Contributions to Mineralogy and Petrology

, Volume 147, Issue 5, pp 549–564

U-Pb columbite-tantalite chronology of rare-element pegmatites using TIMS and Laser Ablation-Multi Collector-ICP-MS

Authors

    • Dept. of Earth SciencesThe Open University
    • Geological Survey of Canada
  • G. L. Foster
    • NERC Isotope Geosciences Laboratory
    • Department of GeologyLeicester University
  • R. L. Romer
    • GeoForschungsZentrum Potsdam
  • A. G. Tindle
    • Dept. of Earth SciencesThe Open University
  • S. P. Kelley
    • Dept. of Earth SciencesThe Open University
  • S. R. Noble
    • NERC Isotope Geosciences Laboratory
  • M. Horstwood
    • NERC Isotope Geosciences Laboratory
  • F. W. Breaks
    • Ontario Geological Survey
Original Paper

DOI: 10.1007/s00410-003-0538-y

Cite this article as:
Smith, S.R., Foster, G.L., Romer, R.L. et al. Contrib Mineral Petrol (2004) 147: 549. doi:10.1007/s00410-003-0538-y

Abstract

U-Pb isotopic analyses using TIMS and Laser Ablation-Multi Collector-ICP-MS were carried out on columbite-tantalite minerals from three suites of rare-element (Li, Cs, Ta) pegmatites in the Superior Province of Canada. Conventional TIMS analyses of these columbite-tantalite crystals produce scattered data and reverse discordance even after HF leaching of the grains prior to dissolution, possibly reflecting the incomplete removal of the open-system metamict segments during sample preparation. LA-MC-ICP-MS analyses of unleached, primary columbite free from inclusions and alteration give consistent and precise (<0.5%) Pb-Pb ages, demonstrating the utility of this approach. However, normal and reverse discordance is also observed in U-Pb data from LA-MC-ICP-MS analyses. This discordance represents either U-Pb mobilisation during recent weathering, sample preparation and/or an analytical artefact originating from variable elemental fractionation between U and Pb during ablation and ionisation that itself may have its origin in the contrasting metamictization of the dated columbite and the monazite standard used. Best age estimates of columbite from pegmatites in the Superior Province are; 2670±5 Ma for the Pakeagama Lake pegmatite, 2644±7 Ma for the Separation Rapids group, and 2665±8 Ma for the Mavis Lake group. The ages broadly show that the rare-element pegmatites are temporally synchronous with adjacent peraluminous granites.

Introduction

Columbite-tantalite minerals are potentially useful for dating the emplacement age of economically important rare-element pegmatites. Romer and Wright (1992) have shown that columbite-tantalite may accommodate considerable amounts of uranium (circa 500–1000 ppm) and exclude lead almost completely. The U-Pb systematics, however, are typically not straightforward; reverse discordance and excess scatter are very common (Romer and Smeds 1994, 1996, 1997). This complex behaviour could result from columbite-tantalite crystals containing uranium-rich inclusions, alterations and exsolutions, which are documented from several localities (e.g. Cerný et al. 1989; Romer et al. 1996; Tindle and Breaks 1998), and accumulated lattice damage (Romer 2003). Romer and Smeds (1994, 1996) suggested that metamict domains adjacent to U-rich inclusions and in areas with high concentrations of unexsolved uranium are susceptible to U-Pb mobility. The commonly observed reverse discordance of columbite-tantalite analyses could be attributed to loss of uranium from metamict domains (Romer et al. 1996), or alternatively, Pb-implantation and/or mobilisation within α-recoil damaged domains (Romer 2003).

In some instances, the discordance can be removed by leaching prior to analysis (Romer and Smeds 1996). However, if the metamict regions are not dissolved quantitatively or if there have been several disturbances to the U-Pb system and secondary Nb-Ta oxides formed, the data may show significant, non-analytical scatter. In such cases, TIMS analyses could yield intercept ages that do not reflect the true crystallisation age.

Obtaining columbite-tantalite crystals free from U-rich inclusions and/or intergrown secondary Nb-Ta oxides may be difficult or impossible in many pegmatites. The purpose of this study is to employ the spatial resolution of the LA-MC-ICP-MS technique as a method of directly analysing primary zones in columbite-tantalite, avoiding inclusions and alteration, to determine reliable crystallisation ages for rare-element pegmatites.

Rare-element pegmatites and columbite-tantalite mineral chemistry

Granitic pegmatites of the rare-element class are characterised by minerals enriched in the elements Li, Cs, Ta, Y, and Nb. In addition, concentration of non-haplogranite components (boron, phosphorus and fluorine) lowers the solidus temperature and thus magmatic crystallisation to the range 700–500°C (London 1984, 1986, 1987; London et al. 1993). The characteristic minerals of these pegmatites are petalite and spodumene (lithium aluminosilicates), albite, tourmaline, lepidolite (Li-rich mica), beryl, and columbite-tantalite (Nb-Ta oxide).

The columbite-tantalite series [(Fe,Mn) (Ta,Nb)2O6] are the most common Nb-Ta species in rare-element pegmatites. The end members ferrocolumbite (FeNb2O6), manganocolumbite (MnNb2O6) and manganotantalite (MnTa2O6) are all orthorhombic minerals as is ferrotantalite (Fe>Mn) (Ta>Nb)2O6). This latter species is not an end-member however, which is instead ferrotapiolite (FeTa2O6), a tetragonal species of the tapiolite series. All of these species can be represented on the columbite-tantalite quadrilateral, which is used to illustrate compositional trends in rare-element pegmatites (summary in Cerny and Ercit 1985). The common trend followed by columbite-tantalite minerals in rare-element pegmatites is broadly from Fe-, Nb-rich compositions to Mn-, Ta-rich compositions. These are interpreted as fractional crystallisation trends based on the geochemical evidence that Ta substitutes for Nb, and Mn for Fe during the crystallisation of a pegmatite magma (Cerný and Ercit 1985). The effect on the columbite-tantalite quadrilateral is to move from the ferrocolumbite field (lower left quadrant) to the manganotantalite field (upper right quadrant) with progressive chemical evolution.

Complex zoning patterns occur in columbite-tantalite, which can be seen to good effect using backscattered electron images. Variations in Nb and Ta content (inferred from mean atomic number variations) are readily observed as darker and lighter regions within individual crystals. Oscillatory zoning is considered to be a primary feature, produced by magmatic growth of columbite-tantalite (Cerny et al. 1992). In contrast, convolute and ‘patchy’ zoning are secondary features, which overprint the primary zoning (Cerny et al. 1992; Abella et al. 1995). These secondary zones are thought to be produced by an abundance of highly reactive residual fluids that subject pegmatite minerals to partial or complete replacement during the late stages of pegmatite formation (London 1986; Wood and Williams-Jones 1993).

Analytical techniques

Electron Microprobe

Chemical analyses and backscattered electron images were obtained on a wavelength dispersive Cameca SX100 electron probe microanalyser at The Open University. Samples were prepared as polished mineral mounts in epoxy resin. An operating voltage of 20 kV and probe current of 20nA were used for quantitative analyses. Following microprobe analysis, grains used for TIMS analysis were removed from the resin and crushed into small fragments. The grains used for LA-MC-ICP-MS analysis were retained in the resin blocks.

