Comment: Hf-isotope heterogeneity in zircon 91500

Chemical Geology 233 (2006) 358–363www.elsevier.com/locate/chemgeo

Short communication

Comment: Hf-isotope heterogeneity in zircon 91500

W.L. Griffin ⁎, N.J. Pearson, E.A. Belousova, A. Saeed

GEMOC Key Centre, Department of Earth and Planetary Sciences, Macquarie University, NSW 2109, Australia

Received 12 December 2005; received in revised form 4 March 2006; accepted 8 March 2006

Abstract

A database of N600 analyses of the zircon “standard” Harvard 91500 shows considerable heterogeneity in 176Hf/177Hf; thedistribution is essentially bimodal with major peaks at 0.282284±22 and 0.282330±29 (2σ). Although the zircon shows a widerange of 176Yb/177Hf and 176Lu/177Hf, there is no correlation of 176Hf/177Hf with either parameter. This isotopic heterogeneitylimits the degree to which 91500 can be used to evaluate the precision or accuracy of different treatments of mass bias and overlapcorrections for in situ analysis of Hf-isotope compositions in zircons, or differences between solution and in situ data.© 2006 Elsevier B.V. All rights reserved.

Keywords: Zircon; Hf isotopes; 91500

Fig. 1. Cumulative-probability plot (isoplot; Ludwig, 2003) of

1. Comment

Fragments of the Canadian zircon crystal known asHarvard 91500 (Wiedenbeck et al., 1995) have beenwidely distributed for use as a U–Pb and Hf-isotopereference material, with about 110 g having beenavailable. Despite this wide distribution and relativeabundance, many laboratories have access to relativelysmall volumes, often a few chips. Ideally, a referencematerial should be isotopically homogeneous, and suchhomogeneity has been either implicitly assumed ortested on limited material by a number of workers,several of whomwho have proposed different values for176Hf/177Hf in 91500 (Amelin et al., 1999; Machado andSimonetti, 2001; Goolaerts et al., 2004, Woodhead et al.,2004; Woodhead and Hergt, 2005; Iizuka and Hirate,2005).

The original description of 91500 (Wiedenbeck et al.,1995) presented seven solution-TIMS analyses of 176Hf/

⁎ Corresponding author.E-mail address: [email protected] (W.L. Griffin).

GEMOC LAM-MC-ICPMS analyses of 176Hf/177Hf in zircon91500; values and 2σ of the two main peaks are derived as shownin Fig. 2.

0009-2541/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.chemgeo.2006.03.007

359W.L. Griffin et al. / Chemical Geology 233 (2006) 358–363

177Hf from four fragments; each analysis consumed0.5–1.5 mg of sample. Six of these gave the“recommended” value of 0.282284±3, while one“outlier” at 0.282329±12 was rejected because itdiffered from the analysis of another split from thesame column chemistry process.

We have used 91500 regularly since 1999 as areference material for in situ Hf-isotope analysis (LAM-MC-ICPMS). These analyses have been carried outusing a variety of lasers (266 nm, 213 nm and 193 nm)attached to a Nu Instruments MC-ICPMS (Griffin et al.,2000; Belousova et al., 2006). Most of the analyses havebeen done with a pulse rate of 4 Hz (pulses per second)

Fig. 2. Weighted-average plots (isoplot; Ludwig, 2003) for analyses

and a beam energy of 1 mJ/pulse, producing a spatialresolution of ca. 50 μm; each analysis consumes ca.400–500 ng of zircon or ca 4–5 ng of Hf. We now (mid-2005) have a database of N600 analyses from six large(several mm2) fragments (see Supplementary Data).Some heterogeneity in 176Hf/177Hf was noted by Griffinet al. (2000), who showed a spread in the data (n=60)beyond 3σ of the typical internal precision, and thisheterogeneity has later become more obvious as morefragments were analysed. Most fragments are homoge-neous and have 176Hf/177Hf indistinguishable withinerror of the “recommended” value of Wiedenbeck et al.(1995). However, some fragments give clearly different

in each of the two main peaks of 176Hf/177Hf shown in Fig. 1.

