increase in absorbance. At higher p H levels the decrease in absorbance is attributed to incomplete conversion of ammonia to trichloramine, and t o some destruction of chloramines as well as chlorine and hypochlorite by nitrite ion, which becomes a stronger reducing agent. The results of a temperature study are shown in Figure 4. The series of curves, which cover the concentration range of the calibration curve, indicate that samples should be between 23” and 29” C. before a determination is made. Figure 5 illustrates the relationship between color development of the blue starch-triiodide complex and time. Maximum absorbance in most cases is reached in about 10 minutes and then remains constant. Table I lists the results of standard additions of ammonia to natural and treated nater samples. Each value is an average of five consecutive determinations. The samples contained from 0 to 112 parts per billion of ammonia prior to addition. ilmnionia samples were prepared containing a series of substances which might be suspected of causing interferences. The results of this study are given in Table 11. Aromatic and aliphatic amines interfere a t concentra-
0.1 0
1 0
7I 5
IO
15 2 0 2 5 MINUTES
30 35
40
Figure 5. Color development of blue starch-triiodide ion complex as a function of time
tions of 1.0 p.p.m. and below. Iodide ion can be tolerated up to about 1.0 p.p.m., while bromide ion, as shown in Figure 2 , must be removed completely. Iron interferes above 2.0 p.p.m. Cations such as cobalt and copper interfere only when their color masks the blue starch-triiodide complex. Examination of the calibration curve (Figure 1) reveals a negative deviation to Beer’s law. This behavior was observed in a previous instance when
the starch-triiodide ion complex was used in colorimetric methods (4) and also in the standard chlorine curve prepared here to check the trichloramine recovery. A possible explanation may be that more than one triiodide ion is needed per helical unit of starch to produce a blue complex. From the calibration curve, the molar absorptivity calculated for ammonia as trichloramine on reaction with the cadmium iodide-linear starch reagent was 53,200. LITERATURE CITED
Am. Public Health Assoc., Am. Water Works Assoc., Water Pollution Control Federation, “Standard Methods for the Examination of Water and Wastewater,” 11th ed., pp. 62, 88, 167-75,
(1)
1961. (2) Bolleter, LT. T., Bushman, C. J., . 33, 592 Tidwell. P. JV.* h . 4 ~ CHEX (1961). (3) Lambert, J. L., Ibid., 23, 1247 (1951). ( 4 ) Ibid., p. 1251. (5) Lambert, J. L., Rhoads, 8. C., Ibid., 28, 1629 (1956).
RECEIVEDfor review July 16, 1962. -4ccepted October 1, 1962. Taken in part from a dissertation to be presented by Fred Zitomer to the Graduate Faculty of Kansas State University in partial fulfillment of the requirements for the degree of doctor of philosophy. Study supported by Public Health Service grant WP00235-02.
Rubidium Determination in Rubidium-Doped, Thalliated, Sodium Iodide Scintillation Crystal Using Sodium Tetraphenylboron KENNETH J. JENSEN Chemistry Division, Argonne National laboratory, Argonne, 111. To determine milligram amounts of rubidium in sections of a rubidiumdoped, thalliated, sodium iodide scintillation crystal, rubidium is precipitated and estimated as rubidium tetraphenylboron after prior separation of thallium on an anion exchange column. Potassium contamination seriously affects the determination. A flame photometric method was used to estimate potassium included in the rubidium tetraphenylboron precipitate. This permitted a correction to b e made, which allowed a distinction between potassium originally present in the crystal and small amounts of potassium spuriously introduced during the analysis. Techniques used to estimate, eliminate, or set a limit on possible errors in the determination due to cesium and ammonium ions and to large amounts of sodium chloride are given.
