Nondestructive neutron activation analysis of small samples of

Small Samples of Witwatersrand Ore for Gold. P. W. de Lange, W. J. de Wet, J. Turkstra, J. H. Venter. Atomic Energy Board, Pelindaba, Pretoria, South ...
0 downloads 0 Views 401KB Size
shows this not to be the case for copolymers prepared under widely different reactor conditions. This is strikingly illustrated in copolymers 200 and 201 of Table I1 where sample 201 contains only half as much comonomer as sample 200 and yet it is less crystalline. If all of the crystallinity data in Table I1 are plotted as a function of VP branching alone, no correlation exists. This points to the need for an additional parameter in defining the relationship between copolymer branching and crystallinity. This additional parameter turns out to be the hydrocarbon branching of the system which, when added to the copolymer branching, gives the linear relationship depicted in Figure 2. It will be noted that the intercept of Figure 2 is near 50 branches per 1000 carbon atoms. Samples 206 and 220 are more highly branched than this, and therefore possess no crystallinity or melting point. Points for these two samples are shown on the abscissa axis of Figure 2, but cannot be included in Figure 1 since they possess no melting point. Having established the relationships shown in Figures 1 and 2, of what value are they in the general characterization of ethylene copolymers? Again, it should be emphasized that VP-ethylene copolymers are unique, in that they are the only readily available systems where both methyl and comonomer branching may be measured. In other ethylene copolymerse.g., vinyl acetate, methyl acrylate-only the comonomer content can be measured by infrared techniques. If one assumes

all comonomers behave identically in their effects on crystallinity and melting point, then Figures 1 and 2 may be used in conjunction with IR comonomer measurements on non-VP copolymers to obtain hydrocarbon branching in them. For example, a certain vinyl acetate copolymer was analyzed by I R and found to contain a comonomer content equivalent to 20 vinyl acetate branches per 1000 carbon atoms. DTA crystallinity measurements showed it to be 3 9 z crystalline. From these measurements and Figure 2 it can be seen that this copolymer contained 10 hydrocarbon branches per 1000 carbon atoms. A limitation to this method is that it is limited to total branchings less than 50 per 1000 carbon atoms. Fortunately most commercial copolymers are in the range of applicability, ACKNOWLEDGMEh'T

The authors wish to express their gratitude to Dr. H. D. Anspon for advice and encouragement, and to Dr. F. E. Brown for preparation of the many polymer samples. Appreciation is also expressed for Mr. Bob Bartholomew and Mr. Gene Rouse for much of the experimental work. RECEIVED for review August 16, 1967. Accepted November 17,1967.

Nondestructive Neutron Activation Analysis of Small Samples of Witwatersrand Ore for Gold P. W. de Lange, W. J. de Wet, J. Turkstra, J. H. Venter Atomic Energy Board, Pelindaba, Pretoria, South Africa

A NUMBER OF TOPICAL PAPERS have appeared on the nondestructive activation analysis of gold (Au) in ore samples. De Silva (1) experienced that the inhomogeneity of the mother sample, although crushed to better than 200 mesh, was such that the specific activity of 19*Au from neutron activation showed deviations of more than 1000 when he took 200-mg samples from the mother sample for activation. At the Tashkent Conference on Activation Analysis ( 2 , 3) satisfactory activation of gold in ore samples was obtained when finely ground 30-mg samples were irradiated for 6 hours in a neutron flux of 8 X 10l2n cm-2 second-' and then allowed 5 to 8 days for decay. An average deviation of + 12 is stated in the Nuclear Science Abstract (4) of the Tashkent Conference. All these authors used NaI(T1) scintillation gamma-spectrometry. The general approach in the South African Mining industry (5) is to use samples of 60 to 100 grams for fire-assay quantitative analysis because of the large grain size of the gold present

z

(1) J. G . de Silva Filho, A. Abrao, F. W. Lima, Publ. I.E.A. 98, Inst. Energia Atomica, Sao Paulo, Brazil (1965). (2) G . A. Perezhogin, I. P. Alimarin, First All-Union Co-ordinating Conference, Tashkent, October 24-28, 1962; E. M. Lobanov, Ed., Israel Program for Scientific Translations, 1966, p. 55. (3) E. M. Lobanov, I. A. Miranski, V. F. Pozychanuk, D. G. Saifietdinov, A. A. Khaidarov, Zbid., p. 60. (4) E. M. Lobanov, I. A. Miranski, M. M. Romanov, A. A. Khaidarov, Nucl. Sci. Absrr., 18, Abstr. 43432 (1964). (5) C. H. Coxon, Corner House Laboratories, Johannesburg, private correspondence, June 1967

in the crushed ore. Gold grains tend to flatten out under continuous ball mill crushing and do not break up. N o doubt exists that neutron activation analysis of gold in ore samples can be accomplished, but sample inhomogeneity and interference by other gamma peaks in the analysis have prevented the widespread application of such activation analysis. PRINCIPLE OF THE METHOD