Thermal Ionisation Mass Spectrometry (TIMS)

All isotopic analyses using TIMS were performed at GFZ Potsdam, Germany, using the procedures described by Romer and Wright (1992) and Romer and Smeds (1996). All samples analysed by the TIMS method were leached in warm 20%HF for 20 minutes before ion exchange chemistry. Leaching removes surface contamination, inclusions such as feldspars and sulphides, and most importantly dissolves metamict parts of the columbite (Fig. 1). Columbite dissolves easily in hot 40%HF. Lead and uranium were separated using ion-exchange chromatography as described by Tilton (1973) and Manhès et al. (1978). Lead and uranium were loaded on separate single Re filaments using silica-gel emitter and H3PO4 (Gerstenberger and Haase, 1997) and analyzed at 1200°–1260°C and 1350–1400°C, respectively, on a Finnigan MAT262 multicollector massspectrometer. Pb was analyzed using static multi-collection on Faraday collectors, with 204Pb analysed using ion counting. Uranium was analysed using Faraday collectors.
Fig. 1

SEM images of columbite-tantalite grains after leaching in 20% HF for 20mins. a,b Sample SS134 Pakeagama Lake pegmatite. c Sample 99FWB74 Big Mack pegmatite, Separation Rapids group. Note the etching of fission tracks that reflect the distribution of U and, thus, the distribution of α-recoil induced damage in the crystal lattice

Laser Ablation-Multi-Collector-ICP-MS

The LA-MC-ICP-MS analyses were performed at the NERC Isotope Geosciences Laboratory (NIGL, Keyworth, UK) using the procedure outlined in Horstwood et al. (2003), which is briefly summarised here. In order to preserve the replacement zones in columbite-tantalite, grains used for LA-MC-ICP-MS analyses were not leached. After cleaning to remove surface contamination, grain mounts of columbite-tantalite were placed into the ablation cell (ca. 30 cm3) of the New Wave Research, Microprobe II, 266 nm, Nd:YAG laser ablation system, in an Ar environrnent. A large but weak laser beam (~80 μm in diameter, ~1 J/cm2) was briefly rastered over the surface targeted for analysis to remove any further surface contamination. A more powerful but smaller beam (approx. 30×20 μm, 35  J/cm2) was then rastered over an area of 35×40 μm on the surface of the target crystal at a speed of 30 μm/sec. A pit approximately 50×50×10 μm was excavated and the ablated material was carried in a stream of Ar gas to the ICP-MS source. The analytical procedure included a 1 minute on-peak-zero blank acquisition prior to firing the laser, an ion counter-Faraday gain measurement, an ion-counter monitor of the Hg level in the Ar carrier gas, a measurement of mass 204 by ion-counting, and an assessment of instrumental Pb and Pb-U mass bias using a mixed Tl-235U tracer solution (cf. Horn et al. 2000) that was simultaneously aspirated through the laser cell. When the laser was fired, Tl and Pb isotopes and 238U were measured simultaneously in the multiple Faraday array of the ThermoElemental Axiom MC-ICP-MS (9 faraday collectors—one wide faraday to measure 238U, and one axial ion counting detector). A single analysis typically takes less than 4 minutes to complete.

The ability of LA-MC-ICP-MS to generate precise and accurate (<1%) Pb-Pb isotopic data and ages for a number of mineral matrices has been convincingly demonstrated recently by Willigers et al. (2002), notably also using an Axiom MC-ICP-MS. Our analytical methodology is slightly different to that outlined in Willigers et al. (2002), in that we apply a common Pb correction based on the measured 204Pb signal and Stacey and Kramers (1975) model Pb. The precision and accuracy of the approach, as noted in Horstwood et al. (2003) is dependent on the 207Pb signal strength and varies from ca. 3% at 3 mV down to <0.1% at 100 mV. lt is important to note that no matrix correction is required and machine induced mass bias is corrected by using simultaneously aspirated Tl.

Within-run Pb/U fractionation at the site of ablation is typically removed by rastering the laser across the sample (Parrish et al. 1999; Li et al. 2001) to maintain focus at the sample surface and a constant particle size (Guillong and Gunther, 2002). Instrumental Pb and Pb-U mass bias is corrected using the mixed Tl-235U tracer solution (Horn et al. 2000). However, despite these precautions and corrections, relic fractionation must still be corrected for by comparison to a standard of known age (in this case a 554 Ma monazite; Horstwood et al. 2003) analysed under identical instrumental and ablation conditions. This fractionation is distinct from within-run fractionation that often characterises LA-MC-ICP-MS analyses and is termed ‘plasma-induced’ fractionation. The reproducibility of the 206Pb/238U ratio of the standard is quadratically added to the internal error of each analysis reflecting the uncertainty of this correction (typically ~2–3% 2σ). Due to the lack of a well-characterised columbite-tantalite ablation standard, a monazite standard was used in this study. lt has recently been suggested that the dominant cause of fractionation using a 266 nm Nd:YAG laser is the incomplete ionisation of large aerosol particles (Guillong and Günther 2002). Since the particle size distribution is a fundamental property of the laser wavelength, there exists the exciting possibility that exact matrix matches are not required to quantify plasma-induced Pb-U fractionation in LA-ICP-MS analyses. Horstwood et al. (2003) have shown that a monazite standard can adequately correct for U-Pb fractionation in zircon. However, should this not be the case for columbite-tantalite we would still expect the Pb/U ratio of repeated analyses of a columbite of exactly the same Pb/U ratio to cluster with the same precision as the ablation standard (providing similar intensity). The cluster will, however, be offset from concordia, and lie on a zero-age chord. In other words, the Pb/U data will be precise but inaccurate by a uniform amount. An important observation, however, is that although in the monazite standard rastering removed all within-run fractionation, a number of columbite analyses were fractionated. This fractionation was no more than 20% in any analysis (the Pb/U ratio increases with time) and was corrected for using the intercept method of Kosler et al. (2001). Nonetheless, this observation is indicative of contrasting behaviour between the columbite and monazite during laser ablation, the significance of which will be discussed later.

Geological setting of the rare-element pegmatites

The samples used in this study are taken from three pegmatite groups within the Superior Province of Canada (Fig. 2). This mid- to late-Archaean craton is comprised of a series of linear granite-greenstone terrains interpreted as remnants of island arcs, separated by metasedimentary terrains, which may be analogous to accretionary prisms (Langford and Morin 1976; Card and Ciesielski 1986). U-Pb geochronology has shown that crust-forming magmatic and tectonic events in the Superior Province began at 3.0Ga and culminated with the Kenoran Orogeny at 2.7 Ga (Corfu and Davis 1991). Peraluminous granites and associated rare-element pegmatites occur along subprovince boundaries and regional scale shear zones, where granulite-grade metamorphism led to partial melting of volcano-sedimentary sequences (Cerny and Meintzer 1988; Breaks and Moore 1992; Pan and Breaks 1997).
Fig. 2

Location of pegmatites sampled for U-Pb geochronology, subprovinces of the Superior Province after Card and Ciesielski (1986); Thurston (1991)

Separation Rapids pegmatite group

The Separation Rapids pegmatites include beryl- and petalite-bearing pegmatites hosted within the Separation Lake greenstone belt. The pegmatites share mineralogical and geochemical characteristics with the peraluminous, S-type granite Separation Rapids pluton (Tindle and Breaks 1998, 2000). However, U-Pb columbite dating has shown that some regional rare-element pegmatite fields are unrelated to their assumed parental granite (Romer and Smeds 1996). Thus, one aim of this study is to determine if the Separation Rapids pluton and rare-element pegmatites were synchronous. The pluton has a U-Pb monazite age of 2646±2 Ma (Larbi et al. 1999).

