X-Ray Fluorescence Determination of Strontium in Biologic Materia Is by Direct M atrix-Tra nsmitta nce Correction MARVIN GOLDMAN and R.
P.
ANDERSON
Radiobiology Project, School of Veterinary Medicine, University of California, Davis, Calif. A quantitative method for strontium x-ray fluorescence determination of food and bone is described for weighed specimens mixed and briquetted with organic binders. Transmittance of the briquets as measured with monochromatic Sr Ka radiation provides accurate corrections for the Sr Ka emission of fluorescent x-rays derived from specimens of differing x-ray opacity. Specimens from 0.050 to 2 grams containing from 20 to 2000 pg. of Sr per gram have been analyzed and show the emergent fluorescent x-ray intensity to b e linearly related to the transmittance of the briquet. Rapid quantitation of Sr can thus b e performed on dry samples without prior knowledge of matrix composition with an expected relative error of less than 5%.
S
can be determined quantitatively by several modern techniques, including arc emission spectrography (1I ) , neutron activation analysis ( I ) , flame spectrophotometry (with and without enrichment by cation exchange) (5, 6, 16), atomic absorption spectrometry ( 4 ) , and x-ray fluorescence spectrometry ( 1 4 ) . While each of these methods has proven reliable and accurate, they all require standardization of technique in the use of reference samples of similar chemical composition and physical state to the sample under investigation. Matrix interferences add to the overall effort of quantitation and occasionally hinder accuracy, sensitivity, and precision. Interelement absorption and/or enhancement effects within the matrix are often compensated by addition of internal standards or chemical purification and dilution ( I O ) . This report demonstrates that quantitative s-ray fluorescence determinations of trace concentrations of strontium in biologic materials can provide rapid and accurate measurement of strontium in food and bone (2, 4 , 7 , 1 4 ) . The present method is applicable to dry biologic materials containing strontium and ,offers the advantage of direct determination of matrix influence on each specimen. In biologic materials, the presence of traces of strontium represents that of a n rlement of intermediate atomic number TRONTIUM
71 8
ANALYTICAL CHEMISTRY
within a matrix of relatively low effective atomic number. Thus x-ray absorption within the matrix will increase in proportion to increasing strontium concentrations (12). The degree of internal absorption is related to the mass absorption coefficients of each of the elements within the matrix ( 3 ) . When the elemental composition of the matrix is known, the mass absorption coefficients of the matrix may be calculated from the sum of the coefficients, weighed for the concentration of each element present in the sample. Mass absorption coefficients have been estimated in minerals by the measurement of the Compton scattering ( I S ) . A practical measure of the mass absorption coefficient can be obtained by measuring the transmittance of a monochromatic beam of x-rays by the sample under investigation (15 ) . For such an approach, we have chosen the wavelength of the element being analyzed. This restriction provides a practical solution of the matrix problem by quantitating interelement absorption effects a t the wavelength of the emergent analytical line. If the sample and the spectral energy of the exciting x-ray beam are kept constant, an integrated mass absorption coefficient for any matrix not infinitely thick may be established for the selected set of conditions. Therefore
111,
= e-rz =
e -! m P
(1)
where I , is the incident x-ray beam intensity, I is the transmitted x-ray beam intensity, p / p is the mass absorption coefficient in sq. cm. per gram and m is the mass of the sample in grams per sq. cm. For strontium, transmittance of the K a wavelength of 0.875 h. (14.1 k.e.v.) has provided a practical measure of the mass absorption coefficient, permitting interelement absorption corrections to be calculated without further knowledge of the elemental composition of the biological sample. Excitation of fluorescent radiation by the incident beam in briquets ot varying density follows Beer's law when an effective wavelength is used to describe the polychromatic x-rays which interact with the briquet. Therefore, in a given specimen, a total absorption
956 J 6
coefficient, the sum of the incident polychromatic beam and subsequent fluorescent coefficients influences fluorescent intensity and, hence, sensitivity. The ratio of the total absorption coefficient and the fluorescent absorption coefficient in a sense characterizes the specimen. If both monochromatic x-ray transmittance ( I / I o ) and sensitivity (c.p.s./ pg.] are independently measured in the same specimen, the relation of the t n o measurements is proportional to their absorption coefficients. I n the case 01 the sensitivity factor, the matrix effect on the ability of the polychromatic incident beam to excite fluorescence and for the fluorescence to be quantitated by measuring its intensity, are both included in the measurement. I n an approach analogous to that of Leroux, Lennox, and Kay (9), for diffractionabsorption relationships, u e have determined that a proportionality exists between transmittance and sensitivity if the sole variable is the elemental composition of specimens of equal grams per sq. cm.; since the exponential terms in both absorption coefficient espressions cancel out when expressed as ratios. ~
EXPERIMENTAL
Apparatus. h Phillips vacuum x-ray spectrometer was utilized in these studies and its operational data are summarized in Table I.
