Determination of Microgram Quantities of Strontium in Solution Evaluation of Flame Spectrophotometric Method ALBERT E. TAYLOR and HAROLD H. PAIGE ldaho State College, Pocatello, Idaho Hinsvark and associates ( 3 ) . I n the latter case the minimum quantity of strontium measured was over 200 p.p.ni. I t way decided to investigate the following factors which may affect the determination by this method: wave length, type of detection unit, type of fuel and its pressure, and interferences by a few common elements.
Because of the widespread occurrence of small quantities of strontium, its chemical resemblance to calcium, and its occurrence in the bones of man and animals, there is a need for reliable methods to detect and measure microgram quantities of this element. A study of flame photometric methods including two different fuels, acetylene and hydrogen, and different lightmeasuring devices has been carried out. A few interfering elements have been included in the study. A statistical evaluation of the results indicates the general superiority of a photomultiplier detection unit (Photovolt Model 520-M), acetylene as the fuel, and a wave length of 681 mp with a Beckman Model DU flame photometer attachment. Average standard deviations of 0.5 p.p.m. or less in the range 0 to 10 p.p.m. strontium have been obtained. The method can be readily applied to solutions containing strontium in this range after preliminary experience with the equipment and interferences.
EXPERIMENTAL
Method. The interferences of five different ions-calcium, sodium, magnesium, aluminum, and ferric-on the determination of strontium in the range of 0 to 10 p.p.m. strontium were studied a t three fixed ratios to the quantity of strontium taken. Of these three ratios, the first was chosen aft,er consulting Vnited States Geological Survey publications ( 4 ) to obtain the maximum ratios of these ions which occur in natural waters throughout the United States. This maximum was substantially increased, ranging from approximately threefold for sodium ion to 66-fold for ferric ion. The other two ratios were chosen for purposes of determining interferences: one represented an arithmetic ratio-e.g., 0 to 10 p.p.m. strontium plus 1000 p.p.m. test ion; the other represented equal parts per million of each, strontium and test ion. S o solutions were examined with combinations of these interfering elements. The solutions were examined with a Beckman DU quartz flame spectrophotometer, utilizing as detection units the phototubes and a Model 520-11 Photovolt photomultiplier photometer. Each of the fifteen solutions, as well as the strontium alone, was atomized int,o both hydrogen and acetylene flames and intensities n-ere measured a t each of two wave lengths, 461 and 681 mp. .1 graph \vas plotted for each set and the average standard deviation of strontium was determined for each throughout the range of 0 to 10 p~p.m.strontium as determined from the straight line most nearly connecting the experimentally derived points.
S
TRONTIUM occurs widely to the extent of about 0.02yo in the earth’s crust, and on the average there appears to be about 0.5% as much strontium as calcium in agricultural soil. Because of its chemical resemblance to calcium, and hence because of its availability to plants in appreciable quantities and its occurrence in the bones of man and animals, there is need for reliable methods to detect and measure microgram quantities of this element. I survey of available literature failed to disclose any thorough investigation for a rapid determination of strontium in solution a t concentrations of 0 t o 10 p.p.m. The flame photometric method xould seem to be most promising for this purpose. I t s use is mentioned by Beckman Instruments, Inc. ( I ) , and
Table I.
Determination of Strontium Using Photomultiplier Unit 10
Sr found, p.p.m.