Four samples were chosen that represent the different mineralogical, chemical, and structural pegmatite types. Several pegmatitic segregations containing the assemblage quartz + K-feldspar + muscovite + albite±columbite, occur within the Separation Rapids pluton. A single euhedral columbite grain with dimensions of 1×3x5 mm was extracted from a pegmatitic assemblage at the southwest margin of the pluton (sample 96–86b).

Tindle and Breaks (1998, 2000) have divided the Separation Rapids pegmatite group into two chemical suites, viz. Fe-suite and Mn-suite, based on the occurrence of Fe- and Mn- varieties of columbite-tantalite minerals. Sample 94–24a is a single euhedral columbite (1 mm across) collected from an undeformed Fe-suite, beryl-bearing pegmatite. Marko’s pegmatite is a Mn-suite, petalite pegmatite that is also undeformed. Euhedral columbite grains (<1 mm) occur as inclusions within muscovite in the wall zone of Marko’s pegmatite (sample 94–44).

Localised ductile deformation has affected some of the pegmatites within the Separation Rapids group. A characteristic example is the Big Mack pegmatite, a large petalite pegmatite (30 by 250 m) that displays isoclinal folding and recrystallisation of original pegmatite minerals (Breaks et al. 1999a). Sample 99FWB74 is a 1×1×10 mm single grain, collected from a recrystallised quartz+petalite assemblage within the core zone.

Pakeagama Lake pegmatite

The Pakeagama Lake pegmatite is located within the North Spirit Lake greenstone belt, which forms part of the North Caribou Terrain (Thurston, 1991; Fig. 2). The pegmatite has exposed dimensions of 50×250 m. Breaks et al. (1999b) have described the mineral assemblage, which includes: beryl, petalite and spodumene (lithium aluminosilicates), columbite-tantalite, tourmaline, and lithium mica species. Heterogeneous brittle—ductile deformation parallel to the regional foliation (approximately NW—SE) is associated with shear-sense movement along the nearby Bear Head fault. Sample SS137 was collected from the K-feldspar+petalite zone, which forms the majority of the pegmatite body. This is a euhedral grain of tantalite with dimensions of 6×4×2 mm, which occurred as an inclusion within a petalite megacryst. Sample SS134 is a euhedral tantalite grain with dimensions of 4×2×2 mm, collected from a lepidolite-rich vein that cross-cuts the spodumene+quartz core zone. These veins contain an assemblage of Li-rich mica (lepidolite), quartz and Nb-Ta oxides, interpreted as part of the latest stages of crystallisation from melt+F-rich fluid (Breaks et al. 1999b; Smith 2001).

Mavis Lake group

The Mavis Lake pegmatites are spatially and chemically associated to the Ghost Lake batholith, a peraluminous S-type granite (Breaks and Moore 1992). The Ghost Lake batholith and Mavis Lake group pegmatites are situated between the greenstone-rich Wabigoon Subprovince and the tonalite-gneiss dominated Winnipeg River Subprovince (Fig. 2). The emplacement age of the Ghost Lake batholith is constrained by a U-Pb monazite date of circa 2685 Ma (D. Davis, unpublished data; reported in Breaks and Moore 1992). The Mavis Lake group includes Fairservice pegmatite #1, with exposed dimensions of 12 by 76 m. Sample 92DF-5 is a 3×2×2 mm columbite grain selected for U-Pb isotopic analysis, taken from the primary assemblage, spodumene-albite-quartz unit of Fairservice pegmatite #1.

Results

Separation Rapids pegmatite group

Pegmatitic granite (Sample 96–86b)

A single euhedral columbite grain (1×3×5 mm) was analysed from one of the pegmatitic segregations within leucogranite of the Separation Rapids pluton. Electron microprobe analyses (Table 1, Fig. 3) show the relatively primitive ferrocolumbite composition of 96–86b and backscattered electron images (Fig. 4a) reveal a lack of zoning in the grain and the presence of small inclusions of cassiterite.
Table 1

Representative electron microprobe data of columbite-tantalite minerals, structural formulae based on 6 oxygens

Separation Rapids pegmatites

Pakeagama Lake pegmatite

Mavis Lake group

pegmatitic granite

Fe-suite

Marko’s peg. Mn-suite

Big Mack peg.

Kspar+petalite zone

vein cross-cutting core zone

Fairservice peg. #1

Sample

96-86B

96-86B

94-24A

94-24A

94-44

94-44

99FWB74

SS137

SS137

SS134

SS134

92DF-5

92DF-5

CaO

0.00

0.00

0.03

0.03

0.00

0.00

0.04

0.00

0.02

0.01

0.02

0.02

0.06

0.01

FeO

14.03

10.04

13.15

13.14

7.45

7.73

12.86

13.87

1.02

0.94

0.74

0.57

5.71

7.37

MnO

6.28

10.28

6.90

7.28

10.11

9.90

3.35

3.78

14.65

14.01

15.34

14.42

11.78

10.06

TiO2

0.63

0.42

1.21

1.21

2.50

2.94

0.54

0.30

0.10

0.00

0.02

0.00

0.94

0.60

Nb2O5

63.81

63.05

63.96

70.61

51.42

49.92

24.15

37.72

26.55

17.89

27.32

17.50

44.59

37.86

Ta2O5

13.76

15.01

11.23

7.39

25.16

26.27

58.22

42.91

57.30

66.03

56.34

67.85

34.06

40.77

SnO2

0.26

0.00

0.03

0.02

0.79

0.34

0.44

0.03

0.16

0.16

0.16

0.14

0.57

0.17

WO3

0.35

0.08

0.51

0.63

1.08

1.25

0.58

0.37

0.19

0.24

0.18

0.19

0.07

0.18

PbO

0.48

0.58

0.30

0.29

0.27

0.55

0.12

0.15

0.18

0.10

0.11

0.09

0.24

0.19

UO2

0.27

0.00

0.00

0.02

0.27

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Sb2O3

0.00

0.01

0.00

0.00

0.00

0.00

0.00

0.00

0.01

0.02

0.02

0.02

0.00

0.00

Bi2O3

0.00

0.00

0.00

0.00

0.05

0.45

0.00

0.00

0.00

0.00

0.00

0.00

0.00

0.00

Sc2O3

0.00

0.00

0.00

0.00

0.61

0.68

0.10

0.00

0.00

0.04

0.00

0.00

0.03

0.00

Total

99.88

99.47

99.99

100.62

99.71

100.02

100.41

99.14

100.18

99.45

100.23

100.79

98.03

97.21

Ca

0.000

0.000

0.002

0.002

0.000

0.000

0.003

0.000

0.001

0.001

0.001

0.002

0.004

0.000

Fe

0.703

0.507

0.646

0.631

0.388

0.404

0.784

0.796

0.062

0.060

0.044

0.036

0.317

0.428

Mn

0.319

0.526

0.343

0.354

0.534

0.524

0.207

0.220

0.900

0.911

0.938

0.929

0.662

0.592

Ti

0.028

0.019

0.053

0.052

0.117

0.138

0.030

0.016

0.005

0.000

0.001

0.000

0.047

0.032

Nb

1.727

1.720

1.767

1.834

1.449

1.411

0.796

1.170

0.871

0.621

0.891

0.601

1.338

1.188

Ta

0.224

0.246

0.181

0.115

0.426

0.446

1.154

0.801

1.130

1.378

1.106

1.403

0.615

0.770

Sn

0.006

0.000

0.001

0.000

0.020

0.008

0.013

0.001

0.005

0.005

0.005

0.004

0.015

0.005

W

0.005

0.001

0.008

0.009

0.018

0.020

0.011

0.007

0.004

0.005

0.003

0.004

0.001

0.003

Pb

0.008

0.009

0.005

0.004

0.005

0.009

0.002

0.003

0.003

0.002

0.002

0.002

0.004

0.004

U

0.004

0.000

0.000

0.000

0.004

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Sb

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.001

0.001

0.001

0.000

0.000

Bi

0.000

0.000

0.000

0.000

0.001

0.007

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Sc

0.000

0.000

0.000

0.000

0.033

0.037

0.007

0.000

0.000

0.003

0.000

0.000

0.002

0.000

Total

3.024

3.029

3.007

3.004

2.994

3.005

3.007

3.013

2.981

2.986

2.992

2.981

3.005

3.021

Mn/(Mn+Fe)

