Marine sediment sample preparation for analysis for low concentrations of fine detrital gold

GEOlOGICAl SURVEY CIRCUlAR

United States Department of the Interior

ILLUSTRATIONS TABLES INTRODUCTION

Analytical Methods

Samples were prepared for analysis either in a mobile sedimentology laboratory in the field or in the U.S. Geological Survey laboratories in Menlo Park, Calif. Nearly all the gold analyses were made by a wet chemical method (Lakin and Nakagawa, 1965) in conjunction with atomic absorption spectrophotometry, although a few were made spectrographically. Most of the atomic absorption analyses were conducted in the field in a mobile laboratory.

We gratefully acknowledge the assistance of F. J. Swanson, David D' Armond, and R. G. Winkler with the preparation of the sediment samples and the help of Alan Chleborad and Francis Michaels with the atomic absorption analysis. E. M. Baldwin and Sam Boggs, Jr., of the University of Oregon, contributed stream sediment samples for the study. Offshore samples were taken witr. a grab-type bottom sampler from the U.S. Geological Survey research vessel Polaris under the direction of G. A. Rusnak.

Particle Sparsity Effect

The analytical data can reacily be misinterpreted where samples cortain a limited number of gold particles. A 500-gram sample of sand containing 0.5 ppm gold in flakes weighing 5 X 10~ g (approximately 0.03 mm thick and 0.1 mm in diameter) would contain 50 flakes. If the sample were split into 250 2-g portions, none of which c0ntained more than a single flake of gold, 50 portions would contain one flak'e and the remaining 200 would contain no gold. In other words, in analyzing this sample, the chances are 80 percent that no gold will be detected; if f'. gold-bearing split were analyzed, its gold cortent would be a misleading 2.5 ppm.

This circumstance we have termed the particle sparsity effect, wherein the detrital gold content of a given sample, as determined by analysis of a split, depends not S'J much on the actual gold content of the s·ample as on whether or not a random flake occurs in the analyzed portion. Obviously, nf'.tural samples are more complicated than the hypothetical sample discussed above, as they contain several sizes of gold flakes and any one portion could readily contain more than one flake. The principle, however, of the particle sparsity effect does apply to natural samples as noted by Pardee (1934, p. 34) and as may be seen from table 1. The gold content of splits from a single sample of unprocessed sand differed as much as 29 ppm (table 1, sample M660C-56); the analyses are not reproducible, nor is any of them representative of the initial sample. The particle sparsity effect is most sig:'lificant in samples containing relatively little ~·old. As the content increases, the effect will de.dine if particle size remains constant. It is unquestionably important in the range of cor~entra tions that we have found in beach 2nd offshore sand.

!I .1

(Griggs, 1945) to dilute the gold below the detection level. For such samples, further concentration can be made magnetically. A hand magnet readily removes the magnetite, and a Frantz magnetic separator (Gaudin and Spedden, 1943) removes ilmenite and chromite when set at 0.6 amps, with a forward slope of 25 ° and a side slope of 15 °. Some of the gold in the southern Oregon black sands reportedly is coated with iron oxide (Pardee, 1934, p. 38). This coating, however, seems not to increase significantly the magnetic susceptibility of the gold, which occurs almost entirely in the less magnetic fraction (table 5). The occurrence of gold in the magnetic fractions of sample M660C-56a is thought to result from incomplete magnetic separation, a conclusion supported by the results of analysis of M660C-56b, a carefully processed split of the same sample.

A few concentrates-even aft~r sieve, gravimetric, and magnetic separation-are too large for a single analysis. Such concentrates can be split into analyzable portions with a riffle-type sample splitter. Analy-sis of all the portions establishes the total gold content; analysis of one provides a value presumably representative of the total concentrate. The particle sparsity effect, howev~r, may still apply if the sample contains v~~ry little gold. Splits of such concentrates show less consistency in their values than do concentrates of large splits of a single sample (table 6). Therefore, analysis of the tot~.l concentrate provides a better gold determination than does analysis of a presumably repre.,.~ntative split from the concentrate. It is most important that the analyst utilize the entire p'lrtion of concentrate submitted for the analysis, lest the particle sparsity effect reappear even after the most careful concentration of th~ gold.

A concentrating technique for marine sands is suggested in figure 2. This m~thod requires modification depending on the size of the gold and the sorting of the sedimert. Beach, offshore, and marine-terrace sand"' of southern Oregon contain very little gold coarser than 0.124 mm. These sands lie tens of miles from the inland original source of the gold; therefore, their detrital gold may be expected to be very fine. In beach sand in Alr~ka, however, gold less than 2 miles from the source (D. M. Hopkins, oral commun., 1966) is of similar grain size. Additional data may show that the grain size of gold in beach and shelf sands is everywhere smaller than 0.124 mm.

Gold in stream sediments shows less consistency in size (table 2). The concentration technique, obviously, must take the size factor into account. In stream sediments, for example, the material greater than 2 mm may be eliminated by sieving, clay-size particles may be removed by washing and decanting, and the light fraction may be separated gravimetrically, thus concentrating the gold in the heavy intermediate grain-size fraction. In the Snake River, Idaho, the finest detrital gold flake observed by Hite (1933b, p. 698) was approximately 0.01 mm across and 0.002 mm thick. Such a particle would weigh about 0.0033 millionths of a gram, and 300 of these

b------

b------

c------ particles in a gram of sample (or 600 in 2 g) would constitute 1 ppm, enough to eliminate the particle sparsity effect and the n~ed for concentration. Our studies, however, have not yet established the presence of any appreciable amount of such extremely fine gold.

