Geochimica et Cosmochimica Acta, Vol. 69, No. 8, pp. 2153–2163, 2005
Copyright © 2005 Elsevier Ltd
Printed in the USA. All rights reserved
0016-7037/05 $30.00 ⫹ .00
doi:10.1016/j.gca.2004.10.012
Accurate measurement of silver isotopic compositions in geological materials including low
Pd/Ag meteorites
S. J. WOODLAND,1,* M. REHKÄMPER,1 A. N. HALLIDAY,†,1 D.-C. LEE,2 B. HATTENDORF,3 and D. GÜNTHER3
1
Institut für Isotopengeologie und Mineralische Rohstoffe, ETH Zentrum, Sonneggstr. 5, CH-8092 Zurich, Switzerland
2
Institute of Earth Sciences, Academia Sinica, 128 Academai Road, Section 2, Nankang, Taipei 115, Taiwan
3
Laboratorium für Anorganische Chemie, Wolfgang-Pauli-Str. 10, ETH Hönggerberg, CH-8093 Zürich, Switzerland
(Received October 9, 2003; accepted in revised form October 21, 2004)
Abstract—Very precise silver (Ag) isotopic compositions have been determined for a number of terrestrial
rocks, and high and low Pd/Ag meteorites by utilizing multicollector inductively coupled plasma mass
spectrometry (MC-ICPMS). The meteorites include primitive chondrites, the Group IAB iron meteorites
Canyon Diablo and Toluca, and the Group IIIAB iron meteorite Grant. Silver isotopic measurements are
primarily of interest because 107Ag was produced by decay of the short-lived radionuclide 107Pd during the
formation of the solar system and hence the Pd-Ag chronometer has set constraints on the timing of early
planetesimal formation. A 2␴ precision of ⫾0.05‰ can be obtained for analyses of standard solutions when
Ag isotopic ratios are normalized to Pd, to correct for instrumental mass discrimination, and to bracketing
standards. Caution must be exercised when making Ag isotopic measurements because isotopic artifacts can
be generated in the laboratory and during mass spectrometry. The external reproducibility for geological
samples based on replicate analyses of rocks is ⫾0.2‰ (2␴).
All chondrites analyzed have similar Ag isotopic compositions that do not differ significantly (⬎0.3‰)
from the ‘terrestrial’ value of the NIST SRM 978a Ag isotope standard. Hence, they show no evidence of
excess 107Ag derived from 107Pd decay or, of stable Ag isotope fractionation associated with volatile element
depletion within the accretion disk or from parent body metamorphism. The Group IAB iron meteorite
samples analyzed show evidence of complex behavior and disturbance of Ag isotope systematics. Therefore,
care must be taken when using this group of iron meteorites to obtain chronological information based on the
Pd-Ag decay scheme. Copyright © 2005 Elsevier Ltd
nuclide 107Pd decays to 107Ag with a half-life of only 6.5 Ma
and in stellar nucleosynthesis it can be produced through both
the s- and the r-process pathways. Many iron meteorites of the
groups II, III and IV show good correlations between excess
107
Ag and Pd/Ag ratios. These correlations are thought to be
due to the presence of live 107Pd in the early solar system and
in situ decay of this short-lived nuclide (Chen and Wasserburg,
1996). Palladium is both more refractory and more siderophile
than Ag. Hence, volatile element depletion in the solar nebula
and/or planetary core formation may have caused the requisite
early Pd/Ag fractionations necessary for the system to be useful
as a chronometer. The best examples of this are the iron
meteorites, where the Fe-Ni metal typically has much higher
Pd/Ag ratios than co-genetic sulfides, due to the highly siderophile nature of Pd. Isochrons can thus be constructed using
these phases.
Early Ag isotopic measurements were carried out by TIMS
(Chen and Wasserburg, 1983). The precision of TIMS analyses
was limited to ⬃⫾1‰ to 2‰, however, due to the difficulties
in applying an instrumental mass bias correction because Ag
has just two naturally occurring isotopes. Such data were sufficient for the study of meteorites with high Pd/Ag ratios, which
developed large Ag isotopic anomalies. However, more precise
measurement techniques are required to resolve the very small
differences in Ag isotope compositions that are expected for
meteorites that display low Pd/Ag. Carlson and Hauri (2001)
addressed this problem and extended the Pd-Ag chronometer to
low Pd/Ag materials using MC-ICPMS (multiple collector inductively coupled plasma mass spectrometry). By utilizing
1. INTRODUCTION
Silver is a trace element and is found at the ppb level in most
terrestrial and extra-terrestrial materials. The exceptions to this
are rare ore-deposits where Ag may occur as a native metal. In
such deposits, the Ag is rarely pure and often forms alloys with
Au, Cu, Te and Sb (Frueh and Vincent, 1967). Silver most
commonly occurs in nature as a univalent cation and it is a
moderately volatile element that can display both siderophile
and chalcophile behavior.
Silver has just two naturally occurring isotopes, 107Ag
(51.4%) and 109Ag (48.6%) and a value of 107Ag/109Ag ⫽
1.07638 ⫾ 0.00022 was determined for the Ag isotope reference material NIST SRM 978a in carefully calibrated mass
spectrometric measurements. Analyses of native Ag metals
from globally distributed mines have shown, that 107Ag/109Ag
can vary by up to ⬃0.6‰ in terrestrial ore samples, presumably
due to mass dependent stable isotope fractionations that occurred at low temperatures (Hauri et al., 2000). This indicates
that Ag isotopes may be a useful geochemical tracer, for
example, in economic geology studies.
High 107Ag/109Ag ratios of up to almost 10 (Chen and
Wasserburg, 1996) have been determined for certain meteorites, but these variations are thought to be due to radiogenic
decay of 107Pd rather than stable isotope fractionation. The
* Author to whom correspondence should be addressed (woodland@
erdw.ethz.ch).
†
Present address: Department of Earth Sciences, Parks Road, Oxford
OX1 3PR, UK.
2153
2154
S. J. Woodland et al.
Table 1. Chondritic meteorite Pd-Ag data.
Pd conc.
(ng/g)
Chondrite samples
Class
Sample weight (g)
␧107Ag
␧107Ag repeat
Allende (1)
Allende (2)
Allende (3)
Allende (4) (leach exp.)
Leach A
Leach B
Allende (5)a
Allende (6)a
Murchison
Guarena
Ausson
Tennasilm
Mezö Madaras (1)
Mezö Madaras (2)
ALH 84081 (1)
ALH 84081 (2)
Abee (1)
Abee (2)
Abee (3)a
Indarch
CV3
CV3
CV3
CV3
1.000
1.001
0.816
3.000
(AQR digest)
(HF-HNO3 digest)
0.102
0.114
0.506
0.501
0.502
0.551
0.618
1.159
0.601
1.206
0.199
0.202
0.205
0.202
–2.6
–0.5
–0.1
–0.6
–1.2
746
–1.9
–0.3
–2.9
686.6
688.2
567
865
534
CM
H6
L5
L4
L3
LL6
EH4
EH4
Ag conc.
(ng/g)
108
Pd/109Ag
87
4.73
80.8
80.9
113
57
117
4.67
4.68
2.76
8.36
2.53
538
64
4.62
510
52
5.39
249
259
249.0
275
2.10
1.93
1.93
1.68
–1.3
–0.4
–1.6
0.3
–2.9
–0.5
0.2
0.6
3.4
3.0
0.6
–1.4
–0.6
–2.7
0.5
3.4
0.7
–1.7
–0.4
–2.4
948
910
820.0
840
The difference in Ag/ Ag between a sample and bracketing NIST SRM 978a standards is expressed as ␧ (1␧ ⫽ 1 part in 10,000). Where
possible samples analyzed in duplicate (i.e., Mezö Madaras), or triplicate (i.e., Abee) to assess reproducibility. Many sample solutions were run twice
(i.e., ‘repeat’) to confirm that no instrumental problems occurred during the initial analysis. 108Pd/109Ag was determined using ID and MC-ICPMS
(samples marked a), otherwise 108Pd/109Ag was measured directly on sample solutions without chemical processing using an Element 2 ICPMS.
107
109
admixed Pd to correct for the instrument-induced mass discrimination of Ag, they were able to achieve a reproducibility of
⬃⫾0.13‰ for the determination of 107Ag/109Ag.