1740
ANALYTICAL CHEMISTRY
A
determination of small amounts of rubidium (1 to 10 mg.) in sections of rubidium - doped, thalliated, sodium iodide scintillation crystal, NaI(Tl,Rb), has been attempted in a n effort to resolve a discrepancy in the half life of rubidium-87 found by different workers. This half life is of great interest because of its usefulness in geological dating. A summary of 17 published determinations of this half life is given by Beard and Kelly (1). RlcSair and Wilson (9) have since reported another determination. The rubidium content of three sections of a NaI(T1,Rb) crystal was determined by the procedure described here, and the values were used with integral counting rates for rubidium-87 to determine its specific activity, from which the half life was determined ( I ) . Although this paper is concerned with CCURATE
the details of the determination of rubidium in a particular substance, the discussion of the effects of potassium, cesium, ammonia, thallium, and large amounts of sodium chloride on the determination and the methods which allow an accurate analysis in their presence should prove useful in the determination of rubidium in other materials. RUBIDIUM DETERMINATION I N PURE SOLUTIONS
Sodium tetraphenylboron has been used as a reagent for the determination of certain alkali metals since its discovery by R i t t i g et al. ( I S ) , and details for the quantitative, gravimetric determination of potassium using this precipitant have since been enumerated by a number of workers ( 2 , 6, 8, 11). The precipitation conditions recom-
mended by Cluley (2) for the determination of potassium a t pH 6.5 have been applied t o the determination of rubidium. The results obtained on standards, using the modified procedure described below, are shown in Table I. Reagents. Rubidium standard solution was prepared from Matthey Specpure rubidium chloride by weighing t h e dried salt (110' C.) and diluting t o volume to yield a solution of known rubidium concentration. T h e concentration was verified by addition of excess perchloric acid t o a n aliquot of the solution, careful evaporation t o dryness, and final ignition to rubidium perchlorate a t 250" C . This assay indicated the salt to contain 99.95% rubidium chloride. A spectrographic analysis indicated Li < 4 p.p.m., S a -100 p.p.ni.. and K -200 p.p.m. Cesium \vas determined by neutron activation to be -100 p.p.ni. Sodium tetraphenylboron solution, O . I M , was prepared from Baker Analyzed reagent sodium tetraphenylboron essentially as clescrihed by Cluley ( 2 ) .
The p H of this solution (and filtrates from determinations of rubidium with sodium tetraphenylboron) could not be determined with pHydrion paper. Solutions 0.1 to 0.003.11 in sodium tetraphenylboron, when tested with this paper. always indicated a n apparent p H of about 3, even though bromocresol purple indicator and a pH meter indicated the p H to be 6.5. A solution of an organic salt of a strong base in a saturated aluminum hydroxide solution \Todd not be expected to exhibit a pH of 3 . Authors have varied in their recommendations regarding the preparation of sodium tetraphenylboron solution to be used for the precipitation of potassium tetraphenylboron. One reason for this variation may be the amount of potassium determined-i.e., when relatively large amounts of potassium are precipitated, the small amount of reagent decomposition product included in the potassium tetraphenylboron precipitate is insignificant, but when smaller amounts of potassium are determined the amount is appreciable. Varying brands of reagent (or even lots of the same brand) may yield solutions of varying turbidity. Nevertheless, the use of aluminum hydroxide to clarify the sodium tetraphenylboron precipitating reagent as originally recommended by Kohler (S), and used by others ( 2 , 61, yields a crystal clear solution and. as t h e work described here shows, a very satisfactory precipitant for the determination of rubidium. Saturated rubidium tetraphenylboron wash solution was prepared by vigorously stirring a small amount (>9 mg.) of rubidium tetraphenylboron in 500 nil. of water for at least 2 hours and
then filtering through a fine-porosity, sintered-glass filter crucible. Deionized water, prepared by passing distilled water through a mixed bed ion exchange column, was used. Redistilled nitric acid and hydrochloric acid, reagents ordinarily employed in spectrographic work, were used in the analysis of the crystals. All DreciDitations were carried out a t 21" *-2o c. Procedure. T h e solution containing 1 t o 10 mg. of rubidium is diluted t o 90 ml. a n d 3 drops of 0.1N acetic acid and 3 drops of bromocresol purple indicator [0.01% (w./v.)] are added. T h e solution is then adjusted t o pH 6.5, using O.1N sodium hydroxide. Eight milliliters of 0.1M sodium tetraphenylboron solution is added dropwise, with stirring, over a period of about 3 minutes. T h e precipitate is allowed to s t a n d for about 30 minutes a n d then filtered through a 15-m1., fine-porosity, sintered-glass filtering crucible. After transfer, the precipitate in the crucible is washed five times with small (1- t o 2-ml.) increments of wash solution. The total amount of solution used in the transfer and wash operations is 30 to 50 ml. The precipitate is finally washed with -1 ml. of water. The crucible containing the precipitate is carefully wiped with an acetone-soaked absorbent tissue, dried for an hour a t 110' C., then cooled for 2 hours (covered, but exposed t o the atmosphere), and weighed. The precipitate is then dissolved out of the crucible with acetone. About 25 ml. of acetone sprayed in small increments into the crucible with suction applied, is sufficient to dissolve and wash 25-mg. amounts of rubidium tetraphenylboron completely out of the crucible. The crucible is then wiped, dried, and reweighed as above. The difference in the two weights represents the amount of rubidium tetraphenylboron determined.
