analytical methods for determining gold

8/17/2019 Analytical Methods for Determining Gold

http://slidepdf.com/reader/full/analytical-methods-for-determining-gold 1/48



GEOLOGIAN TUTKIMUSKESKUS GEOLOGICAL SURVEY OF FINLA

Tutkimusraportti

4

Report of nvestigation

4

sko Kontas

ed.)

ANALYTICAL METHODS FOR DETERMINING GOLD IN GEOLOGICAL SAMPLES

Espoo

993



Kontas, Esko ed.),

1993.

Analytical methods for determining gold in geological

samples. Geologian tutkimuskeskus, Tutkimusraportti eolog ical Survey of Finland

Report of bwestigation 114 41 pages, 16 tables, 1 appendix.

Key words GeoRef Thesaurus, AGI): chemical analysis, gold, sample preparation,

methods, spectroscopy, atomic absorption, techniques, neutron activation analysis,

accuracy

sko

Kontas Geological Survey of Finlmd P.O.Box 77 SF-96101 Rovmtiemi Finland

ISBN 95 1-690-476-9

ISSN 0781-4240



ONTENTS

Foreword

skoKontas

4

Ana lyttcal methods for determining gold in geological samples

slwKontas

Determ ination of gold by atom ic absorption after lead fire assay separation

iitta Juvonen and P. J. Vaananen

13

Determination of gold by aqua regia-potassium bromate digestion methyl

isobutyl-ketone extraction and flame atomic absorption

.Noras 17

Determ ination of gold and palladium by aqua regia digestion dibutylsulphi-

de-di-isobutyl-ketone extraction and flame atomic absorption

U. Penttinen 21

Determ ination of gold palladium and platinum by aqua regia digestion

dibutylsulphide di isobutyl ketone

extraction and flameless atomic absorption

.Ojaniemi 25

Determ ination of gold and palladium by aqua regia digestion stannous

chloride-mercury coprecipitation and flameless atomic absorption

E.Kontas

29

Neutron activation analysis of gold in geological samples

R Rosenberg Maija Lipponen and Riitta Zilliacus

33

Gold concentrations of some reference samples iscussion

E.Kontas

9

APPENDIX: The effect of sample weight and digestion and separation

method on the results of gold determinations.



FOREWOR

This collection of papers documents the analytical methods on which geological gold

studies in Finland are mainly based. It includes a general review of the problems associated

with the determination of gold and details of six procedures used by the laboratories of

the Geo logical Survey of Finland the Geoanalflcal Laboratory of Outokumpu Metals

Resources the Research Centre of Rautaruukki Co. and the Reactor Laboratory of

Technical Research Centre of Finland. Being very sim ilar to those for gold the analytical

methods for palladium and platinum are also described. Finally the gold contents of five

reference samples analysed by each laboratory are presented and their applicability for

different purposes is discussed. As the procedures have all been reported previously or

are otherwise well known no new scientific features are presented. How ever many of

the m ethods have been developed and improved over the years and thus a close look

at the procedures may reveal useful information for chemists confronted by problems

when analysing nob le metals. Because the correct selection of analytical methods is crusial

in gold studies the topic is worth a document of its own.

I thank all chemists and all the other persons who have contributed to this work in

one way or another and so made publication possible. In particular I thank Dr. Pekka

A.

Nurmi and Olli Lehto Geological Survey of Finland who reviewed the report and

made many useful suggestions and Pentti Noras Geological Survey of Finlan d and Pertti

Hautala Outokumpu Oy for critical discussions and valuable comments. The English

of the m anuscript was corrected by Gillian Hakli.

Rovaniemi

25 2 1992

Esko ontns

Geological Survey of Finland



Analytical methods for determining gold in geological samples

Edited by Esko Kontas

Geologian tutkimuskeskus, Tutkimusraportti

Geological Survey of Finland Report of Investigation

114 , 5-11, 199 3

ODS FOR DE

GOLD

IN

GEOLOGICAL SA

by

sko

Kontas

K o n t a s

E

1993

Analytical methods for determining gold in geological

samples. Geologian tutkimuskeskus, Tutkimusraportti Geological

Survey of Firzland Report

of

Investigation 11 4, 5-1 1, tables,

Because of the heterogenous mode of occurrence, or nugget effect , the

preparation and representativeness of analytical samples pose special

problems i n assaying gold. Field samples of gold ores and their host rocks

shou ld be crushed to a grain size of at least 95% below 2 mm o r preferably

below 1 mm b efore splitting and subsequent milling to analysis grain size.

About 100 g of powder is needed for analyses and possible control analy-

ses.

The six analytical procedures presented below are those most often used

for gold studies in Finland. The principles of the procedures, including

sample weight, digestion, separation, determination, detection limits and

capacities, are presented in Table 1

Table 1. The principles of the analytical procedures.

Sam ple Digestion Separation Determi- Detect. Capacity

weight nation limit

g ppm samples lw

25-50 lead fir e assay FAAS 0.1 150

20 aqua regia-KBrO, MIBK FAAS 0.05 200

10 aqua regia DBS-DIBK FAAS 0.05 300

5 aqua regia DBS-DIBK GAAS 0.02 400

1 aqua regia SnC1,-Hg GAAS 0.001 50 0

0.6 EN AA 0 .003 570

(Abbreviations: AAS atomic absorption spectrometry, FAAS flam e AAS,

GAAS graphi te furnace AAS, MIBK methyl i sobutyl ke tone , DBS dibutyl

sulphide , DIBK di-isobutyl ketone, ENAA instrumental epithermal neutron

activation analysis).

For the evaluation of methods, five internal reference sam ples prepared

by the Geological Survey were analyzed several times with the above

procedures. Th e relative standard deviations varied from below 10% up to

90%, being, a s expected, largest for the methods that used the sm allest

sample weights.

The classical lead fire assay method (with a sampl e weight of 5 0 g) is

suited best for inventories of gold deposits and for assaying gold concent-

rates. The methods with sample weight of 5-20 g and aqua regia diges-

tion are appropriate for exploration analyses of gold ores and for geoch em-

ical prospecting. Owing to their low detection limits and high capacities,

the last two methods in the above table are suitable for the geochemical

mapping and prospecting on a regional scale and for basic research.



Geologian tutk imuskeskus, Tutkimusraportti

eological Survey of Finland Report of Investigation

114 1993

Esko Kontas

Key words GeoRef Thesaurus, AGI) : chemical analysis , gold, sample

preparat ion, methods, spectroscopy, atomic absorption, techniques, neutron

activation analysis, accuracy

Esko Kontas Geological Survey of Finland P.O.Box 77 SF-96101

Rovanierni Finland

INTRODUCTION

There are three factors due to which exploration

for go ld may be viewed almost as its own branch of

science. The first is the rarity of gold, the second is

its heterogenous, nugget like mode of occurrence of

the metal in the nature, and the third is its high eco-

nomic value. As a result samp ling, sample prepara-

tion and analytical methods have to meet more strin-

gent requirements tha n other element.

The average crustal gold content is estimated

0.0035 ppm Li Yio 1966). The lower limit of the

abundances

in

gold ores is about a thousand times

higher. For geochemical exploration and for studies

of the richest ores, the analybcal methods of gold

should cover a concentration range of 104 to 102

PPm.

The gold concentrations in common geological

materials are as follows:

1 Igneous rocks median): ultramafic 0.0032, mafic

0.0032 and granites 0.0023 ppm Wedepohl

1969-1978),

2) sedimentary rocks median): limestones 0.005,

sandstones and quartzites 0.005 and shales 0.004

ppm Wedepohl 1969-1978),

3) soils average): 0.002 ppm Brooks 1972) and

4) fresh water median): 0.002 ppb Turekian 1977).

The m ineralogical mode of occurrence of gold in

nature varies widely but native gold is predominant

Boyle 1979). Even in very low concentrations it

usually appears as discrete particles of considerable

size, and is rarely, if ever, uniformly distributed. The

discrete particles tend to segregate easily due to their

high density relative to other minerals Brown

Hilchey 1974). Being soft and malleable, gold grains

are not easily reduced in milling; moreover in in-

tense milling the grains may adhere to the walls of

milling vessels and thus cause losses of gold from

the sample Harris 1982, Riddle 1983, Bu rn 1984).

Because of this uneven distribution, the main prob-

lem in the gold analyse and one which is o ften very

difficult to resolve, is the representativeness of sam-

ples. Owing to its high economic value, gold should

be determined with high accuracy and reliability. For

instance, the gold i n many base metal concen trates

has a major effect on their price, and even small er-

rors in assays may have m arked economical signif-

icance when translated into masses of m aterial.

In practice, the analytxal capacities, detection

limits and operating costs of procedures also have to

be taken into account. Therefore no single procedure

can effectively and economically meet all these re-

quirements. For one procedure, the representative-

ness of an analybcal sample may be good enough but

the detection limit may be too high and capacity too

low; for another, the representativeness may be poo r

but the capacity high, and so on . Furthermore, there

are very many procedures that are very similar in

principle and chemistry but very different in practice.

The difference is particularly marked in th e analyt-

ical apparatus and other facilities. The classical f ~ e

assay with lead collection requires only a simple

fusion furnace and a microbalance. Two advanced

instruments often used are the rather simple and

cheap flame atomic absorption spectrometer FAA S)

and the more complex and expensive graphite fur-

nace atomic absorption spectrometer GFA AS). A

tool that is rapidly gaining popularity is the very

powerful plasma mass spectrometer ICP-M S) Date

et al. 1987, Jackson et al. 1990). Most powerful of

all is the nuclear reactor, which

s

used for the neu-

tron activation analysis NAA) Hoffman et al. 1978,

Hoffman and Brooker 1982). The following takes a

look at all these procedures except the p lasma-based

techniques.

S MPLE REPRESENT TIVENESS

Gold occurs in nature most commonly as native

ver, copper or platinum-group metals. Some gold

grains or as the main com ponent of alloys with sil-

and gold-silver tellurides, stibnites, selenides and



Geologian tutkimuskeskus, Tutkimusraportti eological Survey of Finland Report of I~zvest igation114 993

Analytical methods fo r determining gold in geological samples

bismuthides are also known. But the predominant ore

minerals are native gold, aurostibnite and various

tellurides (Boyle 1979).

Gold tellurides are brittle and readily grindable

minerals (Burn 1984). However, gold deposits also

contain native gold as g rains, which are not pulver-

ized

in

milling. These grains may be extremely vari-

able in size. Furthermore, gold can be incorporated

in comm on sulphide minerals. Discrete inclusions,

smaller than 0.1 pm, are termed invisible gold , as

they are not detec table by optical and scanning elec-

tron microscopy. The concentrations of invisible

gold range from less than 0.5 ppm to more than 100 0

ppm in sulphide grains from 12 different ore depos-

its, as determined by ion-probe microanalysis (Cook

Chryssoulis, 1990). The most important gold

carriers are arsenopyrite and arsenic-rich varietes of

pyrite. In the extended gold province of Wit-

watersrand, the average grain size of gold is 80 pm

(Boyle 1979). In placer deposits gold grains may

weigh thousands of grams although weighths of a

few milligrams are more common.

The representa tiveness of samples containing gold

particles has been theoretically and experimentally

studied by Clifton et al. (1969). According to them:

It can be shown mathematically that the number of

gold particles in the sample is the only factor cont-

rolling the precision of the chemical analysis. If the

following assumptions are valid:

1) gold particles are of uniform mass,

2) gold particles comprise less than 0.1 of all parti-

cles,

3

the sample contains a total of over 1000 particles,

4) ana lyt~ cal rrors are disregarded

5) go ld particles are randomly distributed within the

sample,

a precision of S O % at the 95% confidence level will

be achieved when the sample taken for analysis

contains twenty particles of gold. For reconnaissance

studies, a sm aller sample containing fewer particles

of gold may suffice. It is important, however, to note

that as the expected number of particles per sample

falls below five, the chance of having no gold parti-

cles in a given sample greatly increases.

In fact the above assumptions are never valid,

especia lly those concerning uniform grain sizes and

analytical errors. Nevertheless, they provide a practi-

cal starting point for estimating the representative-

ness of a sample, as long as something is known

about the grain size. If the size distribution of the

gold particles is unknnown and analyses of splits are

not available, an adequate sample size can be deter-

mined by assuming that all the gold in the sample

occurs uniformly in grains as large as the maximum

size which has an effect on the total gold content

(Clifton et al. 1969).

The analytical representativeness of samples with

discrete particles has been studied at lenght by Gy

(1982). The standard deviation of the fundamental

sampling error for particulate materials can be esti-

mated theoretically from som e basic properties of the

material to be sam pled. Gy (1982) derived the follo-

wing equation for making the estimation:

where is the relative standard deviation of the

fundamental sampling error,

M,

the mass of the sam -

ple,

M

the mass of the lot to be sampled, Z and C

the sampling constants and d the dimension of the

largest pieces in the lo t to be sampled. Because

M

is

large related to MS he equation can be simplified:

where Z is defined as Cd3. The sampling constant C

for the particular material to be sampled contains

four parameters characteristic of the material:

C = f g l c, where f is the particle shape factor, g the

size range fac tor, 1 the liberation factor and c th e

composite factor. A pproximations or calculations for

these factors has been briefly and clearly presented

by Minkkinen (1987). The sampling constants C or

Z

can also be estimated empirically by determining

the relative standard deviation of the sampling error

and calculating it from the original equation. This

requires a number of analyses of subsam ples having

the same size for fine materials and analyses of indi-

vidual fragments for coarse materials (Gy 1982).

On the basis of Gy's theory of sampling,

Minkkinen (1987) developed a computer program

SAMPEX)

for solving practical sampling problems.