Table I.
Instrumental Settings Scintillation (750
Counter (voltage)
X-ray tube target, voltage, and current Analyzing crystal X-ray path Collimators, entrance, exit Pulse height, baseline Sr Bragg angle, 28 Sr background angle; 28 Peak and background measurements Transmittance measurements,
III,
vdc) Tungsten, 5O-kv., 45 ma.
LiF Air 2-inch X 5-mil; 1inch X 20-mil Integral, 3 volts 25.15"
24.15", 27.15' Time to record 1,024,000 counts each Time to record 128,000 counts
with and without a briquet in the beam, as in Figure 1.4. The briquets were suspended in 0.25-mil X y l a r envelopes for ease of measurement without prejudice of sensitivity (2). The specimens were then measured in the conventional manner a t the 25.15' 28 Bragg angle as well as two adjacent background angles a t 24.15' and 27.15' 28 as in Figure l B , since the hackground continuum from light matrices is quite variable a t these energies (8). The diffracting crystal and x-ray tube potential used produced a background response curve which was essentially linear over the energy band from 0.842 A. to 0.945 A. which includes the Sr K a characteristic wavelength. Therefore! interpolation between the two "off angles'' permitted estimation of the Sr Ka background. I n this case two thirds of the difference in count rate between the two off-angle rates was added to the lower energy (at 27.15' 2e) count rate. The net intensity a t 25.15' 28 was then calculated for each series of briquets and the typical loss in net intensity with increasing concentration of biological material in the binder was demonstrated. Five series of 2-gram samples were prepared such that the weight per cent of binder and Sr-containing specimen was varied but known. The Sr concentration of each specimen was then used to detwmine the sensitivity factor, in c.p.s.ipg. Sr.
A. J
/ /
/
/ B.
Figure 1. Schematic arrangement for x-ray transmittance and fluorescence measurements Plaster of paris excited by x-rays provides source of Sr Ka radiation. Transmittance of unknown sample i s measured as c.p.s. in Sr Ka beam ( I ) before exit collimator divided by primary beam intensity (I,) measured as count rate without sample in beam path 6. Sample Sr fluorescent radiation measured in conventional manner with unknown now in sample position, excited b y x-ray tube emission; Sr Ka i s net c.p.s. determined b y scintillation counter
A.
Reagents. T h e materials measured for strontium include canine bone ash, dried dog food (a kibbled grain compound), food ash, a n d a Sr standard of argillaceous limestone which was cert'ified by the National Bureau of Standards. T h e strontium content of the various samples has been subjected to referee evaluation. T h e x-ray determinations were performed by the method of internal standards ( 7 ) . The strontium-containing materials tested were briquetted when necessary with a binder such as a detergent (Tide), soap (Ivory Snow), or powdered sugar. These binders were found to be essentially free of strontium. Several grams of commercial grade plaster of paris were used as a reliable Sr K a source in the specimen transmittance evaluations, since it contains about 0.1% Sr impurity. Procedure. T h e st rontium-containing materials were ground to at least 200 mesh a n d dried prior to use (100' C. for 24 hours). Aliquots of each specimen were then thoroughly mixed with varying quantities of binder such t h a t t h e total weight of each preparation, specimen and binder, was 2.000 grams. T h e 2-gram mixtures were then compressed into briquets of 31.78 mni. diameter by a steel die in a hydraulic press. The arbitrary choice of a 2-gram briquet ensured adequate mechanical stability. Thus, the mass per unit area for all samples
RSULTS AND DISCUSSION
was maintained a t 0.253 gram per sq. cm. The use of an organic bindere.g., sugar-does not interfere with possible subsequent analytical steps such as thermal ashing or dilutions. The analysis of the specimens was accomplished in two straight-forward steps as illustrated in Figure 1. I n the transmittance measurements, 'the prepared briquet was positioned at right angles to the exit collimator, thus intercepting the 0.875 A. beam reflected by the LiF crystal prior to detection by the scintillation counter. Transmittance was determined with the goniometer fixed on the plaster Sr K a Bragg angle as the time to record 128,000 counts
Table
Material Dog bone ash Ad
B Dried dog foodd Limestone lad a
II.