Table 1Summary of data for zircon 91500

Reference Method n 176Hf/177Hf±2σ 176Yb/177Hf±2σ 176Lu/177Hf±2σ

Wiedenbeck et al. (1995) S-TIMS 6 0.282284±60 0.000288±3Amelin et al. (1999) S-MC-ICPMS 6 0.282320±28 0.000302±29Goolaerts et al. (2004) S-MC-ICPMS 59 0.282302±8Woodhead et al. (2004) S-MC-ICPMS 3 0.282306±8Woodhead et al. (2004) LA-MC-ICPMS 93 0.282293±28Griffin et al. (2000) LA-MC-ICPMS 36 0.282297±44 0.000295±50Machado and Simonetti (2001) LA-MC-ICPMS 5 0.282270±123 0.0088±20 0.000308±34Woodhead and Hergt (2005) LA-MC-ICPMS 125 0.282296±28Woodhead and Hergt (2005) S-MC-ICPMS 3 0.282306±8 0.000311 (na)Iizuka and Hirate, 2005 LA-MC-ICPMS 35 0.282321±46 0.0073±33 0.000295±119This report LA-MC-ICPMS 632 0.282307±58 0.0115±50 0.000317±54

360 W.L. Griffin et al. / Chemical Geology 233 (2006) 358–363

values of 176Hf/177Hf. More importantly, in terms ofinterpreting the spread of published data for 91500,some fragments dominated by the “recommended”value contain domains of distinctly different 176Hf/177Hf. These variations are significant for the use of91500 as a reference material for solution analysis, aswell as in situ analysis; individual fragments taken forsolution work may lie anywhere in the spectrum ofisotopic composition demonstrated by the in situ work,producing a biased result.

A cathodoluminescence (CL) study of several chipsof 91500 from the original aliquots (Belousova, 2000)did not reveal any obvious internal structure of the typedescribed by Wiedenbeck et al. (2004). The chips usedfor the analyses described here have been repeatedlyrepolished and are now so thoroughly pitted that furtherpolishing and CL imagery are impractical. We therefore

Fig. 3. Covariation of 176Yb/177Hf and 176Lu/177Hf in GEMOC analyses of zi

cannot make any meaningful correlation betweenpossible internal structure and the observed isotopicheterogeneity. However, as noted below, there is nocorrelation between 176Hf/177Hf and compositionalfactors (especially HREE concentration) that commonlyare reflected in variations in CL colour or intensity(Belousova et al., 2002).

A summary of our LAM-MC-ICPMS data for 176Hf/177Hf in 91500 is shown in Figs. 1 and 2. The data showa major peak at 0.282284±22 (2σ), which correspondsto the “recommended value” derived from the selectedTIMS data of Wiedenbeck et al. (1995). The remainderof the data are strongly skewed to higher values, with amarked subsidiary peak at 0.282330±29 (2σ); thiscorresponds, coincidentally, to the “outlier” (0.282329±12) reported by Wiedenbeck et al. (1995). Intermediatevalues may be real, but may also reflect overlap of the

rcon 91500. Error bar shows typical internal (within-run) 2σ precision.

361W.L. Griffin et al. / Chemical Geology 233 (2006) 358–363

laser beam across domains corresponding to the twomain peaks. The overall mean value of 176Hf/177Hf=0.282307±58 (2σ).

By comparison, the long-term average and variancefor 176Hf/177Hf of our LAM-MC-ICPMS data for theMud Tank zircon is 0.282523±43 (2σ, n=2190); this issimilar to the values given by Woodhead and Hergt(2005) for solution analysis (0.282507±6, n=5) andLAM-MC-ICPMS analysis (0.282504±44, n=158).This dataset has a single peak and the variance mayrepresent the long-term reproducibility of the techniquefor a homogenous zircon, taking into account the range

Fig. 4. Covariation of (a) 176Yb/177Hf and (b) 176Lu/177Hf with 17Hf/177Hf i(within-run) 2σ precision.

of instrumental (laser and mass spectrometer) operatingconditions. However, the overall 2σ value is two to threetimes the typical within-run 2σ precision. This sort ofspread normally would be regarded as evidence ofheterogeneity and Woodhead and Hergt (2005) havesuggested that some parts of Mud Tank are morehomogeneous than this (0.282506±26). It is possiblethat few natural zircons are isotopically homogeneous atthe scale of laser ablation, but our data indicate that MudTank is at least more homogeneous than 91500.

The mean value of 176Hf/177Hf derived by combiningall of our LAM-ICPMS analyses for 91500 (0.282307±

n GEMOC analyses of zircon 91500. Error bars show typical internal

Fig. 5. Histogram of GEMOC analyses of 176Lu/177Hf in zircon 91500;cf. Woodhead and Hergt (2005).

