X-Ray Fluorescence Analysis Using Ion Exchange Resin for Sample Support Determination of Strontium in 0.1M Calcium Acetate Solutions ROBERT L. COLLIN Cancer Research Insfifute, New England Deaconess Hosaifal, Bosfon, Mass.
b An analyticul method for the determination of trace amounts of an element in solution using x-ray fluorescence has been developed. The solution to b e analyzed i s passed through an ion exchange column and the resin is removed, pressed into a pellet, and inserted into an x-ray spectrograph where the fluorescence intensities are measured. Variations caused b y errors in weighing, loss of resin, and inhomogeneities in the pellet are corrected for b y an internal standard which can b e easily introduced directly onto the column. The method is illustrated with the determination o f strontium in 1 0-ml. samples of 0.1M calcium solutions. In a series o f 16 such samples containing from 20 to 1000 pg. of strontium (2 to 100 p.p.m.) the standard deviation of the points from a straight line was 5.2% o f the amount o f strontium present. If background corrections are made, the method can b e used for solutions having wide variation in major constituents.
process which is likely to be inconvenient for trace analysis. Recently van Xiekerk, de Wet, and Wybenga have developed a method for trace analysis of solutions (6). Our method was developed independently and, although based on the same principle, makes use of somewhat different techniques. Instead of using a batch procedure the resin is formed into a column and the solution to be analyzed is poured through it in the manner commonly used for chromatographic work. In this way an amount of resin can be used which will absorb all the cations (or anions) in the solution and one is not troubled by unfavorable equilibrium constants for the exchange reactions. A measured volume of solution containing an internal standard can also be added directly to the resin column. Instead of using a liquid cell to hold the sample, the resin is pressed into a solid pellet which can be used directly in the spectrograph and saved for future reference. EXPERIMENTAL
I
with certain experiments on bone mineral a method was needed for determining microgram amounts of strontium in 5- to 10-ml. samples of 0.1-Vcalcium acetate. Since x-ray fluorescence had proved useful in determining the strontium content of strontium-calcium phosphate mixed crystals (Z), an attempt was made to cxtend the technique to solution analysis and in the process develop a general method for the determination of trace amounts of an element in solution. Analysis of solutions directly by x-ray fluorescence is common (1) but the elcmcnt determined must be present in concentrations greater than 0.01% iii most systems. For trace analysis a concentrating procedure is usually nrrdcd, and both ion exchange membranes (3) and evaporation onto filter paper (4) have been proposed. The ion eschange membrane method has the disadvantage of both a long equilibration time and a small exchange capacity, while the filter paper method involvea evaporation of the solvent, a N CONNECTION
Materials. Calcium acetate was prepared from Matthey (Johnson, Matthey & Co., Ltd., London, England) Specpure calcium carbonate certified to contain only 4 p.p.m. of strontium. An excess of the carbonate was boiled n i t h acetic acid, then filtered and diluted to the desired molarity. A strontium stock solution was made up from weighed amounts of analytical grade strontium nitrate accurately diluted n i t h O.1M calcium acetate. Standard samples for analysis were prepared by taking aliquots of the strontium stock solution and diluting to 10 ml. with the 0.1ilI calcium acetate. The analytical grade boric acid was in finely powdered form and a 2-gram pellet of this material showed no SrK, fluorescence. Barium chloride was reagent grade, and the rubidium chloride internal standard was made from Fisher Scientific Co. C.P. grade rubidium chloride. I n the amounts used in the experiments, the rubidium chloride introduced no detectable strontium into the samples. Dowex cation eschange resin, 50W-X8, was mashed on a Buchner funnel with HC1 and then with distilled water to neutrality.
