per square inch of air pressure through atomizer 11, e r r o r s caused by hydrostatic pressure became app r e c i a b 1e using t h e apparatus shown in Figure 6. Consideration of these errors led to designing the app a r a t u s s h o w n in Figure 7 , i n which 5 PSI.) d i f f e r e n c e of t h e height of the solution before and after atomi(J z a t i o n , a s well as Figure 6. Apparatus for Deterhydrostatic pressure, mining Rates of Atomization was held to a minimum. Atomizer ratios for gelatin solutions were not determined because of coagulation of the gelatin and clogging of the atomizer tip during the relatively long time interval required to empty the bulb (Figure 7 ) as compared with the short time required for flame intensity measurement. For this reason ratio- for the
gelatin solutions were measured with the viscometer, giving values estimated to be up to 5% higher than if they had been measured in the usual manner. I n correcting for viscosity, it was assumed that apparent concentration of metal found was inversely proportional to the time of atomization. The amount of metal found was multiplied by the correction ratio calculated for the particular solution to obtain the value corrected for viscosity. The corrected apparent concentration values and data used for their calculation are listed in Tables 111, IV, and V. The corrected values are plotted a* curves A in Figures 2 to 5. As with curves B all data were plotted but only the curve for potassium was drawn. However, the corrected values for urea solutions were omitted because of lack of room on Figure 4 and a dotted line was drawn to indicate potassium values. EFFECTS O F SPRAY PARTICLE SIZE VARIATIONS CAUSED BY SOME ORGANIC SOLUTES ON FLAME INTENSITY
I t is conceivable that the organic substances causing interferences might change the nature of the spray entering the flame and thus affect the flame intensity. Therefore, the spray particles from a series of solutions containing the foreign substancea were collected, examined, and compared.
Table 1.. Corrections for 1-iscosity Interferences Caused by Contaminants in Solutions Containing 100 P.P.M. Calcium Concn. of Interfering Solute Sucrose, molarity
0.0000 0.0250 0.0500 0.1000 0,2500 0,4000
0,6000 0.8000 1.000
Flame Intemity
Calcium Found, P,P.M,
Correction Ratio
Corrected Calcium Found, P.P.M.
100.0 90.4 83.6 70.8 50.0 38.3 29.1 20.5 15.5
100 0 89.5 82 3 69 3 48 0 36 6 27 5 19 1 14 i
1.000 1.004 1.010 1.030 1.130 1.290 1.580 1.963 2.496
100.0 89.9 83.1 71 4 5; 2 47 2 43 4 37 5 36 7
100.0 89.9 81.6 66.9 46 3 36 1 31 6 23 0 17 3
100.0 89.0 70.1 65.3 44.5 34.5 30.0 21.5 16.0
1.000 1.010 1.030 1,062 1.185 1.340 1.470 1.897 2.531
100 0 89.9 81 5 69 3 52 7 46 2 44.1 40 8 40 5
100 96 91 81 72 64 58 55 45 40 32 27
100 0 93 5 91 2 80 3 71 3 62 3 57 0 53 5 44 2 39 2 31 3
1.000 1 ,002 1,004 1.007 1.012 1.015 1,020 1.027 1.047 1.061 1,070 1.161
100 95 91 80 72 63
0 i 6 9 2 2
04 46 41 33 30
9
1.000 1,023 1.045 1.093 1,199 1 320 1 474
100 0 95 6 93 5 92 7 91 8 94 4 98 0
Urea, molarity 0 0000
0 0 0 0
0
0500 1000 2000 4000 6000 8000 000 600 000
0 1 1 2 3 000 4 000 Gelatin, mg ./ml 0 0
0 5 1 0 2.0 4.0 6.0 8 0
.
0 0 9 8 8 1 8 3 9
8 9 7
100.0 94.0 90.5 85.8 78 2 73.1 68 2
26 1
100 0 93 5 89 3
84 76 71 66
8 6 5 5
ps
1
3 6 5
3
Figure 7 . Modified Apparatus for Determining Rates of Atomization through Atomizer I1
Apparatus Modifications. I n order to make conditions iiniildi
to those encountered when flame measurements were made. the spray was collected above the burner with the burner unlighted. The chimney above the burner was removed to gain access to the burner and steam was passed through the colling coil surrounding the burner to prevent condensation of the atomized particle,. Atomizer I1 was used and the spray chamber heater was .et at 100 volts and the air pressure set at 25 pounds per square inch. Collection of Spray Particles. Sucrose, glucose, urea, and gelatin solutions were each atomized into the heated spray chambet. The resulting spray particles were collected on clean microscope slides placed upon a frame built above the burner a t about the height of the spectrophotometer n indow. h small card stopped the spray and was removed when the sample was to be taken. Thiq technique ensured an identical position of each slide above the burner and accurate control of the time of collection. The time was varied somewhat to collect samples with easily observable particles evenly distributed over the slides. Approximate collection times for solutions of each contaminant &ere2 second- for
c
A
Effect of Sucrose on the Spray Particle Size of Solutions Atomized into Flame E’ipure 8.