When a 1- to 2-gram gold-bearing ore sample is placed in an ORR-type nuclear reactor such as SAFARI-1 with a thermal neutron flux of 1013ncm-2 second-', the following reaction is the most important: l9lAu (n, y) lgaAu l?&

(1)

lQ8Auhas a half life of 64.8 hours and decays with a single gamma of 412-keV after /3-emission. The activation cross section is 98.8 barns for thermal neutrons and it has a resonance integral of 1558 barns with a major resonance of 4.9 eV. Uranium (U) is present in most of the Witwatersrand ores and in a reactor neutron spectrum the following reaction can be obtained: 238U (n, y)239U 239Np l?& 239pu (2) (a, = 2.7b) (23.5m) (2.35d)

a

Neptunium-239 emits a number of gamma rays of which the 277-keV transition is convenient for analytical exploitation. VOL 40, NO. 2, FEBRUARY 1968

451

-I

I

I

I

I

239 190 Au

?tlkrV

..

76 As 00

120

559 krV

110

I50

CHANNEL NUMBER

315 k*V 333 krV.

Figure 2. Graphical analysis of the 412-keV gamma peak in the decay of lg8Au

CHANNEL NUMBER

Figure 1. Section of the gamma spectrum of No. 11ore sample (9.45 dwt/ton) 4 days after an irradiation of 2 minutes in a neutron flux of 3 X 1013n cmd2 second-’ Thorium-232 is also present in many Witwatersrand ore samples and, in a short irradiation in a nuclear reactor, the following reaction can be utilized to determine the Th content: 232Th(n, y) 23Th + B,? 233pa----f 087 (3) (ua = 7.4b) (22.lm) (27.4d) A suitable y transition of 310 keV occurs in the decay of protactinium-233 that can be utilized for analysis. Other elements such as manganese, arsenic, and sodium present in Witwatersrand ore will be activated and gamma rays due to manganese-56, arsenic-76, and sodium-24 will be detected. EXPERIMENTAL

Sample Preparation. Because of previous success with methanol wet-mixing of uranium ore samples (6, 7), this method was again applied in an effort to obtain well-mixed samples with a uniform distribution of the various grain sizes of gold. Hand stirring of a wet mixture containing oxide powders and metallic filings used as a filler in epoxy resin has produced uniform mixtures as proved by neutron activation of the various elements in the filler material (8). An amount of 10 ml of methanol was added to about 50 grams of a mother composite ore sample, which was ground to better than 200 mesh. The wet mixture was stirred in a mortar with a pestle for about 0.5 hour. The mixture was not stirred or handled unnecessarily after it had dried. Activation Procedure and Analysis. Three samples of about 1 to 1.5 grams each, taken from a 9.45 dwt of Au per ton of mother sample (No. l l ) , were sealed in quartz ampoules and irradiated as one batch together with a pure gold standard (1 dwt/ton = 1.71428 ppm). A 2-minute irradiation was done in the hydraulic rabbit (thermal neutron flux = 3 X 10‘3 n crn-%econd-’) of SAFARI-1, an ORR-type reactor. Four such batches of No. 11 composite sample were irradiated. Three similar batches were prepared from another well-mixed composite ore sample (No. 6) which had 0.33 dwt Aulton. (6) P. W. de Lange, Trans. Proc. Geol. Soc. S. Africa, 59, 240 (1956). ( 7 ) P. W. de Lange, ANAL.CHEM., 31, 812 (1959). (8) P. W. de Lange, C. B. Bigham, Nucl. Applications, in press.

452

0

ANALYTICAL CHEMISTRY

0, 21

I

I

30

40

I

1 50

60

C N A N N E L NUWIIER

Figure 3. Graphical analysis of the 277-keV gamma peak in the decay of 239Np Triplicate samples of two other composite samples (Nos. 1 and 9) and No. 11 were irradiated later. Gamma-ray emission was detected with the ampoule placed 1-cm below the Ge(Li) detector. This detector was a 5-cc coaxial Ge(Li) diode (Princeton Gamma Tech, Princeton, N.J.) mounted in a cryostat and cooled with liquid nitrogen, An uncooled TC 130 Tennelec Preamplifier, a TC 200 Tennelec amplifier and an R.I.D.L. bias amplifier (only the cut amplifier section), were used to obtain low noise amplification of the diode output signals. The spectrum analysis was done on an Intertechnique 400 channel analyzer bypassing the built-in linear amplifier of the analyzer. RESULTS AND DISCUSSION

Figure 1 shows a y-spectrum (10-minute count) obtained with No. 11 sample, The decay time was 4 days after an irradiation of 2 minutes. Background Determination. A graphical analysis to obtain the true count under the 412-keV gamma peak of I98Au is shown in Figure 2. A similar analysis of the 277-keV

412 kU

I

. .