The method involves estimating the sampling con-

stant C . For well-characterized materials,

C

can be

estimated from the material properties; for unknown

materials it can be evaluated experimentally. The

program can be used to solve the following prob-

lems: the minimum sam ple size for a tolerated rela-

tive standard deviation of the fundamental sampling

error; relative standard deviation for a given sample

size; the maximum particle size of the material for a

specified standard deviation and sample size; the

balanced design of a multi-stage sampling and sam-

ple-reduction process; and samplin g for particle size

determination.

When the grade is low and the gold particles are

large the sample size required increases very rapidly.

However, normal laboratory instruments and other



Geolo gian tutkimuskeskus Tutkimusraportti

eological Survey of Finland Report of I?zvestigatio~t

114 1993

Esko Korltas

facilities often restrict the use of large analytical

samples. In many production geared, commercial la-

boratories sam ple weights range from 10 g to 30 g.

For very inhomogeneous materials it may be best to

use physical methods of concentration for the deter-

mination of coarse-grained gold and the chemical

analysis of fine -grained gold (Nichol 1985). Howev-

er, the more complex procedure the higher are the

operation costs.

Mathematical determination of the representative

sample weight always calls for some information

about the mode of occurrence of gold and its mine-

ralogy.

In practice, when gold explorations begin in a new

area, laboratories have to rely on assumptions. No

data are available, not even on the possible existence

of gold.

Since the precision and accuracy of analyses are

highly dependent on the mode of occurrence of gold

in samples, chemists need all the information geol-

ogists can give them about the samples. With goo d

cooperation between geologists and chemists, it is

possible to choose the methods that are the most ad-

vantageous and economically viable for each pur-

pose.

SAMPLE PREPARATION

The physical preparation of rock samples involves

drying, crushing, splitting and pulverizing. Because

of the hete rogeneous distribution of gold , the normal

rules for safely reducing sample weight do not apply

(Gy, 1979). The laboratory should therefore work

with the complete sample submitted as far as possi-

ble (Ridd le 1983). The Geo logical Survey of Finland

obtained acceptable accuracies and precisions for

gold in different types of ore and their host rocks

using the following procedure : Bulk samples (2-5

kg) were crushed in two steps in a jaw crusher to a

grain size of 95 less than

2

mm. The samples were

then split in a rotatory tube divider to obtain a

subsample of about 100 g, which was then milled for

analysis in a swing mill. T here were 137 samples in

all and their concentration range was from back-

ground to ore grade (Nurmi et al. 1991). Samples

weighing 20 g were analysed for gold both by fire

assay and by GFAAS after aqua regia digestion. The

correlation coefficien t obtained between these deter-

mination~was 0.93 (Appendix).

Geoanalytica l Labora tory of Outokum pu Metals

Resources (Olli Lehto, P.O. Box 74, SF-83501

Outokumpu, Finland, pers. commun.) uses the fol-

lowing procedure for preparing rock and drill core

samples: Samples are crushed in a jaw chrusher to a

grain size of 90 less than 4 mm. All coarse mate-

rial is th en crushed in a roller mill to a grain size of

95 less than

1

mm. The sam ple is split in a rotatory

tube divider into eight fractions. Depending on the

amount of the sample one or more fractions are

pulverized completely in a swing mill to obtain

about 60 g powder.

Based on the above methods of preparation, a rec-

ommended procedure might be as follows: Before

the first splitting a whole field sample is crushed to

a grain size of 95 less than

1 mm;

a portion of

about 100 g of the crushed samp le is then pulverized

for analysis.

A jaw crusher is a co st-effective appara tus univer-

sally used for crushing rocks. It is quick and easy to

use and contaminates the sample only very slightly.

However, the grain size achieved is at most 95 less

than 2

mm.

Nevertheless, a jaw crusher is fully ade-

quate fo r precrushing. A grain size of 95 less than

1 mm is easily obtained by precrushing material in a

roller mill, but then contamination of samples by

rollers is a considerable problem. Samples, particu-

larly those containing metallic gold, need frequent

and very careful cleaning because the gold grains can

smear and stick to the surface of the rollers. disk

mill, which is very effective for fine crushing down

to grain sizes of approximately 0.15 mm and is fre-

quently employed at laboratories specializing in the

appraisal of gold ores. The disks, however, wear

easily when hard rocks are milled and produce con-

tamination by iron and its alloys, which is harmful if

the same powder is used for other geochemical stud-

ies (Olli Lehto, pers. com mun.).

Grinding ore-grade samples may smear rather than

grind the gold grains. Thus the grinding vessels con-

taminated with gold whereas the sample is slightly

depleted in gold (Riddle, 1983, Burn, 1984).

E.

Ojaniemi (Rautaruukki CO,Research Centre, Raahe,

Finland) performed quantitative experiments to test

the smearing of gold on the surfaces of grinding ves-

sels in a swing mill as follows: Three sam ples with

varied gold concentrations were milled for different

length of time. After each milling the grinding pots

were washed with water before milling sterile quartz.

The gold concentrations of the milled quartz were

determined to check whether or not any gold had

smeared (Table 2 .



Geologian tutkimuskeskus, Tutkimusraport ti eological Survey o Firrland Report o brvestigatiorz 114, 1993

The relative depletion of gold during the usual

milling time of about 60

S

is insignificant. But the

contamination of subsequent samples may become

significant if samples of ore grade and of back-

ground contents are crushed and milled in the same

batch and with the sam e equipment.

Soil sam ples are usually dried and sieved to specif-

ic fractions. The nugget effect of sediment samples

is much more problematic than that of ore samples

because the grade is generally low whereas the gold

grains are often relatively coarse (Clifton et al.,

1969). Plant and other organic samples are usually

ashed before digestion. Warren and Horsky (1986)

have successfully determined gold and thallium

directly o n pulverized plant samples after nitric acid

and aqua regia digestion.

~ n a l ~ t i c a lethods for determining gold in

s a m p l e s

Ta b l e

2.

Go l d c o n c e n t r a t i o n s i n

20

g p o r t i o n s o f c l e a n i n g q u a r t z

(mi l l ing t ime 60 S) i n r e l a t i o n t o t h e m i l l in g t i m e s o f t h r e e s a m -

ples.

M i l li n g Am o u n t Au i n q u a r t z a f t e r c l e a n i n g

t ime of of quar tz

s a m p l e S a m p l e 1 S a m p l e 2 Sa m p l e 3

S g PPm PPm PPm

30 20 0 .016 0.003 0.024

60 20 0.014 0.007 0.020

120 20 0.030 0.073 0.087

240 20 0.056 0.119 0.127

Sa m p l e 1 = copper concent ra te wi th 32.0 p p m A u ,

s a m p l e

2

= a l b i t is i z e d s c h i s t wi t h

1.2

p p m A u a n d

s a m p l e 3

=

a lb i t i zed schi s t wi th 13.5 p p m A u .

GOLD RING CONT MIN TION

Experiences has shown that a gold ring may be

a

source of heavy contamination, especially during

sampling but also during sample preparation. Gold

ring contamination becomes significant when the

gold co ncentration of samples is a t the ppb level. For

example, gold concen trations of 500-600 ppb were

found in a fine fraction of till when samples were

handled after sampling by a person wearing a gold

ring. A lso rather soft materials such as plant samples

are easily contaminated by a ring. The plastic cap-

sules used in neutron ac tivation analysis may pick up

significant quantities of gold from a ring. After slight

contam ination, Au contents of 15-170 ng were

found on capsules (Kontas 1990). It should be re-

membered that the neutron activation method m ea-

sures all the gold in a capsule regardless of whether

the gold is outside or inside the capsule.

REVIEW ON N LYTIC L METH ODS

The physical preparation of samples is usually

Neutron activation analysis is the only method that

followed three stages:

allows gold to be determined directly on samples

with a relatively low detection limit without any

1 decomposition of the sample,

preconcentration. H owever, in most activation labo-

2)

separation and preconcentration of the gold and

ratories, the direct method restricts the size of the

3) measurement.

sample, making it too small and unrepresentative.

Decompositon of samples

Fusion

Fire assay for gold uses reductive fusion and the

classic collection by lead in a procedure that has

been used since ancient times to concentrate and iso-

late the noble metals (Beamish Van Loon 1977,

Van Loon 1984). Lead oxide is added to the fusion

pot. The oxide is reduced to lead m etal in the fusion

process, which then quantitatively ex tracts and

collects any gold or silver in the sample. More re-

cently, nickel sulphide has also been used as a col-

lector in the fire assay of noble metals (Robert et al.

1971).

cid treatment

Mixtures of hydrochloric and nitric acids or aqua

regia are most commonly used in acid treatment

procedures. A normal mixture is HCl+HNO,

=

3+1

but others can also be used. Even if aqua regia dis-

solves silicate minerals only to some degree, partial

attack with aqua regia usually results almost total



Geologian tutkimuskeskus Tutkimusraportti

eological Survey

of Finlmzd, Report of htvestigation

114

1993

Esko

Kontas

dissolution of gold from geological materials when regia Rubeska et al. 1977). Hydrobrom ic acid with

samples are roasted Signiholfi et al. 1984). Silicate bromium is also useful in dissolving gold from roast-

samples can also be digested with hydrofluoric and ed samples Thom pson et al. 1968).

perchloric acids before gold is dissolved in aqua

Separation and preconcentration

In the conventional fir assay, lead is separated

from the fusion slag, hammered into a cube or button

and placed on a heated cupel of bone ash or magne-

sia. This in turn is placed in a cupellation furnace

where the lead i s oxidated into lead oxide, which is

absorbed by the cupel leaving behind a small gold-

silver alloy bead that contains all the gold to be de-

termined. Gold is determined either gravimetrically

or instrumen tally. In the ~zickel ulphide

fir

assay

procedure the NiS button obtained by fusion is

milled and dissolved in hydrochloric acid. Gold and

the other noble metals precipitate as sulphides or

metals.

The p recipitate is filtered, rinsed and dissolved for

measurement AAS , ICP, ICP-MS) or the noble

meta ls are measured directly from the precipitate by

neutron activation Hoffman et al. 1978).

Solvent extraction

The gold i n a solution may be separated and con-

centrated with liquid-liquid extraction. Numerous

solvents and complexing agents are in use and sum-

maries of the methods have been given by Beamish

and Van Loon 1977). The most common method

involves extrac tion of gold II1) from a hydrochloric

or hydrobromic acid solution with methyl isobutyl

ketone MIBK). In Rubeska et al. 1977), acid attack

is followed by extraction into dibutylsulphide in to-

luene and in Parkes and Murray-Smith 1979) into

dibutylsulphide in di-isobutyl ketone.

oprecipitation

In accordance with its character, metallic gold is

easily precipitated by reductants. Since solutions are

highly d iluted in relation to gold, a coprecipitant is

necessary. Tellurium is the usual coprecipitant and

stannous chloride Fryer and Kerrich 1978) or

hypophosphorous acid the most common reductant

McH ugh 1983 ). Stannous chloride-mercury copreci-

pitation is another attractive method for separating

gold Kontas 1981, Kontas et al. 1986) and some

other readily reduced elements such as the platinum-

group metals, silver, tellurium and selenium Niska-

vaara and Kontas 1990). n advantage of

coprecipitation is that the analyses can be performed

in water solutions, which are more stable and easier

to handle than the organic solutions and solvents.

Measurement

In the classical lead cupellation method gold is

determined g ravimetrically. The silver-noble metal

bead or prill obtained by fusion and cupellation is

treated with dilute nitric acid to separa te the go ld as

pure metal, which is then weighed. The detection

limit is generally 0.1 ppm. Fire assay separation has

not changed much down the ages, but gravimetry

has, being first replaced by optical emission spectro-

metry and then by atomic absorption spectrometry.

The method is easy to apply and is used for gold ores

and concentrates. Various separation and pre-con-

centration methods can be combined with different

instrumental measuring methods. N owadays atom ic

absorption instruments are very popular Van Loon

1985). If a detection limit of 0.05 ppm is sufficient,

flame atomic absorption is a cost-effective method.

Graphite furnace atomization readily permits a ppb

level, but measuring is slower and the equipm ent is

more expensive than in flam e atomization. Emission

spectrometry had became alm ost obsolete until the

new plasma emission and plasma m ass instruments

gave it a new lease of life. The most sensitive measu-

ring method, plasma mass spectrometry, permits

ultra trace contents to be determined. Neutron activa-

tion analysis is now adays also widely applied, and if

it is com bined with a suitable preconcentration meth-

od, a small sample size does not impair its

representativeness. The sensitivity of X-ray analysis

is very low. However, based on X-ray analysis an

interestingmethod has been developed for reconnais-

sance determination of go ld in the field: gold is dis-

solved from a sample into a cyanide solution and

adsorbed and concentrated on the surface of a carbon

disk. Gold is then determined from the disk with a

portable X-ray instrument ASOM A Instruments,

Austin, Texas).



Geologian tutkimuske skus, Tutkimusraportti eological Survey of Finland Report of Investigation 114, 1993


REFERENCES

Beamish F. E.

1966.

Analytical Chemistry of the Noble

Metals . Pergamon Press Inc. Oxford, 609 p.

Beamish F. E. Van Loon J. C.

1977.

Analysis of Noble

Metals . Overview and Selected Methods. Academic

Press, New York, 327 p.

Boyle R. W. 1979. The Geochemistry of Gold and I ts De-

posi ts. Bull. Ge ol . Surv. Can. 280, 584 p.

Brooks R. R. 1972. Geobotany and Biogeochemistry in

Mineral Explorat ion. Harper and Row, New York, 290 p.

Brown

B.

