International Journal of Mass Spectrometry 285 (2009) 49–57 Contents lists available at ScienceDirect International Journal of Mass Spectrometry journal homepage: www.elsevier.com/locate/ijms Fractionation and mixing in a thermal ionization mass spectrometer source: Implications and limitations for high-precision Nd isotope analyses Rasmus Andreasen a,b,∗ , Mukul Sharma a a Radiogenic Isotope Geochemistry Laboratory, Department of Earth Sciences, Dartmouth College, 6105 Sherman Fairchild Hall, Hanover, NH 03755, USA b Department of Earth Science & Engineering, Imperial College London, South Kensington, London SW7 2AZ, United Kingdom article info Article history: Received 19 December 2008 Received in revised form 9 April 2009 Accepted 10 April 2009 Available online 19 April 2009 Keywords: Thermal ionization mass spectrometry Neodymium Mass fractionation Exponential law Rayleigh law abstract We present a large dataset of normal Nd standard analyses to evaluate the adequacy of commonly used “laws” to correct for the mass dependent isotope fractionation introduced in a thermal ionization mass spectrometer (TIMS) source, and to assess if the assumption of homogenous sample evaporation and ion- ization from one sample domain on the filament can be considered valid when obtaining high precision (better than 5 ppm, 2) 142 Nd/ 144 Nd ratios for studies in geo- and cosmochemistry. The exponential law is fully adequate to correct for the mass fractionation at the current level of obtainable precision and sur- prisingly performs better than the Rayleigh law. Our modelling shows that the observed correlations in the isotope ratios that remain after data have been corrected using the exponential law are a consequence of correlated uncertainties in counting statistics. These correlations are therefore not from residuals resulting from inadequate correction. Application of the exponential law, however, assumes evapora- tion from a single homogenous domain on the filament—a condition which was impossible to maintain even under optimal sample loading and heating conditions. While a majority of samples showed an increase in heavy/light isotope ratio with time (normal fractionation), many samples showed the reverse trend (reverse fractionation) indicating evaporation and mixing from variably depleted domains on the filament. Our modelling suggests that up to 50% of the calculated external reproducibility (=standard deviation of the population,  p of n independent measurements) can be explained by the ion emission from multiple domains of somewhat different isotopic composition on the filament. The fractionation behaviour of a sample is not necessarily a good indicator of the extent of domain mixing as mixing effects in data collected during reverse fractionation are similar to those of data collected during normal frac- tionation before and after periods of reverse fractionation. Domain mixing effects in 142 Nd/ 144 Nd isotope ratios can be assessed by examining variations in stable 148 Nd/ 144 Nd and 150 Nd/ 144 Nd ratios. The 148 Nd and 150 Nd data from recent reports of 142 Nd deficits in terrestrial samples suggest that the 142 Nd anoma- lies are likely produced from domain mixing during analysis rather than from the decay of short-lived 146 Sm during the early history of the earth. It is imperative to measure all the isotopes of Nd to the utmost possible precision and to examine domain mixing effects by normalizing the data using multiple isotope pairs. © 2009 Elsevier B.V. All rights reserved. 1. Introduction The intent of this study is to evaluate the extent to which the accuracy of highly precise measurements of Nd isotope ratios is affected by our lack of understanding of the mass dependent isotope fractionation and mixing that takes place in the mass spectrome- ter source during analysis. Recent improvements in multi-collector thermal ionization mass spectrometry have renewed interest in the ∗ Corresponding author at: Department of Earth Science & Engineering, Imperial College London, South Kensington, London SW7 2AZ, United Kingdom. Tel.: +44 0 20 7594 6474; fax: +44 0 20 7594 7444. E-mail address: r.andreasen@imperial.ac.uk (R. Andreasen). coupled 146 Sm– 142 Nd and 147 Sm– 143 Nd radioactive decay system as a chronometer for early silicate differentiation within plane- tary bodies [1–10]. The new generation mass spectrometers (mainly ThermoFinnigan Triton) have minimized sources of random errors and systematic bias during analysis (see [11,12] for a complete discussion of uncertainties). These improvements have permit- ted routine measurements of Nd isotope ratios with an external reproducibility (=standard deviation of the population,  p of n inde- pendently loaded standards) that is similar in magnitude to the internal precision (=standard error,  m ) of a single standard. How- ever, repeated measurements of rock samples, in general, display isotope variations that are much larger than expected from repeated analyses of standards (cf. [9]). Since Nd standards are solutions of metal salts with a matrix that is likely different from Nd separated 1387-3806/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.ijms.2009.04.004