Neodymium isotope fractionation in the mass spectrometer and the issue of 142Nd anomalies in early Archean rocks

Abstract

Efforts to search for 142Nd excesses in ancient rocks with high apparent ε Nd 143 over 10 years ago yielded conflicting results and indicated that better mass spectrometers/techniques needed to be devised to resolve the anomalies. Recent searches again give apparently conflicting data: Caro et al. [Caro, G., Bourdon, B., Birck, J.L., Moorbath, S., 2003. 146Sm– 142Nd evidence from Isua metamorphosed sediments for early differentiation of the Earth's mantle. Nature 423, 428–432.] use a new generation Thermal ionization mass spectrometer (Triton) and find that all rocks from 3.7 to 3.8 Ga Isua Greenstone Belt, West Greenland give +15 ppm excesses in 142Nd/ 144Nd ratio over Ames Nd standard. In contrast, Papanastassiou et al. [Papanastassiou, D.A., Sharma, M., Ngo, H.H. Wasserburg, G.J., Dymek, R.F., 2003. No 142Nd excess in early Archean Isua gneiss IE 715-28. Lunar Planet. Sci. Conf., p. 1851.] using another Triton find that an Isua felsic gneiss (IE 715-28; initial ε Nd 143 = + 4.0 ) contains no 142Nd excess over CIT nNd-β standard. Whereas it is possible to imagine early Archean rocks with variable 142Nd anomalies there appears to be a critical issue of the manner in which the two sets of data were reduced. Papanastassiou et al. (2003) used the exponential law to correct for mass dependent isotope fractionation. Caro et al. (2003) found that they needed to use a second-order correction to remove the observed positive correlation between exponential law normalized 142Nd/ 144Nd and 150Nd/ 144Nd ratios. Such a correlation between average ratios of a suite of samples could arise from non-ideal mass fractionation correction if the samples are variably fractionated. It could also be a result of ion optical aberrations. In this paper we investigate if, during thermal ionization, there is a systematic fine structure to the observed mass dependent isotope fractionation of Nd isotopes, which violates the exponential law. We performed nineteen high precision Nd isotopes analyses of CIT nNd-β. The data were obtained for a limited range of mass fractionation (±0.1% per amu), which permited to isolate the fine structure from ion optical effects. The raw data were corrected for mass fractionation using the Rayleigh and exponential laws. Both the Rayleigh and exponential law normalized data sets give an external reproducibility of 6 ppm (2 σ). We also find that while there is no positive relationship between exponential law normalized 142Nd/ 144Nd and 150Nd/ 144Nd ratios, a positive correlation exists between 148Nd/ 144Nd and 150Nd/ 144Nd; the latter is found to be associated with ion beam growth and possibly related to effects arising from fractionation and mixing of different reservoirs on the filament. We conclude that if the data are obtained for a limited range of fractionation and the mixing between enriched and depleted reservoirs on the filament is minimized (by controlling the beam glowth), the exponential law (or Rayleigh law) would provide adequate correction for mass fractionation so that a second-order correction would not be necessary. No ion optical aberrations were observed at 6 ppm level. By analyzing enriched standards we are certain that we can resolve a 142Nd anomaly of ≥+10 ppm. We confirm that the level of 142Nd excess in IE 715-28 cannot be resolved from the baseline nNd-β standard. As IE 715-28 is a metasedimentary rock likely reflecting derivation from older continental crustal material with a high average initial ε Nd 143 , the absence of 142Nd anomaly in this rock indicates that highly depleted ( f Sm/Nd > 0.3) reservoirs were present in the late Hadean (<∼4.25 Ga). Such reservoirs were short-lived.