Simultaneous Determination of Oxygen and Silicon in Meteorites and

Chem. , 1964, 36 (3), pp 559–564 ... Determination of silicon in rocks by fast-neutron activation analysis using ... Nuclear Applications 1967 3 (12...
0 downloads 0 Views 711KB Size
Simultaneous Determination of Oxygen and Silicon in Meteor;ites and Rocks by Nondestructive Activation Analysis with Fast Neutrons JAMES WING Argonne National laboratory, Argonne, 111.

b The weight ratio!; of oxygen to silicon, per cents of oxygen, and per cents of silicon in Plairiview, Karoonda, Atlanta, Murray, Orgueil, Manbhoom, Vigarano, Richardton, and Mokoia meteorites and in G-*' and W-1 rocks have been determined nondestructively by activation with 14-m.e.v. neutrons. These quantities were obtained by simultaneously counting the gamma activities of nitrogen-1 6 and aluminum-28 produced in the 0l6(n, p)N16 and Si28(n,p)AI28 reactions, respectively, using quartz as the reference. Two counting methods were used: one following the nitrogen-1 6 and aluminum-28 activities by two single-channel analy.rers; the other recording the y-energy spectra of the irradiated samples by a 200channel analyzer. 'The results are compared with thofie obtained by other methods. The sensitivities of the present method ore 1000 p.p.m. for oxygen and 250 p.p.m. for silicon, each with an estimated error of 570.

S

EVERAL ANALYTICAL methods for

the direct determination of oxygen and silicon involve des;ruction of the samples during the analyses. For the samples which are extremely difficult to obtain in large qdantity or to replace, destruction of the samples would be objectionable. The present paper reports a nondestructive method for the simultaneous determination of oxygen and silicon by activation using 14-m.e.v. neutrons. Several available meteorites (Plainview, Karoonda, Atlanta, Murray, Orgueil, Manbhoom, Vigarano, Richardto i, and Mokoia) and G-1 and W-1 rocks were chosen as the samples. The oxygen contents in these materials were not measured directly; they were obtained by alloting the amounts of oxygen necessary to form the standard oxides of the elements not present in the free state, and by adding the amount necessary to form FeO from the iron not present as sulfide or free Fe 119). The silicon contents of these materials were determined for the purpose of comparing the present results with those obtained by other methods.

The determination of silicon or oxygen invarious materials by activation analysis using 14-m.e.v. neutrons has been studied by several workers (1, 12, 16). The nuclear reactions involved in the present method are 0l6 ( n , p ) P and Si28(n,p)A128. The y-energy spectra of the irradiated meteorites and rocks showed the presence of only the prominent peaks of the 6 to 7 m.e.v. of N16 and 1.78 m.e.v. of Alz8. By simultaneously counting the activities of N16 and A128, the ratio of 0 to Si, per cent of oxygen, and per cent of silicon could be determined in one single run, with Si02 as the reference material. The short half lives of N16 (7.3 seconds) and A128 (2.3 minutes) also permitted the following of the decay of the radioactivities as well as the repeating of many runs a t short intervals. The constituents present in meteorites and rocks with abundances greater than 0.02% were studied for interferences. The chemical compositions of all the samples used in the present experiment have been determined by other workers (4, 5, 8). EXPERIMENTAL

Figure 1 is a schematic diagram of the equipment used. Sample Transfer System. A polyethylene tubing with '/2-inch i.d. was used for transferring the capsule containing the sample. One end of the tubing was connected with a Lucite section to catch the sample capsule for irradiation. The other end was connected with a funnel which fitted in the mouth of the well in the S a 1 detector. The Lucite section was firmly held a t a position to receive maximum neutron flux. The other end of the Lucite section was connected with compressed air and vacuum. The sample capsule was transferred from the irradiation site to the detector by compressed air; the travel time was 1.8 seconds. The reverse flow of the capsule was made with the use of vacuum. The sample capsule consisted of a polyethylene (Agilene HT) plug and a polyethylene tube which fitted inside the l/rinch well of the NaI detector. The plug fitted the polyethylene tube tightly, and no diffusion of oxygen and