The strontium content of the specimens detertnined by x-ray fluorescence is compared to referee values in Table 11. The samples prepared plus the binders used are shown in Table 111. Note that ah increase in the fractions of sample bone ash or limestone standard in binder, soap, sugar, detergent, or dry dog food, is accompanied by a corresponding nonlinear decrease in sensitivity and transmittance ( I / I o ) . In addition, Sr concentration (gg. &/%gram sample) does not appear to influence sensitivity (c.p.s./pg. Sr) a t equal transmittance (Z/Z0). h plot of these two matrix related factors, transmittance us. sensi tivi t y , provides a linear relationship which ip fact is a strontium calibration graph for
Sr-Containing Materials-Intermethod Analysis Mg. Sr/gram material
Flame spectrometrya
Neutrod activatian Activation Analyaib
0 423 0 229
0 36* 0 23b
0 32' 0 204
1.90
1:iiP
1.880
...
...
'
X-ray fluoreacencec
NBS value
0 384 0 212 0,025 1,945
1.9
..
Average of 2 or more determinations.
* Single determination (Laboratory A).
c
d
Average of 2 determinations (Laboratory B). Samples used in present study, Table 111. Internal standard method.
VOL.
37, NO. 6, M A Y
1965
719
Table 111.
Specimen plus binder in briquet Bone ash plus detergent
Bone ash plus sugar
Limestone l a plus sugar
Limestone l a plus dry dog food Limestone l a plus soap
briquet sufficiently dense to be opaque to the 14.1-k.e.v. strontium line would, under these instrumental conditions, be expected to have a sensitivity of only 0.40 c . p . s . i p g . Under the experimental conditions described above, subst,itution of the factors determined by the linear regression fit to the data into Equation 2 results in Equation 3.
Sensitivity and Transmittance of Strontium-Containing Briquets
Binder, gram 0 200
Specimen weight, gram 1 1 1 0 0 0
0 500 1 000
1 1 1 1 1
500 600
Sr 681 576 384 192 154 76.8 38.4 384 192 154 76.8 38.4 19.2 2334 1945 972.5 583.5 389.0 194.5 97.3 434 242 146 972,s 389.0 194.5 97.3 /.l&
800 500 000 500
400 200 0 100
800
900
000 1 500 1 600 1 800 1 900
1 000
0 500 0 400 0 200 0 100 0 050 1 200 1 000 0 500 0 300 0 200 0 100 0 050 0 200 0 100 0 050 0 500 0 200 0 100 0 050
1 950 0 800 1 000 1 500 1 TOO
1 800 1 900 1 950 1 800 1 900 1 950 1 500 1 800 1 900 1 950
TransSenmitsitivity Briquet Backtance c.p.s. / p g net ground Sr Sr Ku c.p.s. 1/ I , 0 12 2 14 1456 2438 2 40 1384 257 1 0 11 3 01 1154 2934 0 15 3566 0 21 3 98 765 4 24 0 24 3790 6$2 4232 4 88 0 28 3i5 4482 0 32 5 05 194 0 21 3 67 1410 3262 1131 0 39 5 89 4653 1001 j227 0 44 6 52 6142 690 8 98 0 55 390 0 63 10 16 7780 10 21 0 68 I96 8473 0 24 3 42 3391 7972 4 00 3815 0 25 7788 5544 6056 6 23 0 42 4544 0 52 7 79 6192 7088 3407 0 57 8 99 1941 7786 9 98 0 66 8361 1049 0 69 10 78 8 05 3493 0 49 6195 0 54 6937 8 85 2141 13 .i% 0 57 9 26 7246 0 37 5 89 5733 5248 8 30 0 50 3225 6806 9 19 1787 7575 0 57 925 8105 0 60 9 51
+ 0.40
Sr, = 15.0 ( I 'Io)
(3)
To determine the unknown p g . Sr/gram in 2 gram briquets, the sensitivity equation (2), specimen weight, and net fluorescent intensity are utilized. Rearrangement of these values to solve for strontium concentrations in terms of micrograms of strontium per gram of specimen, where the briquets are maintained at total weight of 2 grams, results in the following Equation 4; pg. Sr'gram =