X-Ray Fluorescence Apparatus. -4 General Electric XRD-3 spectrograph with LiF crystal, krypton proportional counter, and a tungsten target x-ray tube operated a t 45 kv. and 50 ma. was used for measuring the fluorescence intensity. A special steel face plate with a round hole 5/8 inch in diameter was made for the sample holder, and the resin pellet mas pressed against this by a plastic plate held in position with a spring. Sample Preparation. A sample of 1.5 grams of the wet resin (Elbout 50 weight % moisture) was weighed out to the nearest 0.1 gram, suspended in water and poured into a 10-mm. borosilicate glass tube to which a 5-mm. exit tube had been attached. The resin formed a column with a total exchange capacity of 0.0038 equivalent on top of a glass wool plug. I n a fern esperiments chromatographic tubes with sinteredglass disks were used; in general, these were much more convenient. After the water level had dropped to the level of the resin, the sample was added along with 1 ml. of a standard rubidium chloride solution containing 280 fig. of rubidium per ml. When the water level had again dropped to the level of the resin, 5 ml. of distilled water was added and allowed to drain through the column. The resin was then washed out into a sintered-glass funnel (M porosity) with a reverse stream of distilled water. While suction was being applied, the resin was stirred n-ith a glass rod to thoroughly mix particles from various parts of the column. When dry, the resin was rinsed with acetone and air was drawn The through for about 5 minutes. resin was scraped out of the funnel into a 5-ml. plastic vial and 1 gram of boric acid (neighed to the nearest 0.1 gram) n-as added to serve as a binder along with a '/?-inch plastic ball. The stoppered vial and contents were shaken on a Wig-L-Bug for 3 minutes to mix the resin with the boric acid. The homogeneous powder was poured into a 3/4-inch diameter cylindrical steel mold and pressed a t 45,000 p.s.i. to give a pellet 3//16 inch thick. A number of esperiments were carried out using a batch method with 1.5 grams of resin. The resin and sample, diluted to 100 ml., were stirred for l/2 hour on a magnetic stirrer with a Teflon-covered stirring bar. VOL. 33, NO. 4, APRIL 1961
605
Table I. Analysis of Standard Strontium Solutions
Sr Found,
Sr Added, /Iff.
20 30
50 70 89 99 149 199 298 397 497 596 695 795 894 993
zBr/zRb
Pg*
0.692 0.718 0.727 0.745 0.767 0.785 0.855 0.922 1.0313 1.160 1.302 1.412 1.520 1.661 1.808 1.887
20 42 48 63 80 95 151 205
298 ~~-
395 509 598 683 796 913 977 ~
~-
~
Table It. Replicate Analyses of a 0.1M Calcium Acetate Solution Containing Approximately 95 pg. Strontium per 10
MI. 8r Found, Pg. 94.3 96.3 93.8 84.8 95.1 99.7 88.9 86.5
ISr/zRb
0.7843 0.7868 0.7837 0.7724 0.7852 0.7911 0.7775 0.7746 0.7775 Mean z B r / z R b strontium.
88.9 =
0.7814
Or
92.0
pg.
Fluorescence Analysis. T h e intensities a t the SrK, position, 25.10'28, and at the RbK, position, 26.60°29, were determined by setting the proportional counter arm at the desired angle and measuring the time necessary to accumulate 16,384 counts. Each intensity was read four times and the results averaged. Background corrections were usually not made and the ratio I S r / I R b was taken as a n indication of the amount of strontium present. IS?and I Rare ~ the measured intensities at 25.10'20 and 26.60'20, respectively.
approximately 95 pg. of strontium per 10 ml. of 0.1M calcium acetate. Nine 10-ml. aliquots were analyzed for strontium by the procedure described above and the results are given in Table 11. The standard deviation was 5.1 pg. of strontium. The completeness of cation pickup by the column was tested in three experiments. In the first, a 10-ml. solution of 0.1.V calcium acetate containing 1000 pg. each of strontium and rubidium was passed through a column and the effluent in turn passed through another column. No strontium or rubidium was detected on the second column in three separate runs, and this indicated that the pickup in the first column had been greater than 99%, since the minimum detectable amount of strontium and rubidium on the second column was estimated to be 10 pg. In the second experiment, various volumes of 0.1M calcium acetate were passed through the column and the effluent was tested for the presence of calcium with ammonium carbonate solution. It was necessary to pass 18 ml. of the calcium solution through before there was a positive test for calcium in the effluent. This is well over the 10 ml. used in the determinations and hence the exchange capacity of the resin exceeded the total cations in solution by a considerable factor. The effect of dilution on the pickup by the column was tested in the third experiment by diluting IO-ml. samples of a 0.1M barium chloride solution containing 100 pg. of strontium and 280 pg. of rubidium to 50 ml. and 100 ml. with distilled water. These two solutions, as well as the undiluted solution, were passed through columns and the strontium and rubidium intensities were read on the resin pellet. There were no significant differences in pickup from the different solutions. Corrections for background could be made but me have favored efforts to keep the background constant rather
RESULTS AND DISCUSSION
A series of 16 samples varying from 20 to 1000 pg. of strontium in 10 ml. of 0.1M calcium acetate was analyzed and the results are given in Table I. A straight line of the form [Srl =
+
MIS~/-IRLJ b
(I)
was fitted to the data by least squares, where [Sr] is the number of micrograms of strontium in the sample and hence the total amount of strontium on the resin. The values of m and b were 800.2 and -533.3 pg., respectively. The standard deviation of the points from a straight line was 5.2% of the amount of strontium present. A solution was made up to contain
606
ANALYTICAL CHEMISTRY
Table 111. Comparison of Column and Batch Methods for Strontium and Rubidium in 0.1M Calcium and 0.1M Barium Solutions
(Solutions containing 84 pg. strontium and approximately 280 p g . rubidium.) Column Batch 0.1M Ca Isr - background 51 55 counts/ counts/ ZRb
- background
Isr - background ZRb
- background
sec.
