ously described, the N G interference with TA was minimized by thermally decomposing the NG. The retention times of several of the most likely nitrated derivatives of resorcinol and 2-NDPA were measured. The relative retention times are shown in Table 11. As shown by these data, all of the components studied were separated from TA, DMS, resorcinol, and 2-NDPA or the compounds in question did not elute from the column. Precision Study. The precision study was conducted using extracts obtained from two different propellant formulations, 1 and 2. Because these propellants did not contain DMS, a known amount of this compound was added to the propellant 2 extracts. The recovery of the added DMS was then determined. A typical chromatogram for these propellant extracts is shown in Figure 1. The data obtained from the precision study are shown in Table 111. Two different propellant formulations were analyzed to emphasize any effects of different ratios of the components on the results. These data showed that the concentration of any one component did not influence the results of another component. The coefficient of variance for each of the two pro-
pellants was essentially the same-Le., TA, 0.03%; DMS, 0.02%; resorcinol, 0.01 %; and 2-NDPA, 0.01 %. In addition, data in Table 111 show the close agreement obtained between the G C and present IR-TLC ( 4 ) methods, about *2% relative or less between all the components. Because the method would be applied to large numbers of samples, the precision calculations were based on peak height instead of the more accurate area measurements by triangulation to decrease the time required for calculation of the results. The results definitely indicated that NG was not a problem because low coefficient of variance for each component was obtained. These data were confirmed by analyzing samples with widely varying NG/TA and NG/stabilizer ratios as shown in Table IV. These results proved that the effect of NG on the stabilizers had been removed and, further, the interference of NG with TA had been eliminated. RECEIVED for review July 15, 1968. Accepted September 16, 1968. (4) G. F. Macke, J. Chromatogr., (in press).
Solution of Blank Problems in 14-MeV Neutron Activation Analysis for Trace Oxygen S. S. Nargolwalla, E. P. Przybylowicz,' J. E. Suddueth, and S. L. Birkhead National Bureau of Standards, Washington, D . C. 20234 THE ACCURACY of trace oxygen determinations by 14-MeV neutron activation analysis can be seriously affected if the oxygen contribution from the container is significant. If the count from the blank is merely subtracted from the total sample-in-container count, errors as large as 100% may be introduced. Solutions by other workers (1-4) are specific to the individual systems being used and are not of general applicability. With a general purpose activation analysis facility, an attempt has been made to provide a generalized solution to the problem. In our previous approach (3, leak problems, associated with the encapsulation technique for loading steel samples in polyethylene containers under a nitrogen atmosphere, introduced imprecision in about 2 0 z of the experiments. The present technique gives consideration to the sample-in-container geometry and any attenuation effects which may arise during irradiation and counting. The application of this method does not presuppose the availability of low-oxygen containers, but can be used in any analysis where the blank problem is considered to be significant. 1 Research Associate from Eastman Kodak Co., Rochester, N. Y .
(1) 0. U. Anders and D. W. Briden, ANAL.CHEM., 36, 287 (1964). (2) R. F. Coleman, Iroi Steel Inst. (London), Spec. Rept. 68. ( 1960). (3) K. G. Broadhead and H. H. Heady, ANAL.CHEM.,37, 759 (1965). (4) H. F Priest U.S. Army Materials Research Agency, Watertown, Mass., private communication, June 1968. (5) S. S. Nargolwalla, M. R. Crambes, and J. R. DeVoe, ANAL. CHEM., 40, 666 (1968). (6) F. A. Lundgren and S. S. Nargolwalla, ibid., p 672.
168
ANALYTICAL CHEMISTRY
This study contributes to the state of the art in the following ways: determination of the attenuation of the activity from the container by samples of different diameters; definition of a geometrical model which quantitatively expresses the observed attenuation by the samples; and design of a flow-through container for solid samples which reduces the capsule blank contribution and eliminates the need to encapsulate samples in a nitrogen atmosphere provided the propelling gas in the pneumatic system is dry nitrogen. The solution of these blank problems considerably improves the reliability of the 14-MeV neutron activation technique for the precise and accurate determination of trace oxygen. EXPERIMENTAL
Equipment. The facility at NBS consists of a 2.5-mA beam current Cockcroft-Walton neutron generator, a dualsample pneumatic system with a rotating sample assembly (6), a sequence programmer, and a detector system of two 4 inch by 3 inch NaI(T1) detectors coupled to a 400-channel pulse height analyzer. For a quantitative study it was necessary that a container be constructed of a material with macro amounts of oxygen-viz. nylon (ca. 12% oxygen) identical to the polyethylene vials normally used. The samples were steel rods containing small amounts of oxygen ( < O S % ) relative to the nylon container. The flow-through container was a two-dram polyethylene capsule perforated by 5 / l ~ - i ndiameter ~h holes so that nitrogen could flow freely around the sample during transfer. Consequently the void volume is free of oxygen. Procedure. Verification of the experiment was done by comparing results using the previous technique ( 5 ) with those obtained with the flow-through container. The effectiveness of the perforated container was evident in the reproducibility of the blank determinations and the determination of oxygen in seven rods of valve steel using both methods.
A
Parallel Neutron Beam
Figure 2. Geometric model: neutron parallel beam Point
-1
1-
,8047 in
Figure 1. Geometric model: neutron point source Fraction of container surface in shadow of sample
Fraction of container surface in shadow of sample
=
=
sin-'
(x> A
i?
