Académique Documents
Professionnel Documents
Culture Documents
J 1751-908X 1998 tb00691
J 1751-908X 1998 tb00691
The platinum-group elements (PGE) and gold have Les lments du groupe du platine (PGE) et l'or ont
been determined in twenty international rock refe- t analyss dans vingt matriaux de rfrence
rence materials by inductively coupled plasma-mass internationaux, par ICP-MS aprs une prconcentration
spectrometry (ICP-MS) after pre-concentration by a par fusion au sulfure de nickel. Il a t possible
nickel sulfide fire assay. It was possible to achieve d'obtenir des limites de dtermination de l'ordre de
determination limits for a 50 g sample that ranged 1 pg g-1 (Rh) 23 pg g-1 (Au). Par comparaison
from 1 pg g -1 (Rh) to 23 pg g -1 (Au). Compared to avec des valeurs certifies et recommandes dj
published certified and recommended values for publies, la mthode a dmontr sa justesse.
rock reference materials, the trueness of the method Nanmoins, nous avons observ des variations
was found to be good. However, in some cases we importantes des concentrations situes dans la
observed large deviations for all elements in the gamme des 10 ng g-1, pour tous les lments dans
sub 10 ng g -1 range within individual reference certains aliquotes. Nos rsultats montrent que les PGE
sample splits. Our results show that the PGE and Au et l'or sont distribus de faon inhomogne dans les
are inhomogeneously distributed in the reference matriaux de rfrence que nous avons analyss
materials analysed here, where they are present in lorsquils sont prsents des concentrations trs
low concentrations, using 50 g test portions. basses. Ceci est dtectable sur des aliquotes de 50 g.
The highly siderophile platinum-group elements Several methods for separation and pre-concentration
(PGE) Ru, Rh, Pd, Ir and Pt as well as Au are becoming of the PGE are commonly used. Overviews describing
increasingly important as powerful indicators of specific these methods are given in two monographs (Van Loon
geological processes. Due to developments in analyt- and Barefoot 1991, Potts 1992) and in one article (Hall
ical technology (e.g. inductively coupled plasma-mass and Bonham-Carter 1988). The determination of the
spectrometry), it is now possible to determine this coher- PGE is mostly performed using three steps. After
ent group of elements in the pg g -1 concentration sample decomposition, a pre-concentration step is
range within common rocks. The increasing attention implemented followed by the determination of the ele-
being paid to the PGE means that it is important to ments using a variety of instrumental methods. The
expand data sets of existing international rock reference three most widely used decomposition methods are
materials to facilitate the comparison of published PGE acid decomposition, fusion with an alkaline flux and
concentrations by different methods. fire assay. All these methods require a chemically com-
plex treatment to guarantee complete dissolution of all
Due to their low concentrations in common rocks, it is possible PGE-bearing phases and to prevent loss of
not possible to determine directly the PGE quantitatively. volatile elements.
5 g CaF2 50 g sample
60 g Na2B4O7 5 g Ni-powder
30 g Na2CO3 5 g sulfur
on
1150 oC 60 min off
ground
200 ml HCl button
2 parts 1 part
10 ml
HCl H2O2
90 oC
16 h
188
GEOSTANDARDS
NEWSLETTER
The Journal of Geostandards and Geoanalysis
was carried out in a fume cupboard. After approxi- instrument was optimized on 115In and 205Tl masses.
mately fourteen hours, the dissolution was complete Data were acquired and calculated using PQ-vision
and the cold solutions were vacuum-filtered through 4.0 Software. The isotopes used for PGE and Au deter-
+
glass micro filter discs (Schott, porosity 4) fitted mination were 101 Ru (corrected for 61 Ni 40 Ar and
+
within screw filter devices. The PGE-containing resi- 64 Ni 37 Cl , other interferences were found to be insigni-
dues were washed carefully with HCl (40% v/v, ficant), 103Rh, 105Pd, 193Ir, 195Pt and 197Au. A PGE multi-
twice-distilled). After placing the filter discs in 100 ml element stock solution was prepared from 1.000 g l -1
beakers, the solids were dissolved with a mixture of Johnson-Matthey single element solutions at a concen-
2 parts HCl (25 ml, 40% v/v, twice-distilled) and one tration of 10 g ml-1 in 2 % v/v HCl. This stock solution
part H 2 O 2 (12.5 ml, 30% v/v, analytical grade). was found to be stable over several months.