W. Hilchey G. R. 1975. Sampling and analysis

of geochem ical materials for gold.683-690 in Geoche-

mica1 Explorat ion 1974, ed. by I . L . Elliott and W K. F1-

etcher. E lsevier S cientific Publishing C O, Amsterdam

1975 , 720 p .

Burn R. G.

1984.

Factors affecting the selection of the

methods of gold analysis . Mining M agazine 150, 5 , p .

468.

Clifton E.

H.

Hunter R. E. Swanson F. J. Phillips R.

L. 1969. Samp le s ize and meaningful gold analysis . U. S.

Geol. S urv. Prof. Paper 625-C, 17 p.

Cook

N.

J.

Cryssoulis S. L.

1990.

Conce ntration of in-

visible gold in the common sulf ides. Can. Mineral . 28,

1-16.

Date A. R. Davis A. E. Cheung

Y . Y.

1987. The poten-

t ial of f i re assay and inductively coupled plasma source

mass spectrometry for the determination of platinum

group elements in geological mater ials . Analyst , 112, 9 ,

1217-1222.

Fryer B. J. Kerrich R.

1978.

Determination of precious

metals a t ppb levels in rocks by combined wet chemical

and flameless atomic absorption method. Atom. Abs.

N ew sl . 1 7 , 1 , 4 - 4 .

Gy P. M. 1982. Sampling of Particulate Materials. Theory

and Pract ice. Elsevier Publishing CO , Amsterdam, 431 p.

Harris J. F.

1982.

Sampling and analyt ical requirements for

effect ive use of geochemistry in explorat ion for gold.

53-67 in Precious Metals in the Northern Cordillera ed.

by A. A. Levinson. Published by the Association of

Explorat ion Geochemists , 1982, 214 p.

Hoffman E. L. Naldrett A. J. Van Loon J. C. Hancock

R. G. V. Manson A.

1978.

The determination of al l

platinum-group elements and gold in rocks and ore by

neutron activa tion analysis after precouce ntration by a

nickel sulf ide f ire assay technique on large samples.

Anal. Chim. Acta 102, 157-166.

Hoffman E. L. Brooker E. J. 1982. The determination of

gold by neutron activation analysis.69-77. l Precious

Meta ls In the Northern Cordil lera ed. by A. A. Levinson.

Published by the Associat ion of Exploration Geochem ists

1 9 8 2 , 2 1 4 p .

Jackson S. E. Fryer B. J. Gosse W. Healey D. C.

Longerich H P. Strong D. F.

1990.

Determination of

the precious me tals in geological mater ials by inductively

coupled plasma-mass spectrometry ( ICP-MS) with nickel

sulphide f ire assay collect ion and tel lur ium copreci-

pitation. Chem . Ge ol. 8 3, 1 19-132.

Kontas E.

1981.

Rapid determination of gold by flameless

atomic absorption spectrometry in the ppb and ppm rang-

es w ithout organic solvent extract ion. Atomic Spectrosco-

py, 2, NO 2, 59-61.

Kontas E. Niskavaara H. Virtasalo J.

1986.

Flameless

atomic absorption determination of gold and palladium in

geological reference samples. Geostandards Newslet ter

Vol. 10. No 2, 169-171.

Kontas E.

1991.

Gold contamination of the f ine f ract ion of

t i l l dur ing sampling and sam ple preparat ion. J . Geochem .

Expl. 39, 289-294.

Li T. Yio C-L.

1966.

The abundance of chemical ele-

ments in the ear th 's crust and i ts major tectonic units .

Sei. S in. Vol. 15 , No 2, 258-272.

McHugh J . B. 1983. Determination of gold in water in the

nanogram range by electrothermal atomization af ter

coprecipitat ion with tel lurium. Atom ic Spectroscopy Vol.

4, No 2, 66.

Minkkinen

P. 1987.

Evaluation of the fundamental sam-

pling error in the sampling of par t iculate sol ids. Anal .

Chim . A cta, 19 6, 237-245.

Nichol I.

1985.

Gold Exploration 8.-10. May 1985. Cou rse

Notes, Depar tme nt of Geological Sciences , Q ue er s Uni-

versi ty , Kingston, 51 p .

Niskavaara H. Kontas E.

1990.

Reduct ive

coprecipitat ion as a separat ion me thod for the determina-

t ion of gold, pal ladium, plat inum, rhodium , s i lver , seleni-

um and tellurium in geological samples by graphite fur-

nace atomic absorption spectrometry. Anal . Ch im. Acta,

231, 273-282.

Nurmi P. A. Lestinen P. Niskavaara H. 1991. Geo-

chemical character is t ics of mesotherm al gold deposi ts

in

the Fennoscandian shield and a c ompe r ison with selected

Canadian and Austral ian deposi ts . Geol . Surv. Finland,

Bullet in 35 1, 1 01 p.

Parkes A . Murray-Smith R.

1979.

A rapid method for

the determination of gold and palladium in soi ls and

rocks. A tomic Abs. N ewsl. Vo l. 1 8, No 2 , 57-58.

Riddle C. 1983. Analytical me thod s for gold. 272-278 in

The Geology of Gold in Ontar io, ed. by A. C. Colvine.

Ont. Geol . Surv. Miscel laneous Paper 110, 450 p.

Robert R. V. D. Van Wyck E. Palmer R.

1971.

The

collection and determination of the noble m etals in ores

and concentrates by the fusion technique using nickel

sulphide as a collector . National Inst i tute for Metal lurgy

(South Afr ica) Repor t 1371.

Rubeska I. Koreckova J. Weiss D.

1977.

The determi-

nation of gold and palladium in geolo gical mater ials by

atomic absorption af ter extract ion with dibutylsulf ide.

Atom ic Abs. New sl. V ol. 16, No 1, 1-3.

Signiholfi G. P. Gorgoni C. Moham ed A. H

1984.

Comprehensive analysis of precious metals in some

geological standards by f lame less A.A. spectroco py.

Geostand ards New sletter, Vol. 8, No 1, 25-29.

Thompson C. E. Nagakava

H.

M. Vansickle G. H.

1968.

Rapid analysis for gold in geologic mater ials .

U

S.

Geol. Surv. Prof. 600-B, 130-132.

Turekian K. K.

1977. Geochemical distribution of ele-

ments . 111 Encyclopedia of Science and Technology, 4th

edn. 627-630. McG raw-Hill, New York.

Van Loon J. C.

1984.

Accurace determination of the noble

metals . I . Sample decomposit ion and method s for separa-

t ion. Trends Anal . Chem . 3, 10, 272.

Van Loon J. C.

1985.

Accurace determination of the noble

metals . 11. Determination methods. Trends Anal . Chem .

4, 1, 24.

Warren V. H. Horsky S. J. 1986. Thall ium, a

biogeochemical prospecting tool for gold.

J.

Geochem.

Expl. 26, 215-221

Wedepohl K.

H.

Editor

1969 1978.

Handbook of Geo-

che mist ry, Vo l. 11-4. Sprin ger Verla g, Be rlin.






Geological Survey of Finland Report of Investigatioti

114 13-16 1993

D IN GEOLOG MA TERIALS

EAD FIRE ASSAY SE ARAT ION

by

Riitta Juvonen and Paavo J. Vaananen

Juvonen, R. Vaananen,

P.J.

1993 Determination of g old in geological

materials by atomic absorption af ter lead f ire assay separation. Geologian

tutkimuskeskus, Tutkimusraportti

eological Survey of Finland Report

of Investigation 114 13-16 3 tables.

Key words (GeoRef Thesaurus, AGI): chemical analysis, techniques, gold,

atomic absorption, reagents, accuracy

Riitta Juvonen and Paavo J. VaanZinen Geological Survey of Finland

SF-021 50 Espoo Finland

INTRODUCTION

Analysis of gold and silver with the classical

fire assay technique h as been in use at the Geo-

logical Survey of Finland for decades. The

method with its many variations is well docu-

mented in numerous publications and monog-

raphes (e.g. Haffty et al. 1977, Beamish et al.

1977, Moloughney 1986). The method is widely

used all over the world not only for analysing

gold but also for silver, platinum and palladium.

It permits use of a large sample, which is neces-

sary because of the uneven distribution of gold

in geological material (Moloughney 1986).

The dem and for gold analyses in our labo rato-

ry has increased. We have therefore studied the

method carefully, seeking to m ake it faster and

easier to perform without sacrificing the quality

of the results. The following describes the met-

hod as it is currently performed at the laboratory

of the Geological Survey of Finland.

REAGENTS AND APPARATUS

1) Naber N -41 H , muffle furnace with a vent on

4) Atlantic-Schmelztiegel GmbH, fusion cruci-

top of the furnace,

bles, roasting dishes and cupels,

2) Perkin-Elmer Model 5000, atomic absorption

5) Merck, fusion flux reagents and lead oxide

spectrophotom eter equipped with AS 50, (PbO), extra pure grade,

autosampler

6) Merck, HNO, (G ) , HCI (37 ) and AgNO,,

3) Perkin-Elmer Model 2380 with HGA-500

reagent grade,

graphite furnace equipp ed with AS 4 0 autosampler,

7) BDH Chemicals Ltd, gold, platinum and



Geologiau tutkimuskeskus Tutkimusraportti eological Survey

o

Finland Report

o

I~west igat ion

114 1993

Riitta Juvonen and Pnavo

J.

ViiZi~zlinen

palladium standard solutions and tartrate and 0. 6 kg of glass powder are mixed

8) Fusion flux: 0.8 g N a2C 03, 1.3 kg K 2C0 3,

well in a ball mill.

1.0 kg Na2B,0,, 1.3 kg of potassium hydrogen

EXPERIMENT L

A 25 g sample of ground rock is mixed well

with 50 g of lead oxide (litharge) in a plastic

bag. A bout 130 g of the fus ion flux is added and

the contents of the plastic bag are mixed well. A

known amount of silver is added in the form of

a silver nitrate solution. The plastic bag is

placed into a fusion pot and the pot is trans-

ferred to a preheated furnace at 1100°C. The

sample is left in the furnace for hour, after

which the melt is poured into an iron mould to

cool. Most of the slag is hamm ered off, and the

lead regulus is further cleaned by soaking in

10 HCI. The regulus is hammered into the

form of a cub e, brushed clean and cupelled. The

lead is o xidized by air to le ad oxide at a furnace

temperature of 940°C. The liquid lead oxide is

absorbed by the magnesite cupel and a small

bead of silver is left behind. It contains the gold

and silver and the platinum-group metals to

varying degrees. Twelve samples are fused and

cupelled at a time, using the same furnace type

for both operations.

The silver bead obtained by fire assay is flat-

tened and placed in a 10-m1 graduated tes t tube.

A small glass bead is added to avoid overboiling

the solution. HNO, (6 5 ), 0.5 ml, is added and

the test tube is warmed carefully in a water bath

for about one hour. The tube is cooled and 1.5

m1 of HC l (37 ) is added. The solution is agi-

tated, allowed to stand overnight at room tem-

perature, and warmed very carefully, avoiding

overboiling, on a water bath for about 30 min-

utes. Wh en the dissolution is complete, the tube

is cooled and filled to the mark with 6 M HCl.

Gold standards are prepared to contain the same

amoun t of acids as the sam ples. Silver need not

be added to the standard solutions. Gold is de-

termined by flame atomic absorption. The solu-

tion obtained is also used for determining plati-

num and palladium by graphite furnace atomic

absorption.

THE LE D FIRE SS Y METHOD

Extraction of noble metals into lead

The well ground sample is melted together

with lead ox ide and a reducing fusion flux mix-

ture, in which gold and the platinum-group

metals together with added silver are reduced

and dissolved in the simultaneously reduced

metallic lead. Quantitative extraction of the

noble metals into lead requires a complete reac-

tion between the sample and the fusion flux. The

sample must be finely ground and well mixed

with the reagents before fusion. The viscosity of

the flux should be s uch that the noble metals are

dissolved by the metallic lead as it sinks to the

bottom of the fusion pot. If the flux is too fluid,

the lead will sink too fast to bring the noble

metals down with it. If the flux is too viscous,

lead globules will remain in the slag, resulting

in low metal values.

The best possible fusion flux has been devised

for each rock type (Haffty et al. 1 977). Howev-

er, in routine work it is not generally possible to

prepare the optimum fusion mixture for each

sample. The flux, mentioned above, in routine

use at the laboratory of th e Geological Su rvey of

Finland, generally works well. To control the

fu-

sion process, a check is kept on the weight and

the appearance of the lead regulus. Reference

samples are frequently analysed along with the

sample series. Samples with more than

2

sul-

phur are roasted at 600°C in special roasting

dishes before fusion.



Geologian tutkimuskeskus, Tutkimusraport t i eological Survey

o

Fiidmrd Reporr of Iirvestigntiora 114. 1993

Determinat ion of gold in geological materials by atomic absorpt ion after lead f i r e assay separat ion

SCORIFIC TION OF THE LE D REGULUS

The lea d regulus m ay be further purified by a

process called scorification if the sample con-

tains hig h concen trations of interfering elements,

mainly copper and nickel. In the fusion process

copper, cobalt and nickel are partly reduced

along with the lead and noble metals. Cobalt

will not dissolve in the lead, but copper and

nickel will.

If the sam ple contains appreciable amounts of

copper or nick el, further purification of the lead

regulus is necessary. The regulus along with

about 1 g of borax is put into a shallow

scorification crucible, which is transferred to a

furnace with ample air flow, at 1000°C. In the

furnace, part of the lead is oxidized and simul-

taneously copper and nickel are oxidized and

dissolved in the slag. The scorification is re-

peated, and metallic lead is added until a clear,

smooth and malleable lead regulus is obtained.