0.312

0.509

0.347

0.359

0.579

0.565

0.209

0.216

0.936

0.938

0.955

0.963

0.676

0.580

Ta/(Ta+Nb)

0.115

0.125

0.093

0.059

0.227

0.240

0.592

0.406

0.565

0.689

0.554

0.700

0.315

0.393

Fig. 3

Compositions of columbite-tantalite from the Pakeagama Lake pegmatite, Separation Rapids pegmatites and Fairservice pegmatite #1, Mavis Lake group. Fractional crystallisation produces increasing Mn/(Mn+Fe) and Ta(Ta+Nb)

Fig. 4a-d

Backscattered electron images of columbite-tantalite grains from the Separation Rapids pegmatite group. a Sample 96–86b ferrocolumbite from a pegmatite pod, bright inclusions are cassiterite. b Sample 94–24a, ferrocolumbite (Fe-suite). c Sample 94–44 manganocolumbite from Marko’s pegmatite (Mn-suite). d Sample 99FWB74 ferrotantalite from the deformed Big Mack pegmatite (Fe-suite), note the lack of zoning.

All ages and concordia diagrams were produced using the Isoplot/Ex program v.2.22 (Ludwig 2000); all age uncertainties are at the 2σ level. Four fractions of columbite 96–86b were analysed by TIMS (Fig. 5a). Concentrations of U are uniformly low, 59–71 ppm (Table 2). Two reversely discordant fractions and one concordant fraction lie along a discordia (MSWD=0.22) with an upper intercept of 2629.1±1.5 Ma and a negative lower intercept of −303±250 Ma. The fourth fraction plots away from the discordia, and has been omitted from the regression. The intercept age is younger than that of the host Separation Rapids pluton (2646±2 Ma, Larbi et al. 1999). The anomalous lower intercept age, indicating that the columbite U-Pb system was disturbed after crystallisation, may be due to fractionation of 206Pbrad and 207Pbrad during leaching, which could reflect effects of recoil implantation and preferential mobilization from structurally damaged areas (e.g. Mattinson 1997), as well as the thermal history of the rock that takes its expression in the preferential annealing of early formed damages over later ones (cf. Davis and Krogh 2000; Romer 2003).
Fig. 5a-d

Concordia plots of TIMS analyses for columbite-tantalite grains from the Separation Rapids pegmatite group. a Sample 96-86b, fraction 901 is omitted from the regression. b Sample 94-24a, fraction 880 is omitted from the regression. c Sample 94–44, fractions 896 and 898 are omitted from the regression. d Sample 99FWB74.

Table 2

U-Pb isotopic data from TIMS analyses

Sample

Run

Weight

U

Pb

Th/Ua

measured ratios

atomic ratiosb

apparent ages

(mg)

ppm

ppm

206Pb/204Pb

207Pb/206Pb

208Pb/206Pb

206Pb/238U

207Pb/235U

207Pb/206Pb

206Pb/238U

207Pb/235U

207Pb/206Pb

Separation Rapids pegmatite group

96-86b

901

0.111

62.4

33.4

0.024

2732

0.1805

0.0188

0.5177

12.583

0.1763

2690

2649

2618

±1

96-86b

902

0.074

67.3

35.4

0.028

6462

0.1788

0.0128

0.5142

12.560

0.1771

2675

2647

2626

±2

96-86b

903

0.231

70.6

37.5

0.037

8649

0.1784

0.0140

0.5176

12.643

0.1771

2689

2653

2626

±1

96-86b

904

0.045

59.3

30.6

0.035

3313

0.1807

0.0196

0.5050

12.353

0.1774

2635

2632

2629

±1

94-24a

877

0.078

511

272

0.034

23485

0.1800

0.0108

0.5190

12.858

0.1797

2695

2669

2650

±1

94-24a

878

0.040

447

240

0.034

12450

0.1805

0.0120

0.5233

12.966

0.1797

2713

2677

2650

±2

94-24a

879

0.099

419

227

0.030

16440

0.1799

0.0104

0.5287

13.074

0.1794

2736

2685

2647

±1

94-24a

880

0.079

215

115

0.039

11530

0.1793

0.0135

0.5228

12.865

0.1785

2711

2670

2639

±1

94-44MU5

893

0.068

1640

732

0.047

19550

0.1783

0.0145

0.4341

10.644

0.1779

2324

2493

2633

±1

94-44MU5

894

0.167

1560

702

0.050

53500

0.1785

0.0144

0.4383

10.783

0.1784

2343

2505

2638

±1

94-44MU5

895

0.224

1290

698

0.043

29420

0.1793

0.0130

0.5280

13.034

0.1790

2733

2682

2644

±1

94-44MU5

896

0.044

1620

870

0.045

54660

0.1784

0.0130

0.5231

12.862

0.1783

2712

2670

2637

±1

94-44MU5

897

0.737

1180

695

0.054

35000

0.1801

0.0157

0.5733

14.225

0.1799

2922

2765

2652

±1

94-44MU5

898

0.043

1690

951

0.049

60430

0.1813

0.0141

0.5459

13.648

0.1813

2808

2726

2665

±1

99FWB74

871

0.149

34.7

19.3

0.069

990

0.1853

0.0522

0.5157

12.309

0.1731

2681

2628

2588

±3

99FWB74

873

0.639

11.2

5.88

0.033

3014

0.1797

0.0201

0.5090

12.353

0.1760

2653

2632

2616

±2

99FWB74

874

0.105

40.9

21.7

0.046

1865

0.1814

0.0303

0.5084

12.272

0.1751

2650

2625

2607

±2

99FWB74

876

0.099

45.4

23.4

0.036

2709

0.1789

0.0220

0.4987

12.009

0.1746

2608

2605

2603

±3

Pakeagama Lake pegmatite

SS134

882

0.060

129

66.7

0.009

10030

0.1802

0.0058

0.5079

12.548

0.1792

2648

2646

2645

±1

SS134

883

0.038

114

59.6

0.011

6814

0.1815

0.0078

0.5126

12.726

0.1801

2668

2660

2653

±1

SS134

885

0.070

564

286

0.007

24570

0.1787

0.0032

0.4989

12.271

0.1784

2609

2625

2638

±1

SS134

886

0.083

55.2

29.0

0.015

5340

0.1827

0.0103

0.5136

12.794

0.1807

2672

2665

2659

±1

aTh/U are atomic ratios.

bLead corrected for fractionation, 15 pg lead blank, and tracer contribution and initial ratios 206Pb/204Pb=13.9, 207Pb/204Pb=14.8, 208Pb/204Pb=33.6. Uncertainties of Pb isotope ratios ratios not involving 204Pb and 205Pb are less than 0.1%. For ratios involving 205Pb and 204Pb the uncertainty is typically less than 0.25% for 206Pb/205Pb and less than 0.3–0.5% for 206Pb/204Pb. Data reduction took into consideration the following uncertainties: measurement errors, 30% for fractionation correction, 50% for blank level, 0.1, 0.05, 0.2 for 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb blank composition, respectively, and 205Pb/206Pb=21.693 for isotopic tracer composition. The 205Pb/235U uncertainty of the tracer is 0.3%. Data reduction was performed by Monte Carlo modelling of 1000 random normally distributed data sets that fit above uncertainty limits, allowing for error correlation when appropriate. Uranium corrected for fractionation and 1 pg uranium blank.