7 Std. dev.
Sr found,
Conditionsa p~p.m. Sr only 0.6-mm. slit 9.85 0.15 7.00 Sr Ca 10.20 0.60 6.95 1000 p.p.m., 461 mp, 0.06-mm. slit 400 X p . p . m . S r . 4 6 1 m ~ , 0 . 0 5 - m m . s l i t 9.93 0.19 7.20 9.90 1 X p.p.m. Sr.,0.6-mm.slit 0.12 6.92 Sr Na 1000 p.p.m.. 0.5-mm. slit 10.15 0.32 6.80 400 X p.p.m. Sr, 1.0-mm. slit, 12 Ib. 02,sIb. HP 9.65 0.11 7.00 1 X p.p.m. Sr 0.5-mm. slit 9.63 0.19 7.33 Sr M g 200 p.p.m., 2.0-mm. slit. 12 lb. 02,6 lb. H2 10.30 0.28 6.95 100 X p.p.m. Sr, 0.5 m m . slit 9.90 0.38 7.45 1 X p.p.m. Sr, 0.7-mm. slit, 10 Ib. 02, 4 Ib. CzHz 9.95 0.21 7.18 Sr .41 100 p.p.ni., 2.0-mm. slit, 12 Ib. 02,7 lb. Hz 10.95 0.44 6.20 50 X p.p.m. Sr, 2.0-mm. slit, 12 lb. 0 2 , 7 lb. Hz 10.35 0.33 6.80 1 X p.p.m. Sr, 0.8-mm. slit, 10 lb. 02, 4 lb. CzHz 10.35 0.49 6.80 Sr Fe 100 p.p.m. 0.7-mm. slit, 101b. 0 2 10.25 0.15 6.80 50 X p.p.m. Sr, 0.5-rnm. slit, 10 lb. 0 2 , 4 lb. CzHz 9.85 0.08 7.05 1 X p.p.m. Sr, 0.6-mrn. slit, 10 lb. 02, 9,65 0.24 7.25 4 Ib. C2Hz Conditions. 681 m p , 8 l b . 0 2 , 4 lb. CzHz except as indicated.
+
+
+
+ +
Std. dev.
Strontium Taken, Parts per Nillion 5 3 Sr Sr found, Std. found, Std. p.p.m. dev. p.p.m. dev.
1
Sr found,
0
Sr found,
p,p.m.
Std. dev.
p.p.m.
Std dev.
0.13
5.05
0.13
3.15
0.15
0.90
0.16
-0.05
0 13
0.33 0.15 0.07
4.90 5.07 5.17
0.13 0.15 0.07
3.20 3.36 3.15
0 31 0.27 0.09
1.00 1.10 1.10
0.39 0.04 0.11
0.05 -0.08
0.16 0 21
0.32
4.90
0.28
3.05
0.42
1.15
0.53
-0.30
0.63
0.13 0.17
510
0.07 0.17
3.16 3.18
0.12 0.19
0.55 1.50
0.06 0.21
-1.75 -0.20
009
5.50
0.16 0.18
4.85 5.20
0.18 0.21
3.00 3.00
0.28 0.15
1.40 1,l5
0.15 0.00
0.45 -0.10
0.17 0.22
0.19
5.30
0.15
3.23
0.25
1.1s
0.16
-0.13
0.34
0.49
4.13
0.26
2.70
0.27
1.20
0.45
-0.20
0 29
0.20
4.80
0.27
2.80
0.20
0.80
0.16
0.36
5.00
0.00
3.15
0.00
1.70
0.36
-0.40
0.43
0.25
5.15
0.26
3.20
0.27
1,35
0.36
-0.20
0.51
0.07
5.13
0.09
3.27
0.05
0.87
0.06
-0.05
0.00
0.21
5.50
0.13
3.30
0.30
1.10
0.35
-1.00
0.34
282
,
0.00
0.30
0 13
0 28
0 55
V O L U M E 2 7 , NO. 2, F E B R U A R Y 1 9 5 5 Table 11. Interfering Elementa None Calcium
Photo Detection Unit Photomultiplier
Wave Length,
1000
Photomultiplier
081
Photomultiplier
681 ti8 1 68 1 461 681 681
p.p.m.
p.p.m. Sr 1 X p.p.m. Sr Sodium
1000
p.p.m. 400 X p.p.m.
Sr
461 681
461 681
Photomultiplier
100 X
Photomultiplier
461 681 461 461 681 40 1 681
100 p.p.m.
50
x
p.p.m. Sr
1 X p.p.m.
6r
Photomultiplier Photomultiplier Red phototube Photomultiplier
461 681 461 415 1 681
Blue phototube Red phototube
401
Photomultiplier
Red phototube Strontium taken, 0 i o 10 p.p.m.