B*

559

*v

b

moo

D

O R " I : CWNTS PER MINUTE PER GRAM SAMPLE OR*PH 11: COUNTS PER ~ M I W T EP€R GRAM SAMPLE

,

I

I

0

100

m

300

C W E L HUWBER

Figure 4. Comparison of the gamma activity of gold in various composite Witwatersrand ore samples, before and after methanol mixing

Figure 5. Section of gamma spectrum of No. 11 ore sample irradiated in a eadmium cover Decay time of 2 hours after irradiation of 3 minutes (thermal), R C d for gold = 4.67

Au dwt/ton values have been obtained from fire assay

peak of 239Npis given in Figure 3. Fortunately, the small peak at 285-keV is also due to 239Npdecay. The peak analysis is not such a sophisticated computerized approach as suggested by Heath (9) and also used by Lamb et al. (10). However, the graphical approach can be applied with ease in industrial laboratories. The estimate of the background under the peak is obtained by the linear extrapolation of the spectrum on the high energy side of the peak to the region under the peak. The steep rising edges of a peak can easily be identified when a linear scale is used. The estimated uncertainty involved in the obtained background for the 412-keV peak is (9) R. L. Heath, Nucl. Imtr. Methods, 43,209 (1966). (10) J. F. Lamb, S. G. Prussin, J. A. Harris, J. M. Hollander, ANAL.CHEM.,38, 813 (1966).

+

=

1013

n cm-*

+2.OX (10-minute count) in the case of the No. 11 sample (9.45-dwt of Au per ton) and i1.0 in the case of No. 6 (0.33 dwt of Au per ton) (30-minute count). Sample Uniformity. The ratios of the gamma peaks at 412-keV ("38Au) and 277-keV (zs9Np) were calculated for the 12 samples taken from sample No. 11 from yspectra measured at cooling times of 2 days and again at 9 days. The necessary corrections for relative decay and irradiation times were made after the estimated background subtractions and weight normalization were done. A similar peak-ratio analysis has been done on activated samples of sample No. 6, after a cooling time of 3 days. Table I lists the results of the three ratio groups. The standard deviation (a) for each peak-ratio distribution is given. In spite of the fact that U and Au appear in the ore in completely different matrixes, the distribution of the ratios in

Table I.

Analysis of Ratios of Peaks of

Au 412keV and Np 277-keV Sample No. 11, irradiated May 30 Measured Measured ratios, June 1 ratios, June 8

. A

0

785

I

I

IO0

100

1.719 1.584 1.603 1.737 1.738 1.780 1.686 1.788 1.605 1.782 1.903 1.633

2.126 2.131 1.948 2.299 2.019 2.053 2.189 2.080 1.994 2.126 1.972 2.327

Av 1.713

Av2.105

Av 6.324

u = f0.085

u = f 0.222

k Y O

I

YD WWL NUMBER

Figure 6. Section of gamma spectrum of No. 11 ore sample irradiated without cadmium cover (similar to Figure 1) Decay time of 4 hours after an irradiation of 1 minute at similar position as for Figure 5

Sample No. 6 irradiated June 16 Measured ratios, June 19 (xW2)

u =

f0.095

(f5.573

(f4.0~)

6.471 6.643 6.371 6.160 6.143 6.417 5.998 6.376 6.324

x

lo-'

(*3.5%)

VOL 40, NO. 2, FEBRUARY 1968

453

all three cases are those of a normal distribution. As it is accepted that the uranium particles are uniformly distributed (6, 7), it is concluded that the methanol-mixed samples are acceptably uniformly mixed. The analysis done when a higher background was present is not significantly different from the analysis done when the peak to background ratio had increased after another 7 days decay. This gives further confidence in the graphical obtainment of the background. Graph I in Figure 4 is a 45 O line on a log-log presentation of counts per minute per gram sample against the known Au content in various Witwatersrand composite samples. Triplicate sample values are given together with the average values for samples 6 and 11 (Table I). Specific Au reference activities were used for normalization. Graph I1 in Figure 4 is a similar graph of counts per 10 minutes per gram sample against the Au content of the various Witwatersrand composite ore samples as received before methanol mixing was applied. Samples of 1 to 2 grams were irradiated. The advantage of wet methanolmixing is clearly illustrated. Nonlinear corrections for difference in sample weights would be a necessity when higher accuracy is desired but in this paper, which serves as a demonstration of the principle of small ore sample activation, nonlinear weight corrections were ignored. The calibration curve above 10 dwt/ton concentrations of Au would not continue to follow a 45 O angle with the X-axis on log-log paper as suppression of the neutron flux can