Fe-suite pegmatite (Sample 94–24a)

This sample is a single euhedral grain (1 mm across) from a beryl pegmatite. Electron microprobe data plots within the ferrocolumbite field indicating that the pegmatite is part of the Fe-suite (Table 1, Fig. 3). Backscattered electron imaging reveals oscillatory zoning, indicating primary magmatic growth (Fig. 4b). There are no inclusions or any evidence for alteration, although the grain contains abundant fractures and micro-cracks. These features precluded LA-MC-ICP-MS analysis as they were unavoidable and increased the common Pb content of the analyses to unacceptable levels (common Pb constituting up to 98% of the measured 207Pb).

Four fractions of columbite sample 94–24a were analysed by TIMS, all fractions are reversely discordant (Fig. 5b). Fractions 877, 878 and 879 define a discordia with an upper intercept at 2653±36 Ma (MSWD=2.5). The 207Pb/206Pb ages of these three fractions range from 2647 to 2650 Ma and define an error weighted mean age of 2649±5 Ma. We interpret this weighted mean to be the emplacement age of the pegmatite. The three fractions have U contents >400 ppm, fraction 877 has the highest U concentration (511 ppm) and is the closest to concordia. A fourth fraction with a comparatively low U content (215 ppm; Table 2) lies off this discordant array, and has a younger apparent 207Pb/206Pb age of 2638.8±1 Ma.

Mn-suite pegmatite (Sample 94–44)

Marko’s pegmatite is a Mn-suite petalite pegmatite. Euhedral columbite grains (<1 mm) occur as inclusions within muscovite books in the wall zone of Marko’s pegmatite. Monazite is also present and occurs as small (<500 μm) inclusions within columbite, which were removed during crushing and hand-picking. Electron microprobe analyses (Table 1, Fig. 3) plot within the manganocolumbite field and backscattered electron images (Fig. 4c) show the relatively homogeneous composition of selected grains.

Six fractions, from three grains of columbite were analysed by the TIMS method. Two fractions are normally discordant and four fractions show variable reverse discordance (Fig. 5c). A regression line using all six fractions (not shown) has a very large scatter (MSWD=232) with upper and lower intercepts at 2646±12 and 186±260, respectively. The excess scatter is mainly due to two fractions (896 and 898) that plot away from the discordia. Excluding these two fractions from the regression yields an upper intercept age of 2644.3±7.1 Ma and lower intercept at 144±140 Ma (Fig. 5c). The upper intercept is interpreted as the emplacement age of Marko’s pegmatite, although there is still significant excess scatter (MSWD=29). Concentrations of U are high and quite variable, between 1180 ppm to 1690 ppm, although there is no correlation between U concentration and discordance.

Deformed pegmatite (Sample 99FWB74)

Electron microprobe data from sample 99FWB74 (Big Mack pegmatite) plots within the ferrocolumbite and ferrotantalite fields (Table 1, Fig. 3), which is a relatively primitive composition for such an evolved petalite pegmatite. Backscattered electron images reveal that the grain fragments lack compositional zoning (Fig. 4d). Together, these observations suggest that the deformation and recrystallisation has homogenised the composition of primary columbite-tantalite (cf. Cerny et al. 1992).

TIMS analysis of four fractions from a single grain (1×1×10 mm) were carried out. All fractions contain low concentrations of U (<50 ppm; Table 2). Fraction 876, which contains the highest U content (45 ppm), is concordant with a 207Pb/206Pb age of 2603±2 Ma (Fig. 5d). Three remaining fractions are reversely discordant and show considerable scatter with 207Pb/206Pb ages of 2616±2 Ma, 2607±2 Ma and 2588±2 Ma (Table 2). These data suggest that sample 99FWB74 has undergone multiple fractionation of the U-Pb system, perhaps related to the deformation event, with an overall loss of uranium. The large range of apparent 207Pb/206Pb ages possibly reflects redistribution and fractionation of U and Pb into different recrystallized phases. The emplacement age of the Big Mack pegmatite is not known.

Pakeagama Lake Pegmatite

Primary assemblage (Sample SS137)

Electron microprobe data show the evolved manganotantalite composition of sample SS137 (Table 1, Fig. 3), indicating the high concentration of rare-elements within the Pakeagama Lake pegmatite. Backscattered electron images of the manganotantalite grain reveal oscillatory zoning indicating primary magmatic growth (Fig. 6a).
Fig. 6a,b

Sample SS137, tantalite grain from the primary assemblage core zone of the Pakeagama lake pegmatite. a Backscattered electron image with locations of rasters. b Concordia plot of LA-MC-ICP-MS analyses

Six raster analyses of the oscillatory-zoned grain using LA-MC-ICP-MS yield Pb-Pb ages that range from 2661.5±13.6 Ma to 2674.2±10.0 Ma (Table 3), with a weighted mean of all six of 2669.6±5.4 Ma (MSWD=2.3). We interpret this weighted mean to be the age of primary tantalite and the emplacement age of the Pakeagama Lake pegmatite.
Table 3

U-Pb isotopic data from LA-MC-ICP-MS analyses

Sample

Uppmb

%f207

Atomic Ratiosa

Apparent Ages (Ma)

207Pb/206Pb

1σ%

206Pb/238U

1σ%

207Pb/235U

1σ%

rho

206Pb/238U

2σ(Ma)

207Pb/235U

2σ(Ma)

207Pb/206Pb

2σ(Ma)