1. 1.OM R. O.5M
D
the two most concentrated, 6 seconds for the two most dilute, and about 4 seconds for the solution of intermediate concentration. Photomicrographs with magnification of 100 diameters, made with dark-field illumination, from the slides are shown in Figures 8 to 11. Someofthelargest particles which are elongated in Rhape arc the result of two or more particles coalescing but those spherical in shape are undoubtedly the original particles. To be certain the particles on the slides were not the result of several particles coalescing, a slide was passed over the burner for only a fraction of a second when 1.OM SUCIOS~ was being atomized. Observation under the microscope of the midely spaced particles thus oolleeted showed they were about the mme size as those in the photomicrograph of 1.OM sucrose ~ p m yparticles. Other sprays were colleoted in the same manner to provide cheeks and in all Cases similar results were obtained. These short-exposure spray mmples did not lend themselves well to photographic reproduction, however, since the number of particles in the field of the microscope was too few to give a representative sampling of the spray. The irregular shape of the urea 8pmy particles was due largely to the crystallization of the urea from the spray droplets within a few minutes after they vere collected, with no observable change in the relative size of partioles. Gelatin solutions containing less than 6 mg. per ml. produced particles not visible a t 100 diameters magnification. Only the spray samples of the solution containing the highest concentra tion, 10 mg. per ml., were reproducible on the photographic plate. However, observation of the sprays from solutions containing less thau this under higher magnification showed that particle size decreased as the concentration of the gelatin decreased.
c.
0.2M
D. O.IM
Addition of Isopropyl Alcohol. Solutions of 0.51K sucrose, each containing 100 p.p.m. of sodium and varying amounts of isopropyl alcohol, were made up and their flame intensities were recorded, using a pure solution of 100 p.p.m. of sodium its a standard. The data obtained are plotted in Figure 12. The enhancement effect of the alcohol was expected, since the 1it.erat,ure has shown that alcohols of lower molecular n,eight under other oonditions enhance the emission of solutions of met,als atomized into the flame ( 1 , 8, O2 10). Samples of spray particle8 from the sucrose-isopropyl alcohol solutions were collected and examined under a microscope to determine if the enhmcemcnt effect of the isopropyl alcohol was accompanied by a decrcaae in particle size. The particle sizes were found to decrease progressively as the concentration of the slcohol W&P incmmcd up to 50% by volume, sucrose concentration being ,held constant. The maximum flame intensity, however, wa8 reachod at alcohol concentrat,ions between 5 and lo%, after which it progressively decreased. Evidently an additional factor preventod continuous increase in flame intensity with decrease of particle siee. This factor was shown to be viscosity. The pmsence of 50% isopropyl alcohol by volume lowered the rate ai atomization of 0.5M sucrofie over 50%, while the flame intensity was depressed only lo%, perhaps largely heeausc of the countoracting effect of decreased particle size. The spray particle samples of the sucrase-isopropyl xleohol solutions were stable for only about 10 minutes, since they apparently absorbed moisture with inoreasing rapidity from the atmosphere for they increased in size and often coalesced aft.er that period of time. Since the samples could not be photomicrographed in less than 30 minut,es from the sample collection time because of location of equipment, exact rrproduction of the samples was impossible. However, photomicrographs of the samples were made after absorption of water and they showed that the particle sisea were still considerably less than the sines of spray particles obtained from 0.5M SucroEe containing no isopropyl alcohol, a8 shown in Figure 13. DISCUSSION OF RESULTS
Viscosity. The offcct of the prrsence of sucrose, glucose, urea, or gelatin in aqueous solution of sodium, potassium, or calcium WBPIin every case R diminution in flame intensity. This depression in flainr intensity inerrnsed :LS thc conemt.ration of oramio sub-
V O L U M E 26, NO. 5, M A Y 1 9 5 4
811
R
A
Figure 9. Effect of Glucose on Spray Partiole Size of Solutions Atomized i n t o Flame Z.0M 6. 1.0.M
C.