This study builds on the work of Carlson and Hauri (2001)
and it readdresses some of the difficulties associated with Ag
isotopic measurements. Using a refined approach for the chemical separation of Ag from the bulk sample matrix, it has been
possible to obtain accurate and precise Ag isotope data for
samples with as little as 20 ng of Ag. This enables analysis of
terrestrial samples with low Ag concentration and of lowPd/Ag meteorites such as chondrites and group IAB iron meteorites.
2. ANALYTICAL PROCEDURES
2.1. Sample Preparation
2.1.1. Reagents and materials
All acids used in this study were purified by subboiling distillation.
A second distillation (in a Teflon still) was found to greatly reduce Ag
blanks. The water used was of 18-M⍀ grade from a Millipore purification system (termed MQ water hereafter).
Two different standards were utilized in this study. The National
Institute of Standards and Technology (NIST) Standard Reference
Material (SRM) 978a is an Ag isotope standard, which is supplied in
the form of an AgNO3 salt. This material was dissolved in MQ water
and stored in 5% HNO3 (⬃1000 ␮g/ml concentration) in a lightproof
Teflon bottle. As NIST SRM 978a is the only universally available Ag
standard, it served as the unfractionated isotopic reference material to
which unknown samples were compared. A 1000 ␮g/ml JMC Ag
solution (ICP standard, supplied in 5% HNO3 by Alfa, Germany) was
utilized as a secondary standard.
The instrumental mass discrimination correction required that a Pd
solution be admixed to all sample and standard solutions before isotopic analysis. For this purpose, we prepared a dedicated standard solution with 10 ppm Pd in 3% HNO3. This solution was made up from a
1000 ␮g/ml JMC Pd standard (ICP standard supplied in HCl by Alfa,
Germany), which was first dried down several times with HNO3 to
eliminate all traces of HCl.
2.1.2. Sampling procedures: chondrites
Soft chondritic meteorites were broken into subsamples by wrapping
in thick plastic and hitting with a hammer. More compact samples were
cut using a diamond-edged saw blade saved exclusively for meteorites.
Tests on this saw blade revealed that its contribution to the Ag blank
was negligible. However, sawn chondrites were nevertheless washed in
distilled ethanol and MQ water, then gently leached for ⬍10 min in 1
mol/L HNO3 to remove any contamination. More extensive leaching
procedures were avoided as they selectively dissolve Ag in preference
to Pd. Chondritic meteorites were powdered, avoiding surfaces with
fusion crust or obvious contamination, using an Al2O3 pestle and
mortar under clean air. Sample weights used for each of the chondrites
analyzed are given in Table 1.
2.1.3. Sampling procedures: iron meteorites
The Canyon Diablo sample (USNM 676) consisted of 2 large discrete troilite nodules (henceforth known as CDT) embedded in Fe-Ni
metal. The CDTs were easily broken out of the meteorite slab. Samples
from both these CDT nodules, named CDT1 and CDT2, were analyzed
(Table 2). A large sample (12.76 g) of Fe-Ni metal that did not contain
any obvious sulfide inclusions or the carbon-rich schreibersite rims
which surround CDTs (Carlson and Hauri, 2001) was subsequently
sawn from the same slab (Table 2). This was washed in distilled ethanol
and MQ water before being heavily leached in aqua regia to remove all
saw marks and rust, leaving only a shiny central piece of metal. The
metal must be thoroughly leached because its low Ag content renders
it susceptible to terrestrial contamination (Kaiser and Wasserburg,
1983). The CDTs were subjected to different cleaning procedures to
ascertain if terrestrial contamination was potentially problematic. Subsamples from the same CDT nodule were either gently leached with aqua
regia, or subjected to a much more rigorous cleaning procedure where they
were first abraded with silicon carbide to remove all saw cuts, rust and
surface irregularities, and then treated with aqua regia (Table 2).
The Toluca sample (USNM 75) also contained discrete sulfide
nodules within the Fe-Ni metal. The sulfide analyzed was carefully
Silver isotopic composition of low Pd/Ag meteorites
2155
Table 2. Data for iron meteorite samples analyzed.
Iron meteorite
samples
Group
Canyon Diablo
(USNM 676)
Metal
CDT1 A
CDT1 B
CDT1 C
CDT1 D
CDT2 A
CDT2 B
Toluca (USNM 75)
Metal
Sulfide 1A
Sulfide 1B
Grant (USNM 836)
Metal
Sulfide
␧107Ag
repeat
Pd conc.
(ng/g)
Ag conc.
(ng/g)
0.4
–2.2
–1.5
–0.4
0.4
–5.6
–2.9
0.8
–1.8
3688
17.6
13.9
30.8
522
448
5.45
0.75
0.86
3.6
6.2
5.6
5.2
5.5
4397
21.3
19.4
26.1
398
376
92.6
0.03
0.03
16.86
1.13
160.6
25.0
159.8
24.1
2167
8.2
1.2
12.4
993.2
0.36
Cleaning procedure
Sample weight (g)
(after leaching)
␧107Ag
Leached (27%)
Leached (2%)
Leached (2%)
Si-C abrasion, leached (6%)
Si-C abrasion, leached (3%)
Leached (4%)
Si-C abrasion, leached (19%)
9.30
0.85
0.82
1.62
1.21
1.36
1.14
Leached (26%)
Leached (6%)
Si-C abrasion, leached (22%)
Si-C abrasion, leached (27%)
Si-C abrasion, leached (9%)
108
Pd/109Ag
IA
0.0
1.0
–6.0
–3.5
2.7
1.5
639
442
65.9
0.02
0.02
0.002
0.002
IA
IIIB
All Pd and Ag concentrations determined using ID. Two separate Canyon Diablo troilites (CDT) and a sulfide from Toluca were sampled and
analyzed numerous times to assess reproducibility and the effect that cleaning procedure had on Ag isotope composition.
chiselled from one nodule. Both metal and sulfides were subjected to
the same cleaning procedures as the Canyon Diablo sample (Table 2).
The sulfide in Grant (USNM 836) was much more difficult to
remove than those of Canyon Diablo and Toluca because it was not
rimmed by schreibersite. Hence, it was separated from the surrounding
metal using a dental drilling tool. For this sample, both metal and
sulfide were cleaned using a combination of silicon carbide abrasion
followed by leaching in aqua regia (Table 2). Although the metal was
examined before sampling and no sulfides were visible on the surface,
it was apparent following extensive leaching that some microscopic
sulfides, which are more resistant to leaching, were present within the
metal. This may be common to all large metal samples.
2.1.4. Sample digestion
The digestion and further handling of the meteorite samples was
carried out in laminar flow cabinets in a clean room facility. To prevent
cross contamination, separate Savillex beakers were designated for
terrestrial rocks and meteorites and also for samples with very high Ag
concentrations, such as sulfides. All lab-ware was cleaned predominantly with HNO3 and the Savillex beakers were refluxed with distilled
HNO3 before use, to eliminate any residual HCl.
For silicate rock and chondritic meteorite samples, between 0.2 and
2 g of sample powder were utilized depending on the Ag concentration.
Some samples were digested on a hotplate in Savillex beakers using
HF-HNO3 for 48 h and then dried in the presence of HNO3 to remove
traces of HF, before final dissolution with 10 mol/L HCl. Alternatively,
samples were weighed into 90-mL quartz vessels containing ⬃10 mL
of aqua regia and digested in a high pressure asher (HPA) at 220°C and
230 bar for between 4 and 8 h. This approach was particularly useful
in obtaining good digestions for samples with high carbon content.
Following treatment in the HPA, samples were transferred to Savillex
beakers and a normal digestion procedure using HF-HNO3 as outlined
above was carried out. Following the final dissolution in 10 mol/L HCl,
the samples were dried to be re-dissolved in 0.5 mol/L HCl and
centrifuged shortly before loading onto ion-exchange columns.
A large subsample of Allende (⬃18 g) was crushed and all Allende
analyses undertaken were aliquots of this same powder (Table 1).