Once the evolution of gases had ceased, 5 ml HCl + Calibration solutions with concentrations of 0.001, 0.01,
2 ml H 2O 2 was added. Finally, the solution was pas- 0.1, 1 and 10 ng ml -1, covering the range of expected
sed through a filter paper, transferred to 250 ml concentrations, were freshly prepared from the stock
beakers and evaporated on a hot plate to a volume solution before each analysis.
of 1-2 ml. After adding a further 5 ml HCl + 2 ml
H 2O 2, the solution was again evaporated to 1-2 ml, The investigated rock reference materials
cooled, transferred to volumetric flasks and diluted
with 2% HCl to a volume of 10 ml. Each sample was The investigated rock reference materials in this
independently decomposed at least three times and study came from four different sources. The rock refe-
each dissolution was analysed twice. rence materials TDB-1 (diabase), WGB-1 (gabbro),
WMG-1 (mineralized gabbro) and WPR-1 (altered per-
idotite) were purchased from the CCRMP, CANMET
Table 1. (NRCan), 555 Booth Street, Ottawa, Ontario, Canada,
Typical instrument operating conditions for our VG K1A 0G1. The samples NIM-S (syenite), NIM-D (dunite),
PlasmaQuad PQ 2+ upgraded to the PQ-XS mode NIM-N (norite), NIM-P (peridotite) and NIM-G (granite)
ICP: are rocks from the Bushveld complex and have been
Forward power (at 27 MHz) 1.35 kW prepared by the former National Institute for
Reflected power <7W Metallurgy, South Africa. These samples are available
Coolant gas-flow (Ar) 12.75 l/min
Auxiliary gas-flow (Ar) 0.95 l/min
from MINTEK, 200 Hans Strijdom Drive, Randburg,
Nebulizer gas-flow (Ar) 0.87 l/min Private Bag 3015, Randburg, 2125 South Africa. The
Nebulizer type Meinhardt TR-30-A3 former Zentrales Geologisches Institut, Berlin (former
Sample uptake 1 ml/min
GDR) prepared the samples TS (black shale), GM (gra-
Spray chamber cooled to 1 C
nite), BM (basalt), SW (serpentinite), TB (clay shale) and
Interface: GnA (greissen). Now these reference materials are dis-
Cone-material Nickel
tributed by Breitlnder GmbH, Hans-Sachs-Str. 12,
Acquisition time: 59077 Hamm, Germany. Dr. Noboru Imai from the
Uptake 90 s Geological Survey of Japan made the following
Acquisition 60 s/run (3) samples available: JG-1a (granodiorite), JG-2 (granite),
Washout 180 s
JB-3 (basalt), JSy-1 (syenite) and JMn-1 (manganese
Data acquisition: nodule). Samples can be obtained from Dr. Noboru
Detector type Galileo # 4870 Imai, Geochemistry Dept., Geological Survey of Japan,
Detector mode pulse counting
1-1-3, Higashi, Tsukuba, Ibaraki, 305 Japan.
Peak jumping mode acquisition:
No. points per peak 5 Results and Discussion
Dwell time per point 5120 s
189
GEOSTANDARDS
NEWSLETTER
The Journal of Geostandards and Geoanalysis
Table 2. Table 3.
