CHAPTER 6: RESULTS AND DISCUSSION 6.1 SAMPLING PROCEDURE: UG2 TAILINGS SAMPLE AND RCCs. 6.1.1 The selection and splitting process for the preparation of the UG2 HG FT composite sample (UG2 Tailings sample) As indicated in section 3.4 the majority of chrome present in the UG2 Reef exploited by the platinum producers deports to the tailings stream from the Concentration process. It was logical therefore to target the tailings stream as the primary source for chromitite crystals. A representative bulk sample or composite sample was prepared from monthly composite UG2 Tailing samples for the years of 2006 and 2007. Composite sampling is a technique used to create a representative sample by the homogenization of multiple representative samples. Laboratories custom design composite sampling procedures to ensure that the resultant sample complies with specific objectives and statistical assumptions. Composite samples are generally prepared for the following reasons: • They are more representative of mean concentration than would be achieved by the same number of individual samples. • They may reduce sampling cost. • They may be prepared and stored in order to provide future cross referencing for that particular sample type and grade. • They may also be stored and kept for future validation of techniques where reference samples are not available. • To minimize storage requirements that would otherwise be necessary for numerous individual samples. In total, 14 samples were selected in order to produce a weighted bulk composite sample size of 30 kg which would be sufficient for the various analytical procedures such as Pb-FA and NiS-FA. Sufficient sample was also needed for the 64
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
CHAPTER 6: RESULTS AND DISCUSSION
6.1 SAMPLING PROCEDURE: UG2 TAILINGS SAMPLE AND RCCs.
6.1.1 The selection and splitting process for the preparation of the UG2 HG
FT composite sample (UG2 Tailings sample)
As indicated in section 3.4 the majority of chrome present in the UG2 Reef
exploited by the platinum producers deports to the tailings stream from the
Concentration process. It was logical therefore to target the tailings stream as the
primary source for chromitite crystals.
A representative bulk sample or composite sample was prepared from monthly
composite UG2 Tailing samples for the years of 2006 and 2007. Composite
sampling is a technique used to create a representative sample by the
homogenization of multiple representative samples. Laboratories custom design
composite sampling procedures to ensure that the resultant sample complies with
specific objectives and statistical assumptions. Composite samples are generally
prepared for the following reasons:
• They are more representative of mean concentration than would be
achieved by the same number of individual samples.
• They may reduce sampling cost.
• They may be prepared and stored in order to provide future cross
referencing for that particular sample type and grade.
• They may also be stored and kept for future validation of techniques where
reference samples are not available.
• To minimize storage requirements that would otherwise be necessary for
numerous individual samples.
In total, 14 samples were selected in order to produce a weighted bulk composite
sample size of 30 kg which would be sufficient for the various analytical
procedures such as Pb-FA and NiS-FA. Sufficient sample was also needed for the
64
extraction of chromitite crystals for the mineralogy study. Samples and masses
used are listed in Table 6.1.
Table 6.1 UG2 Tailings samples selected.
SAMPLES: MC UG2 HG FT
WEIGHT (Kg)
January 2006 2
April 2006 2
May 2006 3
June 2006 2
July 2006 2
August 2006 2
September 2006 2
June 2007 2
July 2007 2
August 2007 2
September 2007 2
October 2007 2
November 2007 2
December 2007 3
Total weight 30
Each monthly composite sample consisted of 10 kg sample portions, which were
first tumbled for 30 min to address any segregation of the sample that may have
occurred during the storage period and then the required weight of sample, as per
Table 6.1, was transferred into a clean plastic container for further preparation.
65
The steps followed in preparing homogenised sample splits are as indicated in
Figure 6.1.
The 30 kg composite UG2 Tailings sample was tumbled for 1 hour.
The 30 kg portion was then split into
10 portions of 3 kg each.
Each 3 kg split was tumbled for ½
hour.
Two 3 kg splits were selected randomly and combined to give a 6
kg portion. The 6 kg portion was tumbled for a ½ hour.
Three 3 kg split portions were
selected randomly and combined to give a 9 kg portion. The 9 kg
portion was tumbled for a ½ hour.
Figure 6.1 Summary of the sampling steps.
66
The homogenised sample splits were distributed for analysis to various sections.
