Single mineral dating by the PbPb step-leaching method: Assessing the mechanisms

Pergamon Geochimica et Cosmochimica Acta, Vol. 61, No. 2, pp. 393-414, 1997

Copyright © 1997 Elsevier Science Ltd Printed in the USA. All rights reserved

0016-7037/97 $17.00 + .00

PII S0016-7037(96)00343-2

Single mineral dating by the Pb-Pb step-leaching method: Assessing the mechanisms

R. FREI, ~ I. M. VILLA, 1 Th. F. NAGLER, I J. D. KRAMERS, 1 W. J. PRZYBYLOWICZ, 2'* V. M. PROZESKY, 2 B. A. HOFMANN, 3 and B. S. KAMBER 4

~Gruppe Isotopengeologie, Mineralogisch-Petrographisches Institut, Universitat Bern, 3012 Bern, Switzerland 2National Accelerator Centre, Faure 7131, South Africa

3Naturhistorisches Museum, Bernastrasse 15, 3005 Bern, Switzerland 4Department of Earth Sciences, University of Oxford, Oxford OX1 3PR, UK

(Received May 19, 1996; accepted in revised form on September 25, 1996)

Abstract--Stepwise Pb-Pb leaching (PbSL) has been successfully used to date rock-forming silicates directly linked to metamorphic reactions defining a PT path. The two features of PbSL are an increase of precision and a control on accuracy: the former, by enhancing the 2°rpb/2°4pb and 2°Tpb/2°4pb ratios, and the latter, by revealing heterochemical inclusions via the 2°Spb/2°6pb ratio and checking isotopic equilibrium with the host.

The question of the need for inclusions as a prerequisite enabling PbSL dating was investigated on a centimenter-sized single crystal of museum-quality titanite. We obtained petrographic (optical microscope, SEM, electron microprobe, proton microprobe), chemical (ICP-MS), and isotopic (TIMS) data on pristine and increasingly leached splits of different grain sizes, as well as on leach solutions.

The PbSL age of 1.00 Ga is identical to the concordant conventional U-Pb age. By use of isotopic and elemental correlation diagrams, we were able to resolve three isotopically distinct sources of Pb. Visible inclusions of K-feldspars as micro-crack fillings may contribute to the first 50% of common Pb (2°4pb) release, and visible 5 /zm zircons account, at least in part, for the residue. An additional effect noticed during the first leach steps is related to surface hydrolysis and consequent weakening of metal cation-oxygen bonds, which releases structurally bound common Pb substituting for Ca in the titanite lattice. The radiogenic Pb released during most leach steps has uniform 2°8pb/2°rpb ratios, suggesting that it is derived from a single mineral phase. Further, as different titanite grain sizes produce a different leach trajectory in Pb isotopic space, it can be concluded that this radiogenic Pb is actually derived from the titanite itself. Therefore, PbSL is indeed capable of discriminating common and radiogenic Pb from a single phase.

During leaching, a well defined reaction front is seen to advance into the grains. Inside this front, the titanite is unaltered, and outside it, a silica-gel-like rim is formed, in which electron and proton microprobe data show retention of high field strength (HFS) elements. We propose that radiogenic Pb occurs in the tetravalent state (due to recoil stripping) so that it behaves as an HFS element and is preferentially retained in the gel rim during leaching, relative to common divalent Pb. The mechanism by which a spread in Pb isotope data is obtained during PbSL is, therefore, explained by two processes controlling the release of cations from a mineral grain into a leach solution: (1) an effective surface dependent hydrolysis of metal cations at the inward migrating reaction front whose influence decreases as leaching proceeds and (2) an increasingly dominant remobilization of HFS-elements and radiogenic Pb from the micro-environment of the leached gel-like layer which acts as a selective HFS cation absorber. In conclusion, Pb/Pb dating by stepwise leaching can be effective as a dating tool, even in minerals free from micro-inclusions. Copyright © 1997 Elsevier Science Ltd

1. INTRODUCTION

In the past few years, interest in dating rock-forming minerals, rather than accessory phases, has greatly increased. This is because geologists have become aware that age data from common silicates permit a more direct link between important structural, petrographical, and petrological observations and time, and, therefore, provide better constraints on models of geological processes. Various successful attempts have been made to date metamorphic minerals with the commonly applied isotope systems. However, common rock-

*On leave from the Faculty of Physics and Nuclear Techniques. University of Mining and Metallurgy, Cracow, Poland.

forming minerals often have low parent-to-daughter isotope ratios, and, therefore, the list of those "datable" has re- mained short. Dating of rock-forming silicates is commonly hampered by difficulties in obtaining pure mineral concen- trates because of the necessity of large sample amounts for precise isotope measurements. Intergrowths and small adhering mineral particles in isotopic disequilibrium with the analyzed mineral may place serious limitations on the use of common chronometers, such as U-Pb and Sm-Nd (e.g., Mezger et al., 1989; N~igler et al., 1995a,b; and references therein). Geochronologists have become increasingly aware of the importance of microscopic solid inclusions, which can strongly influence the mass budgets of certain trace elements and, therefore, may completely control--if not simulate--

393

394 R. Frei et al.

D C

1 c m

C

EMP profile

Fig. 1. Schematic sketch of the gem-quality titanite crystal used in this study. Letters defining the different portions of the crystal correspond to the experiment codes listed in Table 1. Slab "C" was used for an EMP-profile across the entire crystal. Hairline cracks filled with quartz and minor K-feldspar can be seen by the naked eye.

the "age" of a host mineral (e.g., Kamber et al., 1996; N~igler et al., 1995a,b, 1996; deWolf et a1.,1996; Lanzirotti and Hanson, 1995; Heaman and Parrish, 1991; Vance et al., 1991; and references therein).

In a number of studies, attempts have been made to re- move these two limitations on U-Pb dating of rock-forming minerals by making use of acid washing prior to complete dissolution, with the aim of removing acid soluble Pb ad- sorbed on the mineral surfaces, dissolving fine grained alter- ation phases (adhering to the surface or occurring within micro cracks), or leaching unsupported radiogenic Pb from noncrystallographic sites (e.g., Ludwig and Silver, 1977; deWolf and Mezger, 1994). Stepwise partial HF attacks to silicates were used in the past: (1) for isolating original lead in feldspars (e.g., Ludwig and Silver, 1977), (2) for removing disturbed zircon domains characterized by very high U concentrations and low 2°6pb/23sU ages (e.g., Mat- tinson, 1994) to reduce discordancy, and (3) to evaluate the contribution of submicroscopic zircon and monazite inclusions in garnet and, consequently, to refine U-Pb and Sm-Nd garnet ages (deWolf et al, 1996; Zhou and Hensen, 1995). In case 2, the method relies on the prerequisite that the U and Pb associated with the portion of zircon dissolved in each step are completely extracted; otherwise, spurious fluctua- tions in the U/Pb ratio from step to step would be generated.

In contrast to these procedures, Frei and Kamber (1995) developed an acid step-leaching procedure (in the following, termed PbSL) whereby only Pb isotopic ratios are measured in successive leach fractions of the same mineral. This proce-

dure is independent of element fractionation during chemical procedures applied in the laboratory, as it deals only with a single element, Pb, whose relevant isotopes are decay products of the coupled U-Pb decay system, and it enables Pb- Pb dating of many single rock-forming minerals. Examples for titanite, garnet, hornblende, and pyroxene were given. Since then, many other successful results were obtained by utilizing this technique, e.g., for staurolite (Frei et al., 1995), orthopyroxene (Kamber et al., 1996; Blenkinsop and Frei, 1996), titanite (Kamber et al., 1995), garnet and epidote (Buick et al., 1997; Vinyu et al., 1996; Frei, 1995), tourmaline (Frei and Pettke, 1996), and even ore minerals, e.g., pyrrhotite (Frei and Pettke, 1996). This technique allows dating of minerals with less favorable U/Pb ratios because the increased data spread in 2°Tpb/2°4pb vs. 2°6pb/2°4pb space improves the precision of Pb-Pb isochrons. Initial attempts relied on the hypothetical assumption that radiogenic Pb in a mineral, produced from the decay of U and Th and, therefore, sited in lattice defects at the end of c~-recoil tracks, would be more easily accessible and leached by acids than initially built-in common Pb. However, Frei and Kamber (1995) observed the reverse effect, namely, increasingly radiogenic Pb isotope compositions in leach solutions from step to step and proposed two alternative explanations, i.e., a chemical selection and/or the presence of (sub)microscopic U-beating inclusions. Under ideal circumstances, given a homogeneous mineral, an isochron in the uranogenic Pb diagram (2°6Pb/Z°4Pb vs. 2°7pb/2°4Pb) would result as the product of mixing of the two Pb components and, thus, allow dating of a single phase. Besides being independent of a multi-mineral approach and, thus, of the prerequisite assumption of attainment of initial Pb isotopic equilibrium, PbSL has a further advantage over other dating tools in that, through tracing of the behavior of thorogenic vs. uranogenic

Table 1. Experiments performed on the Otter titanite.

Sub-sample Code Experiments

A B

C

D E

A1 U-Pb bulk mineral dating B 1 One HBr and two HNO3 leach experiments

(test for reproducibility of the method) on 200-300 ~zm grain size fractions with successive TIMS U-Pb isotope analyses of leach solutions (6 steps)

B2 Two independent sets of leach experiments with HBr and HNO3 and successive removal of leached grains for EMP-EDS and micro-PIXE element mapping

B3 One HBr and one HNO3 leach experiment each on a 100-200 #m grain size fraction (5 steps) with successive ICP multielement analyses of the leachates, complemented by U-Pb and Sm-Nd TIMS isotope analyses for the HBr experiment

C1 Preparation of a polished thick section for EMP-profiling and microscopic analysis

kept as reference material E1 Initial PbSL test run (two leach steps) E2 Complete initial HBr PbSL experiment on

a 50-100 #m size fraction (6 steps)

Single mineral dating by the Pb-Pb step-leaching method

Table 2. Pb isotope data of leach experiments on titanite from Otter Lake.

395

Leach 2°6Pb/ _+2tr 2°Tpb/ _ + 2 a 2°sPb/ --.2~ 2°aPb/ 2-+2cr 2°4pb ̂ _+2~r ̂ 2°4Pb ̂ 2°6pb ̂ Sample Code time 2°4pb (%) 2e4pb (%) :°4Pb (%) 2°6pb (%) rl ~ r2** (pg) # ^ (%) (%)++ (%)++

Initial test

Otter E mix* E1 10' 201 1.2 28.8 1.2 123 1.2 0.61 0.02 0.9982 0.9996 Otter E HBr* E1 3h 2461 11 192 11.0 1171 11 0.48 0.02 1.0000 1.0000

HBr leach experiment, 50-100/~m grain size fraction

Otter E E2 10' 19.99 0.1 15.69 0.1 38.79 0.1 1.94 0.03 0.9490 0.9105 Otter E E2 40' 309 3 36.7 2.6 177 3 0.57 0.14 0.9988 0.9987 Otter E E2 3h 2102 7 166 6.6 985 7 0.47 0.06 1.0000 1.0000 Otter E E2 6h 1634 11 132 11.4 687 11 0.42 0.04 0.9999 1.0000 Otter E E2 13.5h 888 23 76 23.2 317 23 0.36 0.08 0.9998 1.0000 Otter E E2 residue 371 59 40 59.4 73 59 0.20 1.09 0.9990 0.9998

HBr leach experiment, 100-200 #m grain size fraction

Otter B B3 15' 33.72 0.2 16.86 0.2 45.26 0.3 1.30 0.14 0.9687 0.8577 Otter B B3 50' 637 3 60.0 3.1 330 3 0.52 0.03 0.9997 0.9999 Otter B B3 4h 1788 8 143 7.9 839 8 0.47 0.06 0.9999 1.0000 Otter B B3 8h 1385 5 114 4.9 641 5 0.46 0.44 0.9980 0.9958 Otter B B3 residue 379 10 43 10.8 159 11 0.44 0.06 1.0000 1.0000

