Professional Documents
Culture Documents
The Importance of Assay Method and Accuracy - A Scandium Case Study
The Importance of Assay Method and Accuracy - A Scandium Case Study
J A Horton1
1.Consulting Geologist, ResEval Pty Ltd, Brisbane, Qld 4066 Email: jhorton@reseval.com.au
ABSTRACT
For resource evaluation, assay accuracy is critical, whereas precision can consider or be offset by
sampling density.
In quality assurance and quality control (QAQC) analysis, the assessment of the selected assaying
method and its accuracy is often overlooked. Umpire samples provide an opportunity to test
alternative assaying method outcomes and accuracy. However, often umpire sampling programs
only demonstrate reproducible results for the same assay method at a different laboratory.
If certified using a different assaying method, then certified reference materials (CRMs) can help to
verify an assay method for accuracy. However, all commercial CRMs raise questions regarding the
suitability of the CRM mineral matrix and the certification assay method.
Work performed by Platina Resources Limited (Platina) on their scandium project in New South
Wales (NSW), provides some insight into these issues with the comparison of Inductively Coupled
Plasma (ICP) and X-ray Fluorescence (XRF) analytical data. It demonstrates typical small assaying
accuracy issues for Ni and Co in nickel laterites, as well as significant inaccuracy when analysing for
Sc in scandium rich laterite deposits.
Platina refocused their QAQC work on accuracy, and this paper provides several valuable learnings:
All tested assay methods by ICP were biased low for Sc, to at least some extent, and are not
considered accurate.
Alternative method assays were hidden by the structure and implementation of the
commercial database and overlooked initially.
Lack of focus on absolute accuracy undermined an otherwise well-executed QAQC process.
A few umpire assays by alternative methods can be more valuable than all the other QAQC
samples.
CRMs are not a measure of accuracy but a measure of relative performance against a group
of laboratories using the same assay method.
Use of internally sourced matrix-matched standards can be invaluable in the QAQC process.
These learnings have relevance to all mineral projects, as it is common for QAQC programs to focus
on precision and performance without fully exploring assaying accuracy.
INTRODUCTION
The introduction of glass fused XRF analytical methods to commercial laboratories has seen oxide
and industrial miners move away from traditional ICP and atomic absorption spectroscopy (AAS)
analytical techniques applied in general exploration and by sulphide miners. It is a result of both
increased accuracy and availability of full whole rock analysis offered by XRF. In particular, XRF
provides reliable silica analysis, which is essential for metallurgical understanding and mineral
processing. These benefits outweigh the slightly higher analytical cost for XRF and narrower trace
element spread than that offered by ICP using a four-acid digest method.
Platina recently completed a Feasibility Study for their Owendale Sc-Co-Ni-Pt laterite project in NSW,
with exploration and resource definition drilling completed between 2011 and 2018. An initial phase
of exploration between 2011 and 2014 used ICP with four-acid digest as the principal assay method.
A well-structured QAQC program and independent analysis raised no significant concerns at that
stage. A subsequent second phase of exploration between 2016 and 2018, included the change to
borate glass fused XRF analysis. The re-assaying of some previous samples and an internal CRMs
Umpire pulp samples submitted during both periods of Phase 1 drilling in Figure 2 also use similar
ICP methods. The umpire assay results are mostly consistent, though the umpire laboratory in Figure
2a reports slightly lower results for Ni and Co.
CRMs for Sc, Ni and Co that were submitted blind to the laboratory are restricted to one commercial
CRM for Ni and Co and two local CRMs sourced from local material and prepared and certified by a
CRM supplier. The CRMs in Figure 3 indicate some spread which is misleading on a scatter plot as
some sets contain a considerable number of data pairs; however, the plots do demonstrate no
substantial accuracy issues. Though the laboratory also inserted their own CRMs and other
standards that provide more information, there were no commercial Sc CRMs available at the time
and no additional Sc data.
An independent QAQC specialist assessed the presented data along with the PGM assay data and
some ICP method variations and no issues were raised at the time, except for some individual
batches that required further review.
The QAQC and analysis would appear to have been done very well with the usual samples taken
and independent analysis of the results by a suitable professional. Early exploration geologists can
be commended as they identified the lack of a commercial CRMs for Sc and specifically drilled and
sampled material for in-house reference materials, with the preparation and certification of that
matrix-matched material undertaken by a well-known and respected CRM manufacturer and
supplier.
Laboratory A and B ICP analysis – CRMs assays vs. expected values of the day
(scatter plots contain large datasets and presented to in this manner for comparison and to
demonstrate lack of evident bias, these are unsuitable for assessing assaying precision)
Figure 3 - Phase 1 blind CRMs
HIDDEN DATA
The exploration data was stored and maintained in a well know commercial database system that
was initially set up and supported by a specialist database service firm. Day to day data uploads and
exports were completed by company staff. The system stored both primary data and specific tables
The field duplicates for the XRF analyses, Figure 5, can be compared with the field duplicates for
ICP in Figure 1. The mean difference measure is approximately 80 ppm for Ni and Co and is
statistically similar to ICP; however, the early XRF Ni and Co grade are only reported to 100 ppm.
