Goldschmidt's rules, that trace element distribution will be governed by ionic radius and charge, can be seen to be strongly followed by the Rare-Earth-Elements (REE); thus any deviation from a smoothly varying pattern of chondrite normalised concentrations can be regarded as an 'anomaly'. While the REE are dominantly in the 3+ state, Ce and Eu can also be present in the 4+ and 2+ states (respectively) and, where this occurs, their separation from the other REE is not unexpected. Yttrium is also part of the REE group and has an ionic radius which is virtually identical to Holmium (in
6 fold co-ordination). Both occur only in the 3+ state and on the basis of size and charge alone there should be no fractionation between these two elements. The validity of this assumption can be found in a number of data sets of trace element analyses of whole rocks and minerals. There is therefore strong support for the use of Y as an analogue for Ho where heavy REE concentrations are too low to be analysed accurately (Y is 28.1 times more abundant than Ho by weight (Anders and Grevesse, 1989) and, due to its lower atomic number, is often easier to measure than Ho). Equally, where measured, the Y/Ho ratio of analytical measurements should also have a value close to 28.1 and deviations from this value must first be shown to be outside analytical precisions before anomalous behaviour can be discussed.
Over the years many ion microprobe analyses have been made of REE variations in minerals and glasses from igneous, metamorphic and sedimentary rocks. In ion microprobe analysis instrumental discrimination between Y and Ho is possible during ion formation, the transmission of the ions through the mass spectrometer and during the conversion of the ion to a pulse which can be recorded by a computer. While the latter effects may be reduced if all analyses are made relative to a single uniform standard (Hinton, 1995), the number of ions formed, for every atom sputtered, is largely dependent on the major element chemistry of the matrix. Observed Y/Ho ratios using a single standard may therefore be expected to vary from mineral to mineral purely due to ion production. In the first instance all analyses of minerals and glasses have been normalised to the NIST/NBS SRM-610 glass ('500 ppm') standard which contains equal amounts of Y and Ho (by weight). Compilation of recent data from minerals and glasses from igneous, metamorphic and sedimentary rocks gave an average Y/Ho ratio of 27.9 assuming equal amounts of Y and Ho in the standard glass. Thus, somewhat fortuitously, the glass standard lies between extremes in the matrix related variations observed for individual minerals. In the extreme, the ratio of the Y/Ho ion yields measured for zircon are 8% higher than for the SRM-610 glass standard whereas for monazite they are 15% to lower. Other minerals and glasses lie between these values. Y/Ho ratios measured for mineral unknowns were (on average) within 3 - 5% of a standard of the same composition. Thus, despite the fact that the data set includes Y (and Ho) contents varying by over 5 orders of magnitude, with one exception (see below), there was no evidence for fractionation in the Y/Ho ratio. While the precision dropped rapidly when Ho contents fell below 70 ppb the scatter about the mean was not skewed to either high or low values.
It has recently been shown that large variations in Y/Ho ratio exist within fluorites (Bau and Dulski, 1995) that can be attributed to differences in the stability constants (Walker and Chopin, 1967) of Y and Ho when present as fluoride complexes in hydrothermal solutions. In some cases the Y/Ho ratio, although high, was relatively constant whereas in at least one case the Y/Ho ratio was variable. A 250 mg sample size was analysed (Bau and Dulski, 1995) therefore it is possible that the small scale variability was much higher. Microscale variability in the Y/Ho ratios for similar samples will be investigated by ion microprobe. One group of volcanic glass fragments from a deep sea sediment analysed by ion microprobe gave a relatively large spread in the Y/Ho ratio which was skewed to high values (maximum Y/Ho ratio of 50) and therefore gave a higher than usual average Y/Ho ratio (35.6). This contrasted a data set of volcanic glasses from another location, with a similar range in Y contents, which had an apparently random spread of Y/Ho ratios and a value closer to C1 chondrites.
Anders, E. & Grevesse, N., Geochim. Cosmochim. Acta 53, 197-214 (1989).
Bau, M. & Dulski, P., Contrib. Mineral. Petrol. 119, 213-223 (1995).
Hinton, R. W., The Analyst 120, 1315-1319 (1995).
Walker, J.B. & Chopin, G.R., Adv. Chem. 71, 127-140 (1967).