Optimized low reagent blanks (sample Reference values (Govindaraju, 1994) vs.
equivalent concentration) and the concentrations from this work. Data in italics
method detection limits (calculated as 3) are certified values, other values are proposed
[ng g -1 ] Ru Rh Pd Ir Pt Au [ng g -1 ] Ru Rh Pd Ir Pt Au
Reagent blanks WGB-1 (CCRMP)
1 0.014 0.001 0.047 0.014 0.027 0.047 Reference 0.30 0.32 14 0.33 6.1 2.9
1 - 0.13 14 0.14 3.6 25
2 0.013 0.001 0.045 0.014 0.026 0.052 2 - 0.16 13 0.15 2.7 1.5
3 0.014 0.001 0.051 0.015 0.027 0.041 3 0.21 0.12 13 0.21 3.1 1.8
4 0.015 0.001 0.046 0.014 0.025 0.055 4 0.23 0.13 11 0.16 2.7 1.7
5 0.18 0.13 14 0.19 3.6 1.5
5 0.014 0.001 0.047 0.015 0.024 0.053 6 0.22 0.15 13 0.21 3.7 1.5
6 0.014 0.001 0.039 0.014 0.025 0.052 7 0.11 0.13 15 0.19 3.0 3.4
7 0.014 0.002 0.044 0.014 0.028 0.050 8 0.23 0.16 14 0.24 4.1 3.5
9 0.19 0.13 14 0.21 4.3 2.6
8 0.015 0.001 0.045 0.015 0.024 0.054
10 0.18 0.15 13 0.21 4.1 2.0
9 0.014 0.001 0.044 0.014 0.026 0.049 11 0.19 0.14 14 0.20 3.9 3.5
10 0.014 0.001 0.046 0.015 0.024 0.053 12 0.22 0.14 13 0.18 2.8 3.4
11 0.014 0.001 0.047 0.014 0.023 0.050 13 0.20 0.15 14 0.24 3.7 3.0
14 0.18 0.13 14 0.18 3.1 1.9
12 0.015 0.001 0.042 0.015 0.025 0.055 15 0.19 0.14 14 0.18 3.4 0.69
13 0.014 0.001 0.045 0.014 0.024 0.052 16 0.19 0.13 14 0.17 3.5 2.4
17 0.17 0.17 13 0.23 3.6 1.8
mean 0.014 0.001 0.045 0.014 0.025 0.051 18 0.18 0.14 14 0.19 4.0 0.78
s 0.0005 0.0002 0.0027 0.0004 0.0015 0.0038 19 0.20 0.13 12 0.18 3.0 2.3
20 0.19 0.14 14 0.18 3.9 1.4
DL 0.0015 0.0007 0.008 0.0012 0.0045 0.0113
21 0.17 0.4 12 0.21 4.3 1.0
22 0.15 0.14 14 0.31 7.1 2.1
23 0.20 0.13 14 0.23 3.4 0.78
24 0.21 0.12 14 0.22 4.1 2.8
25 020 0.14 14 0.23 3.3 0.42
the fire assay mixture. In order to determine the proced- 26 0.18 0.14 14 0.25 6.8 2.3
27 0.29 0.11 11 0.14 2.7 1.5
ural PGE blanks, it was necessary to replace the real 28 0.33 0.15 14 0.21 4.3 2.7
samples with a material that is likely to be free of PGE 29 0.19 0.14 9.8 0.23 3.5 1.7
30 0.23 0.16 13 0.25 3.8 1.0
(such as SiO2 powder), because the fire assay flux is very mean 0.20 0.14 13 0.20 3.8 2.0
aggressive and would otherwise attack the crucible. The s 0.040 0.013 1.1 0.037 1.00 0.87
RSD (%) 19.7 9.2 8.1 8.6 26.5 43.8
PGE-concentrations of the chemicals used in the proce-
TDB-1 (CCRMP)
dure varied between production batches and for this Reference 0.30 0.70 22 0.15 5.8 6.3
1 - 0.30 19 0.15 3.7 hB
reason we were not able to present a generalized total 2 - 0.40 22 0.13 3.5 hB
blank concentration. The only way to minimise the conta- 3 - 0.36 21 0.14 3.4 5.6
4 - 0.31 20 0.098 4.5 4.9
mination was to identify and use the purest reagents, to 5 - 0.34 21 0.13 4.3 3.9
6 - 0.38 22 0.11 4.0 5.4
include at least one reagent blank in each fire assay run, 7 - 0.36 21 0.097 4.2 4.6
8 - 0.34 22 0.13 4.5 5.6
and to analyse every sample at least three times. In 9 0.25 0.29 20 0.11 2.9 5.9