With sequential counting, the empty nylon container and one of polyethylene filled with benzoic acid were irradiated, then counted, with the nylon container first. Photons in the energy region 4.8 to 8.0 MeV were counted. Replicate determinations on the steel rods were made. Comparison of the activity of the loaded nylon container with that of the empty container gave a measure of the attenuation of the oxygen activity in the container by the steel rod.
decreases, the sample shadow is decreased. Geometrical corrections are based on two models, neutron and gamma-ray, neglecting end effects (Figures 1,2, and 3). The neutron beam is considered for point source (Figure 1) and parallel source beam (Figure 2), the actual beam being best described as a nonuniform disk source. For neutron attenuation (Figures 1 and 2), only a portion (arc AB) of the container is attenuated. The total activity induced is:
RESULTS AND DISCUSSION
Attenuation Effects. A geometrical model was developed
to calculate the magnitude of attenuation of oxygen activity for all sample diameters. The experimentally determined attenuation of the container activity is given in Table I. Previous work ( 5 ) has shown that neutron shielding and gamma-ray self absorption can be expressed in terms of a generalized exponential absorption law. The measured activity, C,, of a sample is related to the total possible activity, C t , by: C, = Cl(e-WY
x e-ZRX)
(1)
where p o is the total linear gamma-ray absorption coefficient (cm-'), Y is the gamma-ray attenuated sample thickness (cm), Z R is the total neutron macroscopic removal cross section (an-'), and X is the neutron attenuated sample thickness (cm). When sample matrices differ, the exponential product may differ and C 1must be calculated and used in the comparison. As sample diameter decreases, so does the slope of the correction curve which gives a direct measure of the effective attenuated sample thickness. Applied to oxygen in sample containers, several precautions must be observed. The oxygen is presumed to be uniformly distributed, but for individual atoms the attenuation differs depending on their position. The experimental curves (5) average out those differences for the entire sample for different diameters. In attenuation of I6N, geometrical corrections must be made because oxygen is only in the container walls and sample diameter is less than container diameter. The oxygen in the container walls will, on the average, be shadowed by a full container one half of the time. As the sample diameter
where A t is the theoretical amount of activity with unattenuated neutron beam, Z R is the total sample macroscopic removal cross section (cm-l), and X is the attenuated sample thickness (cm). 2,' and X ' are the same parameters for is taken as unity. nitrogen in the container, where If end effects are neglected, calculations based on a twodimensional model show that the shadowed areas are: 2R[sin-'(r/R)
Point source: Parallel beam :
+ ~in-~(r/l)]
2R sin-' (r/R)
(3) (4)
where R is the inside container radius, r is the radius of the sample, and 1 is the distance from the container to the point source. Attenuation by the sample is only through the angle defined by the sample (Figure 3). The activity at point A is attenuated through the angle [f 2 3 41 for a full capsule. For a sample, the activity will be attenuated only through the angle
+ + +
Table I. Steel rod diam, in. 'I8
'I4 3i8 'I2
Calculated Total Attenuation Correction Factor Exptl attenuation" Point source Parallel beam correction factor 0.9924 0.9694 0.9294 0.8666
0.9928 0.9711 0.9330 0.8726
0.9997 i. 0.0092 0.9714 f 0.0091 0.9378 i 0.0085 0.8862 0.0089
+
Results are weighted means of eight determinations, with weights equal to reciprocal of estimated variances; estimated variances based on Poisson counting statistics and propagation of error formulas. Errors are standard errors of weighed means.
VOL 41, NO. 1, JANUARY 1969
169
c
Table 11. Comparison of Methods Using the Standard Capsule and the Flow-Through Container Oxygen concn, Duma Corrected for attenuation of Uncorrected for Corrected for blank using attenuation of attenuation of flow-through Rod No. blank containe~blank containerc containerd 1 2 3 4 5 6
7
50 ?C 4 30 f 4
94 158 83 92 77 84 82
34 i 4 4414 22 f 5 33 4 28 i 5
*
i5 i9 i6 =t8 f6 f6 f7
58 f 5 61 f 5 51 f 5 55 f 5 63 1 5 66 1 5
I
Luciie Abrorbrr No
--I
I ( T I ) Detector
,#'
,
i'
,
L3
,
+
-
LZ-
66 i 5
Concentrations are weighted means of eight determinations on each rod, with weights equal to reciprocal of estimated variances; estimated variances based on Poisson counting statistics and propagation of error formulas. Errors are standard errors of weighted means. * Nitrogen encapsulation method using standard capsule. c Nitrogen encapsulation method using standard capsule corrected for attenuation for lIrinch steel rod (Table I). d Flow-through container encapsulation corrected for attenuation for 1/4-inchsteel rod (Table I).
L
I
L4
-
L
0
+
+
+ + + + + + +
[l 21. The fractional angle [l 2]/[1 2 3 41 is that portion of the total angle through which 16N activity in the container is attenuated, and 1 - [l 21/[1 2 3 41 represents the angle through which there is no attenuationi.e., e-aoY in Equation 1 is unity. For a sample diameter smaller than the container's internal diameter, the measured activity, B,, is related to the unattenuated activity, B t , by:
(1
-
L[1
L[1 + 21 + 2 + 3 + 41))
(5)
The product (&/A,) (Bm/Bt)represents the total neutron and gamma attenuation properly corrected for geometry. Blank oxygen may be subtracted directly only when the blank is low compared to the sample (ca. 2z) or when the sample diameter/container diameter ratio is