One 6 kg portion was retained for extraction of the chromitite crystals. Two 9 kg
portions were submitted to the Assay and IPGM sections of Impala laboratory
respectively for precious metal analysis. The balance of the samples were retained
for further analysis.
The total platinum group metal (TPGM) analysis of the UG2 Tailings composite
sample by Pb-FA collection technique was compared to the average of the
individual TPGM results obtained from weighted Monthly Composite Tailings
samples as shown in Table 6.1 for homogeneity testing. This test was performed
by analysing 6 x 150 g portions of the sample together with the in-house Final
Tailings Quality Control Standard over a period of 5 days. The average TPGM
results are shown in Table 6.2.
Table 6.2 The average TPGM results obtained in mg kg-1 (ppm) for
homogeneity testing using the Pb-FA collection technique.
Weighted Monthly Composite UG2
Tailings samples
TPGM
Mean ± SD
UG2 Tailings composite sample
TPGM
Mean ± SD
1.05 ± 0.10
1.04 ± 0.07
Good agreement was obtained between the average TPGM result of the UG2
Tailings sample compared to the average TPGM result of the weighted UG2
Tailings samples. The TPGM result is within the 95 % confidence limit of the
method. This was not just an indication of homogeneity, but also indicated how
reproducible the Pb-FA collection technique can be when executed under
controlled conditions.
6.1.2 Extraction of the residual chromitite crystals (RCCs)
Having established that the tailings sample was truly representative, one 6 kg
portion of the UG2 Tailings sample was submitted for extraction of the residual
67
chromitite crystals using an alkaline fusion procedure. This procedure is described
in Chapter 5, section 5.9.4. After the extraction of the residual chromitite crystals,
they were further boiled in aqua regia for one hour to remove any entrained PGM
minerals that may have been attached to the grain boundaries in the crystals.
Although most PGEs are profitably recovered from the UG2 ore; minerals such as
chromite, gangue minerals and PGEs associated with silicates are generally lost to
the tailings dams. The UG2 Tailings material is siliceous in character containing
approximately 25% silica (SiO2) and therefore required a basic flux for
decomposition.
Alkali metal carbonates and hydroxides such as Na2CO3, KOH and NaOH are
basic fluxes which attack acidic material and readily form alkali silicates. In the
case of the tailings sample, a basic flux mixture of KOH and Na2CO3 removed
approximately 98% of the silicate composition as indicated by the ICP-OES
analysis of the RCCs. The residual SiO2 composition was indicated to be
< 0.566%.
From the 6 kg composite UG2 Tailings sample, approximately 1.1 kg of RCCs
were recovered from the alkaline fusion procedure, or approximately 18% by
mass. The RCCs were then split into 4 equal fractions for NiS-FA analysis,
microwave dissolution and Te co-precipitation analysis, mineralogy studies and
particle size distribution analysis.
6.1.3 Particle size analysis: UG2 Tailings sample
The old terminology “grading analysis” has been replaced by “particle size”
analysis, which is a more accurate description of the classification of finely
divided material. [1]
Laboratory techniques such as FA-Pb and NiS-FA require a specific particle size
distribution. Prior to evaluation therefore it was critical to confirm that the sample
particle size distribution at least met the minimum criteria. In practice however,
68
particle size analysis is fundamental in the control of Concentrator operations
where the achievement of specific particle size through crushing, milling and
classification is a prerequisite for selective flotation of the PGEs. UG2 Operations
therefore, target a particle size distribution of approximately 70% < 75 μm after
milling and classification.
It was assumed therefore that the UG2 Tailings sample utilized would meet the
particle size criteria, which was subsequently confirmed by the Impala Operations
management. The particle size as mentioned above is more than sufficient for
laboratory techniques such as Pb-FA and NiS-FA collection, which only require a
particle size distribution of 85% < 150 μm. For the most part, fire assay
laboratories have as part of their quality control process, regular confirmation of
the performance of their pulveriser circuits.
