GOLD ANALYSIS - ARE THERE ALTERNATIVE METHODS

OTHER THAN STANDARD FIRE ASSAY FOR THE ANALYSIS?

Eric L. Hoffman,
Activation Laboratories Ltd., 1336 Sandhill Drive, Ancaster, Ontario, Canada L9G 4V5

John R. Clark and James R. Yeager

ACTLABS Inc., 11485 W. I-70 Frontage Rd. N., Wheat Ridge, Colorado, USA 80033

Peter J. Rogers
ACTLABS Chile, Augusto Leguia Sur 98, Dept. 504, Las Condes, Santiago, Chile.

ABSTRACT

The method of choice for gold analysis through time has been fire assay. The first recorded use of the fire assay
technique was that of Erker - Assayer to the Holy Roman Empire. In fact, fire assaying probably even predated that
time. The method used has not changed appreciably with time except perhaps with better quality reagents and
furnaces and now there are a number of potential modern analytical finishes to “read” the fire assay bead other than
the classical weighing of the bead. Time and technological advances, however, have not stood still, and there are
some very viable modern technological alternatives available.

The nature of gold deposits being sought (low grade-high tonnage) and the increasing sophistication of exploration
geochemistry requires better quality data particularly for the lower range of gold values. Modern geological models
have many of these gold deposits being formed by hydrothermal solutions. These solutions have added elements
other than gold or leached out some elements from the host rock so that there is usually a multi-element halo around
the deposit. In many cases this halo is larger or more coherent than gold itself. Analyzing gold by itself rarely will
provide useful details regarding geochemical halos, environmental baseline levels for potentially problem elements
and data on potential metallurgical complications. Multi-element data can provide this information. This paper will
discuss higher technology options for gold analysis and the advantages and disadvantages of each of these methods.
These include INAA (Instrumental Neutron Activation), Aqua Regia Digestion/AA (Atomic Absorption), GFAA
(Graphite Furnace-Atomic Absorption) or ICP/MS (Inductively Coupled Plasma Emission Mass Spectrometry) and
Cyanidation-AA, GFAA or ICP/MS methods.

RÉSUMÉ
CONVENTIONAL FIRE ASSAY PROCEDURE

The basic procedure for fire assay involves mixing an aliquot of powdered sample (10 grams, 15 grams, 30 grams, or
50 grams are the common sizes used) with soda ash (sodium carbonate), borax (sodium borate), litharge (Pb0), flour
(yes - baking flour used to add carbon as a reductant), silica and possible nitre (potassium nitrate). To this mixture
Ag or Pd as a collector can be added in solution or as a foil. The well mixed material is fired at temperatures ranging
from 1000°C to 1200°C (depending on the lab). As the Pb and Ag (or Pd) in the melt settle to the bottom of the
crucible, it scavenges the Au (±Pd, Pt) from the melt.

The hot molten mixture is poured into a mold and when cool, the slag should, but not always does (the sample must
be fluxed properly) separate cleanly from the Pb button which should be a certain weight (again if fluxed properly).
This lead button is placed into pretreated cupels (made from magnesia or bone ash) and the Pb will be absorbed into
the cupel leaving a Ag or Pd bead (depending on collector used) which has collected the gold and some of the
platinum group elements.

The gold is separated from the Ag bead by parting (dissolving in nitric acid) leaving a bead which can be weighed
(gravimetric finish). Alternatively, the entire bead can be dissolved in acid and Au (± Pt, Pd) can be determined by
AA (most commonly used), ICP (inductively coupled plasma emission spectrometry), ICP/MS (for best sensitivity
for Au, Pt, and Pd), or DCP (Direct current plasma emission spectrometry). The bead can also be read directly by
irradiation followed by INAA (for lowest detection limits). A variation of fire assay called metallic screen analysis
is used for samples where coarse gold may be a problem for obtaining representative assays. This involves crushing,
pulverizing and screening the entire sample. The gold is assayed on the +80 mesh (or +150 mesh) material and the -
80 mesh (or -150 mesh) material. Based on a weighted average of the two size fraction components, a final assay is
determined. Details on various fire assay procedures are detailed by Beamish and Van Loon 1977.

FIRE ASSAY ADVANTAGES AND DISADVANTAGES

During the fire assay procedure most steps occur in unmarked crucibles and cupels and the procedure is very labour
intensive. This can easily result in sample mix-ups. Samples can however, be monitored by insertion of blanks,
standards and identified samples so that certain types of mix-ups can be caught (i.e. turning around trays).