Pakeagama Lake pegmatite

SS137-1

25

2.2

0.1809

0.29

0.6304

1.7

15.72

1.7

0.99

3151

108

2860

100

2661

10

SS137-2

41

0.7

0.1815

0.17

0.6006

1.8

15.03

1.8

0.99

3032

106

2817

99

2667

6

SS137-3

63

1.4

0.1809

0.49

0.5423

1.8

13.53

1.9

0.96

2793

100

2717

101

2661

16

SS137-4

58

1.0

0.1823

0.49

0.5581

1.9

14.03

2.0

0.97

2859

110

2752

110

2674

16

SS137-5

40

0.5

0.1819

0.24

0.5751

1.7

14.43

1.7

0.99

2929

100

2778

96

2671

8

SS137-6

30

0.1

0.1823

0.23

0.5640

1.8

14.17

1.8

0.99

2883

101

2761

98

2673

8

SS134-1

22

2.4

0.1818

2.11

0.6511

2.1

16.32

3.0

0.71

3233

137

2896

173

2669

70

SS134-2

45

4.8

0.1805

0.18

0.7067

1.5

17.59

1.5

0.99

3446

101

2967

88

2658

6

SS134-3

43

4.0

0.1809

0.19

0.5408

1.3

13.49

1.4

0.99

2787

75

2715

74

2661

6

SS134-4

71

6.8

0.1812

0.49

0.6215

1.9

15.53

2.0

0.97

3116

119

2848

122

2664

16

SS134-5

60

5.7

0.1817

0.49

0.5805

1.6

14.54

1.6

0.95

2951

93

2786

92

2668

16

SS134-6

24

2.5

0.1779

2.03

0.6917

2.1

16.96

3.0

0.72

3389

145

2933

173

2633

68

SS134-7

24

2.4

0.1793

2.08

0.6517

1.7

16.11

2.7

0.62

3235

107

2883

153

2646

69

SS134-8

24

2.2

0.1812

0.34

0.5167

1.0

12.91

1.1

0.95

2685

55

2673

57

2664

11

SS134-9

12

1.0

0.1774

3.39

0.4956

1.3

12.12

3.6

0.36

2595

67

2614

190

2629

113

Fairservice pegmatite #1

92DF-5-1

53

2.2

0.1810

0.48

0.7579

1.7

18.91

1.8

0.96

3636

124

3037

107

2662

16

92DF-5-2

58

0.7

0.1817

0.48

0.7494

1.7

18.77

1.8

0.96

3605

126

3030

110

2668

16

92DF-5-3

88

1.4

0.1822

0.46

0.6222

1.8

15.63

1.9

0.97

3119

113

2854

107

2673

15

92DF-5-4

123

1.0

0.1804

0.45

0.6053

2.1

15.06

2.1

0.98

3051

127

2819

120

2657

15

92DF-5-5

48

0.5

0.1807

0.48

0.7784

1.8

19.40

1.8

0.96

3711

130

3062

111

2660

16

92DF-5-6

111

0.1

0.1795

0.45

0.5524

2.6

13,67

2.6

0.99

2835

147

2727

143

2648

15

92DF-5-7

18

1.0

0.1711

2.87

0.4785

2.9

11.29

4.1

0.72

2521

148

2547

209

2568

96

92DF-5-8

1112

2.9

0.1338

0.46

0.0475

6.4

0.88

7.1

0.91

299

38

638

90

2148

16

92DF-5-9

163

3.4

0.1542

0.49

0.1490

8.2

3.17

8.8

0.92

895

146

1449

256

2393

17

92DF-5-10

29

5.0

0.1731

1.85

0.3845

3.4

9.18

3.9

0.88

2097

144

2356

184

2588

62

92DF-5-11

33

4.2

0.1598

2.19

0.3868

2.9

8.52

3.6

0.80

2108

122

2288

166

2454

74

92DF-5-12

27

9.1

0.1612

2.76

0.3200

5.1

7.11

5.8

0.88

1790

184

2125

248

2468

93

92DF-5-13

49

3.1

0.1719

0.50

0.4882

4.5

11.57

4.5

0.99

2563

232

2570

234

2576

17

92DF-5-14

91

0.8

0.1779

0.46

0.6898

1.8

16.92

1.8

0.97

3382

119

2930

106

2633

15

92DF-5-15

120

0.1

0.1808

0.45

0.6765

1.7

16.87

1.8

0.97

3331

115

2927

104

2661

15

92DF-5-16

161

0.7

0.1799

0.43

0.5660

2.4

14.04

2.4

0.98

2891

137

2752

133

2651

14

aCorrected for common Pb and Pb/U fractionation

bAccuracy of U concentration ~10%.

c%f207 = percentage of measured 207Pb that is common Pb

In terms of U-Pb, five of the raster analyses are variably reversely discordant, and one analysis (raster 4) is normally discordant (Fig. 6b). All six analyses yield a discordia (MSWD=1.02) with an upper intercept of 2673±8 Ma and a lower intercept of −143±210 Ma. This array with an essentially zero-aged lower intercept indicates a recent disturbance to the Pb-U system of the tantalite or analytically produced Pb/U fractionation. This will be examined further in the discussion.

Cross-cutting vein (Sample SS134)

A single grain (4×2×2 mm) from a late-stage vein that cross-cuts the core zone of the Pakeagama Lake pegmatite was split into fragments for TIMS and LA-MC-ICP-MS analyses. Backscattered electron images reveal an oscillatory-zoned core that is partly replaced by a darker zone, which in turn is surrounded by a bright, compositionally distinct rim (Fig. 7a). Electron microprobe analyses confirm the compositional difference as an increase in Ta/(Ta+Nb) within the manganotantalite field (Table 1, Fig. 3). Clearly, more than one event was responsible for the zoning in sample SS134. The oscillatory-zoned core is interpreted as a primary stage of magmatic crystallisation. The tantalum-rich rim may have crystallised from a residual melt+vapour phase within the lepidolite vein (cf. London 1987) or else developed from redistribution of Ta caused by a later disturbance event.
Fig. 7a–c

Sample SS134, tantalite grain from a lepidolite vein cross-cutting the core zone of the Pakeagama Lake pegmatite. a Backscattered electron image with location of raster analyses b Concordia plot of TIMS analyses. Numbers beside error ellipses refer to fraction label as given in Table 2. c Concordia plot of LA-MC-ICP-MS analyses

TIMS analyses of four fractions yield one concordant fraction, two fractions that are only slightly reversely discordant, and another is normally discordant (Fig. 7b). The concordant fraction lies at 2645±1 Ma. Fitting a discordia to all four fractions (MSWD=3.4) yields an upper intercept at 2649±11 Ma and a poorly defined lower intercept of 1263±920 Ma. The large uncertainties of the intercepts originate from the flat intersection of the discordia with the concordia and the clustering of the data close to the upper intercept (Fig. 7b). The 207Pb/206Pb ages of the four fractions scatter between 2659±1 Ma and 2638±1 Ma. They are inversely correlated with U content of the HF-leached fractions (Table 2), which may reflect that disturbance and Pb loss is strongly controlled by the amount of lattice damage accumulated.

Eight raster analyses of the oscillatory-zoned core using LA-MC-ICP-MS yield Pb-Pb ages that range from 2633.2±68 Ma to 2669.2±70 Ma, with a weighted mean of 2660.5±4 Ma (MSWD=0.5). U-Pb data from LA-MC-ICP-MS analyses are reversely discordant (Table 3, Fig. 7c). Eight raster analyses (nos. 1–8) from the oscillatory-zoned core yield a discordia (MSWD=0.31) with an upper intercept of 2663.8±6.4 Ma and a lower intercept of –58±92 Ma. Raster 9 is not used in the regression as this is not located within the oscillatory zoning. The error ellipse from raster 9 broadly lies across the concordia with a poorly defined 207Pb/206Pb age of 2629±113 Ma. Analysis of the Ta-rich rim of SS134 was precluded by a uranium signal (<50,000 cps), too low to be analyzed on a Faraday collector.

Comparing Figs. 7b and 7c there are some differences in the data derived by the two techniques. The TIMS data shows more scatter than the LA-MC-ICP-MS data that lie along a strongly discordant array. However, heterogeneities seen in the TIMS analyses could be hidden by the large uncertainties of LA-MC-ICP-MS data (the error ellipse from a single raster covers the entire spread of TIMS data). The scatter of TIMS fractions suggests multiple isotopic disturbances probably related to recrystallisation of columbite-tantalite with concomitant fractionation of the U-Pb system, which cannot be removed by leaching. Therefore, the discordia may have an anomalous upper intercept and a lower intercept without geological significance (cf. Romer 2003). Since the TIMS data could be mixed analyses of the different zones recognised in the BSE image, the 207Pb/206Pb ages provide only a minimum crystallisation age of 2659±1 Ma. Also, the last isotopic disturbance occurred at least as late as 2638±1 Ma, >20 Ma after crystallisation, and thus would be entirely unrelated to the pegmatite system.