For exam ple, the sco rification of one gramm e of

copper requires 50 g of lead.

CUPELL TION

The lead is oxidized by a process called cu-

placed on the cupels. The regulus melts almost

pellation: the lea d regulus is heated in the furna-

instantly, the lead begins to oxidize and is ab-

ce on a cup el, made of m agnesite, which absorbs

sorbed by the cupel. In the last stages of the

the forming lead oxide. A small bead of silver

cupellation, only a small spot remains in the

containing the noble metals is left.

centre of the cupel. This finally disa ppea rs, giv-

In practice, the cupels are first heated in the

ing off a bright flash of light at the end. Pro-

furnace at 940 C, after which the lead reguli are

longed heating causes losses of silver.

DETERMIN TION OF NOBLE MET LS

In the conventional method of analysis, the

silver bead is flattened by ham mer and anvil, the

flattened bead is weighed, silver is dissolved in

dilute nitric acid, and the resulting flake of gold

is weighed. The difference gives the total

amo unt of silver. Along with the go ld, any rho-

dium and iridium together with small amounts

of platinum and palladium remain undissolved in

nitric acid treatment (Beamish et al. 1977). The

samples should contain about three times more

silver than gold for a good partition of the two.

The addition of silver also increases the size of

bead, making it easier to handle.

Analysis with atomic absorption has replaced

gravimetry in m any laboratories (Kallmann et al.

1970, Moloughney 1986). At the laboratory of

the Geological Survey of Finland, the flattened

silver bead is dissolved in aqua regia and the

gold is analysed by flame absorption and

Ta b l e 1. Gr a p h i t e f u r n a c e p ro g r a m m e f o r p a l l a d i u m a n d p l a t i-

num.

S t e p Fu r n a c e T i m e I n t e r n a l

No.

t e m p e ra t u re R a m p H o l d g a s f l o w

C m l / m i n

Pa l l a d i u m

1

2

3

5

Pl a t i n u m

1

2

5

platinum and palladium by graphite furnace

atomic absorption.




eological Survey of Finland Report of Investigation 114, 1993

Riit tn Juvonen and P a m o J. Vii i ini inen

Table 2. Gold concentrations of five Canadian reference samples.

Sample Weight

F

Number of

This work Recommended

determinations

AU

P Q ~

values ppm

CH-1 10

CH-2 10

MA- l a 5

MA-3 5

GTS-1 10

Data collected over three years.

Table 3. Platinum and palladium concentrations of reference samples.

Sample Weight Number of det's This work Recommended values

Pt Pd Pt Pd Pt Pd

g PPn' PPm PP PPm

SARM-7 25 9 3 3.4550.22 1.4950.21 3.740 1S 3 0

SU la 5

4

0.3850.02 0.37

Data collected over three years.

PRECISION ND CCUR CY

Five go ld ores of the Can adian Reference palladium were determined from the reference

Materials Proje ct were analysed with the present samp les SAR M-7 South African Com mittee for

method but with reduced sample weights. Corn-

Certified Reference Materials) and SU la Cana-

parison of th e results with the recomm ended dian Certified Reference Materials Pro ject). The

values is presented in Table 2. Platinum an d results are presented in Table 3.

REFERENCES

Beamish P

E

Van Loon

J.

C

1977 Analysis of Noble

Kallmann

S

Wobart E W

1970 Determina t ion of

Metals. O verview and Selected Methods. A cademic silver gold and palladium by combined f ire assay atomic

Press, New York, 327

p.

absorption. Talanta 17, 845-850.

Hafty

J .

Riley L

B.

Goss

W. D. 1977 A Manual on

Moloughney

P

E

1986 Assay methods used in CAN ME T

Fire Assaying and Determination of the Noble Metals in for the determination of precious metals. CA NM ET SP

Geolog ical Materials.

U.

S. Geol. Surv. Bull . 1445, 58 p. 86 -lE , 33 p.





Geologian tutkim uskesku s, Tutkimusraportti

Geological Survey of Finlmzd Repori of Investigation 114 17-19 1993

by

P. Noras

Noras

P.

1993. Determination of gold by aqua regia-potassium bromate

digestion, methyl isobutyl ketone extraction and f lame atom ic absorption.

Geologian tutkimuskeskus, Tutkimusraportt i Geological Survey of

Finlmzd Report of bwestigation 11 4, 17-19, 1 table .

Key words (GeoRef Thesaurus, AGI): chemical analysis, gold, techniques,

atomic absorption, sample preparation, reagents, accuracy

P. Noras Geological Survey of Finland SF-02150 Espoo Finland

INTRODUCTION

In the late 1960s the need arouse for a rapid

method to determinate gold, particularly owing

to the increase in geochemical exploration,

where both speed and cost are of paramount

importance. Thompson et al. (1968) introduced

a method based on Br,-HBr digestion, methyl

isobutyl ketone (MIBK) extraction of AuBr, and

AAS determination of Au. The method, de-

scribed here uses aqua regia and potassium bro-

mate for digestion before extraction of Au-

halogenide complexes and determination of Au

in very much the same way as the method of

Thompson et al. The present method was deve-

loped and used extensively by the Mineral Pro-

ject of the U.N. Development Programme in

Colombia on which the author was employed in

1974-77. Practical improvem ents were ma de to

the procedure in the Geological S urvey of Finland .

The method, which allows sample weights of

up to 20 g to be used, offers a detection lim it of

0.05 ppm Au in the sample, and accepts most

types of material encountered in geochemical

exploration. MIBK extraction has the advantage

over the conventional f ire assay method that it

saves labour and materials expence, and can be

operated by less trained technicians. The disad-

vantage is that it cannot be us ed to ex tract other

precious metals. Moreover some interference is

caused by high contents of iron in the MIBK

phase.

RE GENTS ND PP R TUS

1) Reagents: HCI 37 , HNO, 65 , KBrO,, grade,

methyl isobutyl ketone (MIBK), all reagent

2) apparatus: Furnace for calcination (m aximum



Geologian tutkimuskeskus Tutkimusraportti eological Survey of Firrlmld Report of Investigation

114 1993

P .

Noras

temperature preferably 1000°C and

Elmer Model 460 equipped with three-slot bur-

3) atomic absorption spectrophotometer Perkin-

ner head.

S MPLE PRETRE TMENT ND DIGESTION

If sample s are to be pulverized, avoid grin- to a 250-m1 Erlenm eyer f lask , add 25 m1 of

ding them too fine to prevent the formation of concentrated HC1, 5 m1 of concentrated HN O,

em ulsions during the extraction stage. and 0.5 g of KBrO,, shake the flask, stopp er it

W eigh 20 g of dry sample powder into a wide lightly and allow it to stand at room temperature

roastin g dish. To destroy organic matter, calcine overnight. Then add 0.5 g of KBrO,, shake the

sedim ent samples at 600°C for 2-5 hours, and flask again and place it in a water bath at 90°C

rock samples at 700°C for one hour to decom-

for one hour.

pose sulphides. Transfer the calcined sample

EXTR CTION

Coo l and dilute the sample solution by adding of the flask helps to separate th e two phases.

170 m1 of w ater a nd shaking the flask. Then add

Difficulty in breaking up the emulsion in the

20.0 m1 of MIB K, stopper the flask and shake it acid phase m ay be e ncountered with very fine

vigorously for 2 min. Let it stand until the clear grain sizes. It is recommend ed that such sample

organic phase has settled out. Rapping the neck

solutions be fi l tered before the M IBK i s added.

INSTRUMENT L CONDITIONS ND DETERMIN TION

O F

GOLD

Gold is determined by AAS in an air-acetyle-

water is aspirated. The spectrome ter is calibrated

ne flame at a wavelength of 242.8 nm. A three-

using standard MIBK solutions of gold prepared

slot burner head is em ployed, and the flame is

in the same way as the unknown samples. The

adjusted to lean blue when aspirating MIBK.

working curve is linear, typically up to 2 pg A u

Th e flame is just beginning jump up when the

in 1 m1 of MIBK.

DISCUSSION

The precision and accuracy of the method

were studied by determining Au in some inter-

national stan dard reference sam ples and in-house

reference samples. D ue to the high gold value of

the international materials, sam ple weights were

limited to

2

g (Table 1).

At higher concentration levels (more than

0.15 ppm) the me thod shows good agreement

with recommended values and acceptable pre-

cision. Un fortunately, for assessing the precision

of the method at lower levels, the data mainly

derive from the 2-g sam ples. The detection l imit

(3xb) of the method i s estimated to be 0.05 ppm

Au in sample.

The only serious interference is due to the

high contents of acid soluble iron in the sample.

Iron extracted from an acid solution to MIBK

gives false absorption at the measuring wave-

length. It has been shown that 15% acid soluble

iron in a sample is cri t ical . Fortunately, the iron

content is not that high in most sample types,

although the critical concentration may be

reached in pyrite ores. To overcom e the possible

interference, the MIBK phase should be washed

with 0. 1 M HCl solution until free of yellow co-

lour.



Geologian tutkimuskeskus, Tutkimusraportti eological Survey of Finland Report of Investigation 114, 1993

Determination of gold by aqua regia-potassium bromate digestion, methyl isobutyl ketone extraction and flame atomic absorption

Table 1. Gold concentrations of some reference samples determined by the above method and the recommended values (in ppm).

S ample Weigh Number Au Stand. Recommended

number of det s dev. value1

P P ~

USBM

G 2

10

USGS

2 .6

10

GXR- 1

2

Gladney and Burns (1984).

The determination limit of this method is

fairly acceptable for rock samples, alluvial sedi-

ments and residual soil samples. However, the

method is certainly not sensitive enough for

studying regional geochemical patterns e. g. till

samples. For lower levels of Au and Ag, Bratzel

et al.

(1972)

have described a method in which

digestion a nd extraction are similar to the above

but determination of Au is by carbon-rod AAS.

The daily rate of the method run by two tech-

nicians is

30-50

effective determinations. The

method does not need highly trained staff or

sophisticated instruments and thus s easily

applicable in developing countries. Since

MIBK

is a flammable chemical and harmful for health,

proper fume cupboards and ventilation are

essential.

REFERENCES

Bratzel M . P Jr. Chakrabarti C. L. Sturgeon R. E. elemental concentration data for the United States Geo-

MacIntyre

M .

V. Agemian H.

1972. Determination of

logical Survey s geochemical exploration reference sam

gold and silver in parts per billion or lower levels in

ples GXR-1 to GXR-6. Geostandards Newsletter Vol. 8,

geolog ical and metallurgical samples by atomic absorp- No. 2, 119-154.

tion spectrometry with carbon rod atomizer. Anal Chem Thompson C. E. Nagakava

H.

M. 82 Vansickle G H.

44, 372.

1968.

Rapid analysis for gold in geological matrials. U.

Gladney E. S Burns C. E.

1984. 1982 Compilation of S. Geol. Surv. Prof. P. 600-B, 130-132.






Geological Survey of Finland Report of Investigation 114 21-23 1993

DETERMINA A DIGESTION

EXTRACTION

by

U. Penttinen

Penttinen U.

1993.

Determination of gold by aqua regia digestion, dibu-

tylsulphide-di-isobutyl ketone extraction and flame atomic absorption.


Geological Srmey of

Finland Rep ort of Investigation 114

21-23, 1

table.

Key words GeoRef Thesaurus, AGI): chemical analysis, techniques, gold,

palladium, atomic absorption, sample preparation, reagent s, accuracy

U. Penttinen retired

Outokurnpu Me tals Resources

Geoan alytical Laboratory

P O

ox 74 SF-83501 Outokunnpu Finland

INTRODUCTION

Prospecting for the gold and platinum metals

Outokurnpu Oy. The use of dibutylsulphide in

has increased greatly, giving rise to a need for a gold determination was first described by

fast and sufficiently accurate analytical method. Yudelevich et al. in 1970. The method has since

Introduction of the method described below has been refined b y Rubeska et al. 1977) and

definitely improved the effectiveness of the Parkes and Murray-Sm ith 1 979). W e developed

prospecting for gold and platiniferous deposits our gold and palladium determinations from

undertaken by the Exploration Department of these methods i n 1980.

THE PRINCIPLES OF THE METHOD

sample should be well-homogenized and

or, alternatively, the sample should be subm itted

preferably large. Organic material and sulphide

to total digestion with hydrofluoric and nitric

sulphur are eliminated by roasting at 600°C

acids. The gold is dissolved in aqua regia. Th is

Strong and Murray-Smith 1974). The graphite, is preceeded by hydrochloric acid treatment to

if present, should then be burned in the oxygen

dissolve the iron precipitates on the surface of

stream at 600°C. If gold occurs as inclusions in

the gold particles. The gold and the palladium

silicate particles, very fine grinding is needed

are extracted from 2M hydrochloric acid solu-




eological

Survey

of Finland Report of I?n~estigation

14 1993

U . Penttinen

tion w ith dibutylsulphide dissolved i n di-iso- from the washed and centrifugued organ ic ex-

butyl ke tone. Th e gold and palladium concentra-

tract. The other elements do not interfere with

tions are determined by flame atomic absorption

the determination.

RE GENTS ND PP R TUS

1) HCl 37%, HNO 65%, reagent grade, 3) Atomic absorption spectrophotometer: Perkin-

2) Dibutylsulphide solution: Dissolve 17.4 m1 Elmer Model 500 0 and

dibuty lsulphid e in 50 0 m1 of di-isobutyl ketone,

4) Shaking machine: Buhler Model B.