HBr leach experiment, 200-300 #m grain size fraction

Otter B B1 10' 29.34 0.4 16.31 0.4 42.36 0.4 1.44 0.02 0.9926 0.9958 Otter B B1 40' 425 3 45.1 3.5 228 3 0.54 0.03 0.9993 0.9999 Otter B B1 3h 1302 7 109 7.2 640 7 0.49 0.08 0.9998 0.9999 Otter B B1 6h 2189 11 173 11.0 1030 11 0.47 0.17 0.9999 0.9999 Otter B B1 13.5h 2403 9 189 9.4 1132 9 0.47 0.03 1.0000 1.0000 Otter B B1 residue 1809 21 147 21.3 869 21 0.48 0.03 0.9999 1.0000

HNO3 leach experiment (1), 200-300 #m grain size fraction

Otter B B1 10' 26.45 0.7 16.11 0.7 41.17 0.7 1.56 0.04 0.9905 0.9975 Otter B B1 40' 339 3 39.2 2.7 187 3 0.55 0.04 0.9940 0.9999 Otter B B 1 3h 1167 4 99 3.9 580 4 0.50 0.06 0.9998 0.9998 Otter B B1 6h 2180 10 173 9.7 1029 10 0.47 0.03 1.0000 1.0000 Otter B B1 13.5h 2151 17 170 17.0 995 17 0.46 0.03 1.0000 1.0000 Otter B B1 residue 1869 22 150 22.3 862 22 0.46 0.03 1.0000 1.0000

HNO3 leach experiment

Otter B Otter B Otter B Otter B Otter B

Contaminants

Otter B crack filling +

(2), 200-300/zm grain size fraction

B1 10' 33.66 0.7 16.63 0.7 43.20 0.7 1.28 0.03 0.9951 0.9982 B1 40' 338 2 39.1 2.2 186 2 0.55 0.02 0.9994 0.9999 B 1 6h 2209 11 176 11.1 1051 11 0.48 0.03 1.0000 1.0000 B1 13.5h 2434 15 191 14.7 1135 15 0.47 0.03 1.0000 1.0000 B1 residue 1925 21 153 21.4 811 21 0.42 0.03 1.0000 1.0000

bulk 102 1 21.5 1.3 70.6 1 0.69 0.03 0.9983 0.9995

1833 135 1 41.7 1.1 1228 3840 4 27.9 29.2 719 11200 8 16.3 49.5 278 8760 6 6.3 15.5 338 3770 6 7.7 4.6

2709 45.5 0.2 51.6 0.8 483 2360 3 9.5 5.6

1336 7640 7 25.5 48.4 514 12800 11 9.8 31.5 187 14400 10 3.6 12.7

18 17000 22 0.4 1.0

2437 34.9 0.7 49.4 0.6 477 2290 3 9.7 4.7

1245 7130 5 25.2 43.5 508 12300 10 10.3 33.8 245 12900 17 5.0 16.2

21 23000 23 0.4 1.2

2186 83 2 422 1500 3 528 11800 11 284 13800 14

22 23500 23

Errors are 2 standard deviations absolute. Fractionation amounted to 0.068 _+ 11%/AMU (n = 85) on NBS 981 Pb standard. Total procedural Pb blank <200 pg. * Mix = 1.5N HBr-2N HC1 12:1 mixture; 8N HBr; both used in the initial experiment.

r~ = 2°6Pb/2°4pb vs. 2°7pb/2°~pb error correlation. ** r2 = 2°6Pb/2°4pb vs. 2°sPbfl°4Pb error correlation. ^ Values calculated from data in Table 6. + Sample consisted of quartz, K-feldspar and impurities of fitanite. ++ In percent of total amount recovered during leaching.

Pb in the different leach solutions, various mineral internal inhomogeneities can be detected (Frei and Kamber, 1995). It even allows one to assess whether or not inclusions were in isotopic equilibrium with their host as this crystallized. Various leach spectra, in combination with bulk data of individual minerals from a paragenesis, can be used to delineate the extent of intermineral isotope exchange during mineral growth (cf. Kamber et al., 1996; Niigler et al., 1996; Vinyu et al., 1996; Frei and Kamber, 1995; Frei et al., 1995; Blen- kinsop and Frei, 1996; Frei, 1996; Frei and Pettke, 1996).

In summary, PbSL is able to discriminate real bulk U / P b ages from apparent ones by comparing step-leaching trajectories in Pb isotopic space because it is possible to catalogue inclusions and to assess their mutual Pb isotopic equilibrium. Subsequent work has very much stressed the extreme and partially dominant input of Pb from micro-inclusions in such a large number of cases that the question arises as to whether the presence of inclusions is actually a prerequisite for the isotope spread of apparent PbSL isochron ages of silicates dated. This would limit the usefulness of this method to

396 R. Frei et al.

Table 3. ICP and TIMS analyses of step solutions and residues from Otter titanite (experiment B3).

Amounts leached (#g) Amounts leached (#g)

Element HNO3 HBr 15' 50' 4 h 8 h Res. Conc.* 15' 50' 4 h" 8 h Res. Conc.* Step blind Blind 1 2 3 4 5 (ppm) 1 2 3 4 5 (ppm)

Fe 1 0.2 7 95 264 68 l0 7200 6 80 256 76 18 7300 Ti 0.04 <0.02 83 2080 5400 1022 1372 162000 61 1750 3740 934 2640 152000 A1 2.6 2.6 41 398 1018 278 44 28900 28 300 962 296 67 27500

0

Ca 7.16 1.018 98 2400 6640 1740 272 181000 73 2020 6440 1942 414 181000 0

Si <0.1 <0.1 57 248 134 35 9 7900 41 320 117 58 11 9100

U 0.004 0.008 0.3 4.6 11.2 2.9 0.3 310 0.3 3.4 11.7 3.1 0.5 320 U** 0.3 4.06 10.7 2.6 0 290 0.2 3.5 10.6 3.0 0.6 300 Th 0.068 0.024 0.7 6.1 15.7 3.6 0.4 430 0.5 4.8 15.5 3.8 0.7 420

Pb <0.1 <0.1 0.14 0.84 2.46 0.54 0.36 71 0.22 0.52 1.74 0.46 0.88 64 Pb** 0.18 1.30 2.07 0.65 0.20 72 Pb**c 0.138 0.092 0.054 0.021 0.03 5 Pb**r 0.04 1.21 2.02 0.63 0.18 66

La <0.02 <0.02 0.6 7.9 23.0 6.1 0.9 630 0.4 6.8 22.4 6.8 1.6 630 Ce n.d. n.d. 4.0 51.0 124.2 36.4 5 3600 2.4 35.4 118.8 36.0 8 3300 Nd n.d. n.d. 3.9 43.6 123.4 32.2 4 3400 2.4 35.6 120.6 34.2 7 3300 Nd** n.d. n.d. 4.5 57.2 121.4 33.7 4 3600 Sm n.d. n.d. 0.7 8.9 26.6 6.8 1.0 720 0.4 7.0 25.8 7,6 1.4 700 Sm** n.d. n.d. 1.0 12.8 27.8 7.8 1.1 820 Gd n.d. 0.016 0.8 8.2 24.4 6.3 0.9 660 0.6 6.2 22.6 6.6 1.4 620

Y <0.01 <0.01 1.8 28.6 87.0 23.4 3.7 2400 1.4 23.8 84.0 26.4 5.8 2400 Sr n.d. n.d. 0.2 1.3 3.2 0.8 0.1 93 0.2 0.9 3.1 0.9 0.3 89

Nb 0.002 0.006 1.6 5.6 14.2 0.5 51.2 1200 1.3 3.6 0.6 1.4 70.6 1300 Zr n.d. n.d. 0.1 1.8 4.8 1.2 10.7 300 0.1 1.4 4.3 1.0 10.0 280 Lu n.d. n.d. 0.02 0.38 0.93 0.26 0.01 26 0.02 0.27 0.91 0.28 0.04 25 Hr n.d. n.d. 0.01 0.40 0.89 0.14 0.55 32 0.03 0.23 0.84 0.17 0.51 30

Relative release ratios +

Ti/Si 1.2 7.2 34.5 25.0 127.5 1.3 4.7 27.4 13.9 208.9 Ca/Si 1.3 7.4 38.2 38.2 22.7 1.4 4.9 42.5 25.9 29.5 AI/Si 3.1 7.0 33.1 34.5 21.0 2.9 4.1 35.9 22.3 26.8 Ti/Fe 0.50 0.98 0.91 0.67 6.12 0.44 0.97 0.65 0.55 6.40 Pb**/Ti 4.95 1.40 0.86 1.43 0.33 Pb**c/Ti 53.9 1.4 0.3 0.7 0.6 Pb**r/Ti 1.31 1.42 0.92 1.50 0.32 Fe/Ca 1.97 1.03 1.04 1.02 0.96 2.21 1.04 1.04 1.02 1.16 Fe/AI 0.82 1.08 1.18 1.12 1.02 1.01 1.22 1.21 1.17 12.6 Ca/AI 0.38 0.96 1.04 1.00 0.98 0.42 1.07 1.07 1.05 0.99 Fe/U** 0.93 0.88 0.93 0.99 1.11 1.07 0.87 0.92 0.96 1.24 Pb**/Ca 4.64 1.36 0.78 0.93 1.86 Pb**c/Ca 32.0 0.9 0.2 0.3 2.1 Pb**r/Ca 1.27 1.41 0.86 1.01 1.83 Sm**/Ca 2.30 1.19 0.93 1.00 0.87 Nd**/Ca 2.30 1.21 0.93 0.99 0.82 U/Pb** 0.40 0.77 1.27 0.99 0.41 0.24 1.61 1.44 1.55 0.15 Th/Pb** 0.62 0.79 1.26 0.93 0.35 0.43 1.68 1.61 1.50 0.14 Ca/Th 0.33 0.89 0.96 1.09 1.46 0.32 0.95 0.95 1.16 1.43 Zr/Pb**r 0.40 0.32 0.52 0.41 13.30

# filtered, TiO2 in suspension; ** analyzed by isotope dilution TIMS; Si concentration of residues are affected by SiF4-1oss during HF a + relative release ratios, normalized to the ratio in the unaltered titanite. * Concentration in titanite (cummulative amounts of steps 1-5 relative to titanite weight). res. = residue; conc. = concentration; min. = mineral; c = common; r = radiogenic. blind = blind solution for blank levels (in #g) in same acid volume used for leaching; n.d. = not detected.

Single mineral dating by the Pb-Pb step-leaching method 397

Table 4. Sm-Nd isotope results of leach solutions and the residue from the 4N HBr step-leach experiment B3 on titanite.

Leach Sm Nd 1475In] |43Nd/ Sample Code time (ppm) (ppm) 144Nd 144Nd _+2or . . . .

Otter B B3 15' 0.05 0.22 0.1380 0.512247 _+26 Otter B B3 50' 0.64 2.86 0.1357 0.512268 ___18 Otter B B3 4 h 1.39 6.07 0.1386 0.512274 ___17 Otter B B3 8 h 0.39 1.68 0.1403 0.512262 _+17 Otter B B3 - - 0.05 0.22 0.1463 0.512343 ___21

cases where Pb isotopic equilibrium was reached between inclusions and host as this crystallized.

We assessed this issue of whether the presence of equili-

brated micro-inclusions, a sufficient condition to enhance the spread in the Pb-Pb isochron diagram, is also necessary, or if, on the contrary, the PbSL technique will produce an internal isochron in inclusion-free U-bearing minerals and, if so, by what mechanism. We carried out a multi-method analytical study on a single titanite museum crystal with gem-like characteristics in an attempt to ensure that most, or all, of the U resides in the mineral and to minimize the probability that inclusions affect the PbSL results. Stepwise leaching was carried out in duplicate on dry grain splits, followed by U-Pb and Sm-Nd isotope analyses of the ieachates and residues. Residues of various leaches, as well as fresh grains, were analyzed by electron microprobe in backscattered electron mode (EMP- BSE) and proton induced X-ray emission (PIXE) element mapping to assess the distribution of major and trace elements and their migration during leaching. Multi-element analyses were done on the leach solutions by inductively coupled plasma (ICP) to obtain data on correlative behavior of elements during the leaching process. The combined results provide a dataset against which the hypotheses on the mechanisms of stepwise leaching can be tested.