Mavis Lake group

Fairservice pegmatite #1 (Sample 92DF-5)

Columbite was analysed from the primary assemblage, spodumene-quartz-albite unit of Fairservice pegmatite #1. Backscattered electron images of this single euhedral grain (3×2×2 mm) reveal oscillatory zoning, overprinted by convolute zoning (Fig. 8a). Electron microprobe analyses plot within the manganocolumbite field (Table 1, Fig. 3), and indicate that the replacement zones have lower Mn/(Mn+Fe) ratios than the oscillatory-zoned core. Clearly, there is a complex history of crystallisation and secondary alteration/replacement recorded in this grain. Wood and Williams-Jones (1993) have suggested that such textures may result from interaction with highly reactive, residual fluids during the late-stages of pegmatite formation.
Fig. 8a–d

Sample 92DF-5, manganocolumbite from Fairservice pegmatite #1 within the Mavis Lake group. a Composite BSE image. Locations of LA-MC-ICP-MS analyses are shown (not to scale). b-d Concordia plots of LA-MC-ICP-MS analyses. b Raster analyses from the primary oscillatory zoned core region. c Analyses adjacent to the core zone d Analyses from convolute replacement zones shown in BSE image

Four LA-MC-ICP-MS raster analyses (1, 2, 3, 4) from the oscillatory-zoned core of sample 92DF-5 yield Pb-Pb ages that range from 2657±15 Ma to 2673±15 Ma, with a weighted mean of 2665±8 Ma (MSWD=0.9). We interpret this weighted mean to be the best age estimate for the emplacement of the Fairservice pegmatite. U-Pb data from these four analyses are all reversely discordant (Fig. 8b). A regression line (MSWD=1.3) through the four analyses defines a trend through the origin with an upper intercept of 2662±25 Ma.

Raster analyses nos. 5, 6, 14, 15 and 16 are sited adjacent to the oscillatory-zoned core (Fig. 8a). These five analyses are all reversely discordant and fall on a regression line (MSWD=2.7) with an upper intercept at 2645±49 Ma (Fig. 8c). The weighted average of the 207Pb/206Pb ages of these analyses is 2651±13 Ma, indistinguishable from the oscillatory-zoned core. Analyses of the convolute zoned region (nos. 7 to 13) are all normally discordant indicating Pb-loss, and yield 207Pb/206Pb ages ranging from 2588 to 2148 Ma (Fig. 8d).

Discussion and Conclusions

Crystallisation ages and secondary disturbance in columbite-tantalite

Previous studies have shown that columbite-tantalite chronology can be used to determine whether geochemically similar peraluminous granite plutons and rare-element pegmatites are temporally synchronous (e.g. Romer and Smeds 1996). This study demonstrates, however, that the U-Pb system of columbite-tantalite can be disturbed after crystallisation, and that this is often associated with formation of secondary Nb-Ta oxides. Columbite-tantalite ages from two of the undeformed Separation Rapids pegmatites are within uncertainty of the U-Pb monazite crystallisation age of 2646±2 Ma for the Separation Rapids pluton (Larbi et al. 1999). However, columbite from a pegmatitic assemblage within the Separation Rapids pluton yields a U-Pb age of 2629±1.5 Ma and a deformed pegmatite has a minimum age of 2616 Ma. These younger ages probably reflect isotopic disturbances to the columbite U-Pb system after crystallisation.

In the Pakeagama Lake pegmatite, LA-MC-ICP-MS analyses of primary tantalite yield a best estimate for the crystallisation age of 2670±5 Ma. Tantalite from a lepidolite-rich vein that crosscuts the core zone has a range of 207Pb/206Pb ages as young as 2638±1 Ma from TIMS analyses, suggesting that isotopic disturbances occurred >20 Ma after pegmatite crystallisation. Such disturbances may be related to later fluids exploiting the same zones of weakness as had been employed by the pegmatites.

Causes of the complexities in the U-Pb system of columbite-tantalite

Firstly, it should be noted that some columbite-tantalite samples have extremely reproducible 207Pb/206Pb ages (e.g. Romer and Smeds 1996; Lindroos et al. 1996). These data were obtained from ~1.0 and 1.8 Ga old columbite that was free of secondary Nb-Ta minerals, and where selective dissolution by HF-treatment was successful in removing disturbed sections quantitatively. However, since most columbite samples are rich in Ti-, Nb-Ta-inclusions and secondary features, obtaining such precise data by TIMS is not always possible.

Reverse discordance in the U-Pb system of columbite-tantalite is caused by preferential U-loss from metamict parts of the grain (Romer et al. 1996). Alpha recoil results in open U-Pb systems on the nano-scale; U-rich domains become normally discordant because they lose more Pb than they receive from adjacent U-poor domains; U-poor domains become reversely discordant by Pb-implantation (cf. Mattinson et al. 1996; Romer 2003). The U-rich zones acquire a higher level of lattice damage with time than the U-poor zones, which results in contrasting dissolution behaviour, depending on the accumulated metamictization of the sample, of different segments of a columbite crystal. Thus, exposure to weak acids in nature or in the laboratory may result in strongly reversely discordant U-Pb systems through the preferential dissolution of U-rich zones. HF leaching has been successfully used in previous TIMS studies to decrease, or remove entirely, this reverse discordance (Romer and Smeds 1996). However, this work shows that this is not always possible—the data presented here are generally discordant and have significant non-analytical scatter, so they remain difficult to interpret. This problem may arise from two features: (i) cumulative radiation damage and (ii) annealing and recrystallisation history of the sample.

Cumulative radiation damage

The elongate volume damaged through an individual α-recoil has a dimension in the order of 20 to 30 nm. The decay of 238U to 206Pb includes 8 α-decays, the decay of 235U to 207Pb includes 7 α-decays. Thus, the volume damaged by the formation of radiogenic Pb from U typically has a size that is much smaller than 250 nm, especially due to the random recoil-direction of each subsequent α-decay. Lattice defects related with fission may influence a volume (track) of 10 to 15 μm length. Fission tracks, however are more than six orders of magnitude less abundant than α-recoil cascades. Therefore, most of the lattice damage originates from α-recoil. Due to the small dimension of the volumes that become metamict by α-recoil, the properties of a mineral and the behaviour of its U-Pb system depends on the number of α-recoils, i.e., the accumulated recoil damage. The recoil-related metamict domains in a mineral that has accumulated only little damage (low U content, young) are enclosed in the healthy mineral. The U-Pb system, thus, remains closed. An increasing amount of accumulated radiation damage results in an overlap of α-recoil damages with other α-recoil cascades or with fission tracks, eventually giving rise to a network of damages that allow the migration of ions over larger distances. Further accumulation of radiation damages eventually may result in the metamictization of the entire crystal with the possible exception of a few domains of distinctly lower U-content (see also Geisler et al. [2001] and Geisler [2002] for corresponding annealing studies on zircon). The contrasting behaviour of columbite may be best illustrated by the U-Pb systematics of samples of different age. Mauthner et al. (1995) obtained highly reproducible U-Pb data from 82 Ma columbite. They washed the samples in HNO3 to remove surface U, but did not employ any leaching procedure. Only two samples with distinctly higher U-contents gave strongly anomalous U/Pb ages. In contrast, Proterozoic columbite gave strongly reversely discordant U-Pb data when washed only with HNO3 (Romer and Wright, 1992; Romer and Smeds 1996, 1997). To obtain concordant U-Pb data, these columbite crystals had to be leached with dilute HF. For most samples, this procedure gave subconcordant to concordant data with little excess scatter. Although sample treatment with dilute acids (HNO3 and HCl) in the laboratory may cause reverse discordance, the treatment with HF generally brings the system back to concordance. Those samples with significant excess scatter had markedly higher U-contents or a complex history. Residual reverse discordance may also be related to insufficiently severe HF-treatment. The Archæan columbite samples presented here generally could not be brought to concordance even after partial dissolution with 20% HF and for most samples, there remained significant excess scatter.