water from the air into the sample inside the capsule was apparent. Sample Preparation. T o ensure homogeneity of the samples, the meteorites and SiOz (high purity quartz) were ground to fine powder with a boron carbide mortar and pestle in a dry argon atmosphere to avoid silicon and oxygen contaminations. G-1, R-1, and the materials for interferences studies-ie., CaO, NazC204, MnOz, Cr03, A1~03,K2Cz04. H20, MgO, NiO, Fe&, COO, Ti02 (all reagent grade), P, S, and activated charcoal-were available in the powder form. All samples were packed in approximately the same volume inside the polyethylene capsules. Triplicate samples were made for SiOa, G-1, W-1, Plainview, and Karoonda. The other meteorites were available in quantity only sufficient for one sample each. Sample weights ranged from 300 to 600 mg. Irradiation Facility. A CockroftWalton accelerator which generated 14-m.e.v. neutrons with a maximum output of 1010 neutrons per second a t point source was used. The neutron output for irradiations did not differ appreciably from run to run (two runs out of 100 differed by less than a factor of two), and it was kept constant during each run of 18 seconds. The neutron output was monitored by a Pilot-B scintillator which counted the alpha particles emitted from the tritium target in the accelerator. The integrated alpha count was registrated on a ratemeter. At the end of irradiation, the deuteron beam in the accelerator was deflected by a change of polarity of the electrostatic deflectors, and generation of neutrons was stopped. At this very same time, the compressed air was turned on to transport the sample capsule to the detector, and counting started. These three automatic operations-the deflection of deuteron beam, the opening of compressed air, and counting-were initiated simultaneously by a common relay, as indicated by the dashed lines in Figure 1. The high voltage supply for the acceleration of the deuterons was turned down during the counting period to minimize any background effects. -4 piece of 0.080-inch cadmium sheet was inserted between the tritium target and the Lucite section to reduce the intensity of the thermal neutrons generated from the accelerator. VOL. 36, NO. 3, MARCH 1964

559

Cockrblt-

Piiyithyleni

'I

Accelerotor

.

S a 5 NO1

12 ,ec

C0""t

time

Phototube Preomplifier

I

6-7 Mer I NIs

IL--1

1 F j + I Compressed

!.2.4 145 count

;

tima

I

I,..,,

I

---

Ft--------iy, o f l w end of irrad.

Relay

S m i n count

timc

--------------I----------------

I I

I

I

I

Deflector

Mostcr

PllSlt I0 I10 irrodialion

Figure 1.

Schematic diagram of the equipment

Five runs were made for each of the meteorite, rock, Fe203, P, and MgO samples and for an empty capsule. Counting Methods. Two counting methods were used in each run. I n method A, a NaI(T1) crystal, 5 inches in diameter and 5 inches thick with a '/l-inch well, was used as the detector. The detector was equipped with a phototube and a preamplifier, all shielded by 4 inches of lead. The signals from the preamplifier were transmitted into two transistorized linear amplifiers (with pile-up rejectors), and the outputs were fed into two transistorized single-channel analyzers. With the aid of an RIDL 200-channel pulse height analyzer, the upper and lower window levels in one of the single-channel analyzers were set to accept the 6- to 7-m.e.v. y activities of N16, and those in the other single10.000

10

h....,

,

I

,

,

I

,

I

,

,

,

-

0

6

12 18 T I M E i in units

01

24 30 36 2.71 S c C o I d I )

42

Figure 2. Decay curves of the 6- to 7m.e.v. y activities of "6 in the irradiated quartz and meteorite samples