Similarly, the sorting of a sediment may determine the concentration technique. The wellsorted beach sands of the Oregon coasts contain very little light material in the smaller than 0.124 mm fraction. Because the gold is nearly always restricted to this size fraction, sieving commonly is the only concentration required.

Sieve, gravimetric, and magnetic separations constitute an effective concentration method for gold. Reproducibility is good, as indicated by analysis of concentrates of several splits of the same original sample (table 6). The gold content of the initial sample as calculated from the analysis will be accurate, providing the concentrate contains all the gold in the sample. Examples of this calculation are shown under "Examples of calculation of gold content of samples from analyses of concentrates."

The concentration method de~cribed herein has several disadvantages. Very large san1ples are difficult to handle with standard laboratory facilities. The method is rather slow, although with proper facilities t"vo technicians can readily process 30 1,000-g samples a day. Currently, we use a stack of duplicate sieves separated by insert pans in orier to process several samples simultaneously. Heavy liquid separation is done in a battery of large funnels. Magnetic separation can doubtlessly be improved by using an electrom~.gnetic separator (Holmes, 1930, p. 89), as separation by hand magnet is particularly tiJne consuming. The method has provided a byproduct of useful information about the size of gold particles and the size distribution and heavy-mineral content of auriferous sand.

Examples of Calculation of Gold Content of Samples from ..Analyses of

Exarnple 1.-M660C-41, a sample of 368 g, yields a concentrate of 0.44 g, that, as analyzed, contains 6.6 ppm gold. What is the gold content of this sample, assuming that the concentrate contains all the gold in the sample?

Gold content of the sample= weight

Exarnple 2.-M660C-39, a sample of 155 g, yields a concentrate of 4.13 g. A 2.02-g split of the concentrate, as analyzed, contains 0.7 ppm gold. What is the gold content of this sample, assuming that the concentrate contains all the gold in the samph~ and that the gold content of the split is representative of that of the concentrate?

Gold content of the sample=weight

Con cent rates

of the portion analyzed times its gold content (ppm) divided by the weight of the initial sample, that is, (0.44 X 6.6) /368=0.008 ppm.

Example 3.-M660P-77, a sample of 2,530 g, yields heavy (sp gr >3.3) fractions that were analyzed for gold with the following results:

0.124-0.088 mm size class, 128 g from which six portions were split for analysis: the 0.124-0.088 mm heavy fraction: a. 2.09 g, <0.1 ppm b. 1.22 g, <0.2 ppm c. 1.13 g, <0.2 ppm d. 1.23 g, <0.2 ppm

I I

0.088-0.062 mm size class, 1.95 g, 18 ppm 0.062-0.038 mm size class, 0.44 g, 8.6 ppm <0.038 mm size class, 0.50 g, <0.4 ppm What is the gold content of this sr.mple? Calculation of the contribution of g')ld from

Maximum gold content of the pore. 1.27 g, 4. 7 ppm f. 1.95 g, <0.1 ppm

tions analyzed= sum of the products of the W3ights of the portions

I .1

Minimum gold content of the portions

Assuming the gold content of the portions analyzed to be representative of the whole 0.124-0.088 mm heavy fraction, this fraction contains between 0.80 and 0.68 ppm gold. The contribution of this fraction to the gold in the total sanlple may be calculated as follows:

Contribution of gold from a fraction

Similarly the contributions of .the

mum contribution is insignificant (0.0001 ppm) The gold content of the initial sample is the sum of the contributions from the different fractions:

Maximum= 0.04 + 0.014 + 0.001 ppm=

The gold content of this sample is between 0.055 and 0.049 ppm.

other fractions are as follows: 0.088-0.062 mm heavy fraction=

0.062-0.038 mm heavy fraction=

< 0.038 mm heavy fraction maxi-

Conclusions

reproducible, nor in general is any single analysis representative of the gold content of the sample. Reliable analysis, however, may be made on concentrates. The method described, which combines sieve, gravime~.ric, and magnetic separations, has generally been adequate for beach and offshore sands. With modification, it may prove effective for analysis of stream sediments.

At the time the concentratiiJg method was being developed, the size of the analyzed portion was restricted to 2 g. Presently the analytical technique is being modified to accommodate 10-g samples (H. W. Lakin, oral commun., 1967). This increase greatly facilitates the concentrating procedure; for example, it will be possible to quickly concentrate a sample to 50 g which can be split into five separately analyzable portions.

The described concentrating method works well in the field within a m9bile sedimentology laboratory and requires no highly spe... cialized equipment. Other methods will almost certainly be required to conce:r1trate detrital gold from sedimentary rocks. Techniques such as flotation or electrostatic separation may prove to be highly practical. Industrial-scale facilities may provide the quickest and most convenient means for concentration, particularly for large samples.

Our data are taken only frmn analyses for detrital gold. The principle of the particle sparsity effect, however, is certain to apply to other minerals of high specific gravity, such as platinum, which will likewiEe require concentration of the initial sample.

This paper deals only with the problem of obtaining an analysis that is representative of the gold content of a given sample. The problem of taking meaningful samples, of course, remains with the geologist.

Where this page came from

This page was imported from U.S. Geological Survey. Published by the U.S. Geological Survey and, as a work of the United States government, in the public domain.

Nobody has written it yet — it is the source material at a new address, which is why search engines are asked to skip it and why no one earns from it. It is up for grabs: take it on, and it is yours to rewrite and to earn from.