Annealing and recrystallisation history of the sample

Recoil-induced open-system behaviour may become permanent if recrystallisation or the formation of a secondary Nb-Ta-phase results in the contrasting behaviour of U-poor and U-rich domains during subsequent processes. Mobilisation of radiogenic Pb in such systems may mobilise later formed Pb to a higher extent than old Pb, resulting in anomalous and scattered 207Pb/206Pb in the residual sample. Such a preferential mobilization of young Pb, as argued by Davis and Krogh (2000) and Corfu (2000), may reflect the higher extent of annealing of old recoil damages relative to young damages, largely reflecting the thermal history of the sample. Thus, the U-Pb systematics of columbite samples that have contrasting thermal history, such as long vs. short residence in the middle crust or a late burial or heating event, may behave differently and may not necessarily yield the same age for coeval samples. The possibility of an incorrect age is exacerbated if U-Pb fractionation during recrystallisation results in a discordant system, that during subsequent growth of radiogenic Pb becomes an open system as a result of increasing recoil-induced metamictization (cf Romer 2003).

LA-MC-ICP-MS analyses of primary, oscillatory-zoned columbite-tantalite yield Pb-Pb ages that are typically within analytical error of each other producing weighted averages with MSWD’s close to or less than 1. However, the corresponding U-Pb ages often exhibit extreme discordance, notably to a larger degree than the TIMS analyses. The grain fragments analysed by LA-MC-ICPMS were untreated. Instead, careful imaging and targeting were used to avoid submicroscopic inclusions and cracks, including the zones of alteration that surround them.

Fig. 9 shows the relationship between discordance and U content in LA-MC-ICP-MS analyses. Within samples SS137 and 92DF-5 (core zone and surrounding mantle), analyses with relatively low U (and Pb) content also have the greatest degree of reverse discordance. The convolute zone of sample 92DF-5 shows normal discordance, but with no clear correlation, and sample SS134 falls in two groups that have a positive correlation between discordance and U content. These data indicate the non-systematic cause of the discordance, but suggest that the reverse discordance is related to U-Pb mobilisation, with a net U-loss. Zero-aged lower intercepts (within uncertainty) for all of the LA-MC-ICP-MS data suggest that U-Pb fractionation and Pb and U loss occurred during either (i) a recent isotopic disturbance or (ii) the analytical procedure. Recent mobilisation of uranium and lead within the groundwater environment or by weathering is certainly possible, especially considering that these rocks may have only reached the surface within the last 2.5 Ma as a consequence of enhanced glacial erosion.
Fig. 9

Uranium content (ppm) versus discordance calculated as percent difference between 206Pb/238U and 207Pb/206Pb ages. U content is accurate to 10–20% and is calculated by comparing the signal intensity of U from columbite with that of the standard. The accuracy has been determined by examining minerals with a known U content (Foster and Horstwood, unpublished data)

Reverse discordance produced as an analytical artefact in LA-MC-ICP-MS analyses could be due to non-matrix matching, i.e. chemical differences between the monazite standard and the columbite sample. Results from analysing other accessory minerals suggest that the effect of not matrix matching is small, generally only a few percent, and typically reproducible (Horstwood et al. 2003). However, the extent of discordance in columbite samples is considerably higher. More importantly, the amount of reverse discordance is not always constant (perhaps coincidently, in some cases it is, e.g. rasters 1 and 2 in sample 92DF-5, Fig. 8b), but can vary by 20–30% or more, suggesting the non-matrix match correction is unlikely to be the sole cause.

Instead, this phenomenon, which is controlled at a scale that is comparable to, or smaller than, the resolution of this technique (50×50×10 μm), may be related to properties of columbite. Given the submicron to decimicron-scale of pits and grooves that are etched by HF-leaching (Fig. 1), such material is a likely candidate for contributing to the reverse discordance observed in LA-MC-ICP-MS analyses of untreated columbite-tantalite. The grooves and pits reflect the position of metamict zones in the columbite. Elemental fractionation in zircon that is both variable with time and anomalous in its magnitude during laser ablation analysis has been linked tentatively to contrasting elemental fractionation of U and Pb in crystalline and metamict material (Horn et al. 2000). Such behaviour may also occur in columbite. Also, there may be a significant contrast in the presence of metamict domains in the monazite standard and columbite samples. Numerous experimental studies have demonstrated that phosphates have distinctly lower activation energy for thermal annealing of decay-related defects than silicates or oxides (Weber et al. 1994; Meldrum et al. 1998; Seydoux-Guillaume et al. 2003). Thus, for a comparable thermal history, similar contents of Th and U, and an identical age, the phosphate eventually shows the lowest extent of defects. Samples with a higher extent of radiation damage may show a contrasting behaviour for elemental fractionation. Elemental fractionation depends on the laser fluence (e.g. Horn et al. 2000) and particle size distribution (Guillong and Günther 2002). Thus, for contrasting metamictization where the structure of the material and its response to ablation (i.e. particle size distribution produced) is variable, it is quite possible that there may be variable elemental fractionation, whether at the site of ablation (within-run fractionation) or within the plasma as a consequence of particle size or particle composition. Therefore, using a common correction factor for samples with contrasting extent of metamictization may eventually result in over- or under-corrected data and the observed variable discordance.

In conclusion, the discordance observed in both methods may originate from one or a combination of (i) the metamict nature of columbite-tantalite, (ii) Pb/U fractionation due to recent weathering, or (iii) analytical effects, including sample preparation. We also note that these effects may vary from sample to sample, and compete and interfere with each other during analysis, all of which would contribute to a non-systematic pattem of anomalous behaviour. In the case of TIMS analyses of these crystals, the scattered data most probably reflects the incomplete removal of the open-system metamict segments during sample preparation. The normal and reverse discordance observed in LA-MC-ICP-MS analyses may be caused by variation in the fractionation between U and Pb during ionisation in the plasma that has its origin in the contrasting metamictization of the dated columbite and the monazite standard used. However, the discordance could also be caused by U and Pb loss during recent weathering. Possible solutions to these problems include electron microprobe analyses and backscatter imaging so that domains with complex zoning patterns and inclusions can be avoided. Laboratory heating of columbite may also improve analyses by eliminating the preferential leaching and fractionation effects by annealing the alpha recoil tracks and fission tracks, as demonstrated for zircon by Mattinson (2000, 2001). Analysis of a well-characterised (structural, chemical and imagery) columbite-tantalite sample, will help to elucidate the U-Pb systematics of this economically important mineral series and greatly improve the interpretation of laser ablation data. In cases where an ideal sample can be found, the TIMS method provides higher precision. Where samples contain abundant inclusions and secondary zones, the LA-MC-ICP-MS technique offers higher spatial resolution and larger data sets, allowing statistical determination of a precise Pb/Pb age.

Acknowledgements

S. Smith acknowledges postgraduate studentship funding from The Open University and the Ontario Geological Survey. Analyses carried out at the NERC Isotope Geosciences Laboratory, Keyworth, were funded by ‘Isotopic Analytical Support’ from the Natural Environment Research Council, UK. Constructive reviews by R. Parrish and an anonymous reviewer greatly improved an earlier version of the manuscript.

Copyright information

© Springer-Verlag 2004