560

channel analyzer were set to accept the 1.78-m.e.v. y activities of A128. The rate of channel drifting was negligible in these two analyzers. The X " 6 activities were counted by a readout scaler preset for 2.4 seconds. iin automatic printer printed out the counts in the scaler a t the end of the 2.4 seconds, reset the scaler in 0.31 second, and then started its next counting. The A128activities were counted by another read-out scaler preset for 12 seconds. Another automatic printer printed out the counts in this scaler a t the end of the 12 seconds, reset the scaler in 0.22 second, and then started its next counting. The 6- to 7- and 1.78-m.e.v. y activities were followed for 2 and 7 minutes, respectively, after the completion of the irradiation. Figures 2 and 3 show the decay curves of the 6to 7-m.e.v. and 1.78-m.e.v. y activities of some of the irradiated samples. Other irradiated meteorite and rock samples have similar decay curves. The time units in the abscissas include the 0.31 and 0.22 second of the reset time in the two automatic printers. The counts between the time units 2 and 17 in Figure 2 and between 6 and 38 in Figure 3 (indicated by arrows) were summed to give the gross activities of N16 and A128, respectively. The activities of the matrix elements for interference study were determined in the same way. With the known chemical composition of the rock and meteorite samples, the contribution of the matrix elements to the total N16 and Aiz8 activities were calculated. Activities due to the empty capsule and matrix materials were subtracted from the gross activities of the meteorite and rock samples. The weight ratios of 0 to Si were calculated from the ratios of net to A128 activities, with quartz as the reference. The stoichiometric ratio of 0 to Si in quartz was assumed to be 2 to 1. The

ANALYTICAL CHEMISTRY

weight % of oxygen and that of silicon were calculated from the net counts of NIBand A128, respectively, which had been normalized to the alpha count on the ratemeter and to the sample weight, again with quartz as the reference. Method A had the advantage that the decays of N16 and A128 y activities were followed, so that activities with y energies similar to those of N1e and A12*,but with different half lives, could be separated. By taking only the counts in the linear portion of the decay curves of the irradiated samples, the contamination of y activities by the matrix impurities was eliminated. Compton scattering of the 6- to 7m.e.v. y of N16 did not affect the A128 peak because the first five counts (1 minute) of the A128 activities were discarded. Channel shifting in the two single-channel analyzers was corrected by calibration with the reference. In method B, the output from one of the linear amplifiers was fed into the RIDL 200-channel analyzer. The y-energy spectrum was printed out by an IBM electric typewriter. At the end of irradiation, the analyzer was first inactivated for 3 seconds (to wait for the sample capsule to get into the detector) by a delay box; then it counted for five minutes. Figure 4 shows the spectra of irradiated quartz and Orgueil. The spectra for the irradiated rock and other meteorite samples were all similar to these two. The counts between channel numbers 36 and 47 and between 120 and 160 as indicated by the arrows in Figure 4 were summed to give the gross ;"\16 and activities, respectively. The activities due to Compton scattering of higher y energies were estimated by integrating the area under a line drawn between the valleys (indicated by the two arrows in Figure 4) of the A128 peak. The activities due to Compton scattering, the empty capsule, and the matrix materials were subtracted from the gross counts. The weight ratios of 0 to Si, per cent

N

0

I~

* b

'*

Ah

++,+A%

1

of oxygen, and per cent of silicon were calculated in the same manner as in method A. Method B has the advantage that channel shifting in the analyzer could be observed. Howerer, y activities with energies similar t o those of W6and AlD, but of different half-lives, could not be separated. For example, the decay curves in Figure 3 indicated short-lived activities of 1.78-m.e.v. in the irradiated samples, and these activities were included in the AlZ8peaks of the spectra recorded by the 200channel analyzer.

I00,000

.. ..

1.78 Mev AI2*

cn

10,000

W

I-

3

zP

.

Y)

z

RESULTS AND DISCUSSION

The average N16 m d A P activities per milligram of sanple material or element for all the sainples are listed in Table I. The average count ratios of N16 to AlZ8for the quartz, meteorite, and rock samples are also presented in this table. All the en1,ries in Table I are normalized to an Ezbitrary unit of neutron output. Note that the count ratios (columns 6 and 7) are not calculated from the datlt in columns 2 to 5, but are averages of individual count ratios from the expe:.imental measurements. The N16 activities due to the constituents in meteorites and rocks other than oxygen are negligible. Since fluorine is absent in the present samples, no interference from the F19(n, a)N16 reaction is expected. Although boron may interfere through the Bl1(n,p)Bel1 reaction (Bell with 14-second half life and 6.8-m.e.v. y), the amount of boron Table I.