Net c.p.s. in 2-gram briquet ( I 'I,) 0.401 r gram specimen used in briquet
+
(4) any 2-gram sample which is not totally x-ray opaque-Le. (Z/Zo > 0). linear regression line, fitted to the data from the 27 calibrated specimens in Table 111, includes the 95yGerror interval (2u) for all the measurements and for any single determination and is shown in Figure 2 . Thus the relative standard deviation ( u ) for a single determination would be about 5% over the range studied. Control of the I , a t 7600 c.p.s. by means of occasional minor adjustments of the detector voltage eliminated day to day fluctuations and reproducibility within 2Yc was achieved. I t should be noted that an excessive amount of plaster powder was ut'ilized for the transmittance studies (2 grams) and presented an infinitely thick sample to the primary x-ray beam. For the rate constants and specimen parameters used, the equation for the line would be Sr, = .4(Zl'10)
+B
11
10
9
8
7 Sr in 2g disc C/8/)4
6
5
*
4
b w
> rc
(2)
3 0
where Sr,
0
=
-4 I 'I,
=
dSrz'd(I/Io)
=
B
=
transmittance of briquet x sensitivity (c.p.s./pg.) when I i I o = 0; ordinate intercept
When the data in Table I11 were fitted to the above equation for a straight line using the I , = 7600 c.P.s., solution of the equation yielded values of .I = 15.0 and B = 0.40. Thus the strontium sensitivit,y in a 2-gram 720
2
sensitivity (c.p.s./pg.) in briquet
ANALYTICAL CHEMISTRY
'
1
0
0
0.1
'1'0
0.2
- Bone + Detergent - Bone +Sugar
- Limestone l a +Soap
A Am-
0.3
Limestone l a +Sugar Limestone l o
0.4
+ Kibbled Food
0.5
0.6
0.7
p.8
0.9
TRANSMITTANCE
Figure 2. Relation of strontium sensitivity (c.p.s./yg. Sr in briquet) to transmittance in several prepared matrices. Data from Table
111
___ . error interval for a single measurement 957i0
- -- 9570
error interval for all data
Because of the relatively simple composition of most biologic materials as compared, for example, to geologic specimens, it is probable that multiple trace element determinations can be performed on the same sample, provided the characteristic wavelengths used for transmittance measurements are of sufficient energy t o traverse the sample. I t is possible that preparations of lyophilized materials and filtered precipitates can be treated in a similar fashion with the increased advantage of concentration. ACKNOWLEDGMENT
The authors acknowledge the advice and suggestions of P. R. Stout relating to the subject matter of this paper.
The authors thank R. J. Della Rosa for the d a t a in Table I1 and C. K. Hui for technical assistance. LITERATURE CITED
Bowen, H. J. M., Cawse, P. A,, U . K . t . Energy Res. Estab. Re 1.4309 (1963). Campbell, J. T., Shargosky, H. I., Nature 183, 1481 (1959). (3) Compton, A. H., Allison, S.K., “Xrays in Theory and Experiment,” p. 513. Van Nostrand. New York. 1935. (4) -David, D. J., Analyst 8 7 , 576; (1962). ’ (5) Della Rosa, R. J., Pool, R., O’Sullivan, J., U . S. At. Energy Comm. R e p t . UCD 108, p. 66, (1963-); (6) . , Elfers. L. A.. Hallback. P. F.. Velten. R. J., ANAL.CHEM.36, ’540 (1964). ( 7 ) Goldman, M., Anderson, R. P., Gee, W., U . S. At. Energy Comm. Rept. UCD 108, p. 75 (1963). (8) Johnson, C. M., Stout, P. R., ANAL. CHEM.30, 1921 (1958).