Allende (1) was digested on a hotplate, whereas Allende (2) and
Allende (3) were digested in the HPA. Allende (4) was a leach
experiment that utilized 3 g of powder. In the first leach step, the
powder was digested in 50% aqua regia for 12 h, which should be
sufficient to dissolve sulfides and metal. After this time, the sample was
centrifuged and the liquid was set aside as Leach A. The residue was
attacked on a hotplate with a mixture of concentrated HF and HNO3 for
24 h to completely digest silicate phases. The sample was then centri-
fuged and the liquid was set aside as Leach B. Some carbonaceous
residue still remained after this step and this was finally digested in the
HPA using aqua regia. For this digestion, a special glass capsule
containing KClO3 was inserted into the HPA vessel to provide additional Cl2, which has proved effective at dissolving resistant PGEbearing phases (Paar-Physica, Application Note). The solution produced
following the HPA digestion was termed Leach C.
The terrestrial pyrite sample, Ag ore, and both metal and sulfide
samples from the iron meteorites were dissolved using warm aqua
regia. Samples were then dried and redissolved twice in 10 mol/L HCl
to convert them to chloride form. After this procedure, the samples
were fully dissolved and produced clear solutions when diluted to 0.5
mol/L HCl for column loading. Large volumes of 0.5 mol/L HCl
(⬃500 mL for Canyon Diablo and Grant) were required, however, to
keep the metal samples of iron meteorites in solution.
2.1.5. Aliquoting and column chemistry
For samples that also required the determination of Ag and Pd
concentrations and Pd/Ag ratios, 10% to 20% aliquots of the 0.5 mol/L
HCl solutions were removed just before the column chemistry and
stored in Savillex beakers. The ion exchange procedure was adopted
and slightly modified from the separation schemes of Carlson and
Hauri (2001) and Kaiser and Wasserburg (1983). Sample solutions in
0.5 mol/L HCl were loaded onto anion exchange columns containing
1.25 mL (wet volume) of AG1-X8 resin (mesh size 200 – 400 ␮m)
cleaned and preconditioned using 20 mL of 9 mol/L HCl and 30 mL of
0.5 mol/L HCl. Silicate samples ⬎1.5 g (such as Allende Leach A)
were split over more than one column. The Canyon Diablo, Toluca and
Grant metal dissolutions were split over five anion columns. After
sample loading, the columns were washed sequentially with 33 mL of
0.1 mol/L HCl, 10 mL of 0.001 mol/L HCl, 5 mL of 0.5 mol/L HNO3.
Silver was then eluted with a further 5 mL of 0.5 mol/L HNO3.
The Ag fractions were dried on a hotplate. If a white residue was
visible at this stage, the anion-exchange separation was repeated, as
described above. After a second pass, the samples were invariably
‘clean’ and several drops of concentrated HNO3 were added and
evaporated to remove any residual HCl. The Ag fractions were then
loaded onto a clean-up column, containing 1 mL of AG50W-X8 cation
exchange resin, in 0.5 mL of 0.5 mol/L HNO3. This resin was cleaned
and preconditioned using 10 mL of 10 mol/L HNO3 and 30 mL of 0.5
mol/L HNO3. With the exception of the iron meteorite metal samples,
which were divided over five cation exchange columns, each sample
was loaded onto a single column. After sample loading, the columns
were rinsed with 11 mL of 0.5 mol/L HNO3 and then Ag was eluted
and collected with 15 mL of 0.5 mol/L HNO3.
S. J. Woodland et al.
1.0812
1.0810
Ag/109Ag (FC)
The clean Ag fractions were gently dried several times in the
presence of a few drops of concentrated HNO3 to remove any residual
HCl. The samples were then diluted to 1 mL with 0.5 mol/L HNO3 and
10% aliquots were removed, spiked with U as an internal standard and
analyzed. Mass scans were performed for each aliquot, using the Nu
Plasma MC-ICPMS, to ensure that no potentially interfering elements
were present and to assess the Ag concentration. If significant quantities of a matrix element were present, the sample was further purified,
by repeating the cation exchange clean-up chemistry. Sample solutions
with sufficiently high (⬎100 ppb) Ag concentrations were further
diluted, such that more than one isotopic analysis could be performed.
In addition, the samples were doped with an appropriate amount of Pd
(using the 10 ppm Pd solution in HNO3), such that the final solutions
analyzed by MC-ICPMS had Pd/Ag ratios of 1 to 2, equivalent to that
of the Ag isotope reference standards.
Large instrumental mass bias effects are observed in plasma source
mass spectrometry, due to the preferential extraction and transmission
of the heavier ions. Silver has just two naturally occurring isotopes and
a correction for the instrument induced mass discrimination is therefore
not possible by internal normalization. Instead, an external mass bias
correction can be employed with MC-ICPMS because elements with
overlapping mass ranges display a nearly identical mass bias (Rehkämper et al., 2001). In this study, the Ag data were externally normalized
to Pd using the exponential mass fractionation law. This approach
assumes that the behavior of Ag and Pd is identical during mass
spectrometry, such that ␤Pd ⫽ ␤Ag, where ␤ is the mass fractionation
coefficient (Marechál et al., 1999). This is probably not strictly the
case, as the long-term reproducibility of 106Pd/105Pd, when internally
normalized to 108Pd/105Pd, is considerably better than that of 107Ag/
109
Ag externally normalized to 108Pd/105Pd (Figs. 1a and 1b; Woodland
et al., 2003). The 2␴ long-term reproducibility that can be attained
for 107Ag/109Ag (JMC standard) when corrected for fractionation
using Pd is ⬃⫾500 ppm, whereas, the 2␴ long-term reproducibility
attained for 106Pd/105Pd, when corrected with 108Pd/105Pd, is just
⫾39 ppm.
107
2.1.6. Preparation of sample solutions for analysis
2.2.2. Instrumental Mass Bias Corrections
Mean = 1.080478 ± 0.00054 (5.0e)
(a)
1.0808
1.0806
1.0804
1.0802
1.0800
1.0798
1.2236
Pd/105Pd (FC)
The above procedure efficiently separates Ag from Pd, as well as Cd,
which could generate isobaric interferences on Pd. Other matrix elements were also almost completely eliminated from the Ag fraction,
with the exception of Ir, Ru and Rh (normally ⬍1% of total Ag signal)
and minor Nb (⬍0.1% of total Ag signal). The iron meteorite metal
samples needed to be passed through the cation exchange columns
twice, however, due to the high levels of Ru present. Yields for the total
procedure were determined to be between 75% and 100%, by comparison of the Ag ion beams obtained for the samples with standards of
known concentration. The total procedural Ag and Pd chemistry blanks
were measured using a similar procedure and found to be ⬍150 pg for
both elements if single distilled acids were used and ⬍15 pg if double
distilled reagents were employed. For a typical sample with 80 ng of
Ag, a maximum blank of 150 pg contributes only 0.2% of the total Ag,
which is insignificant.
The major difference of this procedure to that of Carlson and Hauri
(2001) is that the present method utilizes 0.5 mol/L HNO3, rather than
9 mol/L HCl, to elute Ag from the anion exchange columns. There are
two reasons for this modification. Firstly, HCl appears to cause undesirable behavior of Ag during mass spectrometry and may form insoluble AgCl. As such it is better to eliminate it wherever possible.
Secondly, evaporation of an Ag fraction that was eluted with 9 mol/L
HCl leaves a ‘sticky’ residue in the beaker that cannot be effectively
re-dissolved in the small amount of HNO3 used for loading the cation
exchange columns. As a result, up to 25% of the Ag could be lost
during transfer of the sample solution between columns when using 9
mol/L HCl.
106
2156
1.2234
Mean = 1.222893 ± 0.000047 (0.4e )
(b)
1.2232
1.2230
1.2228
1.2226
1.2224
2.2. Ag Isotope Ratio Measurements
1.0
0.8
2.2.1. Measurement protocols
107
(c)
0.6
e 107Ag
All analyses were performed with a Nu Instruments Nu Plasma
MC-ICPMS at the ETH Zurich. The ion beams of all masses from
105
Pd to 111Cd were monitored in a single cycle using Faraday cups and
1011 ⍀ resistors. This enabled measurement of 108Pd/105Pd, 106Pd/105Pd
and 107Ag/109Ag and the interference free 111Cd isotope. The 108Pd/
105
Pd ratio was used for mass fractionation correction of 107Ag/109Ag,
based on a reference value of 108Pd/105Pd ⫽ 1.18899 (Kelly and
Wasserburg, 1978).
