Electron Microprobe Analysis of Hf in Zircon: Suggestions for Improved Accuracy of a Difficult Measurement

It is not commonly recognized as such, but the accurate measurement of Hf in zircon is not a trivial analytical issue. This is important to assess because Hf is often used as an internal standard for trace element analyses of zircon by LA-ICPMS. The issues pertaining to accuracy revolve around: (1) whether the Hf Ma or the La line is used; (2) what accelerating voltage is applied if Zr La is also measured, and (3) what standard for Hf is used. Weidenbach, et al.'s (2004) study of the 91500 zircon demonstrated the spread (in accuracy) of possible EPMA values for six EPMA labs, 2 of which used Hf Ma, 3 used Hf La, and one used Hf Lb, and standards ranged from HfO2, a ZrO2-HfO2 compound, Hf metal, and hafnon. Weidenbach, et al., used the ID-TIMS values as the correct value (0.695 wt.% Hf.), for which not one of the EPMA labs came close to that value (3 were low and 3 were high). Those data suggest: (1) that there is a systematic underestimation error of the 0.695 wt% Hf (ID-TIMS Hf) value if Hf Ma is used; most likely an issue with the matrix correction, as the analytical lines and absorption edges of Zr La, Si Ka and Hf Ma are rather tightly packed in the electromagnetic spectrum. Mass absorption coefficients are easily in error (e.g., Donovan's determination of the MAC of Hf by Si Ka of 5061 differs from the typically used Henke value of 5449 (Donovan et al, 2002); and (2) For utilization of the Hf La line, however, the second order Zr Ka line interferes with Hf La if the accelerating voltage is greater than 17.99 keV. If this higher keV is used and differential mode PHA is applied, only a portion of the interference is removed (e.g., removal of escape peaks), causing an overestimation of Hf content. Unfortunately, it is virtually impossible to apply an interference correction in this case, as it is impossible to locate Hf-free Zr probe standard. We have examined many of the combinations used by those six EPMA labs and concluded that the optimal EPMA is done with Hf La with the accelerating voltage under 18 keV (e.g. 17 keV is optimal), and also with synthetic stoichiometric hafnon as the standard. We have developed useful standards that are to be distributed to the community for those researchers working on this problem and can be obtained from the second author at jhanchar@mun.ca. The standards include synthetic stoichiometric undoped zircon and hafnon, and synthetic zircon doped with 2 wt. % Hf. Donovan et al. (2002) Probe for Windows: User's Guide and Reference Wiedenbeck, M., et al. (2004) Further characterisation of the 91500 zircon crystal. Geostandards and Geoanatytical Research, 28: 9-39.