AY.
Std. Fuel Dev. CzHz 0 37 Hz 0 28 CzHz 0 39
0 08 0 10
0 0 0 0
90 30 30 06 1 80 0 30
Hz CzHz CzHz CzHz H2 Hs
0.06 2 00
CzHz 0 23 Hz 0.33 CzHz 0 . 4 9 CzHi 0 . 3 0 Hz 0.45 CzHz 0 . 2 8 Hs 0.17 C B H ~0 . 4 0 Hz 0.22 CzH2 0.30 CzHz 0 . 4 8 Hz 0.10 CzHz 0 . 4 5 Hz 0.06 H? 0.35 Hz 0.43 CsHr 0 30 Hz 0.43 Hz 0.37 C2Hz 0 . 2 0
0 0 0 0
36 19 22 39
461 681 461 461 681 46 1 461
Photomultiplier
Sr
a
Red phototube Photomultiplier Red phototube Photomultiplier
Slit Width, mm. 0 06 2 00 0 40
blp
Sr 200 p.p.m.
p.p.m. Sr 1 X p.p.m. Iron
Red phototube Photomultiplier
Blue phototube
1 X p.p.m.
Magnesium
Effect of Interfering Elements
Concn. of Other Element 0
400 X
283
0.06 0.60
0.70 0.06 2.00 0.07 1.80 0 07 0.30 0 50 0.06 1.50 0 30 0.30 0 30 0.80 2.00 0.30
681 681 46 1 681 081
p.p.m. of strontium and was prepared by a 1 to 20 dilution of the original stock with 3N hydrochloric acid. The 5000-p.p.m. stock standards of the test ions were used directly in preparing the final samples for testing, except in those samples containing equal parts per million of strontium and test ion. In preparing the latter, the original 5000-p.p.m. stock solutions were cut to 50 p.p.m. by a 1 to 100 dilution with 3N hydrochloric acid. The final samples for testing were prepared by placing in 50-ml. volumetric flasks, respectively, 10, 7 , 5, 3, 1, and 0 ml. of 50 p.p.m. strontium, adding the necessary amount of test ion to give the desired ratio, and then diluting to 50 ml. with water. These samples were atomized through the flame with the photomultiplier unit at maximum sensitivity (switch in position zl). The intensity of the flame from each sample burned with hydrogen and with acetylene was measured a t 461 and 681 mp. A series of ten intensity readings was taken from each sample. The ten readings were averaged and a graph was made showing average intensity reading versus parts per million of strontium taken (Figure 1 is representative). The apparent parts per million of strontium found can now be read directly from observed flame intensities. Part of these data appears in Tables I and 11.
Figure 2. Apparatus. A Beckman DU quartz spectrophotometer with a Model 9200 Beckman flame attachment was used, both with and without a photomultiplier detection unit. The photomultiplier attachment was a Model 520-11 Photovolt unit, replacing the phototubes of the Beckman instrument in a manner similar to that described by Collier and Barschel ( 2 ) . Reagents. Materials used were of reagent grade. Anhydrous sodium carbonate, calcium carbonate, magnesium carbonate, strontium carbonate, aluminum metal foil, and hydrogen-reduced iron powder were used in the preparation of the stock standards. Exchange resin-demineralized water was used. Procedure. A stock standard containing 1000 p.p.m. of strontium was prepared by dissolving 1.6848 grams of strontium carbonate in 3N hydrochloric acid and diluting with 3N hydrochloric acid to 1 liter in a volumetric flask. Stock standards of 5000 p.p.m. were prepared by dksolving in 3,V hydrochloric acid 2.3044 grams of sodium carbonate, 2.5000 grams of calcium carbonate, 3.4674 grams of magnesium carbonate, 1.0000 gram of pure iron porn-der, and 1.0000 gram of aluminum metal foil, respectively, and diluting with 3 N hydrochloric acid to 200 ml. in volumetric flasks The strontium solution, which was used in the preparation of the final samples for testing, contained 50 21.5
I
I
I
I
I
I
I
I
I
(
A
Strontium plus sodium (400 X p.p.m. strontium)
461 mp, blue phototube 0.3 slit; 10 pounds oxygen, 4 pounds acetjlene