Data were collected with 40 integrations of 10 s each, in blocks of
10. On-peak baselines were measured for 15 s before each block during
deflection of the ion beam using the electrostatic analyzer. The internal
precision of the isotope ratio measurements was typically ⫾30 ppm
(2␴mean) for mixed standard solutions with 50 ppb Ag and Pd. The
instrument was used in conjunction with a Cetac MCN 6000 desolvating nebulizer. The sample introduction system was cleaned with 0.7
mol/L HNO3 for at least 15 min between analyses or until the residual
background contribution was ⬍100 ppm of the total Ag signal. Sample
uptake rates were between 80 and 120 ␮L/min and total ion beam
intensities of 2 to 2.8 ⫻ 10–11 A were achieved for solutions with 50
ppb Ag. This is equivalent to a transmission efficiency (ions registered
by the detectors vs. total atoms consumed in the analyses) of ⬃0.035%.
Average Bracketed Std e Ag = 0.02 ± 0.5
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
Fig. 1. (a) The long-term reproducibility (over a 12-mo period) of
Ag/109Ag for a JMC Ag standard when externally normalized for
instrument-induced mass bias using Pd is ⫾5.0 ␧ (2␴). (b) In comparison, the long-term reproducibility of 108Pd/106Pd for a JMC Pd standard when internally normalized (using 108Pd/105Pd) for mass bias is
only ⫾0.4 ␧ (2␴). This illustrates that externally normalized isotope
data are subject to much greater drifts with time compared to internally
normalized ratios due to differences in the fractionation factors of Pd
and Ag during mass spectrometry. Normalization of the Ag isotope
data to both Pd and bracketing standards (c) overcomes problems of
instrumental drift and hence improves the precision of Ag isotopic data
by 1 order of magnitude to ⫾0.5␧ (from Woodland et al., 2003).
107
Silver isotopic composition of low Pd/Ag meteorites
Silver isotopic ratios corrected for instrumental mass bias using Pd
showed the most significant drift during the first hours of an analytical
session. Hence, data for samples were not collected for at least 4 to 5 h
after the instrument was switched on. To combat this drift, a samplestandard bracketing technique was employed in addition to external
Pd-normalization. With standard-sample bracketing, the externally normalized 107Ag/109Ag ratio of a sample is referenced to the average
fractionation corrected 107Ag/109Ag data of the standards that were
analyzed before and after this sample. The relative difference between
the Ag isotopic compositions of the sample and the standard is then
expressed as a deviation in ␧107Ag:
␧107Ag ⫽ [(107Ag ⁄ 109Ag)sample ⁄ (107Ag ⁄ 109Ag)standard) – 1] ⫻ 10, 000
(1)
This technique considerably improves the precision of the Ag isotope
ratio measurements and the data collected for a JMC Ag solution, when
normalized to bracketing standards, displayed a 2␴ reproducibility of
⫾52 ppm, which is equivalent to ⫾0.5 ␧107Ag (Fig. 1c).
2.2.3. Interference corrections
Neither of the Ag isotopes suffers from direct isobaric interferences
from other elements but 108Pd, which is used for the Ag mass bias
correction, overlaps with the minor isotope 108Cd (108Cd/Cd ⫽ 0.89%,
108
Pd/Pd ⫽ 26.5%). A correction for this interference was applied by
monitoring 111Cd and was found to be accurate for solutions with
Cd/Pd ratios of ⬍0.01, which was the case for all samples analyzed in
this study.
The isotope ratio measurements can furthermore be biased by polyatomic spectral interferences. A number of test measurements showed,
however, that our chemical separation procedures and the mass spectrometric techniques (e.g., use of a desolvating nebulizer) yielded
interferences from oxide, nitride, and argide species that were at
negligible levels for all samples. The molecular ions that are most likely
to be problematic are 93Nb14N⫹ (for 107Ag⫹) and 93Nb16O⫹ (for 109Ag⫹).
Experiments were therefore conducted to assess how much Nb could be
tolerated and the measured 107Ag/109Ag ratios of solutions were found
to be unaffected as long as the Nb/Ag ratios were ⬍20. All samples
solutions, however, displayed Nb/Ag ratios of ⬍0.01 at which level the
Nb-based spectral interferences are completely insignificant.
2.3. Determination of Pd and Ag Concentrations and
Pd/Ag Ratios
The determination of Pd and Ag abundances and Pd/Ag ratios
utilized the minor (10%–20%) solution aliquots that were taken from
the sample solutions before the ion exchange separation. Two different
techniques were used for the concentration measurements.
For most samples with high Ag concentrations (⬎50 ng/g) the
aliquots were analyzed directly, without any chemical separation of Ag
and Pd from the geological matrix. These analyses were performed
with a Thermo Finnigan Element 2 ICP-MS, using an ultrasonic
nebulizer for sample introduction. The mass spectrometric measurements involved monitoring of all Pd and Ag isotopes, of 103Rh which
served as an internal standard, and of Y, Zr, Nb, Mo and Ru isotopes
for the detection and correction of potential spectral interferences. The
Pd and Ag concentrations were determined by comparing the measured
107
Ag and 105Pd ion beam intensities for samples, with those of dilute
elemental standard solutions. The precision and accuracy of this approach was estimated by analyzing 10% aliquots from four different
dissolutions of between 0.8 and 1.0 g of Allende powder. The average
108
Pd/109Ag measured was 4.75 ⫾ 0.33 (2␴). Based on this, the
108
Pd/109Ag ratios determined with this method are estimated to have
an uncertainty of ⬃⫾7%. The accuracy of the technique is confirmed
by the good agreement of our Allende data with the results of Carlson
and Hauri (2001), who determined 108Pd/109Ag ratios of between 4.18
and 4.58 for this meteorite.
The Pd/Ag ratios of samples with low Ag concentrations were
determined by isotope dilution (ID). To this end, the solution aliquots
of each sample were spiked and equilibrated with enriched tracers of
105
Pd and 109Ag. Both elements where then isolated from the geological matrix using an extended version of the anion exchange protocol
(section 2.1.5). With this, the Ag fractions of the samples were first
2157
eluted exactly as described above. The columns were then washed with
15 mL of 10 mol/L HCl (which can be discarded) and another 30 mL
of 10 mol/L HCl for the elution of the majority of the Pd. No further
purification of these elemental fractions was performed for the ID
concentration measurements. Instead, the fractions were evaporated to
dryness, dried down several times with a few drops of concentrated
HNO3, and then doped with appropriate amounts of Pd (for the Ag
fractions) or NIST SRM 978a Ag (for the Pd fractions). The ID
measurements were then performed with a Nu Plasma MC-ICPMS
using external normalization to the admixed elements with the exponential law for mass bias correction. The mass spectrometric techniques
were otherwise very similar to those that were applied for isotopic
analyses of Ag.
Cross-contamination between spiked and unspiked sample aliquots
must be avoided as this could readily generate large apparent isotopic
effects. Thus, the spiked and unspiked solutions were (1) handled in
different laboratories with different sets of lab-ware and reagents and
(2) analyzed with different MC-ICPMS instruments and sample introduction systems.
The precision and accuracy of the ID concentration measurements
were estimated by analysis of two spiked dissolutions of 0.1 g of the
Allende chondrite (Allende 5 and 6). These analyses yielded 108Pd/
109
Ag ratios of 4.67 and 4.68, respectively, which indicates that the
reproducibility may be as good as ⫾1% or better (Table 1). The
108
Pd/109Ag results obtained for Allende with the ID technique display
good agreement with our other data (see above and Table 1) and the
results of Carlson and Hauri (2001). Good agreement between the two
methods was also observed for the enstatite chondrite Abee. Two unspiked
Abee samples yielded 108Pd/109Ag ratios of 2.10 and 1.93 whereas one
spiked sample was found to have 108Pd/109Ag ⫽ 1.93 (Table 1).
3. RESULTS AND DISCUSSION
3.1. Accuracy and Precision of Ag Isotope
Ratio Measurements
An average 107Ag/109Ag of 1.08048 ⫾ 0.00042 was measured in this study for NIST SRM 978a Ag, which is identical,
within error, to the less precise TIMS value of 107Ag/109Ag ⫽
1.0811 ⫾ 0.0017 determined by Chen and Wasserburg (1983).
