Zircon is an important refractory mineral that is ubiquitous in most magmatic, metamorphic and sedimentary rocks. It can provide robust U-Pb age(s) for most geologic processes. The trace elements in zircon are of great importance in the interpretation of zircon U-Pb age and relevant geological significance. Commonly, the trace elements abundance in zircon is analyzed by mass spectrometry methods, such as laser ablation inductively coupled plasma-mass spectrometry (LA-ICP-MS), sensitive high-resolution ion microprobe (SHRIMP) and others, which usually require a large spot diameter (10-50μm). Electron probe micro-analyzer (EPMA) is capable of conducting elemental quantitative analyses at micro scales (< 5μm) with relatively high spatial resolution and analytical accuracy/precision. Under appropriate analytical condition settings, the detection limit (D.L.) and analytical uncertainty (standard deviation) of EPMA can be comparable to other high-sensitive methods (e.g. LA-ICP-MS) with only a small difference of one order of magnitude. Besides, EPMA is an accessible instrument in most facilities and is much cheaper and efficient than other techniques. Hafnium is an important trace element in zircon that often substitutes the zirconium ion, since they possess almost the same effective ionic radius in the eightfold coordination, and the zircon has a complete solid solution with the hafnon (HfSiO₄). The Hf abundance in zircon is of importance in the zircon U-Pb age interpretation, and so is the Zr/Hf ratio—it commonly relates with magmatic differentiation: the higher Hf or lower Zr/Hf ratio domain is indicative of zircon growth from a more fractionated melt. Besides, the Hf-related geochemistry is also informative of certain tectonic settings, in which zircon formed. Titanium is an important temperature sensor in zircon that can be utilized in the crystallization temperature estimate. The Ti cation prefers to substitute Si in the four-fold coordination instead of replacing the eightfold Zr in the zircon lattice. Currently, the Ti-in-zircon thermometer has been widely used in studies of magmatic and metamorphic rocks, although certain discrepancy may occur compared to other approaches in temperature estimate. Moreover, some researchers reported a negative correlation between Ti and Hf contents in the igneous zircons and thus argued for a potential Hf-related temperature sensor similar to the Ti thermometry.

This work investigates the standard zircons conventionally utilized in U-Pb dating: TEMORA (TEM) and Qinghu (QH) zircons of plutonic origin and Plešovice (PLE) zircon of ultrahigh-temperature metamorphic origin. They were provided by the Beijing SHRIMP center (TEM) and Ion Probe Facility Institute of Geology and Geophysics CAS (QH and PLE) as polished epoxy resins targets. Suitable analytical conditions are designed on those standard zircons based on conditional analyses in order to acquire accurate and reliable Hf and Ti contents in zircon.

EPMA model of JEOL JAX-8100 equipped with four wavelength dispersive spectrometers (WDS) was utilized at Peking University. The standard materials for Zr, Si, Ti and Hf quantification were zircon (Zr, Si), rutile (Ti) and pure metal Hf from the national standard samples collection. Based on the results in the conditional analyses, two separate experiments (Exp-Ⅰ and Exp-Ⅱ) with nuanced analytical conditions were designed at 20kV accelerating voltage. In Exp-Ⅰ measurement, the current beam was set at 50nA and the counting time was adjusted to 100s for Ti and 10s for the rest; Hf (Lα) was measured by a LIF crystal in the high-resolution H type spectrometer (LIFH), Zr (Lα) and Si (Kα) were measured by TAP crystals in two separate spectrometers, and Ti (Kα) was measured by a PET crystal. In Exp-Ⅱ measurement, the current beam was set at 300nA, the counting time for Ti was adjusted to 300s and a PET crystal in the high-resolution H type spectrometer (PETH) was utilized; to avoid possible dead time effects in measuring Zr and Hf, EPMA CAL function was performed with fixed Zr and Hf contents using their average values obtained in Exp-Ⅰ. The background values were selected empirically in the conditional analyses. In both experiments, the spectrometer's pulse height analyzer (PHA) was set in differentiation mode (Dif) with empirical HV values in order to filter the high order X-ray signals of Hf and Zr that might potentially interfere with the measured characteristic X-ray spectra of Ti (Kα). As monitors, standard materials of Zr-free rutile and Ti-free zircon were also analyzed at the end of each measurement. In all, the detection limit for Ti was 60μg/g with relatively higher uncertainties and 20μg/g with moderate-low standard deviation errors in Exp-Ⅰ and Exp-Ⅱ, respectively.