Table 2 we present best reagent blanks, expressed as 10 0.23 0.26 19 0.092 4.0 5.0
11 0.34 0.36 22 0.13 4.7 5.7
sample equivalent concentration, as well as the 12 0.38 0.41 21 0.15 3.6 4.4
13 0.31 0.39 21 0.13 3.5 5.4
method detection limits. The instrumental detection 14 0.25 0.30 19 0.12 4.2 5.1
limits (reflecting the ability of the instrument) are signifi- 15 0.44 0.34 21 0.13 3.5 3.6
16 0.48 0.32 21 0.13 3.8 5.3
cantly lower. The detection limit of the method was 17 0.36 0.31 19 0.10 2.9 5.5
18 0.24 0.39 21 0.14 3.8 2.4
calculated as three times standard deviation of the 19 0.33 0.34 20 0.12 4.4 5.1
reagent blanks. The reagent blanks were subtracted in 20 0.31 0.33 20 0.090 4.1 3.9
21 0.27 0.28 22 0.086 2.4 2.1
this work from each PGE data set. However, we 22 0.37 0.29 13 0.11 4.0 5.5
23 0.29 0.32 21 0.12 4.2 5.3
were not able to achieve these low PGE concentration 24 0.48 0.32 20 0.15 3.3 4.1
25 0.35 0.26 22 0.12 2.7 5.0
blanks throughout this work. Thus, we determined
mean 0.34 0.33 20 0.12 3.8 4.8
individual reagent blank values for each batch of che- s 0.079 0.042 1.7 0.019 0.59 1.0
RSD (%) 23.7 12.6 8.4 15.9 15.6 21.1
micals. Consequently, detection limits given in Table 4
WMG-1 (CCRMP)
may be different from those in Table 2. Reference 35 26 382 46 731 110
1 27 26 396 48 720 107
2 24 26 363 47 699 108
Accuracy and precision 3 26 26 401 48 655 108
4 26 26 383 42 719 95
5 28 25 350 48 715 106
The results for three geological reference materials 6 27 26 386 48 718 104
mean 26 26 380 47 705 105
CCRMP gabbro WMG-1, diabase TDB-1 and gabbro s 1.3 0.49 20 2.3 25 4.9
WGB-1 are compared to reference values (Govindaraju RSD (%) 5.1 1.9 5.2 4.8 3.6 4.7
190
GEOSTANDARDS
NEWSLETTER
The Journal of Geostandards and Geoanalysis
Table 4.
PGE concentrations of seventeen international reference rocks
[ng g -1 ] Ru Rh Pd Ir Pt Au [ng g -1 ] Ru Rh Pd Ir Pt Au
mean 9.0 3.0 5.4 1.9 11 0.19 1 <0.002 <0.001 0.021 <0.002 <0.005 <0.020
2 <0.002 <0.001 0.028 <0.002 <0.005 <0.020
GM (ZGI) 3 <0.002 <0.001 0.023 <0.002 <0.005 <0.020
1 <0.002 <0.001 <0.05 <0.001 <0.005 0.24
2 <0.002 <0.001 <0.05 <0.001 <0.005 0.14 mean <0.002 <0.001 0.024 <0.002 <0.005 <0.020
3 <0.002 <0.001 <0.05 <0.001 <0.005 0.23
JB-3 (GSJ)
4 <0.002 <0.001 <0.05 <0.001 <0.005 0.37
5 <0.002 <0.001 <0.05 <0.001 <0.005 0.14 1 0.16 0.053 2.3 <0.01 1.6 0.35
2 0.14 0.045 1.8 <0.01 1.2 0.26
mean <0.002 <0.001 <0.05 <0.001 <0.005 0.22
3 0.10 0.031 1.6 <0.01 1.3 0.58
BM (ZGI)
mean 0.13 0.043 1.9 <0.01 1.4 0.40
1 <0.002 <0.001 0.099 <0.001 <0.005 0.16
2 <0.002 <0.001 0.072 <0.001 <0.005 0.11
JSy-1 (GSJ)
3 <0.002 <0.001 0.12 <0.001 <0.005 0.14
1 0.030 0.010 0.13 0.017 0.11 0.093
4 <0.002 <0.001 0.12 <0.001 <0.005 0.13
5 <0.002 <0.001 0.12 <0.001 <0.005 0.18 2 0.027 0.012 0.11 0.017 0.071 0.096
3 0.032 0.008 0.15 0.015 0.091 0.094
mean <0.002 <0.001 0.10 <0.001 <0.005 0.14
mean 0.030 0.010 0.13 0.016 0.089 0.094
Gn-A (ZGI)