A.
n.
0.SM 0.1M
nt.ance WBR increased. For B given concentration in grams pcr liter, gelatin produced thc grcutest effect,. The minimum amount of each organic contaminant found to cause significant depression in thc flame intemit,y of csch ineta1 is listcd in Table VI. A depression of 2% of thc olrncrved flame intensity i8 conaidered to he significant, inasniuch as the readings were usually made with a precision of %0.5% of their values.
Table VI. Concentration of Organic Solutes Causing a Significant Depression (2%) in Flame Intensity (I" "'C. 1nt,crierin.R Subrtanoe SUClVX
(il"WSe Urea Golatin
Sodium Flame 8.4 3.6 2.0 0.25
,>Or ml.)
I'othssiul" l%,rne 1.9
Caloiulll Flame 1.7
1.8 1.8 0.2
1 8 1.2 0.1
Application of viscosity corrections to the "metal found" in t,he solutions containing the organic contaminants showed that only a part of the interference o m he attributed to viscosity. This is in agreement with thc work of Bills et al. ( 8 ) on the interference of glycerol in sodium determinations. When viscosity corrections were applied to the gelatin interference data, eorrocted C U ~ V C Rapproached that of the solution containing no gelatin. The peculiar shapes of tho corrected curv8s may be partly attributed to thefrtct that viscositycorreetionratiosusedwere compare tively inaccurate because of thc instability of gelatin solutions, previously explained. Thc corrected c u m e ~ for all the glucose and s u c m ~ esolutions level off at the higher concentrations of tho contaminants, showing that interferences other than v i s cmity were a,pparently becoming constant, whereas the viscosity error continued t o increase. The int,erferonceof each of bhe compound8 i8 a sum of several f:tctors and each compound produced its own characteristic interference: Gelatin produced an interference which appeared to be due m o d y to viscosity, while viscosity accounted for a much smaller part of the interference caused by the other contnminants; urea. caused large interference in determination of t,he metals, but little of i t was a cnnsequence of the viscosity of the solution. Spray Particle Size. solution^ containing large amounts of sucrose, glucose, urea, arid galatin produced larger spray particles than solutions containing Imer amounts of each solute. Thc diffcrenecs in spray particlr sim are noticeable among solutions
of higher conccntrxtion, but as they become more dilute the differencos are much more pronounced. The solutions containing isopropyl alaohol, which enhanced flrLme intensity, produced smaller p a t i d e s . The effect of isopropyl alcohol upon the flitmc intensity was a combination of a t least two f a c t o r e a n enhancement effect prohably due to a decrease in particle size and a depressing effect caused by the increase in viseofiity of the solution. Concentrations of alcohol :&hove10% increased the viscosity to such an extent that, the enhancement effoet of the alcohol was overcome by the decrease i n the rate of atomization of the sample. If it KCE assumed that, d l t,ho solvont was evaporatocl from a spray particlo when it passed through thc spray chamber Icaving t,hc solute in one mass, tho f i n d siee of the particles ivould be equal t o the volume of thc dvy d u t o in the particle. A 100 p.p.m. sodium solution droplet of a given 8i.e would he rcduced to an extremely small particle of sodium chloride, while the same size droplet of a LOM SUC~ORCnolution containing 100 p.p.m. of sodium would be reduced to a largcr particle of sucrose and sodium chloridc. Most of the sprays leaving the spray chamher were not dry, howcver, but conaistcd of smsll droplets of fbe part.ially waporated solutions. Substances such as sucrosc in thc solutions deeream tho vnpor pressuue of the solvent and prevent the particles from bring coinpletely reduced to a dry &ate when passing through the spray chamber. Since the vapor prefisure of the solution is prnporbional i o the mole fraction of the solvent,, the particles would tend to bccomc more dry as the amount of the mlute in t,he solutions decreased. Conversely, the presence of isopropyl alcohol inoreases the vapor pressure of the solvent and more of it would evaporate in the spray chamher, resulting in production of smaller particles. This effect would increase with increasing alcohol
ANALYTICAL CHEMISTRY
812
C
A
Figt,re I!O. Effect of Urka on Spray Particle Size of Solutions Atomized into Flame A. 4.OM B . 1.OM
C. 0.5M D . O.%M
Figure 11. Effect of Gelatin on Spray Particle Size of Solutions Atomized into the Flame
concentration. It is reasanable to assume that the spray particles which are in the farm of liquid droplets require a longer time for their metal content to become activated in the flame, thus lowering flame intensity of the metal. Surface tension does not play an important part in changing the spray particles size for solutions containing sucrose, glucose, or urea. The most concentrated solutions of these contaminants have surface tensions which do not differ more than about 2 dynes per cm. from the surface tension of water. However, the addition of isopropyl alcohol lowers the surface tension greatly and this factor would be expected to contribute ti, the production of smaller spray particles which were observed on slides prepared from solutions containing isopropyl alcohol.