6.1.4 Particle size distribution analysis: RCCs
To the naked eye, the residual chromitite crystals appeared extremely fine,
although different particle sizes were apparent. During the initial optimisation
phase using microwave and hotplate digestion, it was evident that not all the
sample dissolved. According to Reddy et al. [71] to obtain reproducible data by wet
chemical attack, a uniform sample mesh size of 200 – 250 must be present
otherwise discrete noble metal minerals may not be completely occluded within
grains and would therefore not be effectively dissolved by acid attack. A mesh
size of 200 – 250 is equivalent to 63 – 75 μm.
Particle size distribution analysis was performed using a laser particle size
analyzer: Saturn DigiSizer 5200. This instrument uses a CCD detector, Mie theory
and provides the highest analytical resolution achievable from laser particle size
analysers. The Saturn DigiSizer automatically measures particle sizes ranging
from 0.1 to 1000 μm quickly and accurately. A crystal fraction of 200 g was
analyzed repeatedly with the entire particle size range recorded and ascribed
median or mean particle diameters. The results were graphically generated
emphasising the capability of the laser scattering particle size analyser in
characterizing the chromite sample, refer to Appendix 1, section 1.1.
69
Statistics support reporting of particle size analysis as percentiles rather than
discrete percentages. Percentiles are a way of tabulating data which falls above
and below a given value [72,73]. The results are presented in this manner in Table
6.3. The 50th percentile (median) gave a value of 276.6 μm at or below, which 50
percent of the observations were found. Similar logic applied to the 10th, 25th, 75th
and 90th percentiles selected.
Table 6.3 Particle size analysis displayed as percentiles for the chromite sample.
r2 = correlation coefficient, a = intercept, Sa = uncertainty in the intercept, b =
slope, Sb = uncertainty in the slope, Sy/x = random calibration uncertainty,
The regression data was assessed as follows:
• The correlation coefficient (r2) measures the linear relationship between
two variables i.e. concentration versus intensity. The statistical data
showed that r and r2 point to almost positive linearity for the isotopes 102Ru, 103Rh, 105Pd, 106Pd, 191Ir, 193Ir, 194Pt, 195Pt 196, Pt and 197Au. Thus, the
linear relationship may have been statistically significant, but did not
prove linearity or adequacy of the fit.
• An ANOVA manipulation was also performed on the regression data to
prove linearity and to test the dynamic range. This implied working in that
region of the calibration curve where the graph starts to plateau. The
93
statistical data showed that the F-values pointed to significant linearity,
since Fcalc > Fcrit for the isotopes 102Ru, 103Rh, 105Pd, 106Pd, 191Ir, 193Ir, 194Pt, 195Pt 196, Pt and 197Au.
• The sensitivity of an instrument is constant within the linear portion of the
calibration graph, but progressively decreases as the calibration line
approaches the horizontal. Thus, the method is analytically sensitive if
b ≠ 0. The analytical sensitivity of the method appeared to be satisfactory
for the isotopes 102Ru, 103Rh, 105Pd, 106Pd, 191Ir, 193Ir, 194Pt, 195Pt 196, Pt and 197Au.
• The calibration was also tested for good precision and whether the range
was acceptable. The precision refers to the measurement replicates per
calibration standard used. Since Sb < Sy/x, good precision was indicated for
the isotopes 102Ru, 103Rh, 105Pd, 106Pd, 191Ir, 193Ir, 194Pt, 195Pt 196, Pt and 197Au.
• When a calibration range is inadequate it is considered unacceptable and
means that extra calibration standards should be included between the
blank and first calibration standard. Since Sa > Sb, the calibration range
was acceptable for the isotopes 102Ru, 103Rh, 105Pd, 106Pd, 191Ir, 193Ir, 194Pt, 195Pt 196, Pt and 197Au.
The statistical analysis performed on the calibration data for the isotopes 102Ru, 103Rh, 105Pd, 106Pd, 191Ir, 193Ir, 194Pt, 195Pt 196, Pt and 197Au produced acceptable
results. The calibration data was accepted for all isotopes of interest and the PGE
analysis proceeded with confidence for the RCCs.
6.7.4 Limit of detection (LOD)
The LOD values are presented in Table 6.19 with concentrations converted to μg
kg -1 or ppb.