2. S A M P L E M A T E R I A L

A gem-quality museum specimen of titanite from the Yates (R. Kretz, pers. commun.) locality (Otter Lake, Can- ada) was selected as the experimental material. The euhedral titanite crystal (museum sample number NMBE A7342; in the following, called "Ot te r A - E " ) was collected from a Ca-pyroxene- scapolite-titanite skarn within an area of the Grenville province, composed of gneisses, marbles, gab- broic, and granitic rocks (Kretz, 1960, 1993). The metamorphic grade of these skarns is identical to that o f the enclosing rocks, namely upper amphibolite facies (Kretz, 1960). The crystal shows macroscopically visible hairline cracks more or less oriented in one direction (Fig. 1, section C) and filled with quartz and minor K-feldspar.

3. E X P E R I M E N T A L P R O C E D U R E S

The crystal was cut into five pieces (Fig. 1 ) on which the experiments listed in Table 1 were performed. The leach experiments in this study, in contrast to the procedure suggested by Frei and Kamber (1995) using both HBr and HNO3 in variable concentrations, were performed with either solely 4N HBr (reducing acid) or 4N HNO3 (oxidizing acid). This was done in order to ensure that possible

differences in the response of certain elements towards leaching, with respect to the kind of leach acid, could be detected. Lead isotope ratios are listed in Table 2, together with the results from successive experiments. The suitability of the Otter titanite for the planned leach experiments was first confirmed by two initial leach steps performed on a 50 mg dry split grain aliquot (experiment E1 ). The two leachates carried dramatically different Pb isotope compositions, with strong HBr recovering highly uranogenic-(2°sPb/2°npb ~ 2500) and thorogenic-(2°Spb/2°apb ~ 1200) Pb after only 3 hours of leaching.

Grain separates from different subsamples of the titanite crystal were obtained by hand crushing of small fragments in an ultraclean boron carbide mortar and successive sieving into the necessary grain- size fractions. The grain fractions were then carefully hand-picked under a stereo microscope to separate adhering, crack-related foreign quartz and K-feldspar. All separates were then repeatedly rinsed in deionized water, and 60-85 mg of the material was transferred to 3 mL PFA teflon vials in which acid leaching was performed. Eleven mg of the finely ground subsample A for the bulk U-Pb dating (code A1) was dissolved with HNOa-doped HF in Krogh-style Teflon bombs for three days. Leaching for the different experimental runs (performed on dry split aliquots of subsamples B and E) involved either 4N HBr or 4N HNO3. For the 50-100/~m (experiment E2) and the 200-300 #m (experiment B1 ) dry-split aliquots (Table 2) a complete PbSL procedure consisted of five leaching steps with 1 mL each of the above acids, in addition to the final dissolution of the residue with 0.5 mL HF. Only four leach steps with both acids were performed on 100-200 #m dry split aliquots (experiment B3 ), and the solutions were used for combined multi-element ICP and TIMS U-Pb and Sm-Nd isotope measurements (Tables 2-4, 6). The time intervals during which the different grain separates were exposed to the acids are listed in the corresponding tables. The leach solutions of experiment B 1, including two successive 1 mL rinse solutions with triple distilled water were pipetted into 3 mL PFA teflon vials spiked with a mixed 2°Spb-23sU tracer and dried on a hot plate. In the case of experiment B3, the leach solutions were diluted to 20 mL with triple distilled water for the ICP analyses. Aliquots of the HBr solutions were used for each TIMS U-Pb and Sm-Nd analyses. These liquid aliquots were spiked with a mixed 2°Spb-z35U and a mixed 1495m-15°Nd tracer.

Lead and uranium were separated using AG l-X8 anion exchange resin and standard HBr-HC1-HNO3 techniques. This procedure added a Pb blank of <200 pg, which had a negligible effect on the corrected data. Pb was loaded together with silica gel and phosphoric acid and measured from 20 #m Re filaments on a 5 collector VG Sector* mass-spectrometer in a static mode. Mass fractionation amounted to 0.068 _+ 0.011%/AMU (2cr, n = 85), determined on repetitive analyses of the NBS 981 Pb standard. Uranium was measured in a Ta-Re-Ta triple filament setting on an AVCO 9000 single cup mass- spectrometer. Errors (reported at the 2or level) and error correlations (r) were calculated after Ludwig (1988). Samarium and Nd were separated by conventional ion-exchange techniques, including re- versed-phase extraction chromatography (Richard et al., 1976), and measured on Ta-Re-Ta triple filaments in the AVCO mass spectrometer. Neodymium ratios were normalized to 146Nd/la4Nd = 0.7219. The mean t43Nd/144Nd Value of the La Jolla standard during the period of measurement was 0.511872 ± 0.000020 (2~r). The total chemistry blank for Nd was below 170 pg.

Grains for micro-PIXE and EMP-BSE (backscattered electron) and element mapping were mounted in epoxy, carefully polished using Pb-free materials and coated with C. Microprobe analyses and major element BSE-mapping were performed with a CAMECA SX- 50 microprobe operated at 15 kV and 20 nA using UO2, PbS, ThO2, and natural and synthetic titanite as standards. Micro-PIXE element mapping and point analyses were done using the proton microprobe at the National Accelerator Centre (NAC) in Cape Town, South Africa. The probe is based on a 6 MV single-ended Van de Graaff accelerator and uses Oxford Microprobe triplet lenses for beam fo- cusing (Tapper et al., 1993; Prozesky et al., 1995). The main aim of using micro-PIXE was to determine the spatial distribution of U, Th, and Pb, which could not be analyzed with EMP due to elevated detection limits for these elements. A 3 MeV proton beam was used, focused to between 3 x 3 #m 2 and 5 × 5 #m 2, and rastered over

398 R. Frei et al.

areas from 1 m m x 1 m m down to 50 #m X 50 #m. Beam currents between 2 and 5 nA were used. The X-rays were detected using a Si (Li) detector positioned at 135 ° with respect to the incident beam. A 40 # m thick A1 filter was positioned between the sample and the detector to reduce the count rate from the major elements (Ca, Ti) to acceptable levels to reduce peak pile-up effects. Total charge accumulated for the rastered areas varied between 9 and 31 #C, and for point analyses, the accumulated charge was 1 #C. PIXE spectra were reduced using GeoPIXE software (Ryan et al., 1990a,b) using titanite composition as obtained from EMP for matrix corrections. Element maps were obtained using a rapid matrix transform method called Dynamic Analysis (DA) (Ryan and Jamieson, 1993), which allows for the on-line production of true elemental images with inherent element overlap-resolution and background-subtraction. These images are quantitative after minor off-line correction. Scanned regions were divided into 64 × 64 pixels. Data contouring was based on inverse squared distance smoothing in which the z- height of the plane (i.e., the concentration of the mapped element) at a smoothing point is the weighted average of the y-values at x- values of the data points, where the weights are the squared Euclid- ean distances across x and y. Table 5 summarizes the average main and trace element concentrations and their mean detection limits determined from the mean X-ray spectrum collected from rastering a 50 × 50/~m area in an untreated titanite grain.

Quantitative ICP analyses of Ca, Ti, Si, U, Pb, Th, and Y were performed at the Paul Scherrer Institute, W~renlingen, Switzerland, and an analysis spectrum of seventy main and trace elements was carried out at the Eidgeni3ssische Materialprtifungsanstalt in Dtiben- dorf, Switzerland, respectively. Blank levels (contained in Table 3) were controlled by blind solutions of 4N HBr and 4N HNO3.

A complete table of EMP-profile data, ICP leach solution analyses, as well as half a dozen additional PIXE element map data files of leached grains not used in this study, are available upon request.

4. RESULTS

4.1. EMP-profile

E M P a n a l y s e s a l ong a t r ansec t o f the Ot te r t i tani te c rys ta l

(cf . F i g . l ) are p lo t ted in Fig. 2a ( m a i n e l e m e n t s ) and b

( t r ace e l e m e n t s ) and r evea l the f o l l o w i n g :

1 ) C o n s i d e r a b l e c o n t e n t s o f A1 ( a v e r a g e o f ~ 3 . 2 w t % ) and

Fe ( a v e r a g e o f 0.7 w t % ) , s u g g e s t i n g oc tahedra l subs t i tu -

Table 5. PIXE analyses of a 50 × 50 # m area within an unleached titanite grain.

Cone. 2~r Element (ppm) (abs.) MDL

Ca 189000 2400 80 Ti 163000 1300 21 Fe 8060 250 4 Sr 72 3 4 Y 2450 50 5 Zr 320 8 5 Nb 1180 20 6 La 550 90 216 Ce 3040 190 265 Nd 3750 400 46 Gd 590 60 21 Yb 220 30 9 Pb 60 4 5 Th 420 8 6 U 270 6 6

MDL = mean detection limit; abs. = absolute

20 a

,

• ,= 16 k / ,~ [ " • , k T i . , ~

- - - 1 5

1 3 1 I I I L I I 10000 12000 14000 16000

profile x-coordinate

b

1.2

0.8

0.4

0.0 r T r T T

10000 12000 14000 16000

profile x-coordinate

Fig. 2. (a) EMP main element and (b) trace element profiles along the transect of Fig. 1. The profile length is about 5 m m (starting and end point x-coordinates are arbitrary, one unit corre- sponds to 1 /zrn). Horizontal lines mark the average concentrations of the elements. Slight enrichment of Fe and Y towards the rim relative to the interior portion of the crystal is observable.

t ion o f t he se e l e m e n t s for Ti ( -- 16.5 w t % ) re la t ive to the

ideal va lue for CaTiSiO5 o f 24 .4 wt%.

2 ) L o w C a c o n c e n t r a t i o n s o f 18.3 w t % re la t ive to the ideal va lue o f 20 .4 w t% for C a T i S i O s , s u g g e s t i n g tha t subs t i tu -

t ion o f a lka l ies , a lka l ine -ea r th , and R E E s in the s e v en -

fo ld si te is opera t ive .

3 ) A nea r s to i ch iome t r i c a v e r a g e Si concen t r a t i on o f - - 1 4 . 0 w t% ( idea l va lue for pu re t i tani te is 14.3 w t % ) .

4 ) Dis t inc t , bu t u n s y s t e m a t i c , va r ia t ions o f up to 6% in the

c o n c e n t r a t i o n o f m a j o r e l e m e n t s a long the t ransec t .


5) Enrichments of Fe and Y in the outermost rim relative to the inner portion of the crystal.

All EMP analyses are characterized by low weight percent totals with an average value of ~98.1 wt% (range from 94.8-100.8 wt%). Low weight percent totals of aluminous titanite have been reported, e.g., by Franz and Spear ( 1985 ), and were explained as the result of A1 compensation by OH, i.e., by the coupled substitution of Ti 4÷ + 0 2- = A13+ + OH. Some of the mass deficit may also be the result of elevated REEs and other trace element concentrations of up to ~0.9 wt% in the titanite analyses (cf. PIXE results, Table 5; ICP solution mass budgets, Table 3), which, due to elevated detection limits, could not be analyzed by the EMP. We followed the suggestion of Franz and Spear (1985) to nor- malize our probe analyses on the basis of three total cations. Excess ( > 1.000) sums of octahedrally coordinated cations (Ti + Al + Fe) suggest that substitution of Si 4÷ with Ti 4+ may also play a role in the tetrahedral sites (Hollabaugh and Rosenberg, 1983). The average sum of anions (O + OH + F) of 4.96 calculated from all the EMP data along the transect can be used as a reliability indicator in that it con- trols the stoichiometry of the crystal with an ideal value of 5.0. The high A1 contents in the studied titanite crystal, formed during upper amphibolite facies metamorphic conditions, is consistent with the observation that Al-rich titanite preferably occurs in high grade metamorphic rocks (Krogh et al., 1982; Smith and Lappin, 1982).