Average

Ln

I2

3 0

"6

and

Method B

0.0338 128.7 i 2 . 1

0.0305 63.61 f 1 . 2

c

Ca Na Mn Cr Ti

AI

Ni

co K Mg Quartz G-1

w-1

Plainview Karoonda Atlanta Murray Orgueil Manbhoom Vigarano Richardton Mokoia

62.75 57.39 43.13 46.70 44.83 53.57 57.42 49.78 43.85 42.02 46'.41

30.86 28.35 22.96 23.13 23.87 28.24 29.88 25.91 22.60 21.61 24.18

I 20

I

I 40

I

I

I

I

60

80

contamination from sample grinding with boron carbide mortar is negligible. The large A12* activities from phos-. phorus are due to the P3l(n,a)AP reaction. Fortunately, the amount of

Method A

S

I

looO

6 - 7 Mev N''

c

X

- Net NI6, counts/mg. Sample Empty capsule 0 (in quartz) Si (in quartz) Fe P

.

1000

0

A128

I

I

I

100

I 120

I

I 140

I

I 160

cx I

&%*,

-

180

200

phosphorus present in meteorites and rocks is less than 0.3% in all the known cases and did not interfere seriously. I n fact, the total contribution from the various elements other than silicon in

Activities in Various Elements and Samples

Net A128, counts/mg. Method A Method B 0.138

N16 counts/A128counts Method A Method B

0.143

457.9 i 3 . 3 3.99 190.4 0.0 4.19 2.56 20.14 6.57 17.15 7.42 2.88 0.92 0.0 1.42 10.40

395.8 i 4 . 1 0.73 163.0

157.5 111.3 76.37 71.38 89.70 62.41 51.37 88.92 69.00 79.40 70.33

135.8 97.56 69.14 62.33 80.50 54.54 44.64 80.15 61.86 70.73 63.17

0.0

2.06 2.46 5.06 0.0

0.0 3.71 1.83 0.62 0.03 0.04 0.0

0.3148 i 0.0025 0.3985 0.5118 0.5647 0.6323 0.5028 0.8762 1.1171 0..5.557 ..

0.6631 0.5293 0.6599

0.1800 f 0.0019 0,2261 0.2932 0.3313 0.3713 0.2965 0.5120 0.6721 0.3292 0.3757 0.3054 0.3777

VOL. 36, NO. 3, MARCH 1964

561

these samples amounts to no more than 2% of the A P activities. The experimental results for the weight ratios of 0 to Si, weight % of silicon, and weight % of oxygen, together with their standard and relative standard deviations, are presented in Tables 11, 111, and IV. Xote again that the ratios of 0 to Si in Table I1 are calculated from the count ratios of NIBto A128 in Table I, and not from the per cent of oxygen and per cent of

Table

II.

Weight Ratios of Oxygen to Silicon in Meteorites and Rocks

Sample

Method A 1.443 f 0.020 or 1.37% 1.854 i 0.016 or 0.88% 2.045 f 0.027 or 1.30% 2.290 i 0.026 or 1.14% 1.821 f 0.023 or 1.24% 3.174 f 0.031 or 0.98% 4.046 f 0.058 or 1.44% 2.013 f 0.033 or 1.66% 2.402 i 0.027 or l . 1 4 y G 1.917 f 0.023 or 1.20% 2.390 i 0.042 or 1.77%

G-1 w-1

Plainview Karoonda* Atlanta Murray* Orgueil* Manbhoom* Vigarano Richardton Mokoia

Table 111.