Our result however, is higher than the value of 1.07916 ⫾
0.00052 reported by Carlson and Hauri (2001), who utilized a
Plasma 54 MC-ICPMS. This small apparent discrepancy between the MC-ICPMS data probably reflects differences in the
relative fractionation factors of Pd and Ag (and hence the mass
bias correction) as both studies utilized the same Pd reference
ratio. Due to such variations in the mass bias behavior, absolute
measured ratios are almost meaningless when external normalization is applied. The primary aim of the present technique is,
however, the accurate and precise determination of isotopic
differences between bracketing standards and unknown samples.
Several approaches were used to assess the accuracy and
precision with which samples can be analyzed. The simplest
technique used data obtained for the JMC Ag standard, which
was analyzed on many different analytical sessions relative to
NIST SRM 978a Ag. On average, the JMC Ag standard was
found to have an isotopic composition of ␧107Ag ⫽ ⫺1.5 ⫾ 0.4
(2␴; n ⫽ 25). As this difference was well defined and constant
over time, analyses of these 2 standards relative to each other
were conducted on a regular basis, as a useful test of machine
performance before analyzing unknown samples.
This simple approach, however, does not consider the problems that can affect the analyses of real geological samples.
These include, in particular, isotope fractionation during laboratory handling, spectral interferences, and matrix-induced
2158
S. J. Woodland et al.
mass fractionation effects. Thus, care must be taken that analytical artifacts, which appear to be natural differences in Ag
isotopic ratios, are not generated. Carlson and Hauri (2001)
noted that there was a systematic shift in their Ag isotope data
of ⬃⫹3 ␧107Ag, for standards, which had been processed in the
laboratory, relative to unprocessed pure standards. Ion exchange chromatography can generate isotope fractionation for
some elements if a full yield is not obtained and external
normalization does not correct for such effects. Column yields
in the experiments carried out by Carlson and Hauri (2001)
were often significantly ⬍100% and so it was not clear whether
the observed isotopic offset was due to mass fractionation on
the resin or generated by matrix effects (Carlson and Hauri,
2001; Carlson et al., 2001). In this study we attempted to find
the source of this systematic shift and several interesting observations were made.
Mixed standard solutions containing Ag and Pd, stored in
lightproof bottles, are unstable if any chloride is present, and
the Ag concentrations and isotope ratios were observed to
change with time, presumably due to the precipitation of AgCl.
In contrast, the concentrations of Pd and Ag in HNO3-based
solutions remained constant for several weeks. However, it was
observed (Woodland et al., 2003) that over time even mixed
Pd-Ag standards in HNO3 displayed altered Ag isotope compositions, with ␧107Ag increasing from 0 to ⫹12 over a period
of 2 mo. This does not appear to be due to slow oxidation
because Ag solutions that were treated with hydrogen peroxide
before analysis did not show any significant shifts in Ag isotope
compositions (Woodland et al., 2003). The observed fractionation may be caused by either changes in the speciation of Pd
and/or Ag, or gradual precipitation of Pd on the walls of
poly-organic bottles (Erik Hauri, personal communication)
Consequently, all analyses utilized samples and bracketing
standards that were diluted freshly on the day of use.
As matrix-induced mass fractionation effects can compromise the accuracy of Pd-normalized Ag isotope data (Carlson et
al., 2001), the consequences of adding different matrices to Ag
standards were investigated. Reasonably high concentrations of
various elements could be added to the Ag standards (e.g., 500
ppb Ba, 500 ppb Ru, 10 ppb U) without causing any significant
Ag isotope effects, as long as the elements were added as
solutions in HNO3. However, when HCl, or an elemental
standard made up in dilute HCl, was admixed to an Ag standard
solution, the measured Ag isotope ratios differed from the true
value by ⬃⫺2 to ⫺5 ␧107Ag. This observation prompted us to
modify the chemical separation procedure of Carlson and Hauri
(2001), such that Ag was eluted from the anion exchange
columns with dilute HNO3 rather than 9 mol/L HCl. By avoiding the use of HCl, this modification not only improved the Ag
yields but also appeared to eliminate matrix-induced shifts in
the measured Ag isotope data. Consequently, Ag standards that
were processed with the column chemistry using the modified
separation technique were found to have the same Ag isotope
composition as equivalent unprocessed standards.
A more adequate evaluation of our methods was obtained by
doping two geological samples (the Allende chondrite and a
ferromanganese crust) from which the natural Ag had been
removed by chromatography, with NIST SRM 978a Ag. These
doped samples were then processed with our standard ionexchange techniques. In both cases, the Ag isotopic composi-
JMC Ag Std
-1.5 +/- 0.4
Abee
-0.5 +/- 2.2
Allende
-1.1 +/- 1.8
SCO-1
-4
-3
-1.0 +/- 2.1
-2
-1
0
1
2
3
e 107Ag
Fig. 2. Ag isotope results for replicate analyses of the JMC Ag
standard, the chondrites Allende and Abee, and the USGS standard
rock SCO-1 (Cody Shale). The ␧107Ag values are calculated relative to
the NIST SRM 978a isotope standard. Replicate analyses of real
geological samples have a reproducibility of ⬃⫾2 ␧107Ag, whereas the
data for the JMC Ag standard (n ⫽ 25) are precise to ⬃⫾0.5 ␧107Ag.
tions of the doped samples were found to be identical, within
error, to the bracketing standards, which were solutions of
NIST SRM 978a Ag. This demonstrates that the ion-exchange
chemistry does not induce any Ag isotope fractionation and that
it efficiently separates Ag from other elements, such that matrix
effects and spectral interferences are rendered insignificant.
Given the complexity of the problems that can afflict Ag
isotopic measurements, the true precision of the analytical
technique can only be evaluated based on results obtained for
replicate analyses of geological samples. The USGS standard
rock SCO-1 (Cody Shale) was analyzed 10 times, where each
result is for a separate dissolution of between 0.5 and 1 g of
rock powder (Fig. 2, Table 3). The SCO-1 standard is rich in
organic-C and it contains ⬃140 ppb Ag, such that it is a good
proxy for carbonaceous chondrites. The Ag data collected for
SCO-1 have a reproducibility of ⫾2.1 ␧107Ag (2␴, n ⫽ 10),
Table 3. Data for terrestrial samples analyzed.
Terrestrial samples
␧107Ag
SCO-1 average of 10 analyses
␧107Ag (2␴ error)
Pyrite
Pyrite (repeat)
Ag ore
AGV 2 (andesite)
–1.0
2.1
–0.4
–1.1
1.9
–3.1
The USGS standard SCO I (Cody Shale) was dissolved and analysed
on 10 separate occasions over the course of 18 mo to assess reproducibility. The Ag ore was not chemically purified prior to analysis unlike
all other samples measured in this study.
Silver isotopic composition of low Pd/Ag meteorites
2159
which is about a factor of 4 worse than the precision of data
collected for pure Ag solutions. Separate dissolutions of the
two chondritic meteorites Allende and Abee were also analyzed
several times to assess the analytical precision and these results
displayed reproducibilities of ⫾1.8 (2␴, n ⫽ 5) and ⫾2.2 (2␴,
n ⫽ 3) ␧107Ag, respectively (Fig. 2, Table 2). Based on these
results, an uncertainty of ⫾2.0 ␧107Ag is assumed in the following for all geological samples that were analyzed in this
study. This is only slightly worse than that quoted by Carlson
and Hauri (2001) (⫾1.3 ␧107Ag), but their reproducibility was
based on data acquired for pure standard solutions rather than
results for geological samples.
3.2. Results for Terrestrial Samples
Four terrestrial geological samples were analyzed, the USGS
standard rocks SCO-1 and AGV-2 (an andesite), a pyrite and a
pure native Ag ore from Pribram, Czech Republic. Assuming
an uncertainty of ⫾2.0 ␧107Ag, the samples all have Ag isotope
compositions that are identical, within error, to the NIST SRM
978a Ag standard. This result may indicate that terrestrial Ag
isotope variations due to stable isotope fractionation may be
rather limited in many environments. The JMC Ag standard,
however, was found to have an Ag isotope composition that is
heavier by ⬃1.5 ␧107Ag compared to NIST SRM 978a (Fig. 2).