4.2. EMP-BSE and Element Mapping

Grains that were removed from each of the five steps of the two parallel PbSL experiments (B2) were mapped for their major elements with the EMP, prior to mapping with the PIXE, BSE and major element maps of one which had been continuously exposed to 4N HNO3 for 6 h, are shown as an example on two scales in Fig. 3a, b. The grain, whose original outline is visible in the BSE image (Fig. 3a), has been transformed into a composite of rhythmically zoned layer concentrically surrounding a core of unaffected titanite. Element distribution maps reveal that all Ca and almost all A1 and Fe have been removed from the leached layer whereas Ti and Si are retained in a restructured zone around the titanite core. Concentric detachments of portions of the shell are noticeable, and they are interpreted as resulting from shrinkage of the leached layer after exposure of the grain to air. A sharp jigsaw-like reaction front between the leached part of grain and the still unaffected titanite in the core is a characteristic feature (BSE image in Fig. 3b). This reaction front migrates inward with time and, thus, with successive leach steps. We have not observed any differences in appear- ance between rims produced by HBr or HNO3 leaching.

Figure 4 shows SEM photographs of a portion of a crushed leached grain. They confirm the observations from the EMP- BSE-and element maps and document the formation of distinct, concentric layers around a central, still intact, core during leaching. In detail, this layer is composed of a smooth outer shell and variably structured spongy, very porous concentric layers in sharp contact with still unaffected material

in the center of the original grain. Micro-crystallites of newly formed phases (mainly TiO2) nucleate in the porous zones.

4.3. PIXE-mapping

In order to obtain a representative PIXE-spectrum of main and trace elements, specifically of U, Th, and Pb, the relevant elements to PbSL, we rastered a 50 × 50 #m-sized area within a titanite grain. The average concentrations of selected elements, calculated from the overall stacked X-ray spectrum, are listed in Table 5. Average PIXE U,Th, and Pb concentrations can be used to calculate chemical ages (Mon- tel et al., 1994). Using the data of Table 5, we obtain a U- Th total Pb age of 999 _ 80 Ma (the assigned error is estimated from propagating the individual concentration errors of all elements). This date is fully consistent with the ages obtained by the isotope dating methods (see below).

A dozen leached grains were mapped with PIXE during this study. Selected element maps from an area of 150 × 150 #m in the interior of a leached titanite grain removed after 3 h from 4N HNO3 are shown in Fig. 5. A leach zone is observed running across the imaged area; it probably formed as the result of acid influx along a crack. The main and trace element distribution in this example is typical of all other analyzed grains. In the leached zone, most elements are below detection limit, with notable exceptions being Ti and Zr. The high Ti concentrations, sometimes locally exceeding 20 wt%, together with the identification of TiO2 as a suspension in one of the leach solutions, suggest its relative enrichment through precipitation as an oxide within the leached layer. This observation is consistent with the energy disper- sive spectra results from EMP mapping (Figs. 3a,b) on the white re-precipitation structures in which only Si and Ti oxides were detected. Other elements (Ca, Fe) clearly define the extent of the unleached zones and show gradually decreasing concentrations in the leached layer. The behavior of HFS elements, such as Zr, Hf, and Nb, parallels that of Si and Ti, in that they are quantitatively retained in the re- precipitated layer and only partially removed into the liquid solution. Lead shows an intermediate behavior: its concentration in the leached zone is lower than in the pristine titan- ire, but not negligible. The U and Th distribution show this effect to a much lesser extent.

4.4. U-Pb, Pb-Pb Isotope Results

Results from PbSL experiments (B1, B3, El, and E2) and a bulk analysis (A1)) are given in Tables 2 and 6. 2°6pb/ 2°4Pb vs. 2°Tpb/2°4pb ("uranogenic") and 2°6pb/2°4pb vs. 2°spb/2°4pb ( "thorogenic" ) plots in Fig. 6 display the results for the E2 and B 1 experiments. Increasingly radiogenic Pb was recovered in the first three to four steps in all runs, with highest isotopic ratios reached after 3 h for experiment E1 performed on a 50-100/zm grain size aliquot, after 4 h for experiment B3 on a 100-200 #m grain size fraction, and after 13.5 h for the coarsest, 200-300 #m grain size fraction. Data for the runs plot along isochrons in the uranogenic diagrams (Fig. 6), defining ages between 994 and 1014 Ma. Pooling all thirty-one leach steps (Table 2), we obtain 997

400 R. Frei et al.

A

Fig. 3. (a) Backscattered electron (BSE) and major element maps of a leached titanite grain which has been exposed to 4N HNO3 for 6 hours. The rhythmic structure consists of Ti precipitates and (re)polymerized silica layers around the still intact titanite core. AI, Ca and Fe are removed from the leached layer and can only be detected in the core. (b) Detail of Fig. 3a showing the reaction front between the leached layer (lower right hand portion of the maps) and the intact titanite core (upper left hand portion of the map). Shrinkage cracks in the leached layer can be identified in the Ti distribution map as black concentric areas.

B

Si

(~

Ti


Fig. 3. (Continued)

402 R. Frei et al.

+__ 10 Ma (MSWD = 4.5). The individual PbSL ages are in accordance with the concordant bulk U-Pb age of 1002 _+ 7 Ma (Table 6) for portion A of the crystal. The bulk analysis gives a 2°rpb/2°4pb ratio of 1100. Measured 2°rpb/2°4pb values of leachates range from 20 to >2400. Minor inhomogeneities are indicated by the relative behavior of thorogenic vs. uranogenic Pb during leaching, which can be traced in the corresponding diagrams in Fig. 6. Linear arrays, i.e., constant U/Th ratios, indicative of a two component mixing are obtained for almost all the B 1 experiments, but in the experiment E2, successive data points define a flat clockwise loop, with the last steps and the residue yielding lower Th/ U ratios than the first three steps. In all the PbSL experiments, this effect is most pronounced in the final step in which HF was used to attack the residue. The observation of sporadic <5 #m zircons as inclusions in the titanite used allows us to postulate that the observed trajectories in the uranogenic vs. thorogenic diagrams may result from the admixture of less thorogenic, highly uranogenic Pb, typical of this accessory mineral. However, the data still define isochrons in the uranogenic diagrams which shows that Pb from these zircons was in isotopic equilibrium with the host titanite as this crystallized.

The leaching behavior of radiogenic Pb (Pbr) vs. initially incorporated common Pb (Pbc) and U in the Otter titanite can be studied using the experiment B 1 data which are listed in Table 6. The step-to-step trajectories and inter-step variations are shown in Fig. 7a-e. From the latter we deduce the following:

1 Inter-step leach trajectories in the 2°6pb/2°4pb vs. 2°4pb array (Fig. 7a) are congruent for all three experiments, indicating independence of PbSL from the kind of leaching acid we used. This graph is a modified mixing diagram (2°6pb/2°4Pb vs. 1/2°4pb) which better visualizes the absolute Pbc concentrations.

2) Fifty percent of the total common Pb in titanite is released during the first short leaching step, and common Pb release predominates over that of radiogenic Pb during the first two leaching steps after which the proportions are inverted (Fig. 7b)

3) Thorogenic and uranogenic Pb are released at a constant ratio (Fig. 7c) after 50 rain cumulative leaching time (step 2), reaching the value of 0.52 characteristic for the bulk analysis (Table 6). The first leaching step is characterized by elevated 2°Spb/2°rPb ratios, characteristic of a more common nature of the recovered Pb and consistent with the low measured 2°6pb/2°4pb and high 2°4pb contents in the initial step solutions (Fig. 7a,b). A proportion of this common Pb is attributable to the admixture of Pb from K-feldspar situated in micro- cracks, as the bulk 2°Spb/2°rpb ratio of a fraction enriched in K-feldspar (~50% Kfs, 50% titanite) is 0.69 and lies between the extreme values ranging from 1.44-1.96 in the first leach solutions of experiments B 1 (Table 6) and 0.52 for the bulk analysis of pure titanite.

4) Uranium and radiogenic Pb are almost proportionally released (i.e., chemical fractionation during the first five leaching steps is very small), as indicated by the quasi-

5)

linear arrangement of data points along a 1.0 Ga reference line in the 2°6pb/2°apb vs. # isochron diagram (Fig. 7d). The residue appears to be shifted towards the right. This is not to be interpreted as a younger age of the small zircon inclusions, as it is an artifact of the chemical fractionation during leaching. Preferential (but congruent) leaching of damaged zones of the zircon inclusions, which are characteristically discordant due to Pb-loss, may also explain the above shifted data point. Note that the reference line was not fitted through the first five leach steps and should theoretically pass through the bulk analysis, rather than the fractionated steps. The difference is not visible by eye as the fractionation is small in the first five steps, accounting for practically all Pb (>99%), and only the residue shows a large shift brought about by superimposing small fractionation on a very small Pb and U concentration. The y-intercept of the reference line yields a 2°6pb/2°4pb value of about 15.6-15.8, a very realistic figure for the composition of common Pb at about 1.0 Ga in the Grenville. This demonstrates that the large proportion of common Pb leached in the initial steps is in fact initially incorporated Pb and not surface contamination. The 238U/2°6pb vs. 2°Spb/2°6pb diagram (Fig. 7e) is able to discriminate between the first two steps, the intermediate steps 3-5, and the residue. Thereby, the first steps reflect their predominantly common Pb signatures, visible also to a lesser degree in the second step solutions. The longer lasting steps 3 and 4 do not show significant chemical fractionation between U and Pb, implying derivation from a single source, "titanite." Again, the residues, due to very small U and Pb concentrations, are distinctly different isotopically from the previous two step solutions.

4.5. Sm-Nd Isotope Results

Sm-Nd isotope data from a set of leach solutions (experiment B3 with 4N HBr) are presented in Table 4. Concentra- tions of Sm and Nd are comparable with independent ICP results (cf., Table 3 ). Except for the residual analysis, results from the step leach solutions do not show significant variations in the Nd isotope compositions while a ca. 3% chemical fractionation between Sm and Nd (~10 times that of the extemal reproducibility) is evident. The slight variation of Sm/Nd ratios, while the 143Nd/144Nd ratios remain the same, indicates that the element fractionation was laboratory induced (i.e., by leaching) and that a single mineral phase has been leached. In contrast to this, deWolf et al. (1996) found a large and correlated increase in 147Sm/144Nd and 143Nd/ 144Nd between dissolution steps of garnet. In the case studied by them, inclusions may have already been attacked by the leaching procedure, i.e., the linear array may have resulted from a two component (host-inclusion) mixing.

In our experiment, the influence of microscopically small inclusions is only seen in the residue. It is characterized by an elevated Sm/Nd ratio and by a higher 143Nd/la4Nd ratio relative to leach solutions. This is compatible with the inter- pretation of U-Pb isotope data of experiment B 1 as resulting


Fig. 4. SEM photographs showing details of the leached layer around a titanite grain. (a, b) Concentric shrinkage cracking takes place while exposing the grain to air, indicating the presence of a hydrated structure. The outer portions of the layer are smooth and probably composed of a silica gel-like material, probably built by re-polymerized silanol groups. (c, d) Details of the inner part of a leached layer, composed of an extremely porous, spongy structure through which liberated cations from the intact titanite core can diffuse easily. Minute whitish crystallites are precipitations of TiO2.

from the presence of small zircon inclusions. In this respect, it is noteworthy that zircon generally exhibits elevated Sm/ Nd ratios. We conclude that stepwise leaching protocols are not a promising dating tool in the Sm/Nd system due both to the Sm/Nd fractionation artefact and the negligible spread enhancement of the 143Nd/144Nd ratio in the absense of inclusions.

4.6. ICP Leach Solution Analyses

Main and trace element analyses of leach solutions from the PbSL experiment B3 were performed to study the cation release during leaching and specifically to trace the leaching response of major (Ca, Ti, Si) elements and their possible lattice substituents (A1, Fe, U, Th, Pb, REEs, etc). Analyses are listed in Table 3. Mass balance calculations (concentration of elements in titanite calculated by adding the total amounts of each element in the individual leach solutions

relative to the weight of titanite grains exposed to leaching) agree well with those from the EMP and PIXE analyses of unaffected titanite, except for those for Si, lost during the final HF attack of the residue in the form of SiF4. The solution from leach step 3 with 4N HNO3 had to be filtered prior to analysis as it contained a whitish suspension, one major component of which was identified by XRD as TiO2. The effects can be noticed in the lower concentrations, mainly of Ti and Pb, in this specific solution relative to those of the 4N HBr experiment, suggesting that these elements are preferentially precipitated in oxidizing media. This inter- pretation is consistent with the findings of the selective and retentive behavior of these elements in the leached layer (Figs. 3a, 5).