This procedure was repeated using first the blue-sensitive phototube at 461 mp, then the red-sensitive phototube a t 681 mp as a detection unit. I n these determinations the selector switch was placed in position 0.1, while the sensitivity control was maintained a t five turns from the extreme clockwise position for the acetylene flame and in the maximum counterclockwise position for the hydrogen flame. Six readings were taken for each sample and were treated as described in the preceding paragraph (Figure 2 is representative). RESULTS
Table I shows the results in the determination of strontium in solution by flame photometry over the range of 0 to 10 p.p.m. This table shows the best combination of conditions (wave length used, type of fuel and its pressure, slit width, and type of detecting unit) for the determination of strontium in the presence of some common interfering elements. I n each series the average error is less than 0.55 p.p.m., and the average standard deviation is less than 0.5 p.p.m. Table I1 includes a statement of operating conditions not included in Table I which gave results having average standard deviation not greater than 0.5 p.p.m. in the range 0 to 10 p.p.m. strontium by flame photometry. The maximum average error in this group is 0.55 p.p.m. DISCUSSION
Figure 1.
Strontium plus 100 p.p.m. iron
681 m p , photomultiplier 0.7 slit, 10 pounds oxygen, 4 pounds acetylene
For the flame photometric determination of strontium, the superiority of the Photovolt photomultiplier photometer over blue- and red-sensitive phototubes has been demonstrated in the range of 0 to 10 p.p.m. for the examples studied. I n addition to the data given in detail in Table I for 16 series, there were 26 others utilizing this method which satisfied the criteria for in-
284
ANALYTICAL CHEMISTRY
clusion in Table I, while only ten phototube series of the 64 studied gave results within these limits. I n addition, only eight series with the photomultiplier as detection unit gave average standard deviations greater than 1 p.p.m., whereas 43 of the series studied with the phototube unit were beyond this value. Acetylene was superior as a fuel in most cases where the photomultiplier detection unit was employed. The wave length a t which the most reliable results were obtained appeared to be 681 mM, with the exception of those involving high concentrations of calcium. No study was undertaken of the effect of mixtures of interfering elements nor of the use of other types of detecting units such as the Beckman photomultiplier attachment. Figures 1 and 2 representing the relationship of relative intensity to strontium concentration indicate the manner in which readings on the two detection units used are translated into parts per million of strontium.
Repetition of the procedure in selected cases, including the preparation of new solutions and standards, gave results agreeing with those tabulated within the stated limits. I t appears unlikelp that the flame photometric determination of strontium by these methods can be carried to a lower range of concentration than that included in this study. LITERATURE CITED
(1) Beckman Instruments, Inc., Data Sheet 2 ( N a y 1952). (2) Collier, H. B., and Bawchel, R. P.. ANAL.CHEM.,24, 1030 (1952). (3) Hinsvark, 0. N.,Wittwer, S. H., and Sell, H. M., Zbid., 25, 320 (1963). (4) U. S. Geol. Survey, Profess. Paper 135 (1924); Water S u p p l y Papers 496 (1923), 638-D(1931), 889-E(1949). RECEIYED for review March 0 , 1954. Accepted November 8, 1954, Based on work done for the Atomic Energy Commission under contract .4T(10-1). 310 with Idaho State College.