The origin of this small difference in isotope compositions is
unclear but it may reflect a stable isotope fractionation that
occurred during production of the pure Ag that was used to
prepare these standards.
3.3. Chondritic Meteorites
All chondritic meteorites analyzed in this study have very
similar Ag isotopic compositions and, considering the errors,
most are identical to the NIST SRM 978a Ag standard (Fig. 3,
Table 1). Furthermore, there are no systematic differences in
Ag isotope compositions between carbonaceous, ordinary, and
enstatite chondrites. The ALH 84081 sample was the only
analyzed meteorite that was a find and not observed to fall. This
meteorite also displays the lightest Ag isotope composition
measured but it is presently unclear whether this is due to
terrestrial alteration or coincidental.
With regard to the Allende leach experiment, it was determined that the first two fractions, Leaches A and B, contained
⬃69% and 31%, respectively, of the total measured Ag,
whereas the Ag content of Leach C was below the detection
limit. This demonstrates that highly resistant mineral phases do
not contain a significant fraction of the Ag budget of the
Allende chondrite. The Ag isotope composition of Leaches A
and B are identical within error to the results obtained for bulk
samples of Allende (Table 1). Carlson and Hauri (2001) likewise found no significant difference in Ag isotope composition
of Allende bulk rock and Allende leaches.
The average result of ␧107Ag, ⫺1.1 ⫾ 1.8 that was obtained in
this study for repeated analyses of Allende (Table 1, Figs. 2 and 3)
contrasts with the mean value determined by Carlson and Hauri
(2001), which is ␧107Ag, ⫹5.2. The reason for this discrepancy is
presently unclear, but it may reflect sample heterogeneity, even
though very similar 108Pd/109Ag ratios were measured in both
studies. Resolvable variations in Ag isotope composition (⬎2
Fig. 3. Ag isotope data for terrestrial samples, carbonaceous chondrites, ordinary chondrites, and enstatite chondrites, all normalized to
NIST SRM 978a. The error bars for the samples are ⫾2.0 ␧107Ag,
which is the typical 2␴ reproducibility obtained for Allende, Abee and
SCO-1. Where samples were analyzed in duplicate, average values are
plotted.
␧107Ag) are not expected in any case, for an initial 107Pd/108Pd of
2.4 ⫻ 10⫺5 (Chen and Wasserburg, 1996), given that all samples
had 108Pd/109Ag ratios of ⬍10 (Table 1). Our results are in accord
with this inference, because (1) most of the analyzed chondrites
display very similar Ag isotope compositions with ␧107Ag ⬇ 0
and (2) the chondrite data define a best-fit line in a Pd-Ag isochron
diagram that yields an initial 107Pd/108Pd of (5.5 ⫾ 8.0) ⫻ 10⫺5
(Fig. 4), which is in accord with the value of 107Pd/108Pd ⫽ 2.4 ⫻
Fig. 4. Pd-Ag isochron plot for the chondrites. The line fit provides
a slope corresponding to an initial 107Pd/108Pd of (5.5 ⫾ 8.0) ⫻ 10⫺5.
Error bars for the data points are ⫾2.0 ␧107Ag and ⫾7% for the
108
Pd/109Ag ratio.
2160
S. J. Woodland et al.
10⫺5 (Chen and Wasserburg, 1996). This and all other isochron
slopes were calculated with the Isoplot software of Ludwig (1991).
The chondrites analyzed cover a large range of compositions,
and the Ag abundances are observed to range from ⬃50 ng/g
for some ordinary chondrites to ⬎250 ng/g for enstatite chondrites. The carbonaceous chondrites Allende and Murchison
display intermediate Ag contents of between 80 and 113 ng/g.
Given that Ag is a moderately volatile element with a half-mass
condensation temperature of 952 K (Wasson, 1985, pp. 250 –
251), these variations will reflect volatile element depletion and
redistribution processes that occurred either in the early solar
nebular (e.g., due to partial evaporation and/or condensation;
Humayun and Clayton, 1995; Larimer and Anders, 1967;
Anders et al., 1976) or on meteorite parent bodies (e.g., by
mobilization during thermal metamorphism; Wood, 1967).
Such processes may generate mass dependent stable isotope
fractionations, and the observation that all samples display
identical Ag isotope compositions despite considerable differences in Ag concentrations is thus significant.
Of particular interest is a comparison of the new Ag isotope
results with the stable isotope data that were previously obtained for K, which is moderately volatile with a half-mass
condensation temperature of 1000 K, and the highly volatile
element Cd (half-mass condensation temperature of 430 K;
Wasson, 1985, pp. 250 –251). Humayun and Clayton (1995)
were unable to identify any K isotope differences amongst
various chondrites, achondrites and the Earth, even though they
display large differences in K concentrations. Based on this, it
was concluded that the primary nebular processes, which are
responsible for the variable volatile element contents of meteorite parent bodies, did not involve reactions (such as partial
evaporation into a vacuum) that generate large kinetic isotope
effects (Humayun and Clayton, 1995). The Ag isotope results
of this study for chondritic meteorites support this conclusion.
In contrast, Wombacher et al. (2003) observed that some ordinary chondrites display large Cd isotope fractionations (ranging
from ⬃⫺2‰ to ⫹3‰ per amu mass difference relative to a
terrestrial standard). The interpretation that these variations
reflect metamorphic redistribution of Cd on the meteorite parent bodies by partial evaporation and condensation (Wombacher et al., 2003) is supported by the constant Ag isotope
compositions for ordinary chondrites. Silver is significantly less
volatile than Cd, and as such it is unlikely to be mobilized by
the thermal processes that led to the redistribution (and associated isotope fractionation) of Cd.
3.4. Iron Meteorites
3.4.1. Results
The iron meteorite Grant (sample USNM 836) has previously been analyzed in the Ag isotope studies of Carlson and
Hauri (2001) and Chen and Wasserburg (1983). This sample
was therefore reanalyzed in the present investigation to evaluate the accuracy of our methods (Fig. 5, Table 2). The Grant
sulfide yielded an Ag isotopic composition of ␧107Ag, ⫹24.5.
This compares very well with the data of Carlson and Hauri
(2001) and Chen and Wasserburg (1983), who obtained results
of ␧107Ag, ⫹24.7 ⫾ 1.3 and ␧107Ag, ⫹25 ⫾ 22, respectively,
for Grant sulfides. The metal fraction of Grant was found to
Fig. 5. Pd-Ag isochron plot for the iron meteorite Grant (IIIB). Black
squares are the TIMS data of Chen and Wasserburg (1983) and Kaiser
and Wasserburg (1983), the gray circles are for the MC-ICPMS data of
Carlson and Hauri (2001) and the gray squares denote the results of this
study. Error bars are smaller than the symbols, when not shown. The
data for the metal samples (with high Pd/Ag ratios) display good
agreement. The sulfide data of all three studies completely overlap on
this plot.
display ␧107Ag, ⫹160.2 in this study. Again, this result is very
similar to the data of Carlson and Hauri (2001), who analyzed
two metal pieces and obtained ␧107Ag values of ⫹150.1 ⫾ 1.3
and ⫹148.2 ⫾ 1.3. Chen and Wasserburg (1983) analyzed a
single piece of Grant metal which displayed ␧107Ag of ⫹144 ⫾
11. When the data for Grant metal and sulfide obtained in this
study are plotted in a Pd-Ag isochron diagram, the slope of the
best-fit line yields an initial 107Pd/108Pd of 1.48 ⫻ 10⫺5. This
result is in excellent agreement with the initial 107Pd/108Pd
values of (1.61 ⫾ 0.70) ⫻ 10⫺5 and ⬃1.7 ⫻ 10⫺5, which were
calculated for Grant by Carlson and Hauri (2001) and Chen and
Wasserburg (1996), based on their data for metal and sulfide
phases. The combined data of this and previous studies thus
define a very precise initial 107Pd/108Pd ratio of (1.50 ⫾ 0.08)
⫻ 10⫺5 for Grant (Fig. 5). Taken together, these results provide
excellent evidence for the accuracy of our analytical methods.
Troilites from the iron meteorites Canyon Diablo (CDT) and
Toluca were cleaned using two different techniques (see section 2.1.3) and subsequently analyzed (Table 2). Both procedures were found to yield identical Ag isotope compositions for
the sulfides, which suggests that both cleaning procedures
provide adequate removal of Ag derived from terrestrial contamination.