167
see.
162 0 . l M Ba 23 23 counts/ counts/ sec.
75
sec.
56
than attempt to measure it each time' This has been accomplished by keeping the x-ray tube voltage fixed and by using solutions for the calibration curves that duplicate, in all major constituents, the solutions being analyzed. Introduction of large amounts of different cations into the solutions would lead to variations in background and hence in the ratio I s J I R ~ . If analyses are to be made where major constituents are not constant, then background corrections, involving a considerable increase in counting time, should be made. A comparison of the column and batch method as well as an illustration of the use of an internal standard when background corrections are made is given in Table I11 for small amounts of strontium and rubidium in 0.l.U calcium acetate and 0 . 1 X barium chloride. In these experiments the background was determined by measuring the intensities at 27.60'29, where no fluorescence line appeared, both on samples and on blanks containing 10 ml. of 0.1JI calcium acetate and 0.121 barium chloride but no strontium and rubidium. The and 1?6.6o0/I~i.6~', ratios 1 2 j . 1 0 0 / 1 2 i . 6 0 0 measured on the blank and the measured intensities a t 27.60'20 on the samples were used to compute the background in each case. The poiver of the batch method is shown by the nearly complete pickup of strontium from both solutions, but its weakness is also revealed by the incomplete pickup of the less strongly bound rubidium in the presence of barium. The recovery of rubidium from the calcium solution by the batch method is essentially complete but the recovery from the barium solution is only 757, of maximum. Thus, for certain solutions where an unfavorable exchange equilibrium occurs, the column method might be preferred even though, in its present form, it is slowvrr than the batch method. Preliminary experiments with columns where the solution is forced through under a pressure of 5 to 10 p.s.i. indicate that the column method can be speeded up considerably so that volumes of the order of 500 ml. can be passed through in less than an hour. The values of (Isr- background) ( I R b - background) in Table 111 are 0.305 and 0.307 for the calcium and barium solutions, respectively. This corresponds to a difference in estimated amount of about 1 pg. of strontium and illustrates the usefulness of the internal standard method, if background corrections are made when wide variations in solution composition are encountered. Sources of Error. The very high background, which arises almost entirely from scattering of the primary x-ray beam by the resin and boric acid, contributes greatly to the error. Any increase in the peak to background I ~ bsensitive ratio would make I B r / more
'
to the amount of strontium present and thus increase the precision of the strontium determination. With our apparatus, 300 pg. of strontium is needed to produce a peak height above background equal to the background. The ratio, I B r / I R b , is in turn affected by experimental errors, the chief of which is caused by the inhoniogeneous nature of the sample. Since the resin occurs as discrete particles and the strontium and rubidium are not absorbed equally on each particle of the column, there n-ill be a statistical mixing error that nill depend on a number of factors including the amount of excess resin in the column, the particle size of the resin, and the fraction of pellet effectively contributing to the fluorescence intensity. In any case, the number of particles containing strontium and rubidium, or both, mill be small, and only R fraction of the total pellet nill be
ACKNOWLEDGMENT sampled by the exciting x-rays. The internal standard does not in general The author is grateful to Paul correct completely for the error because Shanley and Judith Dubchansky for the affinity of the rcsin for the unknown carrying out much of the experimental and standard is not the same, and the work. two elements will not be equally disLITERATURE CITED tributed through the resin. An attempt to assess the relative (1) Birks, S., “X-Ray Spectrochemical Analysis, Interscience, Ken- York, 1959. magnitude of the mixing error in com(2) Collin, R. L., J . Am. Chern. SOC.81, parison to the counting and weighing 527.5 (1959). errors was made by measuring Isr/IRb ( 3 j GEGbb, W.T., Zemany, P. I>, iyature ten times on a single preparation but 176,221 (1956). (4) Pfeiffer, H. G., Zemany, P.I)., Ihzd., grinding, mixing, and repressing the 174, 397 (1964). pellet between readings. The variance ( 5 ) van Niekerk, J. X., de Wet, J. F., in I B r / I R b was 19.41 x 10-5 in this Wybenga, F. T., ANAL.CHEM.33, 213 case, while the variance of ten individual (1961). pellets was 39.26 X 10-5. The variance RECEIVED for revien- July 11, 1960. -4cdue to counting was 7.70 X 10-5. cepted December 20, 1960. Work done Even if counting and preparation errors in part under U. S. Atomic Energy Commission Contract AT(30-1)-901. Also w r e made insignificantly small, the supported in part by Research Grant standard deviation due to mixing C-3003 from the National Cancer Instatistics would be 37* in amount of stitute of the National Institutes of strontium. Health.