Certain matrices, for example, sulphide bearing rocks or rocks with higher levels of base metals must be pretreated
(roasted, leached or fluxed differently) and rocks of differing composition must be properly fluxed to achieve good
recovery of precious metals. In a study by Hall et al. 1989 where they submitted 5 reference standards under the
guise of samples to 15 reputable, widely used commercial labs yielded some incredible variation in gold content. The
worst results by fire assay were obtained from two porphyry copper standards with about 0.5% Cu. Details are
shown in FIGURE 1. Other authors Diamantatos, 1984, 1987 have also commented on the deleterious effect of Cu
on Au collection by fire assay.

TABLE 1 shows results on UMT-1, a CANMET mine tailings standard from an ultramafic host rock which were
submitted for homogeneity testing for Au by fire assay. Thirty two, 30 gram aliquots were fire assayed. First results
ranged from 4 to 25 with a mean of 14. When the fire assay lab was informed these were standards and should be
homogenous they refired 20 aliquots and obtained a higher recovery (range of 26 to 45 ppb-average 38). INAA
analysis returned values of 52 ±4 ppb. The certified value for gold is 48 ±2 ppb. In this example improper fluxing
on both fire assay sets resulted in unacceptably low and erratic results.
 Figure 1. Comparison of analytical techniques
for gold (from Hall et al.,1989)
350

AR-AA
300
FIRE ASSAY
INAA
250

200
Au ppb

150

100

50

PORPHYRY CU C
PORPYRYCU B
SOIL A

SOIL B

SOILC

TABLE 1

Fire Assay Results From 2 Fire Assay-AA Runs of the CANMET UMT-1 Versus INAA Gold Values.
Example of Improper Fluxing Not in Homogeneity of Sample.

First Run Second Run

Fire Assay-AA Fire Assay - AA INAA
30 Gram Aliquot 30 Gram Aliquot 30 Gram Aliquot

14 9 38 40 49
11 13 26 39 49
4 10 37 42 48
10 11 39 34 51
16 6 40 42 49
21 11 45 40 47
13 25 30 41 55
11 19 23 43 53
12 21 35 36 53
8 22 41 37 54
12 15 43 48
16 25 53
12 18 61
11 14 57
15 12 48
14 24 53
AVG = 14 +/- 5 AVG = 38 +/- 6 AVG = 52 +/- 4

Average Fire Assay 14 PPB (32 Aliquots) Range - 4 to 25

Average Fire Assay 38 PPB (21 Aliquots) Range - 26 to 45
Average INAA 52 PPB (16 aliquots) Range - 48 to 61
The fire assay has a method blank which is usually 1-3 ppb for a 30 gram sample but can routinely be much worse
depending on the source of litharge (PbO). Some labs’ blanks are as bad as 50-100 ppb. Erratic gold values can
also occur, probably from contamination of crucibles from material which has accumulated on the roof of the furnace
from “spitting” of samples as they are being fired. This material can then drip back down into latter crucibles fired
through the same furnace. We have seen occasional erratic blanks of 1000-2000 ppb. Homogeneity of the sample is
not always the answer when trying to explain why gold values don’t reproduce (contamination may be a good
possibility). Many of the Geological Surveys in Canada both federal and provincial have switched from fire assay
based methods for regional geochemical surveys to the INAA method several years ago. This was done to minimize
the problem of unreproducable low level anomalies with fire assay and to take advantage of the multi-element INAA
capabilities.

For analysis of gold, the gold must be put into solution (for atomic absorption analysis) or separated from the Ag
(gravimetric analysis). This is called parting. Occasionally when parting the Ag doré bead, the gold particles may
creep up the side of the test tube or parting dish as a result of surface tension and may not even report to the assay
(i.e. your assay says you had no gold but you really did). We have seen this happen with samples containing as much
as 2000ppb although the smaller the Au particle the more often this effect can happen.

The true detection limits for all the fire assay methods reported by commercial labs are highly optimistic and should
realistically be computed by the end user by the formula: detection limits = blank + 3 times the standard deviation of
the blank (Standards Council of Canada, 1997).

Fire assay is generally regarded as the most widely used analytical method for gold, however as in any analytical
method, fire assay is no guarantee of correct results. There are a number of “bogus” laboratories out there that have
special proprietary fire assay methods where they contaminate the samples with reagents used, don’t measure method
blanks or use some inappropriate analytical finishes to produce incorrect results.