Sample G-1

w-1

Plainview Karoonda Atlanta Murray Orgueil Manbhoom Vigarano Richardton Mokoia

562

silicon in Tables I11 and IV. The last columns in Tables 11,111,and IV list the values obtained by other workers using chemical or other methods for silicon determination and by the method mentioned above for oxygen. The data for Plainview are taken from Jensen’s tentative results (6); those for G-1 and W-1 are the arithmetic means of preferred values (4) from many independent determinations; and all others are taken from Mason’s compilation

Method B 1.432 f 0.031 or 2.17% 1.857 f 0.021 or 1.13% 2.099 i 0.067 or 3.21% 2.353 i 0.036 or 1.48% 1.879 i 0.042 or 2.24% 3.244 i 0.039 or 1.21% 4.258 0.076 or 1.79% 2.086 f 0.045 or 2.17% 2.380 i 0.040 or 1.6601, 1.935 f 0.031 or 1.58% 2.393 f 0.054 or 2.25%

*

Average of methods A and B

Other methods

1.438

1.440

1.855

1.836

2.072

2.062

2.321

2.330

1.850

1.832

3.209

3.126

4.152 2.050

I. 4.443 11. 4.290 2.060

2.391

2.330

1,926

2.031

2.391

2.381

Weight Per Cents of Silicon in Meteorites and Rocks

Method A

Method B

34.39 f 0 . 2 9 or 0.84% 24.31 i 0 . 3 5 or 1.45% 16.68 f 0.29 or 1.73% 15.59 i 0 . 3 2 or 2.03% 19.59 f 0 . 2 4 or 1.22% 13.63 f 0 . 1 6 or 1.21% 11.22 f 0 . 1 7 or 1.49% 19.42 i 0 . 2 7 or 1.38% 15.07 f 0 . 2 4 or 1.61% 17.34 i 0 . 1 7 or 0.99% 15.36 i 0 . 1 6 or 1.02%

34.30 f 0 . 5 2 or 1.51% 24.65 i 0 . 4 6 or 1.88% 17.47 i 0 . 3 2 or 1.82% 15.75 i 0 . 5 8 or 3.70% 20.34 f 0 . 3 5 or 1.71% 13.78 i 0 . 2 2 or 1.62% 11.28 f 0 . 1 4 or 1.27YG 20.25 i 0 . 4 6 or 2.27% 15.63 f 0 . 1 7 or 1.10% 17.87 f 0 . 2 3 or 1.29% 15.96 f 0 . 2 7 or 1.72%

ANALYTICAL CHEMISTRY

Average of methods A and B

Other methods

34.35

33.83 i 0.22

24.48

24.51 f 0 . 1 5

17.07

17.24

15.67

15.55

19.96

17.72

13.70

13.39

11.25

I. 10.53 11. 10.75

19.83

18.93

15.35

15.37

17.60

16.01

15.66

15.59

(8). 30standard deviations are available for the data in the last columns, except the per cents of silicon for G-1 and W-1. One observes that many of the results of method A agree with those of method B within their standard deviations. An asterisk in Tables I1 and IV indicates the sample whose difference between the results of methods A and B is significant a t 95% confidence level by t-distribution test. Method B generally gives higher results than method A. Method B also has larger standard deviations than method A. This may be explained by the counting difficulty in method B mentioned above. Because of this systematic uncertainty from matrix impurities, one must give more weights to the results of method A than method B. The average values (column 4 in Tables 11, 111, and 1V) perhaps should be closer to those by method A than by method B. Many of the effects of the matrix elements are eliminated in method A by energy selection of the 7 activities and by half-life discrimination. The average values of methods A and B agree quite well with those obtained by other methods (last columns) in many cases. I n most of the disagreements, the present results for the per cent of oxygen and per cent of silicon are higher than those obtained by other workers. However, the observed disagreements may or may not be real depending upon the experimental accuracies in the other methods. The experimental errors for the per cent of oxygen listed in Table IV are expected to be quite large because of the large error accumulated from the determinations of more than 12 other elements in meteorites and rocks. For this reason, the agreements of our oxygen results with those of other methods for G-1, W-1, Karoonda, and Mokoia are remarkable if not accidental. The precision for silicon determination is only 0.3 to 0.6% by chemical methods and 3.5% by spectrochemical method (3). The reported per cents of silicon from over 40 independent determinations range from 32.70 to 34.81% for G-1 and 23.87 to 25.56% for W-1 (3, 4, 14). The present results for the per cent of silicon for G-1 and W-1 agree with the arithmetic means of the preferred values (4). One of the difficulties in the determination of oxygen and silicon contents in meteorites is surface weathering and contamination of the samples by moisture in the atmosphere and by the earth. For example, as mentioned by Mason (Q), the state of hydration of magnesium sulfate in Orgueil varies with humidity. Another difficulty is the inhomogeneity of the sample materials (4,Q).These sampling difficulties may have existed in the present experiment.