(9) Leroux, J., Lennox, D. H., Kay, K., Ibid., 25, 740 (1953). (10) Liebhafsky, H. A., Pfeiffer, H. G., Winslow, E. H., Zemany, P.D., “X-ray Absorption and Emission in Analytical Chemistry,” Wiley, New York, 1960. (11) Mills, A. A., Can. J . Chem. 4 2 , 73 (1964). (12) Mitchell, B. J. “Encyclopedia of Spectroscopy,’’ G. L. Clark, ed., p. 736, Reinhold, New York, 1960. (13) Reynolds, R. C . , A m . Mineralogist 48, 1133 (1963). (14) Roberts, W. Lf. B., Nature 183, 887 (1959). (15) Salmon, M.L., “Advances in X-ray Analysis,” W. hl. hIueller, ed., Vol. 2, p. 303, Plenum Press, New York, 1960. (16) Wade, AI. A., Seim, H. J., ANAL. CHEM.33, 793 (1961). RECEIVEDfor review July 13, 1964. Accepted March 15, 1965. Based on work performed under the auspices of the U. S.Atomic Energy Commission.
Completely Automated Three-phase Countercurrent Apparatus HERBERT L. MELTZER Departments of Biochemistry, New York State Psychiatric Institutes and College of Physicians and Surgeons, Columbia University, New York, N . Y .
JOSEPH BUCHLER Buchler Instruments, Fort lee,
N. 1.
ZACHARY FRANK Warner Chilcott, laboratory Instruments Division, Richmond, Calif. Pertinent details of construction of an automated three-phase countercurrent apparatus are described. The apparatus is capable of running unattended from the time solute is introduced into the first tube until a measured portion of every output sample has been delivered to an analytical station. Its mechanical performance i s so reliable that an operational test of its fail-safe features has not been possible.
T
countercurrent distribution is a technique for the fractionation of complex mixtures. It differs from the familiar (1, 2 ) two-phase distribution in that two phases move at right angles to each other past the third phase stationed within a two-dimensional array. I n contrast, the twophase method requires that one phase move past a linear assembly of another stationary phase. The theory of the three-phase method has been described ( 4 ) . With either method, when the operation is continued for a number of transfers that exceeds the number of stations in the apparatus, the moving phases, containing fractionated solute, may be run into fraction collectors. The automation of the sequence inHREE-PHASE
volving addition of moving phase, distribution, and collection of the output fractions offers obvious advantages in freedom from error and economy of time. The automated apparatus to be described below can perform a variety of useful functions. Foremost among these is its ability to separate complex mixtures. Although the three-phase method was initially devised to deal with mixtures composed of relatively dissimilar components (S), i t has also been applied to the analysis of mixtures of closely related substances (4). A realistic appraisal of the resolving power of the present apparatus will not be possible until the results of many distributions have been studied. However, a n idea of its theoretical limits may be conveyed to the reader by the following. If the apparatus is run without reloading the fraction collector, so that only 59 sets of fractions are collected, and the most favorable distribution coefficients are assumed for the components of the mixture, then nine completely resolved fractions will be obtained in some of the 1770 vials of the fraction collector and another four completely resolved fractions will be found within some of t h e 120 distribution tubes of the apparatus. If reloading of the fraction collector is allowed,
the number of separable components will, of course, be increased. If 20% overlap of component zones is allowed (a condition that still permits quantitative analysis as well as useful isolation of pure material), then the maximum number of components that can be found in 59 sets of fractions, plus the distribution tubes of the apparatus, is increased to 24. The reader should be cautioned that these are theoretical figures only; the probability of obtaining a solvent system with such favorable distribution coefficients for all of the components seems low. I n addition to the separations and analyses discussed above, the apparatus described below has already proved useful for the detection and isolation of useful quantities of trace components. For example, after loading the apparatus with 1 gram of a substance whose purity was in question, an 0.3y0 impurity was detected and isolated. The third principal use of the apparatus is concerned with the proof of purity of a substance. Since the distribution of a single substance produces several sets of output curves, each of which is obtained under conditions of varying peak concentrations, the presence of a poorly resolved impurity will appear as a distortion of a t least one of VOL. 37, NO. 6, MAY 1965
a
721