One of the main advantages of fire assay is the potentially rapid turnaround time and moderate cost. You, however,
as the consumer should be aware that the fire assay methodologies vary from one lab to another. Some reuse
crucibles (potential contamination), some have lower flux to sample ratios or poorer quality reagents (saves a few
pennies in reagent costs), some decrease fusion time (higher throughout but perhaps not total gold recoveries) some
don’t custom flux based on sample composition (poor recovery). Some don’t check high gold values (slows
turnaround and add costs). Do your own due diligence and ask questions of your lab. There are a lot of variables
involved and fire assay does not guarantee correct results.

INAA (INSTRUMENTAL NEUTRON ACTIVATION ANALYSIS) ADVANTAGES AND

DISADVANTAGES

The INAA procedure is inherently simple. It involves encapsulating the sample powder in a polyethylene capsule,
irradiating in a nuclear reactor and after a decay period (usually 7 days) measuring of the sample for Au + 34 other
elements simultaneously. The method has been described by Hoffman, 1992 in some detail. The suite of elements
obtainable is listed in TABLE 2.
 TABLE 2

Elements and Detection Limits for “Au + 34” by INAA

Au 2 ppb Hf 1 ppm Se 3 ppm

Ag 5 ppm Hg 1 ppm Sm 0.1 ppm
As 0.5 ppm Ir 5 ppb Sn 0.01%
Ba 50 ppm La 0.5 ppm Sr 0.05%
Br 0.5 ppm Lu 0.05 ppm Ta 0.5 ppm
Ca 1% Mo 1 ppm Th 0.2 ppm
Ce 3 ppm Na 0.01 % Tb 0.5 ppm
Co 1 ppm Nd 5 ppm U 0.5 ppm
Cr 5 ppm Ni 20 ppm W 1 ppm
Cs 1 ppm Rb 15 ppm Yb 0.2 ppm
Eu 0.2 ppm Sb 0.1 ppm Zn 50 ppm
Fe 0.01 % Sc 0.1 ppm

The advantages of the technique are many. There is no chemical treatment required (simplicity = reliability). There
is no blank to subtract which results in very reliable low level data. The method is non-destructive which is ideal if
one wishes to reexamine samples which have provided surprises (not expecting gold but found it). The true
detection limit is 2 ppb on rocks, soils or sediments and 0.1 ppb on vegetation.

The method is very cost effective, particularly if other elements like As (0.5 ppm), Sb, (0.1 ppm), Ba (50 ppm) or W
(1 ppm) are required. The cost will typically be below that of a fire assay for Au and say As by wet chemistry. The
package also provides excellent low level Sb analysis (far superior to ICP Sb) and provides a total Ba value (Barite
is not soluble in any ICP methodology other than fusion). Barite has increasingly been recognized as an indicator of
mineralization formed by hydrothermal fluids.

Samples can easily be remeasured to confirm analyses rapidly. The gold value provided will be total gold regardless
of the form it may be in (organic complex or silica encapsulated). The method is matrix independent and suffers
from no interferences. Samples analysed by INAA can range from a fraction of a gram to a kilogram with no loss of
sensitivity.

The disadvantages of the technique include waiting 7 days for decay of sodium prior to the analysis. Occasionally
very high levels of certain elements like Sb (>500 ppm) rare earths (>5000 ppm), uranium (>500 ppm) may cause an
elevation in background and hence an increase in detection limits. In no case will this preclude the actual
measurement of gold but may elevate the level which can be detected say from 2 ppb to 10 ppb. Certain INAA
laboratories use “epithermal” neutrons for irradiation. This may cause a low bias depending on Au particle size due
to an effect called “self-shielding”. In this case the gold particle is such a good absorber of epithermal neutrons that
the neutrons will only activate the outer layers of coarse gold particles. This effect can, however, be avoided by
using thermal neutrons. The final disadvantage is that only certain design of nuclear reactor can be used for
commercial INAA and these reactors have very limited availability in the world.