The relative standard deviations in the present results range from 1 to 3y0 for 0 to Si ratios, 2 t o 5y0 for per cent of oxygen, and 1 to 4 % for per cent of silicon. The smaller relative standard deviations for the ratios are expected since many experimental uncertainties (such as in sample weights, neutron output, and counting and irradiation geometries) canceled in the determination of the ratios. For the same method of counting, the relative errors for the per cent of oxygen are larger than those for the per cent of silicon. This probably result8 from the low ?JIB activities compared with A128. Part of the experimental errors may be systematic because of slight irregularities in the size of the samples. Approximately one third of the activities observed in the irradiated empty capsule was a result of background and electroIiic noises in the power lines. The bacnkground activities increased when a c.,Tclotron near the Cockroft-Walton accelerator was in operation. The accuracy and precision of the present method could be improved by reducing these background activities with a heavier shielding and filter in the power lines, and by running the samples during the cyclotron's rest periods. However, it is not likely that, with these improveml:nts, a precision as good as that of the chemical method for silicon will be achieved. An increase in the N16 and .4lZ8count rates by the use of a higher neutron flux, larger sample, or longer irradiatior may reduce the standard deviations in the present method. But it may introduce counting difficulties] such as dead-time corrections. Longer irradiations also will produce undesirable activities from the matrix elements; in the samples. Packing the samples in a more uniform volume will reduce some of the systematic errors. In method A, an increase in the initial count rate of N16 for improving the precision for oxygen determination may be brought about by a shorter transfer time for the sample capsule. A rihorter delay time will also increase the accumulated counts of XI6 by method B, but will hamper the accuracy for the silicon determination because of the presence of the short half-lived activities in the 1.78-m.e.v. peak. From the obser fed activities of N16 per mg. of 0 and Al28 per mg. of Si (Table I), one deduces the sensitivities in the present method (on the basis of 1-gram sample and the present neutron flux) to be 1000 p.p.m. for oxygen and 250 p.p.in. for silicon, each with an estimated error of 5%. However, for a sample nith these levels of oxygen and silicon contents, the amounts of oxygen and silicon in the capsule material must be considered. The observed N16 and A128 activities due

has a relative standard deviation of 1 to 3y0 for the weight ratios of oxygen to silicon. It may be used to distinguish nondestructively species of meteorites and rocks having 0 to Si ratios differing by more than 6%, as the 0 to Si ratios in these materials vary in a wide range, from 1.4 to 4.8 (4, 5, 8, 9). The minimum weight of each sample should be several hundred milligrams to achieve the necessary accuracy.

to the irradiated capsule alone (Table I) correspond to some 230 p.p.m. of oxygen and 185 p.p.m. of silicon, respectively, in polyethylene. A capsule with less oxygen and silicon than this should be used for the analysis of oxygen and silicon in low concentrations. Table V is a summary of the reported sensitivities and the estimated errors for oxygen and silicon by activation analysis using 14-m.e.v. neutrons. Applications. The present method

Table IV.

Sample G-1 w-1

Plainview* Karoonda Atlanta* Murray* Orgueil Manbhoom* Vigarano Richardton Mokoia*

Table V.