In order to understand the leaching behavior better, relative element release ratios (rrr values) were calculated for some important element pairs (Table 3). The rrr is the ratio of one element to another in the leach solution, relative to

404 R. Frei et al.

15B

48

18

8

0 0

64

8 8

48

48

24

1tl

8

0 0

64

66

4O

24

16

8

0

64

48

2 4

DO 0 8 16 24 32 40 48 68, 84 0

~ E L

84

× ~ E L

D O

e, 16 2 4 32 40 48 66 84 8 t 6 24 32 40 48 56

) , tP~L ~ L

66

48

4O

24 Fe

le~Oo 14ooo 8 I 20O0

[ 3 0 0 0

64

I 16 DO

64

58

48

48

24 U I 400 16 I 300 I 200

8 al ~00 130

0 O 8 16 24 32 40 48 56 84 Q ~ 16 24 32 40 4:8 56 54

X ~ L ~(P~EL

111 m 5oo I 376 I 2~50 J ~25

Fig. 5. PIXE maps of selected main and trace elements in a severely leached portion of a titanite grain, Leaching preferably takes place along micro-cracks (white), leaving behind islands of still intact titanite (darkest shadings) surrounded by thick transition layers (gray shades). Diffusion of liberated elements through the leached layer is indicated by the gradual increase of their concentrations, visualized by increasing gray scales towards the intact titanite relicts. Note that Ti, similar to Zr, is retained in areas where all other elements are removed entirely, and thus reaching concentrations up to 40 wt% (in the intact titanite, it is 18 wt%). This indicates the presence of Ti-rich precipitates, most probably TiO2. Pb is also partially retained in the leached parts of the grain. The rastered area is 150 × 150 ~zm, concentrations are in ppm.


6 4

56

4 8

40

24

16

8

0 0

64

5~

48

4O

24

16

8

0 13

8 ~6

8 16

64

48

4O

24

I I 28OO 16 I I 2100 i l 1400 8 I I 700 D O 0

24 32 ~ ~ ~ ~ 0 8 16 24 32

XPtXEL.

2'4 32

XPtXEL

64

40 48 56

Nd

III 1200 11600 O 0

64.

40 48 ~ 64

56

48

40

24 Zr I I 200 16 I I 150 I I t00 8 1150 O 0 0

la

(~tl 1 6 0 0 1 4 5 0 1 3 0 0 M 150 O 0

2~ 32 40 48 56 64

Fig. 5. (Continued)

that in the original mineral phase. The rrr value of the three main elements Ca, Ti, and AI relative to Si are >>1 for all steps (Table 3). This confirms the general picture obtained from the element maps in which it is clearly visible that over the entire procedure, Si is largely retained in the leached layer. This gel-like layer is even preserved at steady state element release conditions during which the reaction front progresses inward at the expense of the pristine titanite. Fig- ure 8 summarizes the relative release behavior of five element pairs into 4N HBr. With the exception of the first step and some divergence in the final residue, a near-constant relative release rate of all pairs shown in Fig. 8, closely conforming to their stoichiometric relative abundance in titanite, characterizes the intermediate leach period. This sug- gests realization of near steady state release conditions for these cations. Comparison of the different relative release rates during the first leach step yields a release sequence for the main elements from titanite as follows: A1 > Fe > Ca > Ti > Si.

Comparisons between trace elements and the major elements for which they could substitute are quite interesting. The REEs are more efficiently released into the solutions

than Ca at the beginning of a PbSL run (Table 3, Fig. 8). The same is even true for Th, which is released preferentially to Ca during the first leach step, but then becomes enriched relative to it in the residue (Fig. 8). This is consistent with the observation from the corresponding PIXE element maps (Fig. 5) in which Th shows a patchy distribution in the leached layer, suggesting its partial re-precipitation there. Furthermore, slightly decreasing Th/U ratios from step to step reveal that, similar to the REEs, Th is more easily released at the beginning of a leach procedure, whereas U preferentially enters the solution as leaching progresses. The HFS elements Ti, Zr, Hf, and Nb are very inefficiently released at early stages (see Table 3), indicating their preferential re-precipitation within the leached layer. Their strong enrichment in the final residue (especially Zr) further points to inclusions--the observation of rare <5 #m zircons was mentioned above. These few examples indicate that leaching is a complex process involving interaction of the acid, first,with the primary mineral, second, with the leached zone, and, finally, with inclusions.

Figure 9a and b show the rrr value of radiogenic (Pbr) and common (Pbc) Pb relative to Ca (for which Pbc is likely

406 R. Frei et al.

Table 6. U-Pb data of leach experiments on titanite from Otter Lake.

Z06pb / 2OSpb / 207pb / 206pb / 2o7pb /

Leach 2°4pb 2°6Pb U Pbr Pbtot 2°7pb/ _+20- Z°TPb/ _+2or 2t~Pb/ _+2o" 235U _+2o" 238U _+20- 2°6pb +2a -2o" Sample Code time (meas.) (meas.) (#g) (rig) (ng) 2°6pb (%) 235U (%) 23sU (%) (Ma) (abs.) (Ma) (abs.) (Ma) (abs.) (abs.) r +

Bulk, 10.8 mg

Otter A A1 1094 0.52 3.42 779 813 0.0725 0.4 1.682 0.7 0.168 0.6 1002 7 1003 6 999 8 8 0.84

4N HBr leach experiment, 100-200 #m grain size fraction, 61.5 mg

Otter B B3 15' 35 1.30 0.30 47 185 0.0749 18.5 1.236 19.7 0.120 1.3 817 161 728 9 1067 333 426 0.94 Otter B B3 50' 636 0.52 4.06 1199 1292 0.0725 1.3 1.634 1.6 0.153 0.8 983 15 976 7 1000 26 26 0.60 Otter B B3 4h 1785 0.47 10.70 2012 2066 0.0721 0.9 1.618 1.0 0.163 0.3 977 10 972 3 989 18 19 0.54 Otter B B3 8h 1383 0.46 2.94 631 652 0.0725 0.8 1.677 1.0 0.168 0.4 1000 10 1000 4 999 17 17 0.53 Otter B B3 res. 359 0.44 0.34 184 209 0.0730 3.4 0.966 3.8 0.096 0.4 686 26 591 2 1014 68 71 0.82

4N Hbr leach experiment, 200-300 #m grain size fraction, 83.4 mg

Otter B B1 10' 29 1.44 0.14 40 244 0.0604 34.9 1.942 36.7 0.233 1.8 1096 402 1352 24 616 616 1016 0.97 Otter B B1 40' 425 0.53 1.34 306 343 0.0723 1.9 1.721 2.1 0.173 0.4 1016 21 1027 4 995 37 38 0.59 Otter B B1 3h 1300 0.49 12.01 2676 2777 0.0725 1.1 1.678 1.8 0.168 1.3 1000 18 1000 13 1001 23 23 0.78 Otter B B1 6 h 2186 0.47 7.74 1729 1767 0.0725 1.0 1694 1.6 0.169 1.2 1.006 16 1009 12 1000 21 21 0.78 Otter B B1 13.5 h 2400 0.47 3.18 696 710 0.0726 0.8 1.662 1.3 0.166 0.9 994 13 990 9 1003 16 16 0.78 Otter B B1 res. 1807 0.48 0.37 54 56 0.0734 2.3 1.13l 3.4 0.112 2.3 768 26 683 16 1025 47 48 0.73

4N HNO3 leach experiment (1), 200-300 ~m grain size fraction, 83.7 mg

Otter B B1 10' 26 1.55 0.10 27 210 0.0574 55.3 1.752 58.1 0.221 2.9 1028 597 1290 38 507 900 2173 0.97 Otter B B1 40' 338 0.56 1.28 239 275 0.0703 2.1 1.358 2.2 0.140 0.4 871 20 846 4 937 42 43 0.53 Otter B B1 4h 1152 0.50 10.43 2205 2298 0.0720 0.7 1.580 3.1 0.159 3.0 962 30 951 28 987 15 15 0.97 Otter B B1 6h 2177 0.47 7.36 1700 1738 0.0726 0.9 1.753 1.2 0.175 0.7 1028 12 1040 7 1003 18 18 0.65 Otter B B1 13.5h 2148 0.46 3.70 807 825 0.0723 1.6 1.658 2.1 0.166 1.1 992 20 992 11 994 32 33 0.64 Otter B BI res. 1866 0.46 0.56 62 64 0.0726 2.4 0.845 3.0 0.084 1.5 622 19 523 8 1002 48 49 0.61

4N HNO3 leach experiment (2), 200-300 #m grain size fraction, 70.5 mg

Otter B BI 10' 34 1.28 0.21 44 209 0.0646 23.2 1.604 24.5 0.180 2.7 972 238 1068 29 761 425 583 0.53 Otter B B1 40' 338 0.55 0.75 211 243 0.0735 1.7 2.155 1.9 0.213 0.7 1167 22 1243 9 1027 33 34 0.54 Otter B BI 6h 2177 0.47 7.34 1776 1815 0.0730 1.0 1.839 1.5 0.183 1.0 1060 16 1082 11 1013 20 21 0.74 Otter B B1 13.5 h 2432 0.47 4.62 1065 1086 0.0726 1.2 1.756 1.8 0.175 1.2 1029 18 1042 12 1003 24 25 0.73 Otter B BI res. 1791 0.42 0.61 62 63 0.0715 2.4 0.781 3.8 0.079 2.7 586 22 492 13 973 49 50 0.77

* r = correlation coefficient :°7Pb/235U vs. a°6pb/23sU. Data corrected after Ludwig (1988) for fractional (0.068 -+ l l%/AMU, blank (80 -+ 30 pg) and common Pb (composition

Table 2). meas. = measured; Pbr = radiogenic Pb; tot - total; res. = residue.

of first step in experiment E2; c.f.

to substitute in the titanite structure) and Ti and Zr, respectively. Characteristically, and supportive of the isotope re- suits obtained in experiment B3, the rrr of Pbc/Ca of the initial and, less pronounced of the residual step solutions are -> 1, implying preferential and effective release of Pbc at the beginning of the leaching procedure and the recovery of slightly elevated Pb contents in the residue. This is also reflected in Pbc/ Ca < 1 for the intermediate steps. If the release of Pbc were unequivocally related to the release of Ca (for which it is likely to substitute), one would expect to observe rrr of - 1 . The fact the the rrr of Pbc/Ca in the first step is 32 (Table 3) implies that either Pbc is dissolved very efficiently at the leach "front" or that Pbc may be added from another common Pb-enriched source besides titanite. In contrast to Pb~, Pbr is initially very inefficiently released compared to Ca. Also, the residue is characterized by a slightly elevated Pbr/Ca rrr of ~ 1.8 and, therefore, indicates that Pbr (together with minor Pbc) is at least partly retained in the leached layer or in inclusions which are only attacked by HF.

The behavior during leaching of both types of Pb relative to Ti is shown in Fig. 9b. While the rrr values for Pbc/Ti

are even more extreme than for Ca, those for Pbr/Ti remain remarkably close to unity for all leach steps except the residue (for which the likely explanation is given above). This behavior indicates that radiogenic Pb has release characteristics similar to that of an HFS element, as also supported by the leach trajectory of Zr/Pbr (Fig. 9b). These results are in accordance with the observations from PIXE element distribution maps obtained from individual leached grains, which show some retention of Pb in the leached layer.

5. DISCUSSION

The question to be addressed now is twofold: First, is it possible to explain the effects observed entirely by heterochemical mineral phases (inclusions); second, if not, what could be the mechanism underlying a difference in leaching behavior between radiogenic and common Pb in a single silicate phase upon leaching?