MicrodeterminatioR of Cobalt in Biological Materials BERNARD E. SALTZMAN Division of Special Health Services,
U. 5. Department of Health, Education and Welfare, Cincinnati, Ohio
The spectrophotometric determination of low levels of cobalt in biological materials of high ash content has previously been hampered by difficulties with interfering metals and by losses of cobalt because of the occurrence of a precipitate at the pH required for complex formation. .4 separation procedure has been developed in which cobalt is complexed with 1-nitroso-2-naphthol under conditions which prevent the formation of a precipitate, and then is extracted with chioroform. The extract is purified by washing with dilute hydrochloric acid, and ashed to yield cobalt practically free from any interfering metal. Final determination is made by a nitroso H salt method which has been improved to give reliable results and very close adherence to Beer’s law. Recovery of microgram quantities of cobalt added to 25-gram samples of bone was 95%. The effect of interfering metals has been reduced to a negligible level. The procedure makes possible accurate analysis of materials of high ash content.
D
U R I S G the course of a toxicologic study, the need arose for the determination of the normal levels of cobalt in tissue and bone. Since these were extremely low (later found to be of the order of 0.01 to 0.1 p.p.m.), a method was required with the highest possible sensitivity, which would tolerate large quantities of extraneous salts and which would be as free as possible from error due to interfering metals. These requirements were not met (especially for bone samples) by the commonly used nitroso R salt method ( 4 , 6, 7-10, 13, 1 7 ) , in which the reagent is applied directly to a solution of the ashed sample to develop a color with cobalt. Sensitivity was inadequate because the physical limits of the solubility of the ashed salts prevented the concentration of the color to the level required. Attempts to concentrate the cobalt-nitroso R salt complex by extraction with various solvents were unsuccessful. The extraneous salts, such as calcium phosphate in bone samples, caused incomplete recovery of cobalt because of the formation of heavy precipitate, to the point of Rolidification, a t the pH required for color development Finally, the specificity of the method was not adequate for the very low level of cobalt to be determined. Similar difficulties were encountered with spectrographic and other colorimetric methods (1-3, 11, 19). An investigation was therefore undertaken to I
develop a method of analysis which would meet the severe rrquirements imposed by bone samples. I t was evident that most of these problems would be solved if the cobalt could be separated from the ashed salts. Separation by precipitation in acid media with 1-nitroso-2-naphthol (16, 18) was not regarded as suitable for microgram quantities of cobalt because of the well-known problems of supersaturation. Separation by extraction using dithizone (11), 2-nitroso-l-naphthol ( d ) , o-nitrosocresol (S),or I-nitroso-%naphthol (5, 12, l 4 ) , was more suitable for small quantities of cobalt, but only the latter reagent could be used a t a sufficiently low pH to avoid precipitation of calcium phosphate with bone samples. Extraction with 1-nitroso-2-naphthol was therefore considered as a means for separating small amounts of cobalt from high-ash materials. Little work has been reported with this reagent, although the bulk of literature on the microdetermination of cobalt is considerable. Paulais (14)removed impurities from the cobalt extract by washing with dilute sodium hydroxide and adsorption in an alumina column. The purified cobalt complex was then ashed and determined photometrically with the same reagent. In a similar procedure which Nichol (12) claimed was more rapid, iron was removed by a preliminary precipitation as ferric phosphate. Other impurities and excess reagent xere removed from the carbon tetrachloride extract of the cobalt complex by a series of six washes with concentrated hydrochloric acid, alcoholic sodium hydroxide, and water. The purified ext,ract was then concentrated by evaporation and cobalt measured spectrophotometrically. In the study described by Houk, Thomas, and Sherman ( 5 ) , preliminary extractions removed iron as the chloride in isopropyl ether, and copper as its dithizonate. Cobalt was then extracted as the complex with l-nit,roso-2-naphthol in carbon tetrachloride, following which it was ashed, and finally determined spectrophotometrically with nitroso R salt. Unfortunately, this paper waR mainly concerned wit,li nutrition and gave no further chemical details. These reports from the literature represent t,he starting point of this investrigation, the latter being preferred because of the greater specificity and colorimetric sensitivity of nitroso R salt. In the procedure given here, excellent recoveries of microgram quantities of cobalt may be obtained routinely from bone samples as large as 25 grams, and the effect of interfering metals, a very important consideration with high-ash samples, has been greatly reduced.