Weight Per Cents of Oxygen in Meteorites and Rocks

Average of methods A and B

Method A

Method B

48.75 f 1.12 or 2.297, 44.59 f 1.10 or 2.467, 33.51 f 1.02 or 3.04% 36.28 f 1.09 or 3.007, 34.83 f 0 . 9 3 or 2.667, 41.62 f 1.06 or 2.55% 44.61 f 0 . 9 4 or 2 . I l 7 , 38.68 f 0 . 9 4 or 2.447, 34.07 f 1 . 3 4 or 3.92% 32.65 f 0 . 7 4 or 2.26% 36.06 f 0 . 9 8 or 2.717,

48.51 f 1.56 or 3.21% 44.57 f 1.27 or 2.86% 36.10 f 1 . 8 2 or 5.04% 36.36 f 1.59 or 4.367, 37.52 f 1 . 3 1 or 3.49% 44.40 f 1.38 or 3.10% 46.97 f 1 . 3 7 or 2.91% 40.73 f 1.26 or 3.10% 35.53 f 1.19 or 3.35% 33.97 f 1.13 or 3.327, 38.01 f 1 . 0 3 or 2.70%

48.70

44.58

44.99

34.80

35.55

36.32

36.23

36.18

32.46

43.01

41.86

45.79 39.70

I. 46.78 11. 46.12 38.99

34.80

35.81

33.31

32.52

37.03

37.12

Sensitivities for Oxygen and Silicon Determination 14-m.e.v. Neutrons

Author(s) Present Coleman Veal and Cook Steele and Meinke Stallwood et al.

Neutron outNeutron put flux neuRef- neutrons/- trons/erence sq.-cm-second second 1010 1010

2

Sample weight, gram 1 .o

13

108

0.1 100

12

108

6.5

Lbov and Kaumova

6

10'-lo8

1.0

Present Lbov and Naumova Turner

6

109

16

Other methods

48.63

by Activation with

Sensitivity

for oxygen, p.p.m. 1,000

1,000 100- 1000 20-50 10,000 10

Estimated error 5% 7% 10 % 15% 17% 1615%

100 30 1,000

10% 25% 10%

Sensitivity for silicon, p.p.m. 10'0

16

1.0

10'-108 5

x

107

3

250 100

5%

60,000

5%

VOL. 36, NO. 3, MARCH 1964

563

It has been suggested (9) that the results from a direct determination of oxygen in meteorites may be used to check the oxidation state of the iron in the oxidized form in these materials. The present method with the estimated errors for oxygen (Table IV) could be used for thisqurpose only for samples containing more than 60% of oxidized iron. Meteorites (and rocks) have much less oxidized iron than that. The present experimental equipment and counting techniques (methods A and B) may be adopted for the simultaneous determinations of other pairs of major constituents in meteorites and rocks. For example, with longer irradiations, aluminum and silicon in rocks may be determined by counting Mg27 and AIZs( I ) , and magnesium and iron in meteorites may be determined by counting ?;az4 and MnS6. Thus, it seems feasible to apply the present method to the nondestructive analysis of the major constituents in materials similar to meteorites and rocks, such as the moon’s crust (7, 10, 11).

ACKNOWLEDGMENT

For the supply of the sample materials the author is indebted to B. -Mason for the meteorites (except Plainview), to F. J. Flanagan for the rocks, and to K. J. Jensen for Plainview. The help of L. H. Fuchs and K. J. Jensen in obtaining these materials is gratefully acknowledged. The author thanks M. A. Wahlgren, F. J. Flanagan, and E. L. Segel for helpful discussions, F. R. Lawless for counting the samples, and P. W. Cross and B. J. Naderer for the operation of the Cockroft-Walton accelerator and construction of the transfer system. LITERATURE CITED

(1) Caldwell, R. L., Mills, W. R., Nucl. Instr. Methods 5, 312 (1959). (2) Coleman, R. F., Analyst 87, 590 (1962). (31 Faiibairn, H. W., U.S. Geol. Surv. Bull. 980, 1951. (4) Fleischer, M., Stevens, R. E.,Geochim. Cosmochim. Actu 26, 525 (1962). (5) Jensen, K. J., Argonne National

Laboratory, Argonne, Ill., unpublished data, 1963. ( 6 ) Lbov, A. A., Naumova, I. I., Soviet J . A&.Energy Englkh Transl. 6 , 330