Procedure

W eigh 1 0 g of fine ground and well mixed

sample into a porcelain crucible and roast i t at

600°C fo r 2-3 hou rs. Mix as need ed. If the

sample contains arsenic, roast it at first for 2

hour s at 450°C an d then for 2 hours a t 600°C to

avoid partial evaporation of gold. If graphite is

to be eliminated, burn i t afterwards in a tube

furnace i n oxyge n stream at 600°C.

Transfer the roa sted sam ple to a 25 0 m1 beak-

er, ad d 80 m1 of diluted HCl (1+1), cover it with

a wa tch glass and keep it at 80-100°C for two

hour s. Add 2 0 m1 of concentrated HNO,, and

keep i t covered at the same temperature for one

hour, stirring from time to time. Evaporate to

wet salts . If evaporated to dryness, gold may

be reduced t o a lower valence and cannot be ex-

tracted.

Add 1 5 m1 of concentrated HC l and 2 0 m1 of

water. Heat to dissolve the salts. Filter through

a Whatman GFIB glass fibre filter and wash

with water. Rinse into a 100-m1 measuring flask,

cool, fi l l to the m ark with w ater and stir.

Transfer a 50-m1 aliquot to a 100-m1

separatory funnel. Add 5 m1 of dibutylsulphide

solution and shake using the full effec t and sha-

king amplitude of the machine for two minutes.

Let the phases separate. Wash the organic ex-

tract with 25 m1 of 2M HCI. Centrifuge for 10

minutes at 3500 rpm.

Flame atomic absorption of gold is measured

at a wavelenght of 242.8 nm using background

correction. The interga tion time is 0.5 and the

reading an average of ten measurements. Palla-

dium is measured using a wavelength of 244.8

nm and background correction.

For calibration, the e xtraction is don e using 1,

2

and 3 m1 of so lutio n conta ining 10 pglm l of

gold and palladium i n 2M HCl.

CCUR CY ND PRECISION

Th e following samples were analysed to check Mine (Sample B) and

the accuracy and precision of the method: 3) gold ore MA -2, Canm et (Samp le C).

1) Cu-Bi-CO matte, Falconbridge (Sample A),

The results are presented in Table 1.

2) Cu concentrate, Outokumpu Oy, Pyhasalmi

DETECTION LIMIT ND C P CITY

If a 10 g sam ple, a 100 m1 measuring bottle, a palladium is a bout 0 .05 ppm. A team of three

5 0 m1 aliquot a nd 5 m1 of dibu tylsulp hide solu-

laboratory workers can make 60 determinations

tion are used, the detection limit for gold and

a day.



Geologian tutkimuskeskus, Tutkimusraportti eologicnl Survey of Finlr n~d eport of bwe stigatio n

114

1993

Determination of gold by aqua regia digestion, dibutylsulphide di isobutyl ketone extraction and flame atomic absorption

Table 1. Gold and palladium concentrations in some reference samples.

S

ample Number of This Stand. Coe ff. of Recommended

det s work dev. var. value

REFERENCES

Parkes A. Murray-Smith R. 1979. A rapid method for

Strong B. Murray-Smith R.

1974.

Determination of gold

the determination of gold and palladium in soils and

in copper bear ing sulphide ores and m etallurgical f lota

rocks. At. Abs. Newsl. Vol. 18, No 2, 57-58.

tion products by atomic absorption spectrometry. Talanta

Rubeska I. , Koreckova J. & Weiss D. 1977. The determi 21, 1253.

nation of gold and palladiu m in geological materia ls by

Yudelevich I . G., Wall G . A., Torgov V. G. Korda T.

atomic absorption af ter extraction with dibutylsulphide .

M.

1970.

Extraction a tomic absorption determination of

At. Abs. Newsl. Vol.

16

No

1, 1-3.

gold in solutions. Zh . Anal. Khim.

25 , 870 .




Edited by Esko Koutas

Geologian tutkirnuskeskus Tutkimusraportti

Geo logical Survey of Finland Report of Irzvestigation 114 25-27 1993

b

E.

Ojaniemi

Ojaniemi E. 1993 Determination of gold, palladium and platinum by

aqua regia digestion, dibutylsulphide-di-isobutyl ketone extraction an d f la-

meless atomic absorption. Geologian tutkimuskeskus, Tutkimusraportti

Geological Survey of Fi da nd Report of Investigation 114

25-27

2

tables.

Key words (GeoRef Thesaurus, AGI): chemical analysis, techniques, gold,

palladium , platinum, atomic absorption, sample preparation, reagents

E.

Ojanienzi

Rautaruukki CO Research Centre

SF-92170 Raahe

Finland

I N T R O D U C T I O N

The method was developed for the exploration tractions. First gold and palladium are extracted

of platinum-group metal deposits containing from 2M hydrochloric acid solution into dibutyl

gold as well as for the determ ination of gold. sulphide in di-isobutyl ketone (Parke s and

The preparation and dissolution of a sample also Murray-Sm ith 1979). After reduction w ith stan-

allow the determination of palladium and plati- nous chloride, platinum can be extracted into the

num. The method is based on hot aqua regia same reagent (Simonsen 1970). Measurem ents

digestion and two successive liquid-liquid ex- are made by flameless atomic absorption.

R E A G E N T S A N D A P P A R A T U S

Reagents: tone. The solution is kept out of the dayligh t in

a refrigerator,

1) HCl 37 , HNO, 65 , reagent grade, 3) Stannous chloride solution: 10 0 g of SnCl,

2) 0.2M DBS in DIBK: 14.6

g

of dibutyl sul-

dissolved in 1000 m1 of 2M HCl. New solution

phide dissolved into 500 m1 of di-isobutyl ke-

is required daily,




eological

Survey

of Finland

Report

of Investigation

114 1993

E.

Ojanietni

4) Ferric chloride solution: 25 mg Felml in

spectrophotometer equipped with an HGA 500

2 M H CI.

graphite furnace and an A S 4 0 autosampler,

2) Shaking machine: Desaga, tube mixer:

Instruments: Heidoplh.

1 Perkin-Elmer Model 5000 atomic absorption

S MPLE PRETRE TMENT ND DIGESTION

Before digestion, roast the samples at 800°C

for 1.5 h to decompose sulphides and organic

matter.

W eigh 5.00 g of sam ple into a porcelain cru-

cible. After roasting, transfer it to a 50-m1 test

tube hav ing a plug and a magnetic stirrer on the

bottom. Add 20 m1 of 6M HCl and shake well.

Hea t the tube o n an aluminium block for 1 h at

90°C. Remove th e tube from the block and allow

it to cool for ab out 15 rnin at room temperature.

Add 5 m1 of concentrated HNO, and shake well.

Put the tube back on the aluminium block and

leave the solution to evaporate overnight. Re-

move the tube from the block in the morning.

The residue must be pulpy; very dry or roasted

residue does not give correct results. Dissolve

the residue in 35 m1 of 2M HCl, shake the tube

well and make sure that all the particles have

disintegrated. Warm the tube again at 90°C for

30 m in.

EXTR CTION OF GOLD ND P LL DIUM

Remove the solution from the test tube into a

100-m1 measuring bottle. Rinse the tube and fill

the bottle up to the mark with 2M HC1, and

shake. Let the silicates settle overnight, or cen-

trifuge the required aliquot of the solution

3

min, 3500 rpm). Pipette an aliquot of 50 m1

into a 100-m1 separatory funn el and add 5 .00 m1

of DBS -DIBK-solution. Extract Au and Pd into

the organic phase by shaking the funnel for 2

rnin manually or for 10 rnin mechanically. Al-

low the phases to separate. Transfer the water

phase to another 100-m1 separatory funnel fo r

the extraction of platinum. Remove the organic

phase to a 10-m1 test tube for the determ inatio n

of gold and palladium. If necessary, clear the

phases by centrifuging 3 min, 2000 rpm).

EXTR CTION OF PL TINUM

Add 2 m1 of stann ous chloride so lution to the

separatory funnel with the water phase, and,

while stirring, add smaller amounts about

0.5 ml) until the yellowish brown colour of the

ferric iron has disappeared. Add another 5 ml,

stir and let the solution stan d for 10 min. Extract

Pt into 5.00 m1 of DBS-DIB K solution by shak-

ing manually for 2 rnin or mechanically for

10 min. When the phases have separated, trans-

fer the aliquot needed for the measurement into

a 10-m1 test tube. If necessary, clear th e phas es

by centrifuging. If platinum need not be deter-

mined gold and palladium can be extracted

directly from the 2M HCl solution in the test

tube that contains the sample residue.

ST ND RD SOLUTIONS ND C LIBR TION

The stand ard stock solutions of gold and them in small volum es of aqua regia. After

palladium, with concentrations of 1000 pmlml,

careful evaporation, the residues are disso lved in

are prepared from pure metals by dissolving hydrochloric acid to yield a conc entratio n of



Geologian tutkimuskeskus, Tutkimusraportti eological urvey of Finland Report of Investigation 114 1993

Determination of gold, palladium and platinum by

HCl of 0.6 M i n the final solutions. The solu-

tions are kept in dark bottles. T he standard stock

solution of platinum, with a concentration of

10 0 pglml, i s prepared in the same way by dis-

solving pure metal in aqua regia. Before evapo-

ration some potassium chloride is added. The

final solution should be 1.2M in respect of HCl.

Standards are calibrated by preparing a wor-

king solution from the stock solutions. It con-

ta ins Au 0 .5 yg lml , Pd 1 .0 ~gl rn land Pt

2.5 pglm l. Aliq uots of 0.0, 0.25, 0.50, 1 O, 1.50 ,

2.00, 3.00 and 4.00 m1 are pipetted into

separatory funnels. Then 5 m1 of FeC1, solution

is added and the solution is diluted to 50 m1

with 2M HCl. Gold and palladium are extracted

into 5.00 m1 of the DBS-DIBK solution. The

water phases are separated from the organic

phases , and after th e adding of stannous chloride

solution platinum is extracted into 5.00 m1 of

DBS-DIBK solution.

If only Au and Pd have to be determined the

standard solutions are prepared i n the sam e way

as the samples but without adding of FeC1,.

The sample extracts remains fresh i n closed test

tubes for a month and the pure standard solu-

tions for at least two m onths.

Table

1.

Ins trument parameters for Au in D BS-DIBK solut ion

STEP TEM P. R AM P HOLD R EAD R EC

INT.FLOW

C

S S

Tube: pyrocoated graphi te tube , samp le volume

10

p1.

Purge gas: argon.

DISCUSSION

The accuracy of the method was controlled by

od described are 0.02 ppm for Au and Pd and

analysing reference samples SAR M-7 and Kont-

0.1 ppm for Pt.

tijarvi over several months. The results are pre-

Two technicians working together are able to

sented in Table 2.

prepare 4 0 samples and to make 12 0 determina-

The detection limits achieved with the meth-

tions a day.

Table 2. Gold, pa l ladium and pla t inum concentrat ions (ppm) in re ference samples SARM -7 and Kont t i ja rvi .

Element Au Pd Pt

Sample SARM -7 Kont t ija rvi SARM -7 Kont t ija rvi SARM . Kont t i ja rvi

Numb.det 's 15 15 15 15

15 15

Average

0.27 0.16

1.41 5.98 3.87 1.78

Stand.dev. 0.05 0.017 0.08 0.38

0.57 0.16

C. V.

19.0 10.6 5.7 6.4

14.7 9.0

Recomm. val. 0.31 0.17 1.53 6.3

3.74 1.80

REFERENCES

Parkes

A &

Murray Smith R. 1979.

A rapid method for

Simonsen

A

1970.

Determinaton of platinum in basic rocks

the determination of gold and paIIadium in soils and

by solvent extraction and atomic absorption sp ectroscopy.

rocks. At. Abs. Newsl. Vol. 18, No 2. 25-27

Anal. Chim. Acta 49, 368-370.






Geologic al Survey of F inland Report of Investigation 114 29-32

1993

DETERMINATION REG

DIGESTION

by

E.

Kontas

Kontas

E.

1993. Determination of gold and palladium by aqua r eg ia

digest ion, s tannous chlor ide-mercury coprecipi tation and f lam eless atom ic

absorption. Geologian tutkimuskeskus, Tutkimusraportti

eological

Survey of Finland Report of Investigation 114 29-32 2

tables.

Key words GeoRef Thesaurus, AGI): chemical analysis, techniques, gold,

pal ladium, atomic absorption, sample preparat ion, reagen ts , accuracy

E.

Kontas

Geological Survey of Finland

P.

0

Box 77

SF-96101 Rovanierni Finland

INTRODUCTION

Geochem ical prospecting for gold and palladi-

um requires a rapid method of analysis with

sensitivity reaching the ppb level and som etimes

even sub ppb level. This method, which was

earlier reported for gold analyses Kontas 1981),

has now been refined to apply to both gold and

palladium Kon tas et al. 1986). In the original

method the detection limit for gold was about

2

ppb, but it is now about 0.2 ppb for Au and

Pd. Th e m ethod is based o n a sample weighing

1 g and overnight aqua regia digesting at the

room tem perature. Heating is partly compensat-

ed for the long digestion time. G old and palladi-

um are separated from analyte solution by re-

ductive precipitation using stannous chloride as

a reductant and mercury as a coprecipitant

Barnard and Zeeman 1958). For the determina-

t i o n ~ ,he Hg Au-Pd) precipitate is dissolved in

aqua regia and diluted with water. M ercury does

not interfere with the determination of gold or

palladium b y graphite furnace AAS

Centrifugation is used to separate a clear sam -

ple solution after digestion and Hg Au-Pd) pre-

cipitation from the bulk of the analyte solution.