The Ag isotope composition determined for Canyon Diablo
metal in this study was ␧107Ag, ⫹0.6. This result agrees well
with previous analyses of Canyon Diablo metal by Carlson and
Hauri (2001), who obtained ␧107Ag values of ⫹0.6 ⫾ 1.3 and
⫹1.3 ⫾ 1.3, for two separate dissolutions. Four separate pieces
of the Canyon Diablo troilite nodule CDT1 were analyzed and
they yielded an average value of ␧107Ag, ⫺0.8 ⫾ 2.5 (2␴; n ⫽
4, Table 2). Two dissolutions of the CDT2 nodule, displayed
more negative ␧107Ag values of ⫺5.8 and ⫺3.2 (Table 2). The
two CDTs that were previously analyzed by Carlson and Hauri
(2001) exhibited even lower ␧107Ag than those of the present
study, with results of ⫺11.7 and ⫺10.7. The Ag concentrations
of the CDTs determined here and by Carlson and Hauri (2001)
are observed to be fairly similar, with abundances of between
Silver isotopic composition of low Pd/Ag meteorites
2161
⬃330 and 630 ng/g. In contrast, the Pd concentrations of the
troilites vary by ⬎2 orders of magnitude, from ⬃1 to 400 ng/g,
such that the nodules exhibit highly variable Pd/Ag ratios. The
reason for this behavior is unclear but it may reflect incomplete
sulfide-metal segregation during cooling, such that some microscopic grains of Pd-rich metal (or other phases with high Pd
contents) were trapped as inclusions within the nodules.
The Toluca metal analyzed in this study has an average Ag
isotope composition of ␧107Ag, ⫹4.4 and the two dissolutions
of Toluca troilite yielded ␧107Ag values of ⫹5.9 and ⫹5.6. The
metal and sulfide phases of Toluca thus exhibit identical Ag
isotope compositions, within error, even though the metal is
characterized by a 108Pd/109Ag ratio that is ⬎3 orders of
magnitude larger (Table 2).
3.4.2. Discussion
Both Canyon Diablo and Toluca are group IAB iron meteorites thought to be derived from a volatile-undepleted parent
body. Their non-magmatic origin is supported by the presence
of angular chondritic clasts and troilite nodules. The meteorites
of this group may represent rapidly-cooled melt pools produced
during impact events either at separate locations on the same
asteroid or on separate but compositionally similar asteroids,
with a chondritic megaregolith (Choi et al., 1995; Wasson and
Kallemeyn, 2002).
The CDTs analyzed by Carlson and Hauri (2001) display
lower 108Pd/109Ag and lower ␧107Ag than the co-existing
metal. These systematics were thought to reflect in situ decay of
107
Pd and an initial abundance of 107Pd/108Pd ⫽ (2.39 ⫾ 0.26)
⫻ 10⫺5 was calculated from the slope of the metal-sulfide
isochron (Fig. 6a). This initial 107Pd/108Pd is in excellent agreement with the highest previously determined initial 107Pd abundance of 107Pd/108Pd ⫽ (2.40 ⫾ 0.05) ⫻ 10⫺5, which was
derived from an internal “metal” isochron of the Gibeon (IVB)
iron meteorite (Chen and Wasserburg, 1996). Importantly, the
CDT analyses of Carlson and Hauri (2001) gave the lowest
107
Ag/109Ag ratios yet measured for solar system materials
with ␧107Ag values of ⬃⫺11 (Fig. 6a). As the validity of this
result was further corroborated by the Pd-Ag isochron of the
Brenham pallasite, which passes through the low-␧107Ag CDT
compositions (Carlson and Hauri, 2001), it was concluded that
the CDT exhibits the most primitive (initial) Ag isotope composition of the solar system.
The present results demonstrate, however, that the interpretation of the CDT data may be more complicated than previously envisaged. This is because the CDTs analyzed in this
study have Ag isotopic compositions different from those determined by Carlson and Hauri (2001). This observation does
not necessarily invalidate the conclusion of Carlson and Hauri
(2001) that their CDTs record the most primitive Ag isotope
composition of the solar system. The differences between the
Ag isotope data for the CDTs are, however, unlikely to reflect
only analytical artifacts. Therefore, we must examine mechanisms that might have produced variations in ␧107Ag for sulfides of the same meteorite, and these are differences in the
timing of sulfide segregation, sample heterogeneities, stable
isotope fractionation processes, and late re-equilibration due to
diffusion.
Based on the present Pd-Ag data for Canyon Diablo troilite
Fig. 6. Pd-Ag isochron plots for the iron meteorites Canyon Diablo
(a) and Toluca (b). (a) The black diamonds and gray squares denote the
results of Carlson and Hauri (2001) and of this study, respectively. The
variable Ag isotope ratios that have been measured for different CDT
nodules are not correlated with differences in Pd/Ag. The initial 107Pd/
108
Pd values, which can be calculated for the sulfide-metal isochrons of
this study, are 2.3 ⫻ 10⫺6 (CDT1) and 8.4 ⫻ 10⫺6 (CDT2). These
results differ from the initial ratio derived from the CDT-metal isochron of Carlson and Hauri (2001). (b) The sulfide-metal isochron for
Toluca has a nearly horizontal slope because the two phases display
identical Ag isotopic compositions, within uncertainty.
and metal samples, initial 107Pd/108Pd ratios of (8.4 ⫾ 4.2) ⫻
10⫺6 to (2.3 ⫾ 4.2) ⫻ 10⫺6 can be calculated for Canyon
Diablo (Fig. 6a). These troilites display significantly higher
␧107Ag values (by up to ⫹11 ␧107Ag) than those analyzed by
Carlson and Hauri (2001) and this could reflect variations in the
timing of sulfide segregation. If the material that formed the
Canyon Diablo meteorite began as a sulfur-rich metallic liquid
that started to segregate sulfur blebs on cooling, the Ag isotopic
composition would be a function of the time of separation from
the high Pd/Ag metallic liquid reservoir. Assuming that the
Canyon Diablo metal (Table 2) is representative of the metallic
liquid from which the sulfides segregated, one can calculate
that CDT1 and CDT2 segregated ⬃13.2 and 6.4 Ma, respectively, after the troilites analyzed by Carlson and Hauri (2001).
Of course with this model the Pd/Ag of the metal has been
enhanced by sulfide loss relative to the original metallic liquid
reservoir. Therefore, these time-scales are minima. Considering
this, it seems unlikely that these sulfides give useful chronological information because the two CDT nodules analyzed in
this study must have formed in close vicinity to each other and
would not be expected to have such widely differing ages. This
2162
S. J. Woodland et al.
is particularly true because the Canyon Diablo meteorite is
presumed to have formed near the surface of an asteroidal body
where cooling is predicted to be very rapid (Wasson and
Kallemeyn, 2002).
It is conceivable that the Ag isotopic compositions of these
impact-generated irons reflect heterogeneities resulting from
variable mixing between the impactor and the parent-body
regolith. Wasson et al. (1980) postulated that different degrees
of equilibration occurred between the impact melts and the
unmelted residues of the chondritic mega-regolith. However, as
the group IAB precursor materials are assumed to be chondritic
(Choi et al., 1995) and the Ag isotopic compositions of chondrites
appear to be restricted to ␧107Ag values of 0 ⫾ 3 (Fig. 3), it is
difficult to envisage how mixing processes could generate the
strongly negative and variable ␧107Ag signatures of the CDT
nodules.
Stable isotope fractionation of Ag could occur between metal
and troilite (or other minor phases) either during the solidification of the parent body or at a later stage, for example during
cooling, thermal metamorphism, or heating as the result of
atmospheric entry of the meteorite to Earth. However, it appears unlikely that such processes have the potential to generate
relatively large shifts in Ag isotope compositions of ⬎1‰,
when the S isotope compositions of the CDTs remained constant to within ⬃⫾0.4‰ in ␦ 34S (Beaudoin et al., 1994). The
troilite nodules analyzed in this study were, furthermore, spatially separated by no more than 2 cm and yet they record
distinct Ag isotope compositions. This suggests (1) that equilibrium Ag isotope fractionation between metal and sulfide is
either negligible or did not occur at the time these phases
formed or (2) that extremely localized fractionation processes,
which are considered to be unlikely, must have occurred to
account for the Ag isotope systematics of the CDT nodules.
