Dating of metamorphic events by the Sm-Nd garnet (Grt) method is subject to a number of uncertainties. First, Grt may not have been in isotopic equilibrium with the whole-rock (WR) or other minerals (Thöni and Jagoutz, 1992). Second, inclusions predating Grt may preserve their initial isotope signature (Zhou and Hensen, 1995). Third, the question remains unsolved of whether Grt has a closure temperature (Tc) for Sm-Nd below the formation temperature, or whether diffusivity of Pb and especially Nd3+ is so low that isotope signatures are established during Grt growth and ages younger than peak T are due to processes like recrystallization or mineral reactions. Tc's proposed for Sm-Nd in Grt range from similar to that of hornblende to Ar diffusion Maboko and Nakamura, 1995) to 850°C (Jagoutz, 1988; Hensen and Zhou, 1995). According to the data on Y3+ of Röschmann, 1981) diffusion of REE in Grt at T below 1000°C is far too low to explain Nd homogenization within 30 Ma. To address these uncertainties we examined two granulites from the Central Zone of the Limpopo Belt, South Africa: 1) The granitic orthogneiss 93/050 intruded at >2.6 Ga. Grt is magmatic (i.e. >2.6 Ga), showing kidney shaped Qtz inclusions, and grew from incongruent melting of biotite (Bt => Grt + granitic melt or Bt => Grt + K-feldspar + granitic melt). At 2.0 Ga the rock was again heated above 800°C. Given the 600 Ma age difference, a resetting of Pb and/or Nd chronometers would be easily detected. 2) Sample 93/035 represents the 2.0 Ga old, anatectic Grt bearing leucosome from a paragneiss, the protolith of which was formed >2.6 Ga ago. This rock was studied to explore the effects of inclusions in Grt on its Pb and Nd isotopic characteristics. For both samples Nd model ages are 3 Ga and monazites (Mnz) yield discordant U-Pb ages compatible with formation at 2.6 Ga and disturbance ~2 Ga. Stepwise Pb-Pb leaching of Grt revealed the existence of inclusions with a high 208Pb/204Pb ratio, siting inherited Pb older than that of Grt. As a consequence, no Pb-Pb isochron resulted from that method for either of the two samples. A Grt-feldspar pair of 93/035 gave ages of 2.4 Ga and 2.1 Ga by Pb-Pb and Sm-Nd respectively. Theses ages can not represent cooling, as the sample was above 800°C at 2.0 Ga. In contrast, they reflect the inability of bulk methods to resolve the effect of tiny inclusions. A Sm-Nd isochron of 93/050 including Mnz, one Grt>100µ fraction and magnetite 300-840µ gave 2736±59Ma (MSWD:0.44). Feldspar plots on the same line, even though it was surely open at 2 Ga (myrmekitic texture). Four further Grt fractions of 93/050 (50-100µ, >100µ duplicate, >165µ and >500µ) plot between this isochron and a 2.0 Ga reference line drawn through feldspar. Pb-Pb leaching results on Grt and the relative high Nd concentrations of some Grt fractions indicate Mnz inclusions. Mnz found outside Grt matches both the 2.7 Ga isochron and the 2.0 Ga reference line. Thus, in terms of
Sm-Nd, all Grt fractions can be explained as a ternary mixing of Mnz, re-juvenated Grt, and an old component with 143Nd/144Nd >0.5154 and 147Sm/144Nd >0.3. This component may well be solid inclusions other than Mnz. Indeed U-Pb analyses on zircons indicate partial resetting of an initial age similar to that of the Sm-Nd isochron. Further, magnetite proved to be a remnant phase retentive of Sm-Nd and is present as inclusions. A second possibility is that Grt cores mainly retained their initial characteristics, whereas the rims record the 2.0 Ga event. As a rejuvenation process affecting Pb and partly Nd we suggest a subtle mineral reaction typical of near isothermal decompression of felsic granulite: Some Grts of 93/050 exhibit anorthite rims, formed during decompression ¾2.0 Ga (Harley, 1989). These rims represent the exsolution of the Ca component (grossular) from the Grt. In this reaction, Nd and Pb mobility may be linked to the diffusivity of Ca2+ (piggyback diffusion). As the Pb2+ ion is very similar to Ca2+ the Pb-Pb chronometer will be strongly affected by this process. Accordingly, the Grt-feldspar pair of 93/050 yields a Pb-Pb age of 1938 ± 275, indicating a resetting of the system at 2 Ga, in spite of the older inclusions detected by Pb-Pb leaching. To a lesser degree, even the apparent diffusivity of the Nd3+ ion may be enhanced in a mineral reaction involving Ca. Cores of larger Grts however may have remained unaffected, and could thus represent the old component seen in the Nd data. The Grt-feldspar Pb-Pb age is younger than some of the highly varying Sm-Nd two point ages based on one of the Grt fraction and the WR. This result underscores how dangerous it is to constrain a P-T t loop applying a specific Tc to a Grt WR pair. The diffusivity of Nd3+ seems to be of minor importance if compared to the influence of solid inclusions and/or subtle mineral reactions. We suggest that different Grt fractions should always be analyzed to resolve theses influences. A simple 143Nd/144Nd versus 1/Nd mixing diagram can help to trace the isotopic characteristics of the components involved.
Harley, S.L., Geol. Mag. 126, 215-247 (1989).
Hensen, B.J. & Zhou, B., Geology 23, 225-228 (1995).
Jagoutz, E., Geochim. Cosmochim. Acta 52, 1285-1293 (1988)
Maboko, M.A.H. & Nakamura, E., Precambrian Res. 74, 195-202 (1995).
Röschmann, P., Phys. Chem. Solids 42, 337-350 (1981).
Thöni, M. & Jagoutz, E., Geochim. Cosmochim. Acta 56, 347-368 (1992).
Zhou, B. & Hensen, B.J., Chem. Geol. 121, 317-326 (1995).