AQUA REGIA DISSOLUTION - AA, GFAA OR ICP/MS FINISH - ADVANTAGES AND

DISADVANTAGES

This procedure involves digestion from 5 to 50 grams (or possibly higher) of sample in aqua regia (a mixture of
hydrochloric and nitric acid). This must be followed by a solvent extraction (MIBK or DIBK) if AA or GFAA will
be used as an analytical finish. The organic extraction is not necessary for an ICP/MS finish. The reason for this
organic extraction is to leave behind Fe which will cause spectral interference on Au or platinum group metals. The
results may show multiple g/t Au or PGE if this extraction is not done.
The advantages of the technique is that it is the lowest cost of the gold analysis techniques and routine aqua regia 30
element ICP package can be added to the gold analysis at low cost. This ICP add on trace element package is
primarily useful for base metals and Ag. Although As, Sb are included in most labs’ ICP packages the data for low
level <30-50 ppm is rarely very good and in fact for Sb is generally useless. In addition Ba coming from barite will
not report at all to an aqua regia digestion.

Depending on mineralogy, results for gold will usually be lower than fire assay or INAA. This may be due to silica
encapsulation or perhaps the gold mineralogy itself. In the study of 5 reference materials by Hall et al., 1989, they
had found on average that the aqua regia - AA values on the soils were only 58-70% of the INAA values and for
porphyry Cu standards 69% of the INAA values. They did note, however, that ore samples could show recoveries of
100% depending on the nature of the samples. Aqua Regia - AA is widely used in Australia for grade control in the
mine environment and in exploration as it offers the lowest cost choice for gold analysis. Aqua Regia methodologies
vary from lab to lab and so will recovery levels.

Various other acid digestions can be used in place of aqua regia depending on the problem to be solved. These
various digestions are described by Chao and Sanzolone, 1992.

CYANIDATION (BLEG) - ADVANTAGES AND DISADVANTAGES

BLEG (Bulk Leachable Extractable Gold) involves weighing a large sample usually 1 - 5 kilos into a polyethylene
bottle, adding an appropriate cyanide solution (0.25 to 1% NaCN depending on the lab) and agitating (bottle rolling)
for various periods of time ranging from a few hours to several days. The gold can then be extracted into an organic
solvent (to avoid Fe interference) and run by AA or GFAA for better detection limits. Some labs do not do an
organic extraction. This may result in elevated gold values particularly for low gold values. Alternatively the
sample can be analyzed prior to the organic extraction by ICP/MS for best detection limit of 0.05ppb. With the
ICP/MS method Pt and Pd can be analyzed concurrently (to 0.05ppb) as can 50 plus other elements.

The advantage of this technique is the use of very large samples. Depending on the sample material, there may be
excess cyanide consumption (sulphides) or loss of gold by adsorption onto carbonaceous material present (preg
robbing). Free cyanide should be monitored but this is rarely done for exploration samples. The BLEG method has
been used extensively in Australia where it was developed and to a much lesser extent in Africa and Nevada for
general exploration. Evaluation of drill core or rock samples by the Cyanidation method seems to get very little use
until the metallurgical testing stage has been reached. This method can be used quite successfully if there is a gold
nugget effect. A large 1-5 kilogram sample can be leached and as long as the residue is also analyzed to see what is
not extracted this method will provide good quality data.

For low level gold results where anomalies of only a few ppb are significant it is important to use an organic solvent
to avoid Fe interference and the graphite furnace - AA finish or ICP/MS finish (organic solvent not necessary). It is
crucial to know what the method blanks are (as run on your samples) to determine the true validity of the data.
BLEG results will usually not be total gold (silica encapsulated or very large gold particles may not completely
dissolve). Reuse of the bottle can easily lead to contamination. Cyanide methodologies are not standard for
laboratories which run cyanide leach, so results may vary widely from lab to lab.

SAMPLE PREPARATION CONSIDERATIONS

Sample preparation has not been mentioned to this point but is critical to producing “correct” results. The following
comments will not address the actual sampling representivity itself but assumes the sample submitted to the lab is
representative of the material intended to be analyzed.

Today the most widely accepted technique for rock preparation is to use a modern jaw crusher on the whole sample
submitted and crush to about 10-20 mesh on the primary crushing. The sample is then split mechanically either in a
riffle or rotary splitter and a subsample is pulverized in a ring and puck (swing) mill. The sample pulverized could
range from several hundred grams up to 5 kilograms in a single milling.
One of the most frequent problems with sample preparation we have seen involves the dust collection system. In an
attempt to reduce airborne contamination, the laboratory uses a too efficient dust collector which sucks out the fines
preferentially. In one investigation of a mine lab we could actually hear the sample rattling down the dust collector.
This was exacerbated by the use of a disc pulverizer (open system) connected to the same dust collection system.
This caused an apparent increase in gold grade of 30% by preferentially sucking out and discarding the fines. Native
gold will also have more of a tendency to smear out on disc pulverizer plates which could lead to contamination of
subsequent samples.