The role of inclusions in geochronology has been critically discussed, e.g., by N/agler et al. (1996, 1995a,b), Kamber et al. (1996), deWolf et al. (1996), Lanzirotti and Hanson (1995), and many others. It is recognized that inclusions


can massively affect, if not dominate, the budgets of parent and daughter elements used for radiometric dating of minerals, and, thus, can lead to incorrect interpretations. For example, monazite and zircon inclusions in metamorphic garnets have been shown to affect the budgets of U, Pb, Sm, and Nd in that mineral severely, which has raised severe doubts about the reliability of U-Pb and Sm-Nd mineral chronology (e.g., deWolf et al., 1996). Inclusions are omnipresent in most common rock-forming silicates, even in the gem-quality Otter titanite. They may contain common (e.g., feldspar) or radiogenic (e.g., zircon) Pb.

The first step in all our experiments released the least radiogenic Pb. The fact that, in all PbSL experiments, the isotope composition of this first-recovered Pb lies on the PbSL isochron indicates that this component must be in isotopic equilibrium with the host titanite. The elevated 2°spb/2°6pb values measured in the first step solutions, relative to the ones in the successive steps (Fig. 7c), can represent the ratio of built-in common Pb which is not affected by the Th/U ratio of the titanite and, therefore, is not a decisve argument for inclusion Pb. We cannot fully discount a common Pb contribution by a separate included phase, considering that the presence of minor K-feldspar in micro-cracks within the titanite crystal has been documented more. However, this feldspar is present in amounts too small to account for the rather major initial common Pb release. Further, similar effects are observed for other element pairs, such as Fe/Ca, Fe/A1 (Fig. 8), or REEs/ Ca which are not K-feldspar specific. We believe that most of the common Pb released in the first leach step truly comes from the titanite itself.

A slight offset of the residue from a linear array in the 2°spb/2°4pb vs. 2°6pb/2°4pb plot, together with elevated # and Sm/Nd values, strongly indicates the presence of zircon inclusions, which were indeed observed (see above). Simi- larly, deWolf et al. (1996) identified and discriminated zircon and monazite in garnet by using Th/U vs. Nd/U and Srn/Nd vs. U/Nd element discrimination diagrams. It would seem attractive to use also the zircon-specific elements Zr and Hf for this purpose; however, in PbSL, their retentivity in the leached layer precludes this.

As the grain size used in experiments is decreased (runs E2, B3), the Pb isotopic effect of inclusions becomes progressively visible in earlier leaching steps. This may be mainly due to mass balance proportions between leached, relative to unleached, grain volumes and the amounts of inclusions. In the case of the Otter titanite, the Pb isotopes indicate that the zircon inclusions were formed contempora- neously and in isotopic equilibrium with the host titanite at 1.0 Ga, as also supported by Sm-Nd isotope data.

A discrimination diagram capable of discriminating different inclusions is the three-element graph shown in Fig, 7e. In this diagram, points that coincide (such as steps 2 - 4 of separate experiments) represent one and the same carrier phase, inclusion-free titanite. Thus, it becomes possible to identify leach steps carrying an isotopic signature from inclusions. It then becomes abundantly clear that the Pb isotope data spread, which enhances the precision of Pb-Pb dating and makes PbSL so attractive, is observed in leach steps whose isotopic and elemental ratios show no contribution

from any inclusions, and, thus, must be explained as a two component mixing of built-in common and in situ produced radiogenic Pb components of titanite itself.

Following this observation, we have to address the second question posed above, concerning the different behavior of radiogenic and common Pb in a single phase. This discussion has to center around the object attacked (the titanite crystal) and account for our observations of the processes taking place during leaching. A perspective view of the idealized crystal structure of titanite is shown in Figure 10. Chains of Ti-octahedra are crosslinked by isolated Si-tetrahedra. Cal- cium ions are located in sevenfold coordinated sites between the chains. The results obtained in this study, i.e., the combination of solution data and visualization of element distribution in leached titanite grains, clearly indicate that a combined dissolution-leaching reaction with a progressively inward-moving, well defined reaction front characterizes an acid step-leach procedure (see Fig. 3). Our earlier tentative working hypothesis (see Introduction) was that radiogenic Pb, as it resides in lattice defects, would generally be removed more easily from a mineral during leaching than primarily built-in Pb, which substitutes mainly for Ca. This working hypothesis must be wrong for two reasons. First, the release patterns of common and radiogenic Pb are the opposite of what it would predict, and, second, it was based on the notion that the lattice would remain largely intact during leaching-- now, certainly in the case of titanite, shown to be incorrect.

Dissolution experiments of silicate minerals and glasses have led to two contrasting models, suggesting control of dissolution rate by diffusion- or surface-reaction kinetics during hydrolysis reactions. Formation of leached layers up to several 1000 .~ thickness, with preferential release of alkalis, for example, have been observed during dissolution processes (e.g., Chou and Wollast, 1984, 1985, and references therein), and it has been proposed that such layers limit diffusion of reaction products through them. In contrast, several workers (e.g., Schott and Bemer, 1983, and others) did not find leached layers of comparable thickness in experiments. This casts some doubt on the validity of a pure diffusion model. It is generally accepted that detachment reactions (mainly cracking of activated complexes) at the solid/liquid interface control the dissolution rate, but the observation of polymerization and other reactions in leached layers (Bunker et al., 1988; Casey et al., 1988, 1989) does not discount diffusion as an additional possible process during dissolution of minerals.

In our experiments, no diffusion within the primary mineral was observed. Instead, there appears to be a well-defined reaction front which progresses inward into the grain. All main elements except Ti (i.e., Ca, A1, and Fe) are preferentially removed over Si, leaving behind a loose corona structure that approximately preserves the original shape of the grain. A likely explanation for the observed features is that the crystal lattice is transformed at the "front" into a hydrated, amorphous solid due to preferential release of Ca from sevenfold positions and Ti from octahedral ones (both bonded by bridging oxygen from SiO4 tetrahedra, Fig. 10). An EDS-spectmm of a leached layer around a relict titanite grain exposed to 4N HNO3 for over 12 h shows a Si/O ratio

408 R. Frei et al.

200

16o

CIw 120 L

8o

40

200

160

120

80

40

• i • i • i • i • l • i •

• i nite ; / 4N HBr

' 13.5h 994 _. 11 Ma 40 (MSWD=0.74) res.

• 0' o ' . . . . . . . . . . 0 4 0 8 0 1 2 0 0 1600 2 0 0 0 2 4 0 0 2800

2°6pbf°'pb

Ti'taniteB1 ' ' ' ' ' £h' ' ~13.5hi

4 N H ~

,., : ._ cv ~ ' . 1004 _. 10 Ma ; / 40' (MSWD=O.22)

, I . O I . I , I . I , I . I

0 400 8 0 1 2 0 0 1600 2000 2400 2800

2°6pbf°~pb

160

..0

120

8o

f .

• I • I • I • I ' I • I •

~oo 6 h i / : Titanite B1 J "" 13.5h 4N H N O 3 ~ res.

/ 3h3h 10' / 1000 _+ 10 Ma

40 ~,a," 40 ' (MSWD=-I.40)

I , I , I , I , I m I , I ' 400 800 1 2 0 0 1 6 0 0 2000 2 4 0 0 2800

2°6pbf°4pb • I • I ' I " I • I - I ' ~ q • •

200 Titanite 6h _/-*/ 160 B1 ~ 13.5h

12O8o i N H N ~ / / res.

~ 4 0 ' 1014 ___ 13 Ma 40 (MSWD=I.71)

• 0' ' 0 ' . . . . . . . . . . 0 4 0 8 0 1 2 0 0 1 6 0 0 2 0 0 0 2 4 0 0 2800

2°6pbf°'pb

1400

1200

i000

800

600

400

200

1400

1200

i000

800

600

400

200

T t'an;te' ' ' ' ' ' ; / t E2 3 ~ 4N HBr

- . ~ , - - , , , ~ g . ~ , , , , , . . . . 400 800 1 2 0 0 1 6 0 0 2000 2 4 0 0 2800

~°'pbf°'pb

• al it'n'te' ' '

BH1Br 6~2k'13.5h

~ e s .

' 4o, ' 0 ' ' 0 ' ' ' ' ' ' ' . . . . 4 0 8 0 1 2 0 0 1600 2 0 0 0 2 4 0 0 2800

2°6pb/2°4pb

1400

tOO0

800 J~

600 R

400

' I ' I • I • I ' i • I • I , , j

12oo Titanite 65 J

4 N H N O 3 ~ ~ 1 3 . 5 h s .

3h

200

° ' 40'0 ' 80'0 ' 1 2 ; 0 ' 1 6 ; 0 '20~0 ' 2 4 ; 0 "28;0

2°'pbf°~pb

T i t ' a n i t e ' ' ' ' 6h " Y B1 J - 1 3 . 5 h

4N H N ~ ~ , / w res.

200 / ~ 0 ~

• o' 0 ' . . . . . . . . . . 0 4 0 8 0 1 2 0 0 1 6 0 0 2 0 0 0 2 4 0 0 2800

2°6pbf°'pb

1400

1200

1000 ,.~

t 800

"~ 600

400

Fig. 6. Uranogenic and thorogenic Pb diagrams with PbSL step solution data from experiment B1 and E2 performed on different grain size fractions of titanite. Isochrons result in all cases, defining consistent ages of - 1 Ga. The Pb isotope composition of the solution removed after 3 h could not be analyzed in experiment B1.A near ideal linear arrangement of data points in the thorogenic diagrams is reached for the experiments B 1 performed on the 200-300


of ~1 :2 (SiO2), instead of the 1:5 characteristic of pure titanite. A further observation supporting the above mechanism is that shells separate concentrically from each other when leached grains are exposed to the air (cf. Fig. 4). This may indicate shrinkage due to dehydration of such an amorphous, hydrated substance. Similar mechanisms have been proposed, among others, by Bunker et al. (1988) and de Jong et al. (1979, 1980) for silicate glasses and by Hell- mann et al. (1990), Casey et al. (1989), and Helgeson et al. (1984) for feldspars.

In contrast to the many investigations in which leached layers were observed to become only several hundreds of thick during silicate dissolution, our example demonstrates that a leached layer may be preserved entirely throughout the whole grain and may reach of up to 150 # m in thickness. Furthermore, our results imply that this layer is not very rate- limiting with respect to diffusion of major cations through it, but preferentially adsorbs, or retains, HFS elements during the step-leach procedure. More precisely, Ti, Zr, Hf, and Nb, as well as U and Th, are retained to a variable degree within the micro-environment of the leached layer, which may be regarded as a selective cation adsorber, particularly effective for HFS-elements.

The strong predominance of radiogenic over common Pb in the Otter titanite has made it possible to distinguish and trace the different behavior of these two lead components in acid leaching, as shown above. In summary, common Pb is highly mobile, while the leaching pattern of radiogenic Pb is that of a high field strength element (most closely resem- bling Ti) . This observation from solution analyses is consistent with the PIXE element mapping (Fig. 5), where Pb was detected in the leached layer. These HFS-like characteristics may be explained if it is assumed that radiogenic Pb, at least to an important degree, is present in its highly oxidized (i.e., 4+ ) state. The sixfold coordinated ion radius of Pb 4-- is 0.86 ,~, similar to 0.7 for Ti, 0.8 for Zr, 0.79 for Hf, and 0.72 for Nb (Whittaker and Muntus, 1970), which indeed characterizes Pb 4+ as a high field strength species.

Such a high oxidation state of radiogenic Pb can be understood from the decay process itself. Considering, by first approxima- tion, the splitting up of a nucleus at rest into two particles of masses (ml and m2), then it can be shown from the law of retention of impulse that the kinetic energies E of the two particles are related by Exml = F-am2. In the a-decay of nuclides with masses around 200, mflm2 ------ 50 (m2 is the mass of the a-particle = 4). With a total decay energy around 5 MeV (with ca. 1 MeV lost in as 3'-rays), this means that the kinetic energy of the recoiling nucleus is around 100 KeV. This is still more than 4 orders of magnitude greater than the energy thresholds associated with any valence electrons or ionic bonds in crystals. Therefore, it is expected that (1) the nucleus will recoil out of its lattice position and (2) it will be stripped of at least all of its valence electrons. An analogy is seen in fission tracks, which

are considered to have been made by highly positively charged particles (Fleischer et al., 1975). The place where the ion comes to rest becomes a defect in a, presumably, previously charge- balanced area of the lattice. It is, therefore, unlikely that it will be reduced beyond its highest common oxidation state. For Pb, this is +4. Likewise, 2°Tpb, which is produced finally by the/3- decay of 2°7T1, is expected to be in the +4 ionization state. The T1 should be in its highest oxidation state (+3 ) and the/3-decay, of which there is no recoil, will then produce Pb 4+ if the electron configuration is not changed. Thus, it can be argued that all in situ produced radiogenic Pb in a mineral, irrespective of its parent isotope or decay branching, should exist in the tetravalent state in the first instance.