The results suggest a higher intrinsic precision that was rounded when reported by the laboratory.
The oversight in assessing the XRF data in the original QAQC program raises an issue that can
happen on sites where multiple assay methods are used for a primary sample. The databases are
typically constructed to extract the “best” assayed grade for each element using a complicated
database query that accepts the assays by different methods. It is often established during the
construction of the initial database, and the extraction query must be updated each time a new
For the low-grade CRM, all the laboratories are mostly within a 5 ppm tolerance of the expected
value of 75 ppm Sc. The possible accuracy of the XRF method is lost in the rounding of the results
by the laboratories to 10 ppm. Overall, at this low-grade range, ICP appears to perform adequately
for the low iron saprolitic clay matrix.
For the high-grade limonite CRM, the mineral matrix is very different and dominated by Fe and Al
clays and a substantially higher Sc grade. The original certified grade of 555 ppm Sc was reviewed
and adjusted to 595 ppm Sc; it was based on the consistent and tightly arranged results for NAA and
XRF.
Phase 1 drilling results by laboratory A, underperformed against the original CRM expected value
and raised no concerns at the time in the independent specialist review. The even lower grade
performance of laboratory B, late in Phase 1 was only recognised well afterward and was part of the
review of the preferred assaying method at the start of Phase 2. The subsequent evaluation of the
certification results raised concerns over the reliability of ICP with both the full spread of the
certification assays and that a disproportionate number of “outlier” high-grades were excluded from
the original certification. A more critical review of the certification assay results for the internal CRM
and addition of an alternative assay method such as NAA, that was available at the time, may have
uncovered the assaying issues with standard ICP methods at the start of the exploration program.
Instead, considerable investigation and re-assaying have been required to understand and re-
evaluate the deposit eight years later.
The seven percent under-call of the CRM Sc grade is significant, and the umpire samples indicate
this bias is higher at other grade ranges. Overall the results indicate four acid ICP analysis at Lab A
was biased ten percent low, and Lab B bias 21 percent low. These accuracy issues had and still
have a significant impact on the Mineral Resource grade and tonnage.
CONCLUSION
In this case study, the need to critically assess the laboratory analysis method is presented. It
includes the evaluation of the certification data for the CRMs used to determine assaying accuracy.
The study indicates that ICP methods suffer from varying degrees of bias for Sc. This could not be
adequately identified and assessed without using alternative assaying methods such as XRF and
NAA.
A laboratory manager or chemist will be focussed on laboratory precision and performance of their
laboratory against its’ peers for the same assaying method. Though performance monitoring is
required to catch and address batch by batch issues, ultimately a geologist should be more
concerned with absolute accuracy than routine performance.
ACKNOWLEDGMENTS
The Author thanks Platina for allowing publication of the case study and the scandium test work as
well to Robert Mosig for the early support to investigate and address the assaying issues identified
early in Phase 2. The author also acknowledges the input to the test work and paper by Gideon
Steyl, Corey Nolan and Boyd Willis from the Platina project team and Andrew Richmond for his
comments. Finally, the author thanks Shaun Kenny from ALS Brisbane for his support in assessing
the analytical issues with Sc for initially trialing scandium analysis by XRF.
REFERENCES
Jervois Mining Limited, 2014a. EL 7805 Syerston Scandium Project, Re-Assayed and Amended Results from May 2014
Drilling Program. Available at <www.asx.com.au> (ASX:JRV announcement dated 24 July 2014) [Accessed:
30 May 2019].
Jervois Mining Limited, 2014b. Re-Assayed and Amended Results from August 2013 and May 2014 Drilling Program.
Available at <www.asx.com.au> (ASX:JRV announcement dated 30 June 2014) [Accessed: 30 May 2019].
Platina Resources Limited, 2017. Cobalt and Scandium Mineral Resource increases at Owendale. Available at
<www.asx.com.au> (ASX:PGM announcement dated 14 Aug 2017) [Accessed: 30 May 2019].
Scandium International, 2016. Feasibility Study - Nyngan Scandium Project Bogan Shire, NSW, Australia, NI 43-101
Technical Report dated 15 April 2016. Available at <www.sedar.com> [Accessed: 30 May 2019].
Shaw, W J, Khosrowshahi, S, Horton, J and Waltho, A, 1998. Predicting and monitoring variability in sampling, sample
preparation and assaying, in More Meaningful Sampling in the Mining Industry, Bulletin No 22, pp 11-19 (Australian
Institute of Geoscientists:Perth)