(1959). (7) Lee, M. B., Ibert, E., Fite, L. E., Wainerdi, R. E., Trans. Am. Nucl. SOC.5, 278 (1962). (8) Mason, B., The American Museum of Natural History, New York, unpublished data, 1962. (9) Mason, B., “Meteorites,” Wiley, New York, 1962. (10) Monaghan, R., Youmans, A. H., Bergan, R. A., Hopkinson, E. C., Trans. Inst. Elec. Electron, Engrs. NS-IO, 183 (1963). (11) Schrader, C. D., Stinner, R. J., J. Geophys. Res. 66, 1951 (1961). (12) Stallwood, R. A., Mott, W. E., Fanale, D. T., ANAL.CHEM.3 5 , 6 (1963). (13) Steele, E. L., Meinke, W. W., Ibid., 34, 185 (1962). (14) Stevens, R. E., U.S. Geol. Surv. Bull. 1113, 1960. (15) Turner, S. E., ANAL.CHEM.28, 1457 (1956). (16) Veal, D. J., Cook, C. F., Ibid., 34, 178 (1962). RECEIVEDfor review August 12, 1963. Accepted November 11, 1963. Based on work performed under the auspices of the U. S. Atomic Energy Commission.

A Practical Approach to the Self-Shielding Problem in Low-Flux Neutron Activation Analysis OSWALD

U. ANDERS

Radiochemistry Research laboratory, The Dow Chemical Co., Midland,

b Methods to measure the average fast and thermal neutron flux to which an analytical sample is exposed during activation are discussed and evaluated. The widely used method of thermal neutron flux determination by means of externally attached pieces of gold foil is shown to result in errors of up to 500% and more for samples of large macroscopic cross-section materials. The newly proposed method of inserting a thin gold wire as flux monitor into the axis of cylindrical samples by means of a hypodermic needle is found to reduce such large errors to less than 15% for 7-ml. cylindrical samples of 1.45-cm. diameter, if Za 6 6 cm.-’

N

EUTRON

ACTIVATION

ANALYSIS

using accelerators as neutron sources has been gaining considerable attention during the past four years (1, 2, 6, 7, 9,10, 12, 14, 18). Although the first article describing the technique it appeared almost eight years ago (4, is being carried out in only a relatively small number of laboratories on anything approaching a routine basis. 564

ANALYTICAL CHEMISTRY

Mich.

One is thus forced t o speculate about the factors that are retarding the progress of this method which has been described so enthusiastically (16). Some of these factors are obvious and have been discussed before (10). A factor of paramount importance, however, has thus far escaped due attention and is, in the opinion of this writer, contributing much to the slowness of the growth of the method. It is the socalled self-shielding effect. This source of error results from the relative intransparency of larger samples to the bombarding neutrons, if they contain significant amounts of high crosssection elements or larger amounts of elements with cross sections of a few barns. In a ’ recent article by Anders and Briden (3)it was shown that for the case of fast-neutron induced activations by means of a high-threshold reaction the self-shielding effect is a function of the sample diameter times the total macroscopic cross section of the sample. The influence of the scattering component of the total cross section can be explained by the fact that even a single collision of a fast neutron inside the

sample will, in general, eliminate it from the effective flux capable of inducing the reaction in the deeper layers of the sample. Upon elastic and inelastic scattering collisions with light nuclei and inelastic scattering on medium and heavy nuclei, the neutron energy may be reduced t o an energy for which the reaction cross section is significantly smaller than for full energy neutrons, or even below the threshold of the reaction. Scattering will also deflect the neutron from its path and prevent it from reaching the deeper layers of the sample. For comparison between sample and standard, the counting data must be normalized t o the same flux conditions. For fast-neutron activations, it was possible to show that measurement of the flux transmitted by the sample, with a fast-neutron counter placed behind the sample and in line with the neutron source, greatly reduced errors due to self-shielding inside the sample. With thermal-neutron induced activation reactions, however, even multiple scattering collisions do not eliminate a neutron from the effective flux. The method mentioned for fast-neutron flux