The most likely explanation, therefore, builds on the previous studies of Chen and Wasserburg (1990), who observed that
the Ag isotope data for sulfides of other iron meteorites (e.g.,
groups IVA and IVB) are very complex, such that they cannot
be interpreted as resulting solely from in situ decay of 107Pd. It
was suggested instead that the disturbed Pd-Ag systematics of
the troilites are due to postformational diffusion transport of Ag
during secondary heating events, possibly related to episodes of
shock and/or thermal metamorphism (Chen and Wasserburg,
1990). Such processes could have also affected the group IAB
iron meteorites. Late redistribution by diffusion can provide a
reasonable explanation for the observation that the Toluca
metal and troilite samples have, within error, identical Ag
isotope compositions, such that they define a nearly horizontal
isochron with a slope of (⫺1.6 ⫾ 3.0) ⫻ 10⫺6 even though the
phases are characterized by very different Pd/Ag ratios (Fig. 6b,
Table 2). The variable ␧107Ag values of the CDT can also be
explained by diffusion, if it is assumed that they were initially
characterized by highly negative ␧107Ag signatures (␧107Ag ⱕ
⫺11.7) that were modified to various extents by later additions
of radiogenic Ag (with high ␧107Ag) derived from the metal.
4. CONCLUSIONS
Silver isotope ratios have been analyzed to high precision
using admixed Pd to correct for the instrumental mass bias by
external normalization. Application of external normalization
together with standard-sample bracketing enables reproducibilities of ⫾0.5 ␧ and ⫾2 ␧ (2␴) to be achieved for Ag isotope
ratio measurements of pure standard solutions and geological
samples, respectively. The procedure of external normalization
does not correct for fractionations that occur in nature or during
sample processing and care must therefore be taken to avoid the
generation of isotopic artifacts in the laboratory.
Several terrestrial samples, chondrites, and the iron meteorites Grant, Canyon Diablo and Toluca have been analyzed. The
majority of the chondrites have the same isotopic composition
as the terrestrial Ag isotope standard NIST SRM 978a. This
indicates that stable isotope fractionation of Ag did not occur
within the inner solar system during condensation of the solar
nebular and chondrite parent body formation. Troilites from the
Canyon Diablo meteorite display variable Ag isotopic compositions, suggesting that either the sulfides segregated over periods of millions of years or, more likely, that complex open
system behavior was associated with redistribution of Ag between metal and sulfide phases. Toluca also shows possible
evidence of such processes because the high ␧107Ag values of
the troilites are likely to reflect transport of radiogenic Ag from
metal to sulfide. These results indicate that caution must be
exercised when the Pd-Ag decay system is utilized to derive
chronometric information on the formation of group IAB iron
meteorites.
Acknowledgments—The authors would particularly like to thank the
Smithsonian Institution, Washington DC, who kindly provided many of
the meteorite samples used in this work. In addition, we would like to
thank Erik Hauri and two anonymous reviewers whose comments
greatly improved the original manuscript. Thanks are also due to Rick
Carlson for many useful suggestions regarding silver isotope analyses.
Associate editor: R. J. Walker
REFERENCES
Anders E., Higuchi H., Ganapathy R., and Morgan J. W. (1976) Chem.
fractionation in meteorites—IX. C3 chondrites Geochim. Cosmochim. Acta 40, 1131–1139.
Beaudoin G., Taylor B. E., Rumble D., III, and Thiemens M. (1994)
Variations in the sulfur isotope composition of troilite from the
Canyon Diablo iron meteorite. Geochim. Cosmochim. Acta 58,
4253– 4255.
Carlson R. W., and Hauri E. H. (2001) Extending the 107Pd-107Ag
chronometer to low Pd/Ag meteorites with multicollector plasmaionization mass spectrometry. Geochim. Cosmochim. Acta 65,
1839 –1848.
Carlson R. W., Hauri E. H. and Alexander C. M. O. D. (2001) Matrix
induced isotopic mass fractionation in the ICP-MS. In Plasma
Source Mass Spectrometry: The New Millenium (eds. G. P. Holland
and S. D. Tanner), pp. 288 –297. Royal Society of Chemistry.
Chen J. H. and Wasserburg G. J. (1983) The isotopic composition of
silver and lead in two iron meteorites: Cape York and Grant.
Geochim. Cosmochim. Acta 47, 1725–1737.
Chen J. H. and Wasserburg G. J. (1990) The isotopic composition of
Ag in meteorites and the presence of 107Pd in protoplanets.
Geochim. Cosmochim. Acta 54, 1729 –1743.
Chen J. H. and Wasserburg G. J. (1996) Live 107Pd in the early solar
system and implications for planetary evolution. In Earth Processes: Reading the Isotopic Code, Geophysics Monograph 95 (eds.
A. Basu and S. R. Hart), pp. 1–20. American Geophysical Union.
Choi B.-G., Ouyang X., and Wasson J. T. (1995) Classification and
origin of IAB and IIICD iron meteorites. Geochim. Cosmochim.
Acta 59, 593– 612.
Silver isotopic composition of low Pd/Ag meteorites
Frueh A. J. and Vincent E. A. (1967) Silver. In Handbook of Geochemistry
II-4 (ed. K. H. Wedepohl), pp. 47-A-1– 47-O-2. Springer-Verlag, Berlin, Germany.
Hauri E. H., Carlson R. W., and Bauer J. (2000) The timing of core
formation and volatile depletion in solar system objects from highprecision 107Pd-107Ag isotope systematics. Lunar Planet. Sci. Abstr. 31, 1812.
Humayun M. and Clayton R. N. (1995) Potassium isotope cosmochemistry: Genetic implications of volatile element depletion. Geochim.
Cosmochim. Acta 59, 2131–214.
Kaiser T. and Wasserburg G. J. (1983) The isotopic composition and
concentration of Ag in iron meteorites and the origin of exotic
silver. Geochim. Cosmochim. Acta 47, 43–58.
Kelly W. R. and Wasserburg G. J. (1978) Evidence for the existence of
107
Pd in the early solar system. Geophys. Res. Lett. 5, 1079 –1082.
Larimer J. W. and Anders E. (1967) Chem. fractionations in meteorites.
II. Abundance patterns and their interpretation Geochim. Cosmochim. Acta 31, 1239 –1270.
Ludwig K. R. (1991) ISOPLOT: A plotting and regression program for
radiogenic-isotope data. Open-File Report 91-445. U.S. Geological
Survey.
Marèchal C. N., Télouk P., and Albarède F. (1999) Precise analysis of
copper and zinc isotope compositions by plasma source mass
spectrometry. Chem. Geol. 156, 251–273.
2163
Paar Physica application note; high pressure Asher (HPA) decomposition of platinum metals and ferrochromium.
Rehkämper M., Schönbächler M., and Stirling C. H. (2001) Multiple
collector ICP-MS: Introduction to instrumentation, measurement
techniques and analytical capabilities. Geostand. News. 25, 23– 40.
Wasson J. T. (1985) Meteorites. Freeman Press.
Wasson J. T. and Kallemeyn G. W. (2002) The IAB iron-meteorite
complex: A group, five subgroups, numerous grouplets, closely
related, mainly formed by crystal segregation in rapidly cooling
melts. Geochim. Cosmochim. Acta 66, 2445–2473.
Wasson J. T., Willis J., Wai C. M., and Kracher A. (1980) Origin of iron
meteorite groups IAB and IIICD. Z. Naturforsch. 35a, 781–795.
Wombacher F., Rehkämper M., Mezger K., and Münker C. (2003)
Stable isotope compositions of cadmium in geological materials
and meteorites determined by multiple collector-ICPMS. Geochim.
Cosmochim. Acta 67, 4637– 4652.
Wood J. A. (1967) Chondrites: Their metallic minerals, thermal histories and parent planets. Icarus 6, 1– 49.
Woodland S. J., Rehkämper M., Lee D.-C., and Halliday A. N. (2003)
High precision MC-ICPMS measurement of silver isotopic compositions. In Plasma Source Mass Spectrometry: Applications and
Emerging Technologies (eds. G. P. Holland and S. D. Tanner), pp.
338 –350. Royal Society Of Chemistry.