1 <0.1 <0.005 0.055 <0.05 <0.08 9.9 JMn-1 (GSJ)
2 <0.1 <0.005 0.074 <0.05 <0.08 8.0 1 4.4 2.0 1.6 0.59 42 1.1
3 <0.1 <0.005 0.063 <0.05 <0.08 12
2 5.5 1.8 1.6 0.58 40 1.1
4 <0.1 <0.005 0.071 <0.05 <0.08 10
3 5.1 2.0 1.6 0.58 41 1.1
5 <0.1 <0.005 0.073 <0.05 <0.08 14
mean <0.1 <0.005 0.067 <0.05 <0.08 11 mean 5.0 1.9 1.6 0.58 41 1.1
191
GEOSTANDARDS
NEWSLETTER
The Journal of Geostandards and Geoanalysis
1994) in Table 3. The differences between the reference PGE-data of a further seventeen
values and our data for WMG-1 are below 5%, with international reference rocks
the exception of Ru, and therefore show very good
recovery and accuracy. In most cases our results are Our PGE data of a further seventeen international
lower, for example for gabbro WGB-1 and diabase reference materials are compiled in Table 4. Data are
TDB-1, these being the most readily available geolog- grouped according to the supplier of the reference
ical reference materials with the lowest certified PGE material. Mantle-normalized PGE-plots (mantle values
values. The recovery of Pd is good for these reference from Barnes et al. 1985, corresponding to 0.815%
rocks (< 9% difference), but poor for Pt and Au. The chondrite values) have been plotted for CCRMP and
question arises as to whether this is due to inefficient NIM samples (Figures 2 and 3).
recovery of our procedure, sample inhomogeneity or
bias in certified values. The accuracy of certified values
in this low concentration range must be questioned,
taking into account the critical comments of Hall and
Bonham-Carter (1988) concerning the sample homo-
Sample/mantle
geneity and accuracy of PGE values in other reference
materials as well as the poor recovery for all PGE
reported in some other studies (e.g. Reddi et al. 1994).
A recent study (Verma 1996) predicts errors in the
characterisation of certified data for most elements of
between 10% and > 100%.
192
GEOSTANDARDS
NEWSLETTER
The Journal of Geostandards and Geoanalysis
Table 5.
Concentrations from this work vs. values from low PGE and Au values in common rocks, as there could
Sen Gupta & Gregoire (1989) by co-precipitation with be concentration variations of up to 100% both
tellurium using electrothermal vaporisation (ETV) for between and within sample splits.
sample introduction
Our experience is, that it is important to plot all
[ng g -1 ] Ru Pd Ir
this work S.G.&G. this work S.G.&G. this work S.G.&G. PGE data normalized to chondrite or mantle values in
order to evaluate and interpret low PGE-concentrations
NIM-P 9 5.6 5.4 4 1.9 4 in rock samples. This is a promising way of evaluating
NIM-D 1.9 24 0.88 4 0.44 6
the quality of the data and of minimising the risk of
NIM-N 0.53 7 1.1 4 0.090 1
NIM-G < 0.3 7.8 0.76 2 0.031 0.6 misinterpreting results.
NIM-S < 0.3 3 0.067 nd 0.010 2.5
Acknowledgement
nd not determined
193
GEOSTANDARDS
NEWSLETTER
The Journal of Geostandards and Geoanalysis
references
194