It was necessary t o use fine porosity crucibles to retain the rubidium tetraphenylboron precipitate quantitatively. This precipitate is finer grained than potassium tetraphenylboron. The solubility of rubidium tetraphenylboron in water is 18 mg. per liter (6) and i t is recommended t h a t all of the precipitating solution be drained through the crucible before washing, so that rubidium tetraphenylboron is not precipitated from the saturated wash solution by the sodium tetraphenylboron reagent. Filtrates from the rubidium determinations remain clear during the filtration but become opalescent after washing, for this reason. If rubidium tetraphenylboron is dissolved out of the crucible after the crucible and precipitate have been weighed and the crucible reweighed to obtain the weight of rubidium tetraphenylboron. any small particles of lint, reagent decomposition product, silica, fines from an anion exchange column separation, etc., are not counted
Table 1. Rubidium Found in Standardized Solutions of Rubidium Chloride by Recommended Procedure
Rubidium taken, mg. 1.355 1,425
1.057
4.989
5.420 5,300 4.934
Rubidium found, mg. 1.358 1.432 1.392 1.067 1.075 1.052
Per cent recovery 100.2 100.5 97.7 100.9 101.7 99.5 1.067 100.9 Av. 100.2 Std. dev. 1.3%
4.989 5.012 4.961 4.933 5.022 4.989 5.420 5,300
100.0 100.5 99.4 98.9 100.7 100.0 100.0 100.0
99.1 99.9 99.3 Av. 99.80 Std. dev. 0 . 6 %
5,250
4.927 4.900
99.9 100.0
10.68
100.3 100.0 99.7 99.9 Av. 99.97
Std. dev.
0.27,
as rubidium tetraphenylboron. When rubidium is determined in pure solutions, the difference in the weights of the crucibles before the determination and after dissolution of the rubidium tetraphenylboron approximates the error of the two weighings, -0.05 mg. On the other hand, when various operations such as passage through a n anion exchange column, evaporation to dryness, etc., have been performed on the solutions prior to rubidium precipitation, the gain in weight of the crucible after a n analysis is of the order of 0.1 to 0.2 mg. and on one occasion was 1.64 mg. because of the presence of an unsuspected chip of glass. This technique also yields a solution of rubidium tetraphenylboron which can later be examined for possible potassium content. The crucibles were cooled to room temperature in a room where temperature and humidity were controlled (temperature 30" f 1" C. and relative humidity 45 =t 373, and were weighed in this room on a single-pan, semimicrobalance. It was estimated that each weighing could be made to +0.02 mg. When necessary, buoyancy corrections were made to correct for the change in the weight of the crucible as a function of atmospheric pressure change. The need for such correction has been discussed by van Kieuwenburg (12). VOL. 34, NO. 13, DECEMBER 1962
1741
Table II.
Results of Rubidium Determination after Thallium Separation
Test mixture hIg.
90
Thallium
1.082
3.2 1.082 3.2 4.99
16.0
10.82 8
10.82 8.0 4.99 0
4.99 0
-
Rubidium Thallium Rubidium Thallium Rubidium Thallium Rubidium Thallium Rubidium Thallium Rubidium Thallium Rubidium Thallium
Rubidium recovered, mg. ] and the solution of pure rubidium tetraphenylboron (in 50% acetone-50% water) not carried through the procedure (Figure 3, A ) . It was estimated that the average potassium pickup during the course of the analysis was -15 pg. of potassium. I n this determination the rubidium tetraphenylboron precipitates contained a n average of 92 pg. of potassium, which resulted
in the subtraction of 0.84 mg. (potassium tetraphenylboron equivalent of 92 pg. of potassium) from the observed weight of 20.87 mg. of supposedly pure rubidium tetraphenylboron precipitate. DETERMINATION OF CESIUM IN RUBIDIUM TETRAPHENYLBORON
Cesium is a less serious interference in rubidium determination than potassium, because i t is less commonly found and because of a favorable gravimetric factor. The acetone-water solutions of the rubidium tetraphenylboron precipitates from the analysis of the crystals were examined spectrographically and an estimated 0.35% cesium impurity was found. This would represent a 025% error in the rubidium determination] which was not considered significant. For larger amounts of cesium impurity a flame photometric or other method would have t o be used to correct for this source of error in a rubidium determination using sodium tetraphenylboron. RESULTS AND SUMMARY
Table VI.