PbSL experiments are broadly reproducible under equal experiment parameter conditions. The ratio of radiogenic to primarily built-in common Pb released increases with leach duration and appears to be only weakly dependent on whether HBr or HNO3 is used, although the tendency of radiogenic Pb to be precipitated in the porous shell appears to be somewhat greater in leaching with 4N HNO3. The electronegativity of both NO3 and B r - is smaller than that of Pb 4+ and larger than that of Pb 2+ . Thus, no very different behavior between the two acids is expected. Pb release from the titanite (disregarding inclusions) is, then, principally controlled by two processes, i.e., ( 1 ) a relatively rapid surface hydrolysis of the mineral and (2) a much slower removal of HFS-elements (and tetravalent radiogenic Pb with similar characteristics), primarily re-precipitated within the leached layer between the inward progressing, hydrolyzing surface, and the grain boundary.

As intuitively expected, the leaching procedure is faster if a smaller grain size is used (Table 2), but there is no significant effect on the highest 2°6pb/2°4pb ratios measured. However, the relative release rates of common and radiogenic Pb at the early stages of leaching are clearly dependent on the grain size of the leached mineral fraction. Results from the first leach step of experiments E2, B1, and B3 (Table 2) indicate decreasing 2°6pb/2°4pb ratios, thus more common-dominated Pb isotopic compositions, with increasing surface/volume ratios (decreasing grain sizes) of the titanite fractions. Using the above scenario with the two release processes, this can be explained as follows: In a small grain-size fraction, the primary leaching of the grains, being a surface process, is relatively rapid. Therefore, the attendant release of common Pb may be completed early, while radiogenic Pb is still well-retained in the leached layer. With larger grains, there are more remnants of unleached titanite still present at later stages of leaching when radiogenic Pb is already released from the leached layers by the second process. The two release types overlap in time, leading to a more mixed Pb signature of the leach solutions.

A greater surface to volume ratio is also produced by

/~m grain size fraction of titanite, with the exception of the residues which consistently lie below this line. Deviations from a line is more pronounced in the data from experiment E2 performed on the 50-100 #m grain size fraction, indicating that the contribution of foreign lead from minute zircon inclusions, due to its prolonged exposure to the leach acid, is earlier noticeable than in the identical procedure applied to the coarser grain size fraction.

410 R. Frei et al.

3000

2500

2000

~ 1500

1000

500

0

a

res. # 3h

/ 40' d l ~ 10'

4N HBr, 4N HNO3 I I

101 10 2 10 3

2°*Pb (pg)

2.0-

1.5

~'-1 0

0.5

0.0

14

12

10

L

¢x,

o ~

,.Q 104

C 4N HBr, 4N HNO3

surface release 10' &

K-feldspar

23h20' 25h20' 50' 3h50' 9h50' ~ /

i z T

0 5 10 15 20 25 cumulative time (hours)

e ~ res.4N HBr, 4N HNO+

steps (t~l step 2 3-5 -;j

step 1

I

0 1 2

~o,pb/2O+pb

60

50

40

30

20

10

0

3000

2500

20OO

3o ~ 15oo

lOOO

500

b 2o4pb

// \ // ~>, 4N //-- \X HNO3 ' ~ \kkx- / 4N

I I l I I I

1 2 3 4 5 6 7

step

d 13.5h T {

rosopic

0 S ~ 6 ¢ solid inclusions 4 ~°~ (<5 gm zircon)

1 / - ~ + ~ 4N HBr, 4N HNO+ r i i i i

0 5000 10000 15000 20000 25000

I.t (=SU/~Pb)

Fig. 7. (a) 2°6pb/2°4pb vs. 2°4pb diagram with leach trajectories of data from experiment B 1. There is no difference in the response towards leaching of Pb with HBr and HNO3. (b) Percentual release of common (traced by 2°4pb; circles) and radiogenic (traced by 2°6Pb; squares) lead in individual steps of experiment B3. During the first 10 min step, ~50% of the total common Pb is released. Radiogenic Pb release overtakes that of common Pb after step 2


N d / C a o 2 ;

"~ " Fe /Ca ~ Ca /Th

~ intermediate leach solutions •

eL) ~ . . . . 2 . _ ~ : . ~ , ~ . ~ . . . . . .

~. - steady state ~D

\ Fe/A1

\ Ca/A1

" O

9'

I I I I I t I

0 2 4 6 8 10 12 14 16

cumula t ive t ime (hours)

Fig. 8. Relative release ratios of selected element pairs in step leach solutions relative to their abundance proportions in the titanite as a function of time in experiment B3. Steady state release is approximately attained after the first short leaching step, in which metal cations are extensively, but variably liberated by efficient surface hydrolysis.

micro-cracks, which appear to enhance leaching, as acids are able to penetrate along them into the center of grains. This observation is consistent with that by Hellmann et al. (1990) who note that 60 to 90% of released Na and A1 from albite during acid and base leaching is non-stoichiometric and likely to derive from deep leaching along dislocation cores, micro-cracks, and from other " in te rna l" surfaces.

Thus, it appears that the present working hypothesis on PbSL dating, which is based on the postulated valence difference between common and radiogenic lead in a silicate mineral, allows for the explanation of the observations on Pb isotope variations obtained by leaching. The question provid- ing the impetus for the present s tudy- -whe the r the presence of heterochemical inclusions is a necessary prerequisite for obtaining an enhanced 2°rpb/E°4pb and 2°7pb/E°4pb data

sp r ead - - can be answered in the negative. Variations in Pb isotopic ratios are observed in the absence of any chemical indicators that would diagnose heterochemical inclusions,

and we offer an explanation for these Pb isotopic variations. Further, PbSL offers the possibility to check for inclusions by using suitable isotope and element correlation diagrams for the step-to-step trajectories of solution data. Moreover, inclusions which were formed in isotopic equilibrium with their host do not perturb PbSL isochrons, as demonstrated by the presence of minute zircon inclusions, even within the gem-quality titanite from Otter Lake.

The further observation that the U-Pb ages obtained in the PbSL experiments are also consistent and agree with the Pb-Pb age (Table 6, Fig. 7d, except for residue-zircon which appears to have lost radiogenic Pb) still needs to be addressed. This was also observed by Frei et al. (1995) for a Variscan staurolite, where cogenetic inclusions could be the cause. In the case of Otter lake, we are sure that a broadly correct U-Pb isochron age was produced by the titanite itself, i.e., that radiogenic Pb and U seem to behave coherently during all the processes of leaching such as hydrolysis, re- adsorption, and dissolution of the leached rim. This baffling observation is, of course, of practical interest as it would open the door to the application of PbSL dating to young metamorphic minerals. A tentative explanation that can be offered in the framework of the valency working hypothesis is that Pb 4+ and U 4+ are both retained in the leached layer to the same extent. They probably have similar radii in sixfold coordination: Whittaker and Muntus (1970) give values of 1.02 for Pb 4+ and 1.08 ,~ for U 4÷ in eightfold coordination; Shannon and Prewitt (1970) give 0.775 and 0.89 in sixfold coordination. The differences are no greater than between Ti 4+ and Pb 4+, which also display coherent leaching behavior (see above and Fig. 9b). Similar retention-release behavior of radiogenic Pb 4+ and U 4+ in the leached layer might lead to the observed apparent U-Pb age consistency. How- ever, this could only hold as long as Pb is not reduced to Pb 2÷ and U is not oxidized to U 6÷ , both of which are likely to happen at some stage during leaching. In spite of this, the consistency is too good to be entirely serendipitous. While dating by stepwise leaching using U-Pb appears as a cavalier proposition, its possibilities may be worth exploring.

6. CONCLUSIONS

We addressed the mechanism of dating single silicate minerals by Pb-Pb using successive acid leaches on a museum quality titanite.

1) We observe a spread in 2°6pb/2°4pb and 2°7pb/z°4pb ratios in all steps. Even in the three central steps, characterized

(crossing point of the two trajectories marked with a circle), giving rise to increasingly radiogenic Pb compositions in a PbSL run. (c) Time-dependence of 2°Spb/z°6pb in individual step leach solutions from experiment B1. The elevated 2°Spb/Z~Pb values in the first step relative to the values in the successive ones indicate that a surface reaction liberates predominantly primarily built-in common Pb. Subordinate contribution of Pb from K-feldspar within micro- cracks transecting the titanite crystal cannot be fully excluded. (d) Isochron diagram from individual step leach solutions of experiment B1. Except for the residues, a more or less linear arrangement of the data points along the 1.0 Ga reference isochron indicates that U and Pb are chemically fractionated during leaching only to a small extent. For the apparent fractionation of the residue, indicative also for the presence of small zircon inclusions, see text. (e) Chemical correlation graph of experiments B1, B3 and E2. The abscissa is proportional to the Th/U ratio, the ordinate is the U/Pb, ratio• Note the similarity of the three trajectories. Most of the Pbr is released in steps 3-5, and the representative points are all indistinguishable, implying derivation from a single reservoir, "titanite" (see text)•

412 R. Frei et al.

by identical e lement ratios and account ing for > 9 0 % of the radiogenic Pb, the ratio of c o m m o n Pb to radiogenic Pb decreases.

2) We observe the formation of a white gel-like layer surrounding brown intact ti tanite cores.

3) W e observe that the HFS-elements are essentially the only e lements enriched in the gel-like layer. Lead behaves evidently like HFSE.

We interpret these three observat ions as follows:

1) On surface hydrolysis (a t the leaching f ront) , the lattice

o

..~

100.00

10.00

1.00

0.10

0.01

100.00

0 10.00

'-~ 1.00

'.,..~

0.10

0.01

a

PbJCa

0

Pbr/Ca I I I I I I I

2 4 6 8 10 12 14

t i m e ( h o u r s )

b ~ Vbfri /

i Pbfri /

f Zr/Pbr

16

i i t i i i i i

0 2 4 6 8 10 12 14 16

t i m e ( h o u r s )

Fig. 9. (a) Release of common (Pbc) and radiogenic (Pbr) Pb relative to Ca and their respective abundance in the unleaehed titanite as a function of time in experiment B3. The preferential release of Pbc during the first step is indicated by the high relative release ratio of 32 and points to an efficient surface dominated hydrolysis of built-in Pbc in titanite besides possible contribution of foreign Pbc from K-feldspar located in micro-cracks. (b) Relative release rate of common (Pbc) and radiogenic (Pbr) Pb relative to the HFS- elements Ti and Zr as a function of leaching time in experiment B3.

t e t r a h e d r a l si tes: Si

o c t a h e d r a l si tes: Ti, A1, Fe, U

s e v e n f o l d sites: Ca, R E E ' s , Th , P b

Fig. 10. Schematic structure of titanite with the three possible cation sites. Silica in tetrahedral sites is bonded by oxygen bridges to cations in octahedral and sevenfold positions.

is effectively destroyed and Ca, along with originally lat t ice-bound common Pb 2+, is released and removed into the leach solution as the front proceeds.

2) Upon destruction of the mineral lattice at the leaching front, a porous gel-like hydrated Si-and Ti-oxide is produced which forms a zone around the still intact cores. HFS-elements such as Ti, Zr, Hf, and Nb are retained in this gel in the first instance, and radiogenic Pb 4+ follows this behavior.

3) During further leaching the HFS species, including Pb 4-- , are gradually released over 10 h or more.

The release of Pb during leaching is therefore controlled by two mechanisms proceeding at different rates, namely, a fast surface hydrolysis of cations at the reaction front and a slow remobil izat ion of initially trapped cations f rom the leached zone.