4,.
Spectrochemical Determination of Trace Elements in Inorganic Salts Using a ConcentrationPrecipitati on Tech nique RICHARD
L. DEHM, WALTER G. DUNN,
and EDWIN R. LODER’
Industrial laborafory, Eastman Kodak Co., Rochester 4, N. Y.
b A concentration procedure using two quinolinol organic precipitants, 8 (oxine) and 2-mercapto-N-2-naphthylacetamide (thionalide), is used in conjunction with a spectrographic procedure for the simultaneous determination of seven elements in potassium bromide, potassium chloride, potassium iodide, potassium nitrate, sodium bromide, sodium chloride, and sodium nitrate. Copper, iron, lead, manganese, nickel, tin, and zinc are determined in the range of 0.1 to 5 p.p.m. The precipitation is carried out in a hot solution at pH 9; aluminum is used as a carrier element.
-
D
spectrographic examination of the inorganic sodium and potassium salts using direct current arc excitation was found to be too insensitive to detect the majority of the impurities at the level a t which they are normally present, which is usually less than 1 p.p.m. This is probably due to the fact that the alkali metals, having low ionization potentials, lower the average energy of the discharge and suppress IRECT
Present address, hiaumee Chemical
Co., Toledo, Ohio.
lines of elements having high ionization potentials. Owens ( 8 ), in a summary of spectrographic methods for trace analysis, recommends the high-voltage, alternating current arc as the most appropriate spectra excitation source for use with inorganic materials. This source has been used for the analyses of sodium hydroxide (1, 3, l l ) , sodium bicarbonate, and sodium chloride solutions (1, l a ) . A solution alternating current arc technique has been used for the analyses of sodium salts, both organic and inorganic, by first converting them to sodium nitrate (9). These procedures failed to give the desired sensitivities for nickel, lead, and zinc below the 1-p.p.m. level and led to the investigation of concentration procedures. Of particular interest JYas a precipitation procedure, prior to spectrographic examination, developed by Mitchell and Scott (5-7). I n their procedure 14 trace elements were concentrated by precipitation with 8-quinolinol, tannic acid, and thionalide in a solution buffered to p H 5.1, containing aluminum as the carrier element. Heggen and Strock (4)reported a similar procedure using indium as the carrier and as the internal standard. Pickett and
Hankins (IO) coinpared the efficiency of these two elements using radioisotopes and reported no differences in carrier characteristics. The concentration procedure described in this report uses two organic precipitants, 8-quinolinol and thionalide. The precipitation is carried out in a hot solution a t p H 9, using aluminum as the carrier element. The spectrographic procedure employs a direct current arc for excitation of the total ash in a matrix of graphite and lithium carbonate with germanium and bismuth as internal standards, REAGENTS
Oxine (8-quinolinol) solution, 5.0% in 2N acetic acid. Thionalide f2-merca~to-iV-na~hthvlacetamide) solution (K*and K Laboratories, 177-10 93rd Ave., Jamaica 33, N. Y.), 1.0% in glacial acetic acid, prepared just before use. Glacial acetic acid, redistilled using a quartz still. Aluminum solution, 11 grams of A1C12.6H10 Der liter of distilled water. Ammoniui hydroxide, reagent grade, 28% NHB. Matrix Ratio. 50 LizCOa 1.0 GeOz 40 graphite 0.30 Bi203:
+
+
VOL. 33, NO. 4, APRIL 1961
+
607