occur at higher Au concentrations. With proper standards activation analysis of Au concentrations above 10 dwt per ton can be done quite easily. It has been shown by Lobanov et al. (3)that it can be advantageous to use cadmium sample containers. With greatly depressed S6Mn( t 1 / 2 = 2.6h) and 24Na( t l / * = 15.0h) activities, it is possible to do quantitative Au analysis after cooling times of 4 to 6 hours only. Figure 5 is part of a gamma spectrum obtained for 0.5 gram of sample No. 11 after an irradiation in a cadmium cover (0.025 inch thick) and a cooling time of 2 hours with the sample 4 cm away from the detector. Figure 6 is a similar measurement but the sample was not canned in cadmium. Leakproof containers could not be obtained in time so that epithermal neutron irradiation was not utilized further. Simultaneously, with the nondestructive activation analysis of gold ore, quantitative analysis of U, Mn, Th, As, and Na can also be accomplished when a Ge(Li) diode is used for gamma spectra analysis. Absolute activities can be determined when the detector system has been calibrated with standard sources available from the IAEA in Vienna. As a practical measure, it is preferable to use a calibration curve which can be obtained with activated ore samples, the contents of which have been chemically analyzed before irradiation. RECEIVED for review August 3,1967. Accepted September 18, 1967.

Adsorption of Silver on Borosilicate Glass Effect of pH and Time

'

SIR: In a recent paper West et al. ( I ) reported a rather marked increase in the amount of silver absorbed on glass at a pH of 4 compared to that at pH's of 7 and 8. This is contrary to the accounts of Chambers and Proctor (2), Hensley et al. (3),and Hamester and Kahn (4) who observed a direct relationship between pH and the amount of silver absorbed on glass surfaces. When absorption experiments involving silver-1 10m and hydrous ferric oxides were initiated in this laboratory, it became apparent almost immediately that the containers used for handling the solutions adsorbed appreciable quantities of silver at low concentrations (-0.02 ppm Ag+). It thus became necessary to determine the surfaces most suitable for handling silver solutions and keep a close silver balance in all operations. The results of surface tests are listed in Table I. Borosilicate glass, treated with Beckman desicote (1) Foymae Kelso West, Philip W. West, and Frank A. Iddings, ANAL.CHEM., 38, 1566 (1966). (2) Cecil W. Chambers and Charles M. Proctor, Robert A. l a f t Sanitary Engineering Center, Technical Report W60-4, 18 pp

(1960). (3) James W. Hensley, Arthur 0. Long, and John E. Willard, Ind. Eng. Chem., 41, 1415 (1949). (4) Hans L. Hamester and Milton Kahn, USAEC Report SCR-593, 46 pp (1963).

454

ANALYTICAL CHEMISTRY

18772 (an organo-silicon product used to produce hydrophobic glass surfaces) gives lowest absorption values and is cleaned easily. However, untreated glass was found to be satisfactory as well. Although molded plastic surfaces adsorbed less strongly than glass surfaces during short intervals (one or two days) they were more difficult to clean, and over a period of several months adsorbed appreciably more than glass. The values obtained in the material tests are of the same magnitude as those reported by Sotnikov and Belanovskii ( 5 ) and Sotnikov et al. (6) who studied the adsorption of a large number of tracers, including Ag+, on germanium, silicon, and quartz surfaces in the presence of various etching agents. The effect of time and pH on the adsorption of silver on brown borosilicate glass is illustrated in Figure 1. The pH 3.8 solutions were obtained by dissolving AgN03 crystals in distilled water and the pH 1.5 solutions by adding "03 to the AgN03 solutions. By comparison Nalgene bottles adsorbed on half the silver from a 0.02-ppm solution at a pH of 3.8 in about 40 days. In one instance a Nalgene bottle (5) V. S. Sotnikov and A. S. Belanovskii, SoGier Radiochemistry (translation) 8, 159 (1966). (6) V . S . Sotnikov, A. S. Belanovskii, and M. I. Kuznetsova, Ibid.

p. 238.