In Exp-Ⅰ measurement, a negative correlation is observed between Zr and Hf that corresponds with the isomorphic substitution of Zr and Hf in zircon. The Hf abundance in TEM standard zircon is relatively lower than that in PLE and QH standard zircons. However, there is no obvious correlation, neither negative nor positive, between Hf and Ti and is observed in all analyzed standard zircons. Generally, the Hf content in PLE, QH and TEM standard zircons decreases successively, as does the Zr/Hf ratio; while the Ti content is mostly either close to or below the detection limit (60μg/g) that requires Exp-Ⅱ to re-examine further.

Note, compared to other high precision techniques such as LA-ICP-MS, the content of Hf obtained by EPMA is about 10% higher in general and on average, despite a relatively wider variation range. The content of Ti acquired in Exp-Ⅱ by EPMA is quite comparable in the average with the LA-ICP-MS analyses, but in general, the EPMA analyses for Ti fluctuate greatly, similarly to the case of Hf analyses (by EPMA): this perhaps relates to the different order of magnitude of presence for each element (Hf vs. Ti) in zircon. Nevertheless, in general, both Hf and Ti analyses obtained by EPMA show relatively higher abundances in smaller spot analytical conditions, compared to the LA-ICP-MS measurement with larger spot: the latter method de facto measures an approximate average elemental concentration in the spot domain (e.g.32μm), while EPMA measures elements in a much smaller spot (2μm); it also explains the broad variation of EPMA analyses for both elements, since they perhaps distribute unevenly in zircon. To be noted, the Ti content obtained by EPMA indicates a much higher and expectable (Ti-) temperature that seems to best match the actual peak temperature attainment of parent rocks, derived by alternative thermometric approaches, while the conventional LA-ICP-MS analyses often yielded somewhat underestimated Ti-in-zircon temperatures. This is probably a potential key factor that explains the mismatch of Ti-temperature, derived from LA-ICP-MS analyses, with other methodological outcomes, particularly in the studies of ultrahigh-temperature metamorphic rocks: the peak metamorphic temperature inferred from pseudosection calculation or else is often higher than the zircon Ti temperature in practice.

In the line analysis (Exp-Ⅰ), a pronounced zonation of Hf content (increase) and Zr/Hf ratio (decline) is noticed from core to rim in the standard zircons, along their oscillatory annulus under cathodoluminescence, although it remains ambiguous for Ti because of its lower detection limit and poor reliability. The Hf and Zr/Hf zonation in zircon corresponds well with the zircon crystallization from magmatic melt and may provide an implication for melt presence in a relatively high temperature condition. Thus, such a pattern of Hf and Zr/Hf in zircon is possibly also applicable in the high temperature metamorphism that could help to identify the zircon (over)growth environment.

In the grid analysis (Exp-Ⅰ), several small domains of 40μm×40μm and 60μm×60μm sizes were selected, and the move step at 10-15μm and probe diameter of 5μm were adjusted appropriately. The measurement outcomes were then processed by the software OriginPro (ver. 2020b) in linear calculation mode and were plotted in colored contours to simulate conventional map scan results. The built elemental maps of those small domains show that there is no apparent correlation between Ti and Hf or Zr/Hf: it is in accordance with the above conclusions derived from statistic evaluation and line analyses. Nevertheless, PLE standard zircon seems to possess higher Ti content than the others in general, as being revealed above. Also, there is no obvious correlation between the Hf-Ti distribution in zircon and its oscillatory texture under electron or cathodoluminescence imaging, which commonly relate to other trace elements such as U, Th and REE.

In this work a reliable EPMA method for the quantitative analyses of Hf and Ti in zircon is established. Exp-Ⅰ provides a reliable Hf measurement in zircon, while Exp-Ⅱ provides a lower detection limit (20μg/g) and standard deviation for Ti in EPMA analyses that can be utilized further in practice. EPMA analyses with a small probe diameter show a generally higher outcome compared to other large-spot techniques, such as LA-ICP-MS, despite the relatively wide variation range. This could explain why the Ti-in-zircon temperature, acquired by LA-ICP-MS analyses, was often lower than the actual peak temperature attainment inferred by other methods. Thus, EPMA analyses are recommended as an optimal method for elemental quantification in the first place. The Hf content in zircon is indicative of melt presence, from which zircon crystalized; thus, profile (line) analyses for Hf are informative for zircon origin deciphering. In both igneous and high-temperature metamorphic (standard) zircons, there is no obvious correlation between Ti and Hf or Zr/Hf, thus, it remains unclear whether Hf is temperature related as proposed by other researchers.