Centrifugation makes the method faster and

more suitable for handling large batches o f sam-

ples than conventional filtering.




eological

Survey

of Finland Report of Investigation

114 1993

E. Ko~ttns

Reagents, all reagent grade:

REAGENTS AND APPARATUS

1) HCI 37 , HNO, 65 ,

2) stann ous chloride solution: 20 SnC12.2H,0

in 1M HC1,

3) mercurous nitrate solution: 1 mg Hg/ml in

0

M HNO, and

4 mercury chloride solution: 0.5 mg Hg/ml in

5 0 vol- aqua regia.

Standard stock solutions:

Merck Titr isol

Apparatus

1 ) Retsch Rhetoterm roasting apparatus,

2) Beckman centrifuge with receptables for 56

test, tubes (100X16mm),

3) Perkin-Elmer M odel 2280 atom ic absorption

spectrophotometer equipped with a

D,

back-

ground corrector and an HGA 50 0 graphite fur-

nace equipped with an AS 40 autosampler and

4) Perkin-Elmer Model 3030 Zeem an atomic ab-

sorption spectrophotometer equipped with an

HGA 600 graphite furnace and an AS 60

autosampler.

SAMPLE PRETREATMENT AND D IGESTION

Samples with carbon, graphite or large

amounts of sulphides are roasted at 750°C for

2 0 min. If sulp hides ex ist only in m oderate

abundaces, roasting is not necessary.

W eigh g of samp les into plastic tubes

(100x16 mm) or into porcelain crucibles if

roasting is necessary. After roasting transfer

them into tubes. Add 2.5 m1 of conc. HCI and

shake; then add 0.5 m1 of conc. HNO, and

shake. Let them stand overnight at room tem-

perature. In the morning shake them again and

add 4 m1 of water; centrifuge fo r 10 min a t

2000 rpm. Pour the clear, supernatant solutions

into conical glass tubes (100x16 mm) or take

aliquots for further separation of Au and Pd.

If the samples are ashes of organic materials,

reduce the weight of samp les 29-32, g), replac e

HNO, with 0.5 m1 of H202,and the water in the

morning with 4 m1 of 3M HCI.

Separation of gold and palladium

Add 2 m1 of SnCl, solution and 1 m1 of

tr ifugation is not generally required in r insing,

Hg2(NO,), soluti on to the sam ple solutions. Wait

since the Hg precipitate remains on the b ottom

for about f ive minutes and centrifuge the soh-

or the walls of the tubes. Make sure that this has

tions for 10 min at 2200 rpm. Pour the solutions

happened.

away; rinse the Hg(A u-Pd) precipitates and the For organic ashes, use only 0.5 m1 of

tubes by filling them with water and then pour-

Hg,(NO,), solution.

ing it away. One rinsing is often enough. Cen-

Dissolving the Hg Au-Pd) precipitate a nd calibration

Add 0.7 m1 of conc. HCI and 0.3 m1 of conc.

by graphite furnace AAS. The final volume of

HNO, to the tubes containing the Hg(Au-Pd)

the sam ple solutions is 2 m l, because about 0.1

precipitate. Shake the tubes and let them stand

m1 of water is left in the tubes a fter rinsing .

for about one hour. Then add 0.9 m1 of 3M From the stock solutions, the calibra tion stan-

HNO,, s hake the tubes once more and let them dards are diluted with mercury chloride solutio n.

stand overnight. T he analysis can then be done



Geologian tutkimuskeskus, Tutkimusraportti eological Survey of Finland Report of Investigation 114 1993

Determination of gold and palladium by

...

Table 1 Instrument parameters for gold' and palladium2.

Au Pd

S T E P 1 2 3 5 1 2 3 4 5

TEMP. C 200 500 1900 2500 100

RA MP s 15 15 0 1 5

H O LD s

15 15 5 5

REA D 0

R E C

2

BA SELI N E - 25

1NT.FLOW

300 300 0 300 300

' p e r k i n - ~ l m e rMode l 30302 AAS and HGA 600 graphite furnace.

Au

wavelength 242.8 nm, slit 0.7 nm. Pyrocoated graphite tube

with L'vov 's platform , peak area integration, purge gas argon.

'Perkin-Elmer Model 2280 AAS with HGA 500 graphite furnace. Pd wavelength 244.8 nm, slit 0.2 nm. Pyrocoated graphite tube,

peak height integrat ion, purge gas argon.

The concentration range of the standards, which

depend on the detection limits required. The in-

varies from 5 to 1 00 nglml of Au or Pd, and the strument parameters are presented i n Tab le 1.

solu tion volumes injected into a graphite furnace

DIS USSION

The precision and accuracy of the method was

tested for gold by analysing USGS reference

sam ples GX R 1-6 (Tab le 2). Ex cept for the

samp le GXR 1, the recommended values

(Gladney and Burns 1984), were not precise

eno ugh for accuracy to be assesse d. Therefore the

results of Meier (1980) are shown. Meier also

used graphite furnace AAS but after HBr-Br2

digestion and methyl isobutyl ketone (MIBK)

extraction. Recommended values for palladium

were not foun d in the litterature for GXR's and

so reliability could not be evaluated. However,

the precision for Pd determinations observed

from replicate weighings was very good.

Despite good agreement with the values of

Meier (1980), the small size of the analytical

sample is a disadvantage in the above method

and thus the poor representativeness of samples

with coarse gold grains. On the other hand, the

use of small samples is an advantage in

analysing as hes of organic materials. A s we can

assume that gold exists homogeniously in or-

ganic material, there is no need to collect and

burn big sam ples.

The method seems to be very suitable for

determining th e background con tents of gold and

palladium in silicate rocks (Kontas et al., 1986)

and other geological materials. The detection

limits are low enough for many purposes. Con-

tamination by reagents is improbable, because

their amounts are small and they are generally

very pure. Except for platinum-rich samples,

partial attack with aqua regia after roasting

usually ensures almost total dissolution of gold

and palladium from geological materials of

different composition (Signiholfi et al. 1984).

This method is m ainly used fo r the geochemi-

cal exploration of gold and palladium, of which

the latter is a good pathfinder element for all

platinum-grou p metals. The main sam ple materi-

als are fine till fraction (-0.064 mm) and bed-

rock. Stream sediment, peat and vegetable sam-

ples have also been analysed.

Carbon and graphite are the matrix substanc-

es, that cause the worst interference as they

adsorb the gold from solutions quantitatively

(Lakin et al. 1974). Fortunately, even small a-

mounts are easily detected at the digestio n stage,

and new portions can be taken and roasted be-

fore the digestion. The ashes of organic samples

often contain finely divided carbo n which can be

destroyed with hydrogen peroxide. Therefore

hydrogen peroxide is used instead of nitric acid

as the oxidizing agent for ashes.

Samples exceedingly rich in iron oxides or

base metal sulphides are som ewhat troublesome

since sample solutions become concentrated in

iron or other metals, which severely interfere

with the separation. This difficulty ca n b e avoid-

ed by taking a fifth or tenth part of the sample

solution for follow-up analysis. The detection

limits are then five or ten times higher, respec-

tively. In favourable cases the detection limit is

0.1 ppb for gold and 0.5 ppb for palladium. T he

capacity is about 2000 samples per one month

and two persons.




eological Survey of Finland Report of hrvestigation

114, 993

E. Kontas

Table 2. Gold and palladium concentrations of GXR reference samples (aritmetic meanrtstandard deviation). Recommended values

after Gladney and Burns 1984) and the results obtained by Meier 1980).

Au P P ~ Pd P P ~

Sam ple No. of Recomm ended Meier This work This work

det s values

GXR

5

3 1005200 2950 3 1005200

0.1

GXR 2 3 46f 19 22 10.0rt0.3 0.1

GXR 3 3 3-600

3 2.4rt1.7 0.2

GXR

4 3 440rt160

353 419rt14 0.2

GXR

3 80560 7 7.0rt0.9 0.4

GXR 6 5 70rt10 63 86+6 2 2

The method is easy to modify. If necessary

method is run on the basis of a

20

g sample.

digestion can be intensified by heating; the

Results are in good agreement with those ob-

sam ple weight can be increased and an aliquot tained by fire assay method (Appe ndix).

taken for follow-up analysis. Nowadays the

REFERENCES

Barnard

E.

Zeeman

P.

B.

1958.

Die Konsentrer ing en

Spektrgraf iese Bepaling van Edelmetale Gesteendes en

Rotse. Tegnikon 1 1,

2,

63-69.

Gladney E. S. Burns C. E.

1984. 1982 Compilation of

elemental concentration d ata for the United States Geo-

logical Survey s geochemical exploration reference sam-

ples

GXR-1

to GXR-6. Geostandards Newsletter Vol. 8,

No 2, 119-154.

Kontas E.

1981. Rapid determination of gold by flameless

atomic absorption spectrometry in the ppb and ppm rang-

es

without organic solvent extraction. At. Spectr. Vol. 2,

N O 2, 59-61.

Kontas E. Niskavaara H. Virtasalo J.

1986.

Flameless

atomic absorption determination of gold an d palladium in

geological reference samples. Geostandards Newsletter

Vol . 10. NO 2, 169-171.

Lakin

H.

W.

Curtin C. G. Hubert

A.

E. 1974. Geo-

chemistry of Gold in the Weathering Cycle.

U

S. geol.

Surv. Bull. 1330, 8 p .

Meier A.

L. 1980.

Flameless atomic abso rption determina-

tion of gold in geological materials.

J .

Geochem. Expl.

13 , 77-85.

Signiholfi G. P. Gorgoni

C.

Mohamed A.

H. 1984.

Comprehensive analysis of precious metals in some

geological standards by flameless

A.

A. spectroscopy.

Geostandards Newsletter, Vol. 8 , No 1, 25-29.



Analytical methods for determining gold i n geological samples



Geological Survey of Finland Report of bzvestigation 114 33-37 1993

GOLD

b

R. J. Rosenberg, Riitta Zilliacus and Maija Lipponen

R.

J.

Rosenberg Riitta Zilliacus and Maija Lipponen.

1993

Neutron

activation analysis of gold in geochemical samples. Geologian

tutkimuskeskus, Tutkimusraportti eolog ical Surv ey of Fiiinlatzd Re por t

of Irwestigation 114 33-37 2 tables.

The analytical techniques used for the analysis of gold in geochemical

materials at the Reactor Laboratory of the Technical Research Centre of

Finland are described. These are epithermal instrumental neutron activa tion

analysis of solid samples, thermal neutron activation analysis of freeze-

dried water and snow samples and radiochemical neutron activation

analysis of inorganic and organic samples. The detection limits are 3 ppb,

5 ng/l and

0.1

ppb, respectively. The accuracy and capacity of the meth ods

are discussed.

Key words GeoRef Thesaurus,

AGI):

chemical analysis, gold, t i l l , meth-

ods, reliabili ty

R. J Rosenberg Riitta Zilliac~ is nd Maija Lipponen

Technical Research Centre of Firilnnd

Reactor Laboratory

SF-021 50 Espoo Finland

INTRODUCTION

A t the R eactor Laboratory gold is determined

from g eological sam ples on an analytical service

basis. The annual number of samples has been

about 13,000 since 1982. The main customers

are mining companies, geological surveys and

universities in Finland and Sweden.

Three methods are in use: Instrumental

epithermal neutron activation analysis ENA A)

for pulverized samples. With this technique 23

other elements are determined simultaneously

with gold Rosenberg et al. 1982). Thermal neu-

tro n activation analysis NAA ) for water sam-

ples evaporated by freeze-drying. Radio chemical

neutron activation analysis RNAA ) for very low

gold concentrations, especially in biological

materials Zilliacus 1983). Because the numb er

of samples needed in geochem ical exploration is

fairly large the cost must be reasonable. T he low

limit of detection is also critical if all anom alies

are to be detected. Therefore special emphasis

was laid on detection limit and cost in addition

to accuracy. The analytical techniques and the

organization of the work are described in the

following.



Ge olo gia u tutkimuskeskus Tutkimusraportti eological

Survey

of Finland Report of btvestigatiort 114

1993

R

J Rosenberg Riitta Zill iacus and Maija Lipporterr

EXPERIMENTAL

Standards

Tw o differe nt k inds of standard are used. solution of gold is dissolved in ethylene silicate

So lutio ns of gold are used for the analysis of all with the aid of ethanol. This mixture is hy dro-

ex ce pt pulverized geological samples. Stock lyzed with ammonia, dried and ignited so as to

solution s of 1000 ppm are prepared by dissolv- yield a dry silicate po wder homog eneously

in g me tallic gold in aqua regia. Suitable dilu- doped with gold.

tio ns are made with deionized water before use. A comp osite standard containing several other

Pow dered standa rds prepared according to the elements is usually prepared.

procedure of Date (1977) are used for ENAA. A

Equipment

A

Triga Mk I1 research reactor is used for

irradia tions. T he reactor is run for about 7 h per

day from M onday to Friday. Sometimes Monday

is neede d fo r service work. The rotary specimen

rack contains 40 irradiation positions. The ther-

m al flu x is 1.2 xlO1' cm-'s-' and the Cd ratio for

gold is 2. Twenty of the positions are used for

thermal irradiations and twenty for epithermal

irradiations. The epithermal flux is obtained by

using c ontain ers of aluminium 30 cm X 25 cm in

size, lined with 1 mm of cadmium and again

with 0.2 mm of aluminium. T hese containers are

permanently located in the reactor and are only

tak en up to chang e samples. One container holds

32 capsules of 0.5 ml, and thus 640 sam-

ples can be irradiated simultaneously in an

epithermal flux. For RNAA the samples are

irradiated in the central thimble, where th e ther-

mal neu tron flux is 1013 cm-'s-l.