case he explained by viscosity alone, although ratios of times required for atomiaation will carrert for mast of the flame depre+ sion produced by gelatin. The depressing effecton the flame intensity of each metal studied is almost the same per gram of sucrose as per gram of glucme. This indicates that for similar compounds the effect is a function of mam rather than molarity of cont,aminad. Tn Rnme of the cases studied viscosity muses only a small part of the interference, the remainder of the interference being caused by the individual effects of each of the contaminants. Much of the
75r
CONCLUSIONS
T h e effects of mcrose, glucose, ures, or gelatin on the determination of sodium, potassium, or calcium by flame spectrophotometry have been studied. Each of these organic contaminants produces no measurable background illumination a t the instrument settings where sodium, potassium, or calcium was measured. Each organic contaminant produces a progressively greater decrease in flame intensity for each of the metals as the concentration of contaminrtnt in increased. This interference e m in no
"0
10 20 30 ISOPROPYL ALCOHOL,
40
%
0Y
50 60 VOLUME
Figure 12. Effect of Isopropyl Alcohol on Flame I n t e n s i t y of a 100 P.P.M. Sodium Solution 0.5M in Suomse
V O L U M E 26, NO. 5, M A Y 1 9 5 4 interference caused by the organic solutes is apparently due to their effects upon the nature of the ~ p m yentering the flame and their consequent hindering of the activation of the metal by the flame. Partial absorption of the radiation produced by the metal or partial dissipation of flame energy by the contaminant are other possible causes of the obwved effects.
813 effects more rapidly than it decreases flame inteneities m u d by metals. In cases where the chemical composition of the samples vary considerably, the procedure of compounding standard solutions is not practical. The compounding of standards is applicable only when one can accurately predict the composition of the samples. If the sample is viscous, the compound causing the viscosity must be used in making up atandmd8, since that particular compound will probably cause significant interferences because of factors other than viscosity. Indiscriminate use of substances in standards to approximate the chemical and physical characteristics of the sample is likely to result in gross errors. ACKNOWLEDGMENT
This work was financed in part b y a Frederick Gardner Cattrell Grant from the Research Corp. LITERATURE CITED
(1) Berry, J. W., Chamell, D. G., and Barnes, R. B., IND. ENG. CHEM.,ANAL.ED.,18, 19-24 (1946). (2) . , Billa. C. E.. McDonald. F. G.. Niedermever. W.. and Sehwarts, M. C., ANAL. CKEM.; 21, 107&80 (1949):
Figure 13. Effec:t of Isopropyl Alcohol on Spray Particle; Size of a 0.5M Sucrose Solution Ata,mized into Flame W^ "..".-"I -^+L^-l .e ,,.retting " " ~ -yv LCL.sLa. l.lrU.lYY empirically for th e interferences of organic contaminants seen..7 1 ,Mmihla. ~ - ~ ~thmref, ...-~ ...om, . " ,it appetL1S that the best method of eliminating such interferences encountered in actual analysis is to rcmove the interfering substance. dilute the 6amde until interferenee is neelisible. or compoundthe standard to approsimate the compositi&of the sample. Ai it is usually impractical or impossible to remove the interfering substance satisfactorily, it becomes necessary to use one of the other procedurefi. The dilution method is best where the metal is of sufficiently high canoentration t0 give accurately measurable resulk after dilution. Diluting decreases interference
(3) Brown, J . G.. Lilleland, O., and Jackson. R. K., Pmc. Am. Soc. Hmt. Sci.. 52, 1-6 (1948). . . 1530-3I (4) Conrad, A. L.. and Johnson. W. C., ANAL. C H ~ M22, (1950). ( 5 ) Fox, C.L., Jr., Ibid.. , 3 . 2 137-42 (1951). .~ (6) Gilbert, P. T.,Jr., Hawes. R. C.. and Beokman, A. u.. 1otu.. za, 772-80 (1950). (7) Hinsmrk. 0. N.. Wittwer, S. I+., and Sell, H. M., Ibid.. 25,320-2
(1953). (R) Inman, W. R., Ropers. R. A , , and Fournier. J. A,. Ib