The fact that these processes operate at different rates results in an effective separation of common and radiogenic Pb in stepwise leaching of silicates. This makes PbSL dating independent of the contr ibut ion of radiogenic Pb from (omnipresent) inclusions, a l though these can massively affect the budget of parent and daughter elements of commonly applied geochronometers . PbSL provides a test for the presence of inclusions and is capable of verifying whether or not they had isotope exchange with their host as this crystallized. Isotopically equilibrated, and, therefore, more or less con- temporaneous inclusions will not perturb PbSL isochrons.

Acknowledgments~We would like to express our thanks to R. Keil from the Paul Scherrer Institute at Wtirenlingen for the ICP analyses of the leach solutions, and to R. Johner and H. Zweili for the XRD analyses and for help with the SEM, respectively. The titanite specimen NMBE A7342 was donated by the Naturhistorisches Museum Bern. R. Kretz is thanked for help in identifying the sample location


of the experimental titanite. From the side of the National Accelera- tor Centre, we acknowledge H. Schmitt for operating the accelerator, and K.A. Springhorn and J. V. Pilcher for technical and software assistance. Constructive and helpful reviews by C. Garirpy, P. Schweda, A. Lanzirotti, and K. R. Ludwig are greatly acknowledged. This work was funded by the Schweizerischer Nationalfonds through grant 20-40442.94 to JDK.

Editorial handling: K. R. Ludwig

REFERENCES

Blenkinsop T. G. and Frei R. (1996) Archean and Proterozoic mineralization and tectonics at Renco mine (Northern Marginal Zone, Limpopo Belt, Zimbabwe). Econ. Geol. (In press)

Bunker B. C., Tallant D. R., Headley T. J., Turner G. L., and Kirk- patrick R.G. (1988) The structure of leached sodium silicate glass. Physics and Chemistry of Glasses 29, 106-120.

Buick I. S., Frei R., and Cartwright I. (1997) The timing of high- temperature retrogression in the Reynolds Range, central Austra- lia; constraints from single mineral Pb-Pb dating. Contrib. Min- eral. Tetrol. (submitted).

Casey W. H., Westrich H.R., and Arnold G.W. (1988) Surface chemistry of labradorite feldspar reacted with aqueous solutions at pH = 2, 3, and 12. Geochim. Cosmochim. Acta 52, 2795-2807.

Casey W.H. , Westrich H.R., Arnold G.W., and Banfield J.F. (1989) The surface chemistry of dissolving labradorite feldspar. Geochim.Cosmochim. Acta 53, 821-832.

Chou W. H. and Wollast R. (1984) Study of the weathering of albite at room temperature and pressure with a fluidized bed reactor. Geochim. Cosmochim Acta 48, 2205-2218.

Chou W. H. and Wollast R. ( 1985 ) Steady-state kinetics and dissolution mechanisms of albite. Amer. J. of Sci. 285, 963-993.

de Jong B. H. W. and Brown G. E. (1979) Polymerization of silicate and aluminate tetrahedra in glasses, melts, and aqeous solutions- I. Electronic structure of H6Si207, HrA1SiOI7 , and H6A120~-. Geochim. Cosmochim. Acta 44, 491-511.

de Jong B. H. W. and Brown G. E. (1980) Polymerization of silicate and aluminated tetrahedra in glasses, melts, and aqueous solutions- II. The network modifying effects of Mg 2+ , K + , Li + , H + , O H - , F , C1 , H2), CO2, and H3 O÷ on silicate polymers. Geochim. Cosmochim. Acta 44, 1627-1642.

deWolf C.P. and Mezger K. (1994) Lead isotope analyses of leached feldspars: Constraints on the early crustal history of the Grenville Orogen. Geochim. Cosmochim. Acta 58, 5537-5550.

deWolf C. P., Zeissler C. J., Halliday A. N., Mezger K., and Essene E.J. (1996) The role of inclusions in U-Pb and Sm-Nd garnet geochronology: Stepwise dissolution experiments and trace uranium mapping by fission track analysis. Geochim. Cosmochim. Acta 60, 121-123.

Fleischer R.L., Price P.B. , and Walker R.M. (1975) Nuclear Tracks in Solids. Principles and Applications. Univ. Calif. Press.

Franz G. and Spear F. S. (1985) Aluminous titanite (sphene) from the eclogite zone, south-central Tauern window, Austria. Chem. Geol. 50, 33-46.

Frei R. (1996) The extent of inter-mineral isotope equilibrium: a systematic bulk U-Pb and step-leach Pb isotope study of individual phases from the Tertiary granite of Jerissos (Northern Greece) European J. Mineral. 8, 1175-1189.

Frei R. ( 1995 ) A new Pb-dating technique for gold deposits: Exam- ples from Zimbabwe. In Centennial Geocongress, Geological So- cietT of South Africa, Johannesburg, South Africa, April 3-7, 1995, pp. 64-67 (ext. abstr.).

Frei R. and Kamber B. S. ( 1995 ) Single mineral Pb-Pb dating. Earth Planet. Sci. Lett. 129, 261-268.

Frei R. and Pettke T. (1996) Mono-sample Pb-Pb dating of pyrrho- rite and tourmaline: Proterozoic vs. Archean gold mineralization in Zimbabwe. Geology 24, 823-826.

Frei R., Biino G.G., and Prospert C. (1995) Dating a Variscan pressure-temperature loop with staurolite. Geol. 23, 1095-1098.

Helgeson H. C., Murphy W. M., and Aagaard P. (1984) Thermody-

namic and kinetic constraints on reaction rates among minerals and aqueous solutions. II. Rate constants, effective surface area, and the hydrolysis of feldspar. Geochim. Cosmochim. Acta 48, 2405-2432.

Heaman L. and Parrish R. ( 1991 ) U-Pb geochronology of accessory minerals. In Shortcourse Handbook on Applications of Radiogenic Isotope Systems to Problems in Geology (ed. L. Heaman and J.N. Ludden), pp. 59-102. Mineral. Assoc. Canada.

Hellmann R., Eggleston C. M., I-lochella Jr. M. F., and Crerar D. A. (1990) The formation of leached layers on albite surfaces during dissolution under hydrothermal conditions. Geochim. Cosmochim. Acta 54, 1267-1281.

Hollabaugh C. L. and Rosenberg P. E. (1983) Substitution of Ti for Si in titanite and new end-member cell dimensions for titanite. Amer. Mineral. 68, 177-180.

Kamber B. S., Blenkinsop T. G., Villa I. M., and Dalai P. S. (1995) Proterozoic transpressive deformation in the Northern Marginal Zone, Limpopo Belt, Zimbabwe. J. of Geol. 103, 493-508.

Kamber B. S., Frei R., Gibb A. J., and O'Nions R. K. (1996) Granu- lite chronometry-how accurately can we interpret precise ages? Journal of Conference Abstracts, Sixth V.M. Goldschmidt Confer- ence, 31 March-4 April, Heidelberg, Germany, 1, 300.

Kretz R. (1960) The distribution of certain elements among coexist- ing calcic pyroxenes, calcic amphiboles and biotites in skarns. Geochim. Cosmochim. Acta 20, 161-191.

Kretz R. ( 1993 ) Petrology of veined gneisses of the Otter complex, southern Grenville Province. Canadian J. Earth Sci. 31, 835-850.

Krogh E. J., Andersen A., Bryhni J., and Kristensen St.-E. (1982) Tectonic setting, age, and petrology of ecoglites within the upper- most tectonic unit of the Scandinavian caledonides. TromsO area, northern Norway. Terra Cognita 2, 316.

Lanzirotti A. and Hanson G. N. (1995) U-Pb dating of major and accessory minerals formed during metamorphism and deformation of metapelites. Geochim. Cosmochim. Acta 59, 2513-2526.

Ludwig K. R. (1988 ) A computer program to convert raw U-Th-Pb isotope ratios to blank-corrected isotope ratios and concentrations with associated error-correlations. USGS Open-file Report OF- 88-0557.

Ludwig K. R. and Silver L. T. (1977) Lead-isotope inhomogeneity in Precambrian igneous K-feldspars. Geochim. Cosmochim. Acta 41, 1457-1471.

Mattinson J. M, (1994) A study of complex discordance in zircons using step-wise dissolution techniques. Contributions to Mineral. and Petrol. 116, 117-129.

Mezger K, Hanson G. N., and Bohlen S. R. (1989) U-Pb systematics of garnet: dating the growth of garnet in the Late archean Pikwito- nei granulite domain at canchon and Natawahunan lakes, Mani- toba, Canada. Contributions to Mineral. and Petrol. 101, 136- 148.

Montel J.-M., Veschambre M., and Nicollet C. (1994) Datation de la monazite ~t la microsonde 61ectronique. Compte Rendue de l'Acad~mie Science de Paris 318, 1489-1495.

Nagler Th. F., Pettke T., and Marshall D. (1995a) Initial isotopic heterogeneity and secondary disturbance of the Sm-Nd system in fluorites and fluid inclusions: A study on mesothermal veins from the central and western Swiss Alps. Chem. Geol. 125, 241-248.

Nagler Th. F., Holzer L., and Frei R. (1995b) Nd and Pb in monazite inclusions within polymetamorphic garnets: hell or heaven? EOS, Transactions, AGU, Fall Meeting, 11 - 15 December 1995, 76/46, F707 (Abstr.).

Nagler Th. F., Holzer L., and Frei R. (1996) What dominates the Nd and Pb characteristics of garnet? Journal of Conference Abstracts, Sixth V.M. Goldschmidt Conference, 31 March-4 April, Heidel- berg, Germany, 1, 424.

Prozesky V. M.et al. (1995) The NAC nuclear microprobe facility. Nuclear Instrument Methods B104, 36-42.

Richard P., Shimizu N., and All~gre C.J. (1976) 143Nd/146Nd, a natural tracer: An application to oceanic basalts. Earth Planet. Sci. Lett. 31, 269-278.

Ryan C. G. and Jamieson D. ( 1993 ) Dynamic analysis: on-line quantitative PIXE microanalysis and its use in overlap-resolved elemental mapping. Nuclear Instrument Methods B77, 203-214.

414 R. Frei et al.

Ryan C. G., Cousens D. R., Sie S. H., Griffin W. L., and Suter G. F. (1990a) Quantitative PIXE microanalysis of geological materials using the CSIRO proton microprobe. Nuclear Instrument Methods, B49, 271-276.

Ryan C. G., Cousins D. R., and Sie S. H. (1990b) Quantitative analysis of PIXE spectra in geoscience applications. Nuclear Instru- ments Methods' B49, 271-276.

Schott J. and Berner R. A. (1983) X-ray photoelectron studies of the mechanism of iron silicate dissolution during weathering. Geo- chim. Cosmochim. Acta 47, 2233-2240.

Shannon R.D. and Prewitt C.D. (1970) Revised values of effective ionic radii. Acta C~stallographica B26, 1046 1048.

Smith D. C. and Lappin M. A. (1982). Aluminium-and fluorine-rich sphene and clino-amphibole in the Liset eclogite pod, Norway. Terra Cognita 2, 317.

Tapper U. A. S.et al. (1993) High-brightness proton beams at the

NAC nuclear microprobe. Nuclear Instruments Methods B77, 1724.

Vance D., Belshaw N. S., and O' Nions R. K. ( 1991 ) U-Pb cbronom- etry of metamorphic garnets: a comparison of TIMS analysis of garnet and high resolution SIMS analysis of U-rich inclusions. EOS 72, 560.

Vinyu M. L., Frei R., and Jelsma H. A. (1996) Timing between granitoid emplacement and associated gold mineralization: examples from the ca. 2.7 Ga Harare-Shamva greenstone belt, Northern Zimbabwe. Canadian J. Earth Sci. 33, 981-992.

Whittaker E.J. W and Muntus R. (1970) Ionic radii for use in geochemistry. Geochim. Cosmochim. Acta 34, 945-956.

Zhou B. and Hensen B. J. (1995) Inherited Sm/Nd isotope components preserved in monazite inclusions within garnets in leuco- gneiss from East Antarctica and implications for closure temperature studies. Chem. Geol. 121, 317-326.

Single mineral dating by the PbPb step-leaching method: Assessing the mechanisms

Documents