362 W.L. Griffin et al. / Chemical Geology 233 (2006) 358–363

58) is not distinguishable from the solution-derivedvalues of Goolaerts et al. (2004) and Woodhead et al.(2004), nor the average LAM-ICPMS values of Wood-head et al. (2004), Woodhead and Hergt (2005) andIizuka and Hirate (2005), which are based on relativelysmall numbers of analyses (Table 1). The somewhathigher solution value (0.282320±28) of Amelin et al.(1999) may be biased toward the population with higher176Hf/177Hf, as shown in Fig. 1. The isotopic heteroge-neity of 91500 suggests that claims of improvements tothe precision and accuracy of Hf-isotope analysis, basedon either solution analysis, or in situ analysis of limitedvolumes of zircon 91500, may be exaggerated.

Accurate corrections for the overlaps of 176Yb and176Lu on 176Hf are essential to in situ analysis of 176Hf/177Hf (Thirlwall and Walder, 1995; Griffin et al., 2000).In the data reported here, interference of 176Lu on 176Hfwas corrected by measuring the intensity of theinterference-free 175Lu isotope and using 176Lu/175Lu=0.02669 (DeBievre and Taylor, 1993) to calcu-late 176Lu/177Hf. The interference of 176Yb on 176Hfwas corrected by measuring the interference-free 172Ybisotope and using 176Yb/172Yb to calculate 176Yb/177Hf.The 176Lu/175Lu and 176Yb/172Yb ratios were correctedfor instrumental mass bias using the exponential law anda mass bias coefficient calculated using 179Hf/177Hf=0.7325. Griffin et al. (2000) used a value of176Yb/172Yb=0.5865; this was later refined to give anew preferred value for 176Yb/172Yb=0.58669, whichhas been used to reduce all data reported here.

Several authors have claimed improvements to thesecorrections, based in part on assumptions about differ-ences in the reported values of 176Hf/177Hf, 176Lu/177Hfand 176Yb/177Hf of zircon 91500, and the relative mass-bias behaviour of Hf and REE. In 91500, contents of Yb

and Lu are broadly correlated, as in zircons generally(Fig. 3; Belousova et al., 2002), with the exception of asmall subpopulation with low Yb/Hf (Figs. 3 and 4). Ourmean value for 176Yb/177Hf (Fig. 4) is 0.0115±50 (2σ).Iizuka and Hirate (2005) report mean values of 0.00388,0.00778 and 0.00813 for three fragments; the latter twocorrespond to the low-Yb/Hf subpopulation in our data.Our mean 176Lu/177Hf value of 0.000317±54 (2σ)(Fig. 4b) is similar to those reported by other workers,but much higher than a value of 0.0001692 reported forone fragment by Iizuka and Hirate (2005), and higher thanthe mean value (0.000288) reported byWiedenbeck et al.(2004). AlthoughWoodhead andHergt (2005) do not giveaverage values for 176Lu/177Hf, their histogram and rangeof values are similar to ours (Fig. 5). These data attestto considerable chemical heterogeneity within 91500;this heterogeneity is typical of many magmatic zircons(Belousova et al., 2002; Griffin et al., 2002) and limits theconclusions that may safely be drawn from differencesamong reported values.

In our data, there is no bimodality in Lu/Hf cor-responding to the two 176Hf/177Hf populations. The lackof any correlation of 176Hf/177Hf with 176Lu/177Hf or176Yb/177Hf (Fig. 4) attests to the efficacy of the overlapcorrections (Griffin et al., 2000) over the observed rangesof these ratios.

The absence of correlations between 176Lu/177Hf and176Hf/177Hf (Fig. 4) suggests that the isotopic variationswithin the single crystal of 91500 are related to changes inthe composition of the enclosing magma during crystal-lisation. These may reflect the mixing of magma volumeswith different sources, or different degrees of wall-rockcontamination, or the movement of the growing zirconcrystal between two or more such volumes. Such isotopicheterogeneity may not be desirable in an isotopicstandard, but it makes zircons, and especially large onessuch as 91500, ideal for tracking the assembly andevolution of magmas (Griffin et al., 2002; Belousova etal., 2006).

Acknowledgements

Hf-isotope analyses have been supported by ARCDiscovery grants (S.Y. O'Reilly and W.L. Griffin), anARC APD fellowship (EAB) and several industry-sponsored collaborative research projects (Anglo Aus-tralia, BHPB, WMC, DeBeers). We thank E. Beyer anda host of students for contributing data. Comments froman anonymous referee improved the MS. This iscontribution 428 from the ARC National Key Centrefor Geochemical Evolution and Metallogeny of Con-tinents (www.es.mq.edu.au/GEMOC). [PD]

363W.L. Griffin et al. / Chemical Geology 233 (2006) 358–363

Appendix A. Supplementary Data

Supplementary data associated with this article canbe found, in the online version, at doi:10.1016/j.chemgeo.2006.03.007.