Chemical Analyses and Integral Counting on Nal(TI,Rb) Crystals
Av. Rb8' integr a1 counting rate, counts/min.
RbS7 specific activity, counts/min. Rbs7 half life, per mg. Rb X 1O1O years
Mass, Total Rb Crystal g. content, mg. Disk (1.9 cm. diam. X 0.63 cm. high) 6.763 10.03 f 0.10 472 f 9 46.8 f 1.0 Cone (1.9 cm. diam. X 0.63 cm. high) 2.874 0.944 f 0.02 43.8 f 1.8" 46.4 f 1.9 Mounted section (1.9 cm. diam. X 2.54 cm. high) 26.76 14.29 f 0.14 671 f: 10 47.0 f 0 . 8 Weighted av.
5.53 f 0.12 5.58 f 0.23 5.51 f 0.10 5.53 f 0.10
Corrected from original, incorrectly stated value (438 f 18) with author's permission.
1746
ANALYTICAL CHEMISTRY
Rubidium has been determined in three sections of a NaI(T1,Rb) scintillation crystal. The rubidium values obtained were used in conjunction with self-scintillation counting data, obtained at another laboratory, t o estimate the half life of rubidium-87 ( 1 ) . The analytical results and counting data are summarized in Table VI, from the work of Beard and Kelly ( 1 ) . The rubidium values are believed to be accurate t o within 1% of the amount present in the two larger crystals. The value obtained on the smallest crystal, which contained -1 mg. of rubidium, is estimated t o be accurate t o within 2%.
ACKNOWLEDGMENT
Acknowledgment is made to J. P. Faris and E. Huff for spectrographic determinations made during this work and to M. Wahlgren for a neutron activation analysis of the cesium content of the rubidium chloride standard. LITERATURE CITED
(1) Beard, G. B., Kelly, W. H., Nucl. Phys. 28, 570 (1961). (2) Cluley, H. J., Analyst 80,354 (1955).
(3) Crane, F. E., Jr., ANAL. CHEM.28, 1794 (1956). (4) Crane, F. E., Jr., Smith, E. A., Chemist Analyst 49,38 (1960). (5) Dean, J. A., Am. SOC.Testing Materials, Philadelphia, Pa., Spec. Tech. Publ. 238,43 (1958). (6) Geilmann, W., Gebauhr, W., 2. anal. Chem. 139, 161 (1953). (7) Kingsley, G. R., Schaffert, R. R., J. Bid. Chem. 206. 807 (1954). (8) Kohler, M., 2: and. &em. 138, 9 (1953). (9) McNair, A., Wilson, H. W., Phil. Mag. 6, No. 64, 563 (1961).
(IO) Reed, M. G., Scott, A. D., ANAL. CHEM.33, 773 (1961). (11) Sporek, K., Williams, A. F., Analyst 80, 347 (1955). (12) van Nieuwenburg, C. J., Anal. Chim. Acta 20, 127 (1959). (13) Wittig, G., Keicher, G., Riickert, A., Raff, P., Ann. 563, 110 (1949). RECEIVEDfor review May 21, 1962. Accepted August 31, 1962. Sixth Conference on Analytical Chemistry in Nuclear Reactor Technology, Gatlinburg, Tenn., October 11, 1962. Based on work performed under the auspices of the U. S. Atomic Energy Commission.