Measurements are performed with automatic

gamma-spectrometers com prising a Ge(Li) or Ge

detector with auxiliary electronics, a sample

changer, a multichannel analyzer, a microcom-

puter and inputloutput devices (Vanska et al.

1983, Rosenberg et al. 1985). Such a system

measures a series of samples automatically and

simultaneously calculates the elemen tal concen-

trations w hich are printed o n paper and cassette.

Five gamma-spectrometers of this kind are

available fo r activation ana lysis.

Procedure for ENAA

The standards and powdered samples are

weighed into polyethylene capsules with an

inner volume of 0.5 ml. Thus the sample size is

0.5-1 g, depending on the density. The samples

are transported and stored in styrox boxes hold-

ing 100 capsules each. The samples are often

weighed by the customers, and a list of the

weights accom panies the box on its arrival at the

laboratory. 32 capsules, four in a plane, are

wrapped in aluminium foil. One irradiation

series comp rises four standards, 12 control sam-

ples and 144 samples. These are inserted into

cadmium containers. The samples are irradiated

for one week (25-35 h) and measu red after a

decay tim e of 4-5 days with an autom atic gam-

ma-spectrometer. The measurement t ime is 20

min per sam ple. Thus measurement of one series

takes almost two and a half days.

The 41 1.8 keV of l g 8 ~ us used for calculating

the results. The half-life is 2.7 days . peak

with a statistical error of 30 or less is accept-

ed. If a peak is not found, the upper le vel is cal-

culated according to the principles of Currie

(1968). The detection limit for most sample

types is 3 ppb.

The work is organized in such a way that two

series are inserted in the reactor on Monday

morning and two on Thursday m orning. The two

series taken out of the reactor are allowed to

decay and thenloaded onto the sample changer on

Wednesday and Monday, respectively. T hus two

gamma-spectrometers are needed for analysing

576 net samples per week. D uring w eek-ends the

analyzers are free for other measurements. Ex-

cluding the weighing, one person can handle

these samples and wrap them i n aluminium foil ,





114 1993

Neutron activation analysis o gold in geochemical samples

load and unload them into and ou t of the reactor

on floppy discs from which they are cop ied onto

and sample changers, and the prepare the data

a cassette for inpu t into the mem ory of th e com -

files. The sample codes and weights are printed

puter at the start of a measurement series.

Procedure for NAA of water samples

M elted snow and ice as well as water samples

are preconcentrated by evaporation before irradi-

ation. This is performed by freeze drying. The

samples are collected into polyethylene bottles

and then preserved with the addition of 1 m1 of

suprapure nitric acid (Merck) per litre of water.

200 m1 of sample is dropped onto a sheet of

polyethylene foil in a Petri beaker and freeze

dried with a WKF L 0 5 lyophilizer. This takes

about 40 h. Standards and blanks are treated in

the same way. The polyethylene sheets are

wrapped and inserted in polyethylene capsules

for irradiation. Then the samples are irradiated

fo r 25 h in a th erm al flu x of 1 . 2 ~ 1 0 ~ ~ c m - ~ s - ~

After a decay time of five days the samples are

measured with an automatic gamma-spectrome-

ter. The measurement time is 1 h per samp le.

The average blank is ng and thus an effective

detection lim it using 200 m1 samples is 5 ngll.

The number of samples is l imited by th e low

capacity of the lyophilizer. Only 15 sam ples a

week can be dried.

Procedure for RNAA

This method is intended for solid geochemical

samples when concentrations lower than 3 ppb

have to be determined. Because the method is

more expensive than the instrumental ENAA, it

is used more seldom. The method is also suit-

able for plants and other organic materials. The

metho d has been described in detail by Zilliacus

(1983) and in this context will be described only

brief ly . 2 0 0 4 0 0 mg samples a re weighed into

ampoules of quartz and irradiated for 25 h in a

the rmal neu tron flux of 10 i3 ~ m - ~ s - ' .fter a

decay t ime of 2 - 4 days the samples are sub-

jected to chemical separation. 1 mg of gold

carrier is added. Mineral samples are dissolved

in hydroch loric acid followed by aqua regia. The

samp le is evaporated to dryness three times with

aqua regia. Organic samples are first wet-ashed

with n itric acid and hydrogen peroxide and then

treated with aqua regia. After this 2 m1 of satu-

rated boric acid solution are added to the dry

sample followed by 3 m1 of hydrochloric acid.

The sample is diluted to 20 m1 with water and

filtered through 50 mg of activated carbon

through a chimney. The carbon is washed with

Cr-carrier solution and water. The filter papers

are inserted into polyethylene capsules and m ea-

sured with an automatic gamma-spectrometer.

The measurement time is 2 h.

The yield of the chemical separation is deter-

mined by reactivation. It varies between 60 and

90 . The detection limit for gold is 0.04 ng

corresponding to 0.1 ppb in a sample when a

400 mg weighing is used.

The method is rather fast. One person can

make 12 separations per day and the same num-

ber of samples can be measured with on e count-

ing system.

Evaluation of the method

There are a number of possible sources of

cause this is the principal technique in use. T he

error in activation analysis. Some of these can

specific problems of the other techniques are

be avoided almost completely and others only at

discussed at the end of this chapter. The s ubjec ts

unreasonable expense. In the following these discussed here are:

sources are discussed in some detail, and the

overall accuracy of the techniques is evaluated.

1) representativeness of the sample,

The discussion applies mainly to ENAA be-

2) contamination,



Geolog ian tutkimuskeskus, Tutkimusraportt i

eological Survey of Finlnnd Report of Investigation 114, 1993

R.

J

Ros enb erg Riitta Zilliacris and Maijn Lipponen

T a b l e 1. The gold co ncent rat ions in the US Geological Survey geochemical s tandards (GX R 1-6) accord i ng t o t h is w ork (E NAA and

RNA A, RNAA f ro m Z i ll i acus 1983) and other autho rs (in ppm ). Allcott a is the ari thm etic mean of 50 determiuat ions and b the ar i -

thmet ic mean of

76

determinat ions . Maier a was made by f lame and b by graphi te furnace atomic abso rpt ion spect rom etry.

G X R l GXR 2 GXR 4 GXR 5 G X R 6

EN A

3.6k0.2 0.03 3kO.004 0.52k0.02 0.01 l+0.002

R N A A

3.65

0.031 0.52 0.008

Al l co t t 1974 a

2.8 0.038 0.58 0.12

1974

b

3.7 0.056 0.74 0.058

M o t o o k a

1979 3.4 0.65

Kont as i n t h i s i s sue 3.1 0.010 0.419 0.007

M ai er 1980 a 3.0 0.070 0.35

-

1980 b

2.95 0.022

0.353 0.007

G l a d n e y 1984 3.1k0.2 0.046+0.019

0.44k0.16 0.080~0.060

3) neutron flu x distribution,

4)

neutron absorption,

5) competitive reactions,

6) counting geometry,

7)

peak evaluation and

8)

counting statistics.

Th e sm all sam ple size, 0.5-1 g, causes a

problem in some cases because of the

inhomogeneous distribution of gold in many

samples. Bu t, according to investigations by the

authors, increasing the sample size from 1 g to

10 g does n ot significantly alter the situation.

Contamination is a special problem in ana-

lysing gold. The reasons for this are the low

concentrations dealt with and the fact that so

many people wear gold rings. Metallic gold is

soft and contaminates easily. Contamitions of

different kind have been encountered during the

work. In some cases the samples themselves

have been contaminated during preparation and

sometimes the surfaces of the polyethylene

capsules have b een contam inated during weigh-

ing. The only way to avoid this is to control

rigorously that peop le handling the samples are

not wearing gold rings.

The production of the radionuclide on which

the analysis is based depends on the neutron

flux reaching the go ld atom s. Variations in this

flux can be caused by the natural variation of

the flux i n the reactor and neutron absorption in

the sample itself. The neutron flux varies by

tens of percent depending on irradiation posi-

tions. Most of this can be compensated by map-

ping the flux and inserting flux monitors. But it

would be too expensive to measure the flux

separately fo r every sample. The horizontal flux

variation inside a cadmium container is little

more than

5

and because the positions of the

samples in the reactor cannot be controlled the

error caused by this variation is not compen-

sated.

Another problem is neutron absorption in the

sample. This is

a

problem specific to gold. Gold

is usually present in the form of small grains.

Therefore the resonance neutron neutron absorp-

tion in these grains may be significant, even if

the average concentration is small. Because the

grain size cannot be con trolled, this error cannot

be controlled either. In average gold concentra-

tions of less than 1 ppm it has not been of sig-

nificance. But in some cases intercomparisons

with other analytical techniques have shown a

10 negative error in concentrations of 10 ppm

or higher.

Competitive reactions are no problem for

gold. Errors in the counting geometry may be

caused by variations in the sample position

during measurement and variations in sample

size. Sample positioning by the sample changer

is precise, and custo mers are requested to B11 the

capsules to avoid variations i n sample size.

The only possible spectral interference is

caused by 1 5 2 ~ ~ ,ut a europium concentration of

50 ppm is needed to give the number of counts

corresponding to the detection limit, 3 ppb of

gold. Such europium to gold ratios are highly

unlikely. Because the peak of gold is situated o n

an undisturbed smooth part of the Compton

continuum in geological samples, evaluation of

the peak area causes no significant errors apart

from that caused by the counting statistics,

which depends on the sample type and gold

concentration.

When water samples are analysed, two addi-

tional problems arise: the possible loss of gold

during handling and the high risk of con tamina-

tion because of the extremely low concentra-

tions. Investigations with tracers have shown

that no loss occurs through absorption on the

walls of the collection vessel or during evapora-

tion. The polyethylene foils onto which the



Geologian tutkimuske skus, Tutkimusraportti eologica l Survey of Finland Report of Inves tigation

114, 1993

Neutron activation analysis of gold in ge ochem ical samp les

samples are evaporated contain gold and there-

for e a blank cannot be avoided. The blank varies

betw een 0 .6 and 1.2 ng per 200 m1 sample, lead-

ing to a corresponding uncertainity in the analy-

sis.

I n R N A A the special problem is the variable

yield of the separation. This is corrected as

described and does not cause a n additional error.

The results of control sam ples were monitored

to investigate the average precision of the meth-

od. 26 samples were chosen at random from

runs during two months. The result was as fol-

lows. The nominal concentration of the sample

is 0.59 ppm. T he mean of 26 determinations was

0.586 ppm, and the standard deviation

0.028 ppm. Consequ ently, the average relative

prec ision at the 95 confidence level is +9.8 .

Because of the random character of this error

Table 2. Results of an intercomparison conducted by

IAEA

1985). The sample is marine sediment SD-N-112.

This work Other laboratories

the error caused by counting statistics starts

influencing the total precision significantly from

60 ppb downwards. Theoretically a total preci-

sion of +1.7 is reached at 30 ppb and +22 at

10 ppb at the 95 confidence level.

The overall reliability of the method was

investigated by analysing standard reference

samples and by making intercomparisons.

REFERENCES

Allcott

G . H.

Lakin H.

W . 1975. The homogeinity of six

geochem ical explorat ion reference samples. Pp 659-681

n Geoch emical Explorat ion 1974, ed. by

I.

L. Elliott and

W. K. Fletcher. Elsevier Sc ientific Publishing Co. Am-

sterdam, 720 p.

Currie

L.

A. 1968. Limits for quanti tat ive detect ion and

quantitative determination. Anal. Chem. 40, 586-593.

Date A. R. 1977. Preparation of trace element reference

materials by a CO-precipitatedgel technique. Institute of

Geologica l Sciences, Geochemical Division, Analytical et

Ceram ics Unit , Repor t No 101. 16 p.

Gladney E. S. Burns C.

E.

1984. 1982 Compilation of

elemental concentrat ion data for the United States Geo-

logical Su rvey 's geochemical explorat ion reference sam-

ples GXR-1 to GXR-6. Geostandards Newsletter Vol. 8

NO. 2, 119-154.

Intercomparison of trace element measurements in mari-

ne sediment sample SD N 112. International Atomic

Energy Agency, R epor t No 24, 1 8 p.

Meier

A.

L.

1980. Flameless atomic absorption determina-

tion of gold in geological materials.

J .

Geochem. Expl.

13 , 77-85.

Motooka

J. M.,

Mosier E. L. Sutley S.

J .

Viets J.

G .

1979. Induction-coupled plasm a determination of Ag, Au,

Bi, Cd, Cu, Pb and Zn in geologic al mater ials using a se-

lectiv e extraction technique reliminary investigation.

Appl. Spectr. 33, 456-460.

Rosenberg R.

J.

Kaistila M. Zilliacus R. 1982. Inst-

rumental epi thermal neutron act ivat ion analysis of sol id

geochemical samples. J . Radioanal . Chemistry 71,

419-428.

Rosenberg

R.

J .

Vanska L.

1985.

STOAV84, a computer

program for an automatic gamm a spec trometer used for

act ivat ion analysis . Technical Research Centre of Fin-

land, Research repor ts No 415, 49 p.

Vanska L. Rosenberg R. J. Pitkanen V. 1983. An auto-

matic gamma spectrometer for act ivat ion analysis . Nu-

clear Instruments and Methods 213, 343-347.

Zilliacus R. 1983. Radiochemical neutron act ivat ion analy-

sis of gold. Radiochemical and Radioan alyt ical Letters

57, 137-144.



Analytica l methods for determining gold in geological samples

Edited

by

Esko Kontas

Geologian tutkimuskeskus. Tutkimusraportti

Geological Survey of Fidand Report of I~westigation

14 39-41 199 3

GOLD CONCENTRATIONS OF SOME REFERENCE

SAMPLES ISCUSSION

b

E Kontas

Kontas

E 1993.