94
Table 6.19 LOD values calculated for the calibration program used for the PGEs
6.7.5 Assessment of the interference data and scans performed
The determination of precious metals at ultra trace levels in geological and
environmental matrices remains challenging even after the adoption of more
sophisticated instrumentation such as ICP-MS. Although ICP-MS technology has
been further improved following the introduction of the dynamic reaction cell
(DRC) and high-resolution (HR) ICP-MS technology, it can still only minimize
but not totally remove interferences from the matrix solution.
To complete an interference study it was critical to identify possible interferences
which may form part of the matrix. As above, the major components of the
residual chromitite crystals are the elements Cr, Fe, Mg and Al followed by trace
levels of Ti, Mn, V and Si. The matrix of the dissolved RCCs was removed during
the pre-concentration process prior to Te co-precipitation. Based on the RCCs’
sample composition, it was decided to perform scans to identify possible
interference using pure reference standard solutions each of 100 ug L-1 Cr, Fe, Se
and Te. For the scanning of the precious metals pure reference standard solutions
each of 100 ug L-1 Pt, Pd, Au, Rh, Ru and Ir were used. De-ionised water and a
95
procedural blank containing the matrix elements were also scanned together with
the pure standard solutions above. The resultant scans can be found in section 2.2
Appendix 2.
No spectral interference was identified for the PGEs isotopes of interest by the
matrix elements: Cr, Fe, Se and Te. It was evident that these elements were
following the background peak position of each PGE isotope, where intensities in
counts per second (cps) were around 10 2 and 10 3. It again emphasised how
important LOD is in assessing accuracy of analysis at trace levels.
According to Simpson [48] a major concern in ICP-MS analysis is severe spectral
interferences caused by the formation of refractory oxides (ZrO+, YO+, SrO+,
TaO+ and HfO+) and argides (ArZn+ and ArCu+) on all major isotopes of Pt, Pd
and Au [48]. The formation of refractory oxides like ZrO+, YO+, SrO+, TaO+ and
HfO+ was therefore considered. Reports suggest that Hf has been associated with
ZrO+ in autocatalyst substrates and interference by ZrO+ on Pd has been reported [48].
In the platinum industry it is known that when Na2O2 fusions are used during the
dissolution of samples, that Zr contains high concentrations of Hf, which
interferes with some isotopic lines as HfO+. Zr crucibles are used during Na2O2
fusions, which contain Hf. No interference was apparent from Hf, using the Te co-
precipitation process and scanning of the procedural blank did not indicate any
interference either. A scan was also performed containing 10 mg L-1 of pure Hf
standard solution, but no significant interference was apparent at the PGE
isotopes.
The formation of Argides is possible when Ni, Cu and Zr are present in solution.
Although no Cu and/or Ni was detected, it would not have been possible to
remove such interference with current SPECTRO MASS 2000 ICP-MS
technology. Some ICP-MS instruments have DRC technology, which use
ammonia gas instead of Argon to remove Argide interference.
It was found that Au gave erratic data during analysis, possibly due to memory
effects. In response, the flushing time between sample solutions was increased to
96
about 3 min which resolved the problem. The Au memory effect is illustrated in
Appendix 2, section 2.2, Figure 2H. The effect was clearly shown when the Au
solution was flushed, followed by a de-ionised blank after a flushing time of one
minute.
6.8 DETERMINATION OF THE BASE METAL COMPOSITION OF
THE RCCs.
6.8.1 Developing a method for the dissolution of the RCCs prior to analysis
The platinum industry is known for using pressure dissolution (PD) to dissolve
complex metals and it was decided that this technique should be investigated for
the dissolution of the RCCs. A portion of the sample was placed into a glass
ampoule, together with conc. HCl acid and saturated with Cl2 gas at about – 22 0C, using solid CO2. After sealing of the glass ampoules with a blow torch, they
were placed into stainless steel vessels. These vessels were placed into a furnace
and heated to 250 0C for about twenty four hours or until the sample was
dissolved [53]. This technique was successful for the dissolution of the RCCs, but
as a result of problems encountered, it was not used as the dissolution process for
the experiment and an alternate technique was investigated.