Hammermills have also been used, most extensively in Australia several years ago. Widespread usage in Australia
stopped when it was discovered that these mills acted as dry placers and were concentrating gold within the mill.
This could lead to low values or contamination.

Most modern commercial labs will use the ring and puck pulverizer or a variation of this concept.

Conclusions

Fire assay with its multiple analytical finishes is not a foolproof technique to recover 100% of the gold present. As
with all analytical techniques the skills of the assayer or analyst are very important. As a consumer of assays do not
blindly trust the assays you receive. Laboratories can and do make mistakes. Some of these can be easily discovered
if you Do Your Own Due Diligence. Most importantly there are a number of higher tech analytical methods
available which can be very cost effective and provide better quality data particularly at lower Au levels than fire
assay.

This due diligence should begin in the field by collecting field duplicates or when you receive results from the lab,
going back and resampling to determine the reproducibility of the original sampling. Sample preparation is the most
important step prior to analysis. Select a portion of the rejects and pulps prepared by the lab and have a sieve
analysis done at another lab to see if the fineness of the crushing and grinding are meeting your specifications or the
specification the laboratories claim. Unfortunately you will find many of these specifications are not met. How
large a pulp is being prepared? Is it representative? Submit a well mineralized sample(native gold bearing) followed
by a blank (rock you know to be barren). What is the carryover of gold into the blanks? Insist on silica blank
cleaning between samples at least on pulverization to minimize contamination. Do not let the lab compromise on
this stage of preparation. Randomization of sample submittals and monitoring this data may also allow you to find
preparation or analytical bias. Submit the occasional certified reference materials blindly to the lab. This will
provide some information on accuracy. Ask whether the lab is enrolled in the CANMET proficiency testing program
for gold analysis and if so ask to see results of all their proficiency tests.

We have now reached the analytical stage. Do you want low level (geochemical) or ore grade (assay) assessment.
The TABLE 3 below indicates an approximate range of usefulness of each of the techniques. Make sure to specify
the analytical method and sample size you wish to have the laboratory analyze. Finally provide feedback to the lab.
It can only help to improve the quality of your data.
 TABLE 3

Comparison of Effective Range of Analytical Methods for Au and Comparison of

Multi-Element Capabilities, Turnaround and Cost.

Au ppb Possible*
Multi- Turnaround Cost US$
0.01 0.1 1 10 100 1000 10,000 100,000 1,000,000 element Days 30g sample

FA-GRAVIMETRIC NO 1-3 9-12

FA-AA, DCP, ICP YES (Au, Pd+ Pt) 1-3 8-10 (Pt, Pd add $5)

FA-INAA NO 2-4 10-11

INAA YES ”Au+34 8 10-15 (MULTI INCL.)

BLEG-GFAA NO 2-4 14-40 (1 Kg sample)

BLEG-ICP/MS YES “Au+50” 2-4 35 (MULTI INCLUDE.)

(1 Kg sample)

CYANIDATION-AA NO 2-4 6-8

AQUA REGIA-AA YES “Au+30” 1-3 5-7 (MULTI + $5)

AQUA REGIA-GFAA YES “Au + 30” 1-3 7-10 (MULTI + $5)

* Assumes no other samples in the lab.

References:

Beamish, F.E. and Van Loon, J.C., 1977.

Analysis of noble metals, overview and selected methods. Academic Press, New York.

Chao, T.T. and Sanzolone, R.F., 1992.

Decomposition Techniques. Journal of Geochemical Exploration, V44, p.65-106.

Diamantatos, A. 1984.
The accurate determination of gold and silver in ores and concentrates by wet chemical analysis of the
lead assay button. Anal. Chim Acta, V165, p.263-268.

Diamantatos, A. 1987.
Fire assay collection of gold and silver by copper. Talanta, V34, p.736-738.

Hall, G.E.M., Vaive, J.E., Coope, J.A. and Weiland, E.F., 1989.
Bias in the analysis of geological materials for gold using current methods. Journal of Geochemical
Exploration, V34, p.157-172.

Hoffman, E.L., 1992

Instrumental neutron activation in geoanalysis. Journal of Geochemical Exploration, V44, p.297-320.

Standards Council of Canada. 1997.

Guidelines for the Accreditation of Mineral Analysis Testing Laboratories Document. CAN-P-1579