Combustion-Spectrophotometric Method for Determination of Trace Quantities of Sulfur in Metals KEITH E. BURKE and C. MANNING DAVIS Research Laborafory, The lnfernafional Nickel Co., Inc., Bayonne,
,A method which is rapid and highly selective as well as sensitive is described for the determination of microgram quantities of sulfur in nickel, its alloys, and various other metal systerns. The sample is burned in a stream of oxygen in an induction furnace, The sulfur is converted to sulfur dioxide, which is absorbed in sodium tetrachloromercurate(l1) and subsequently determined spectrophotometrically with pararosaniline and formaldehyde. The following variables have been studied to design a routine method for the determination of 0 to 50 p.p.rn. of sulfur: effect of cornbustion time, sodium tetrachloromercurate(l1) concentration, crucible blank, and hydrochloric acid concentration on the sensitivity of the chromogenic system.
I
research increased interest in high purity metals and alloys has brought about a need for accurate and reliable analytical methods for element concentrations below 50 p.p.m. Evidence of this interest is demonstrated by the large number of publications in the past few years. Among these trace elements is sulfur (6, 18). The ASTM has not published a working method for low sulfur-i.e., less than 50 p.p.m.-however, its current interest is evident from the work of a task force on methods for 0 t o 100 p.p.m. of sulfur in steel. The sensitivity and generality desired require separation of sulfur prior to its determination as sulfide, sulfite, or sulfate. Separation can be achieved by combustion (I), by dissolution and subsequent evolution, or by a chromatographic separation. Nydahl (g9) has N METALLURGICAL
N. J.
proposed a chromatographic separation of sulfate prior to the gravimetric finish. This method would require extremely large samples and would probably be suitable only as an umpire method. Luke (20) favored the dissolutionevolution method over a combustion technique because “no way was found to eliminate the high and variable blanks due t o the ceramic crucibles.” His method depends upon dissolution in hydrochloric acid. This is not applicable to many nickel-chromium corrosion-resistant alloys which resist attack by hydrochloric acid alone. Subsequently, special low-sulfur crucibles have been made available which increase the feasibility of a combustion method. The standard ASTM method is based on a combustion and subsequent potassium iodate titration of the evolved sulfur dioxide. The method is designed for the determination of sulfur converted to sulfite, when the content is greater than 100 p.p.m., with an average error of about 10 to 20 p.p.m. (16). Boulin and Jaudon (7) have recently described a combustion-titration procedure that is applicable to most metal samples, including stainless and heat-resistant steels, containing sulfur in the range from 50 to 500 p.p.m. Attempts to improve the sensitivity further by the use of a more dilute titrant fail because of the difficulty in reproducing the proper shade of blue in the starch-iodide system. A photometric system that aids in the reproduction of an end point is commercially available. A coulometric titration has been reported by Hibbs and Wilkins (16). I n this laboratory the problem was to establish a routine method with s a cient sensitivity to detect as little as l to 2 p.p.m. of sulfur in a wide variety of
metal samples. Several chromogenic systems (65) are available for the detection of sulfur in its various valence states: sulfitemethylene blue (12, 87) sulfide - pararosaniline - formaldehyde (21, SO), and sulfate-barium chloranilate (S). The sensitivity of this latter system, even in the ultraviolet (4) region, was inadequate. The large number of experimental parameters of the sulfide-methylene blue system eliminated this as a routine method. Since the evolution of sulfur as sulfur dioxide has been studied by the ASTM, chromogenic systems for detecting sulfite were examined. The fuchsin-aldehyde reaction has been applied to the determination of trace quantities of sulfur dioxide (9,10, 96, 28). Blinn and Gunther ( 5 ) described the reaction as unpredictable and used an indirect ultraviolet determination of sulfur dioxide by means of lead(I1). Their method requires separation of lead sulfide prior to spectrophotometric measurement. The pararosaniline - formaldehyde system is better suited for the determination of sulfur dioxide than fuchsin and formaldehyde (SO), and is capable of detecting 5 to 100 p.p.b. of sulfur dioxide in a 38.2-liter sample of air. The mechanism of this chromogenic system has been elucidated (21). This system was used for the determination of evolved sulfur dioxide in a variety of materials (19). An automatic analyzer (14) for the continuous recording of atmospheric sulfur dioxide was based on the pararosaniline-formaldehyde system. Nitrogen dioxide is the only reported interference. This compound, formed by the breakdown of nitrate impurities, is effectively removed by copper (19). I n the present investigation, microVOL 34, NO. 13, DECEMBER 1962
* 1747