The hardness of the RCCs was between 5.5 and 6.5, which rendered the crystals
resistant to dissolution using normal acidic digestion. Although aqua regia is well
known for the digestion of PGEs, it is ineffective for digesting resistant matrix
phases such as chromite, which are not effectively wetted by aqua regia [71]. In the
chrome industry, acidic mixtures of 1:1 H3PO4:H2SO4 are used to digest
ferrochrome samples for the analysis of chromium. It was decided therefore to
employ a combination of these acids for the dissolution of RCCs using both HP
and MW digestion.
During the optimisation process however, total dissolution of the RCCs was not
obtained with small residual black particles visible inside the beaker. After
performing a particle size distribution analysis of the crystals, it was decided to
micronize the sample to about 95% < 5um, with a view to enhancing dissolution
and possibly releasing any enclosed PGEs minerals within the crystals. The HP
97
and MW digestion techniques were further optimised and after the milling
process, total dissolution was achieved and clear dark green solutions resulted.
The two methods developed are discussed in Chapter 5, sections 5.9.5 and 5.9.6.
The microwave digestion parameters for the Anton Paar GmB multiwave
microwave system are found in Appendix 2, section 2.3.
6.8.2 Base metal composition of the RCCs
A SPECTRO GENESIS ICP-OES was used for the determination of the base
metal elements Cr, Fe, Mg, Al, Mn, Ti, V and Si. The RCCs solutions were
prepared in quadruplicate over 2 days with the reference standard SARM 9
prepared in duplicate. The major and minor base metal concentrations of the
RCCs are shown in Figure 6.6 and Figure 6.7.
Figure 6.6 The major base metal composition of the RCCs as obtained using
HP and MW digestion techniques.
98
Figure 6.7 The minor base metal composition of the RCCs as obtained using HP
and MW digestion.
The HP and MW digestion results were compared using Significance testing
(t – Test). The t-test for independent sample means (equal and unequal variance)
was used and the null hypothesis defined as (H0): μ1 = μ2, i.e. there was no
significant difference between the means of the two methods for the elements Cr,
Fe, Mg, Al, Mn, Ti, V and Si, against the alternate hypothesis (H1): μ1 ≠ μ2, i.e.
there was a significant difference between the means of the two methods for the
elements Cr, Fe, Mg, Al, Mn, Ti, V and Si. The difference in standard deviations
was tested using the F-Test to determine whether to use the t-test for equal and
unequal variance. The HP and MW digestion comparison data are found in
Appendix 2, Section 2.4.
According to the results, there was no significant difference between HP and MW
digestions for the elements Fe, Mn and Ti at the 95 % two-tailed, confidence level
(CL). However, the results reported for HP and MW digestion were significantly
different for the elements Cr, Al, Mg and V at the 95 % two-tailed CL.
The biggest difference reported for the two digestion methods were for the
elements Al, with HP digestion reporting 1.07 % higher than MW digestion, Fe,
with HP digestion reporting 0.58 % higher than MW digestion and Mg, with HP
99
digestion reporting 0.37 % higher than MW digestion. Naturally it would be
expected that MW digestion should be more effective than HP digestion as it has
the advantage of introducing controlled factors such as temperature and pressure
to the digestion process. Surprisingly, higher values were reported for HP
digestion possibly as a result of the greater volumes of acid used which in turn
may have introduced additional impurities. Nevertheless, the precision of MW
digestion was very good when compared to HP digestion, which also suggested
that systematic errors had been introduced during sample preparation using HP
digestion. The precision data shall be discussed in section 6.8.4.
Chapter 3, Table 3.1, listed accessory minerals of the UG2 Reef and good
correlation existed between these minerals and the chemical analysis of the RCCs.
The oxides: Cr2O3, Fe2O3, MgO and Al2O3 represent both the major composition
of chromitite crystals and are reflected in the chemical analysis shown in table
6.7. The UG2 Reef also contains minerals such as ilmenite (FeTiO3), rutile (TiO2)
and ulv�spinel (Fe2TiO3). The chemical analysis for titanium as an oxide is
shown in Table 6.20.
6.8.3 Accuracy of the method
The reference material SARM 9 which was selected for accuracy testing was of
chromite origin and was representative of the RCC matrix. A comparison of the
base metal analysis of SARM 9 to the certified values (C.V.) of SARM 9 is shown
in Table 6.20.
Table 6.20 Comparing base metal results obtained for SARM 9 using HP and