Gold concen trations of some reference samp les is-

cuss ion. Geolog ian tutkimuskeskus, Tutkimusra portti eological Survey

of Finland Repo rt of bzves t igation

114, 39-41, 3 table s.

The gold concentrat ions of f ive reference samples, AA V-1, AAV-2,

AAV-3, SUK -1 and SUK-2 were determined in s ix laborator ies using the

metho ds descr ibed in this volume. The sam ple mater ial was the f ine

fractio n of till -0.064 mm ); all samp les contained metallic gold grains.

The sa mple w eights ranged f rom 0.6 to 50 g, depending on the procedure.

Several determinations were made in each laboratory. When the highest

and lowest outl ier values for each sample were omitted, the fol lowing

rang es of gold concen trations were obtained: AAV -1, 24-120 ppb; AAV-

2, 198-1100 ppb ; AAV-3, 127-1000 ppb; SUK -1, 19-83 ppb ; and

SUK -2, 28-95 ppb. The var iat ions were greatest for the samples with the

lowest weights. However, anomalously high gold values were found in

every determination at al l laborator ies .

From the capacit ies , detect ion l imits and representat iveness of the

analyt ical samples, the best applications of the methods are as fol lows: f i re

assay with samp le weights of 25-50 g for the assessm ent of gold dep os-

i ts , analysis of concentrates and quali ty control when cheap methods are

cal led for ; methods based on aqua regia digest ion and solvent extract ion

with sam ple weights of 5-20 g for the analysis of ores and local geoch-

emical prospecting; the method based on aqua regia digest ion and sepa-

rat ion by coprecipi tat ion w ith a l -g sample weight for regional geochemi-

cal mapping, prel iminary prospecting and som e basic research; instrumen-

tal neutron activat ion analysis NAA) with a sample weight of about 0 .6 g

for regional geochemical prospecting and some specia l analyses.

Key words GeoRef Thesaurus, AGI) : chemical analysis , gold, t il l , meth-

ods, reliability

E. Kontns . Geological Survey of Finland

P . 0

B o x

SF-96101 Rovaniemi . Finland

INTRODU TION

To help readers and clients compare the dif- Rovaniemi), and subsequently submitted to

feren t analytical metho ds described in this vol- analysis a t the six participa ting laboratories.

ume , five reference sam ples were prepared at the

Since laboratories employ several methods for

laboratory of the Geolog ical Survey of Finland analysing gold, the methods w ere divided be-



Geologian tutkimuskeskus Tutkimusraportti eological Survey of Finland Report of bnlestigation 114 1993

E. Kontas

tween the laboratories to cover as many as pos-

sible of the procedures required by this study.

The samples were known to contain varying

am oun ts of gold , at least partly in native grains.

The analysed sample material consisted of the

fin e fraction of till (-0.064 mm ) commo nly

used in prospecting for gold and base metals in

all Fenn oscandian countries, Canada and parts of

th e United Sta tes .

Gold typically occurs in till as native grains

varying greatly in size. The maximum size is

defined by the mesh number of a sieve, but not

absolutely, because the oblong or threadlike

shape of the grains enables even larger grains to

pass the sieve. The representativeness of the

analytical portion is a problem, as it is depen-

dent not only on the grain size but also on the

grade of gold in a sample (Clifton et al. 1969).

In addition, it is very difficult to homogenize

such samples, because the gold grains separate

gravitatively in the sample powder. The fine

fraction of till consists mainly of minerals of

very low density; gold, however, has th e highest

density of all minerals.

S MPLE PREP R TION

The samp les, each weighing abou t 40 kg, ses. Two 250 g aliquots of the fine fraction s

were collected from Outokumpu Oy prospects in were sent to each laboratory for analyses. Th e

Kittila district, Finnish Lapland. The material is homogeneity of aliquots was studied by

till, mu ch of it of local origin. analysing them for base metals (Cu, M n, CO

Samples AAV-1, AAV-2, AAV -3, SU K-l and etc.) . These elements turned out to be very

SU K -2 were dried at 70°C, after w hich the homogeneously distributed in the different bat-

-0.064 fraction was sieved for chem ical analy-

ches.

RESULTS ND DISCUSSION

The precisions are clearly poorest with the

methods which used the smallest sample

weights. In som e the variations are even signifi-

cant with greater sample weights. The real

gold contents of all aliquots were probably not

exactly the same.

Th e results suggest (Table 1) that the analyti-

cal portions of the samples studied in this work

should be about 10 g or more, if relative stan-

dard deviations of f 5 0 % or below are aimed at.

The methods of Kontas used both l-g and 9-g

portions of samples. The results (Table 1) show

that precision improved dramatically when lar-

ger an alytical sam ples were used . The results of

the two fire assay procedures, with 30-g and

50-g samples, showed almost ideal agreement.

Generally, when subsampling is deficient, ana-

lytical results tend to be too low rather than too

high (Ingamells and Switzer 1973, Ingamells

1981) .

Very important characteristics of analytical

methods for gold are detection limit, capacity,

the representativeness of the analytical portion

and the costs per analysis. Th e costs are difficult

to estimate, since the prices of instruments and

other equipm ent vary markedly. Generally, large

capacity and little need for attendance mean low

costs per analysis.

The general trend is that the higher the capac-

ity, the lower the weight (Table 2). At the same

time, the representativeness of the analytical

portion declines, which is a serious drawback of

these effective methods. But different methods

are applicable for different purposes. W hen g old

ores or gold bearing concentrates are analysed,

representativeness, precision and accuracy are

the most important criteria. A detection limit of

0.05-0.1 ppm is usually low enough and a

lower capacity is acceptable. A very high c apac-

ity and a low detection limit are particularly

important in geochemical exploration. T he fea-

sibilities of different procedures are evaluated i n

Table 3.

In general, a low detection lim it is necessary

in basic research because the background values

of gold in geological materials are usually very

low. A small analytical portion is not always a

disadvantage, because the gold at background

levels is often hom ogeneuously distributed.





114, 1993

Gold concentrations of some reference samples iscussion

Table 1. Gold concentrations (Au ppb) in reference samples AAV-I. AAV-2, AAV-3, SUK-1 and SUK-2 determined by different

labora tories. (arithmetic mean standard deviation).

Procedure 1 2 2A

Sample weight g 0.6 1 9

Numb. of det s 8 6 3

AAV- 1 49518

4 8528 3956

AAV-2 398rt262

7455435 699577

AAV-3

4595362 6035522

636567

SUK-1 2757

44rt22

3

059

SUK-2 33rt10

3 858 4551l

Procedures: 1 Rosenberg et al., 2 and 2A Kontas,

3

Ojaniemi, 3A Ojaniemi, lead fire assay, 4 Penttinen, 5 Noras, 6 Juvonen and

Vaananen.

Table 2. Some important characteristics of the analytical met-

hods for gold described above.

Procedure Size of Capacity Sample Det.

of team samples1 weight limit

week

g P P ~

Juvonen and Vaananen

4

150

25-50 0.1

Noras 2 200 20 0.05

Penttinen

3 3 00 10

0.05

Ojaniemi 2 400 5

0.02

Kontas 2 5 00

1 0.0001

Rosenberg et al.

2 570 0.5-1

0.003

Table 3. Evaluation of the best applications of different analyt-

ical methods for gold.

Method Applications

Juvonen and Vaananen Assessment of gold deposit s and

analysis of concentrates, quality

control

Noras Analysis of ores, local geochemical

prospecting

Penttinen Analysis of ores, local geochemical

prospecting

Ojaniemi Geochemical prospect ing, analyses

of gold and platinum ores

Kontas Regional geochemi cal mapping and

preliminary prospecting, basic re-

search

Rosenberg et al. Regional geochemical prospecting

and mapping, special analyses

REFERENCES

Clifton E H . Hunter R . E. Swanson F. J . Phillips R. 547-568.

L 1969 Sample size and meaningful gold analysis. U S.

Ingamells C. 0 1981 Evaluation of skewed exploration

Geol.

Surv.

Prof. Paper 625-C 17 p.

data he nugget effect. Geochim. Cosmochim. Acta

Ingamells C. 0 Switzer P. 1973 A proposed sampling

45

1209-1216.

constant for use in geochemical analysis. Talanta 20



APP END IX. T he effect of sample weight and digestion and separation method on the results of gold

determinations.

Effects of sample weights and digestion and separation methods on the detection of gold were

studied by analysing

137

samples collected from

53

gold deposits and their host rocks Nurm i e t al.,

1991 ). W ith som e exceptions, the samples were taken from drill core profiles: one sam ple from the

orebody itself and one sample from either side of it. Gold contents were determined from samples

weighing

1

and 20 g, using aqua regia digestion and stannous chloride-mercury copr ecipitatio n as

a separ ation method Kontas, this volume). Sample weights of

20

and 25 g were used in th e lead fire

assay method XR AL, Canada, Juvonen and Vaananen, this volume).

The results show that gold ores and their closest haloes in the bedrock can g enera lly be detecte d

with all the sam ple weights used. Therefore a quick and effective method with a sa mple w eight of

1

g would se em to be very useful in reconnaissance prospecting for gold. How ever, eve n the sample

weight of 20 g does not always seem to be reliable for the inventories of gold deposits.

Table 1 Gold concentrations in ppb) in 5 3 gold deposits and their closest haloes in bedrock. Aqua regia digestion method with sam-

ple weight s of 1 and 20 g Kontas , this volume). Lead fire assay with sample weight of 20 g XRAL, Canada) and 25 g when the gold

concentration exceeds 1000 0 ppb Juvonen and Vaananen, this volume).

Method: Aqua regia, GFAAS Aqua regia, GFAAS

Lead fire assay

Sample: g 20 g 20 or 25 g

Gold deposit

Mont Charlotta

New Celebration

MA-la, CANMET

Hoyle Pond

Owl Creek

Kerr Addison

Renabie

Ferderber

Page Williams

Lokkiluoto

_ _

Muurinsuo

- l-

-

Korvilansuo

Ramepuro

-

Kuittila

_ _

- -

Lalleanvuoma

- l-

Sukseton

-

-

n

d. = not determined

#

=

Reference sample, CCRM, M A-l a, recomended value for gold 21400+400 ppb.



T a b l e

1

(continued.) Gold concentrations (in ppb) in 53 gold deposits and their closest haloes in bedrock. Aqua regia digestion

me thod with sample w eights of 1 and 20 g (Kontas, this volume). Lead fire assay with sample weight of 20 g (XRAL , Can ada) and

2 5 g when the gold concentra t ion exceeds 10000 ppb (Juvonen and Vaananen, th is volume).

Meth od: Aqua regia , GFAAS

Aqua regia , GFAAS Lead f i re assay

Sa mple :

1

g

2 0 g

20 o r 25 g

Gold de pos it

K iv ima a

_ l_

_ _

Suur ikuus ikko

Hirvi lavanmaa

_ _

_ I_

Rovase lka

- -

- l -

Sa a t topo ra

_ _

_ _

Sa a t topo ra S

-

-

-

-

I s okuo tko

-

-

- -

Sore t iavuoma

- -

-

-

Si rkka W

_ I_

Bidjovagge

Juomasuo

-

Saynajavaara

-

Sivakkaharju

-

Kontt iaho

_

Makararova

_ _

Laivakangas

- -

- -

Jokis ivu

-

-

-

-

Isoves i

_

- -

Antinoja

-

-

Vesipera

_ _

~ n ~ e s u e v a

_ _

- -

n. d.

=

not determined

# =

Reference sample , CCRM , MA -la , recomended va lue for gold 21400 ~400 ppb.



Table 1. continued.) Gold concentrations in ppb) in 53 gold deposits and their closest haloes in bedrock. Aqua regia digestion

method with sample weights of 1 and 20 g Kontas, this volume). Lead fire assay with sam ple weight of 20 g XRAL, Canada) and

25 g when the gold concentration exceeds 10000 ppb Juvonen and Vaananen, this volume).

Method: Aqua regia, GFAAS Aqua regia, GFAAS Lead fire assay

Samp le : 1 g 20 g 20 o r 25 g

Gold deposit

Kiimala 100 160 170

_ l 1 _

1600 2650 2600

30 60 62

Pir i la S

40 100 60

_r l_

12000

32000 9500

_ t p _ 10

60 56

Kalliosalo

210 280 3 40

_I(_

5000 6700 7500

l1

30 120 90

Kurula 40

80 72

- 1000

1280 1100

100 300 240

Pir i la N

100

240 260

- 1700

3100 2800

-

3

0

40 64

Kaapeliukulma 100

210 240

- I I - 1700

3250 3200

-

70

260 140

Laivakangas

S 200 450 3 60

- 3900

3500 5500

10 120 100

Pohlola

840

410 320

_l _

8800

12800 12500

lt 40

60 49

Ko p sa 890

410 310

2400

3 800 3500

Kangaskyla

290 340 280

2100 2300 2000

-

250 520 340

Osikonmaki E

350 540 550

q v

5800 7000 7000

2060 1280 l20 0

Osikonmaki W

350 620 680

3400 5400 5000

100 270 430

Bjorkdahl 1000 2850 3600

n.

d.

= not determined

= Reference sample, CCRM , M A-l a, recomended value for gold 21400rt400 ppb.

REFERENCE

Nurmi P

A .

Lestinen P Niskavaara H. 1991 Geo -

C an ad ian an d Au s t r a l ian d ep o s i t s . Geo l . Su rv . F in lan d ,

ch em ica l ch arac te r i s t ics o f meso th ermal g o ld d ep o s i t s in B u l le t in

351 101

p .

th e Fen n o sca n d ian sh ie ld , an d a co mp ar i so n w i th se lec ted

analytical methods for determining gold

Documents