References

Amelin, Y., Lee, D.-C., Halliday, A.N., 1999. Early–middle Archeancrustal evolution deduced from Lu–Hf and U–Pb isotopic studiesof single zircon grains. Geochimica et Cosmochimica Acta 64,4205–4225.

Belousova, E.A., 2000. Trace elements in zircon and apatite:application to petrogenesis and mineral exploration. Unpubl.PhD thesis, Macquarie University. 310 pp+ appendices.

Belousova, E.A., Griffin, W.L., O'Reilly, S.Y., Fisher, N.I., 2002.Igneous zircon: trace element composition and relationship to hostrock type. Contributions toMineralogy and Petrology 143, 602–622.

Belousova, E.A., Griffin, W.L., O'Reilly, S.Y., 2006. Zircon morphol-ogy, trace element signatures and Hf-isotope composition as a toolfor petrogenetic modeling: examples from eastern Australiangranitoids. Journal of Petrology 47, 329–353.

DeBievre, P., Taylor, P.D.P., 1993. Table of the isotopic composition ofthe elements. International Journal of Mass Spectrometry and IonProcesses 123, 149.

Goolaerts, A., Mattielli, N., de Jong, J., Weis, D., Scoates, J.S., 2004.Hf and Lu reference values for the zircon standard 91500 by MC-ICP-MS. Chemical Geology 206, 1–9.

Griffin,W.L., Pearson, N.J., Belousova, E., Jackson, S.E., O'Reilly, S.Y.,van Achterberg, E., Shee, S.R., 2000. The Hf isotope composition ofcratonic mantle: LAM-MC-ICPMS analysis of zircon megacrysts inkimberlites. Geochimica et Cosmochimica Acta 64, 133–147.

Griffin, W.L., Wang, X., Jackson, S.E., Pearson, N.J., O'Reilly, S.Y.,Xu, X., Zhou, X., 2002. Zircon chemistry and magma genesis, SEChina: in-situ analysis of Hf isotopes, Pingtan and Tonglu igneouscomplexes. Lithos 61, 237–269.

Iizuka, T., Hirate, T., 2005. Improvements of precision and accuracy inin situ Hf isotope microanalysis of zircon using the laser ablation-MC-ICPMS technique. Chemical Geology 220, 121–137.

Ludwig, K.R., 2003. User's manual for Isoplot 3.00. A geochrono-logical Toolkit for Microsoft Excel. Berkeley GeochronologyCenter, Special Publication No. 4a, Berkeley, California.

Machado, N., Simonetti, A., 2001. U–Pb dating and Hf isotopiccomposition of zircons by laser ablation–MC–ICP–MSA. In:Sylvester, P. (Ed.), Laser Ablation ICPMS in the Earth Sciences:Principles and Applications. Short Course—Mineral. Assoc. Can.,vol. 29, pp. 121–146.

Thirlwall, M.F., Walder, A.J., 1995. In situ hafnium isotope ratioanalysis of zircon by inductively coupled plasma multiple collectormass spectrometry. Chemical Geology 122, 241–247.

Wiedenbeck, M., Alle, P., Corfu, F., Griffin, W.L., Meier, M., Oberli,F., Von Quadt, A., Roddick, J.C., Spiegel, W., 1995. Three naturalzircon standards for U–Th–Pb, Lu–Hf, trace element and REEanalyses. Geostandards Newsletter 19, 1–24.

Wiedenbeck, M., Hanchar, J.M., Peck, W.H., Sylvester, P., Valley, J.,Whitehouse, M., Kronz, A., Morishita, Y., Nasdala, L., 2004.Further characterisation of the 91500 zircon crystal. Geostandardsand Geoanalytical Research 28, 9–39.

Woodhead, J., Hergt, J., 2005. A preliminary appraisal of seven naturalzircon reference materials for in situ Hf isotope determination.Geostandards and Geoanalytical Research 29, 183–195.

Woodhead, J., Hergt, J., Shelley, M., Eggins, S., Kemp, R., 2004.Zircon Hf-isotope analysis with an excimer laser, depth profiling,ablation of complex geometries and concomitant age estimation.Chemical Geology 209, 121–135.