Trace Element Characterisation of MAD-559 Zircon Reference Material for Ion Microprobe Analysis

Transcription

1 Trace Element Characterisation of MAD-559 Zircon Reference Material for Ion Microprobe Analysis Matthew A. Coble (1)*,JorgeA.Vazquez (2), AndrewP.Barth (3), Joseph Wooden (1), Dale Burns (1), AndrewKylander-Clark (4), SimonJackson (5) andcarae.vennari (6) (1) Department of Geological Sciences, Stanford University, Stanford, CA, USA (2) United States Geological Survey, Menlo Park, CA, USA (3) Department of Earth Sciences, Indiana University Purdue University, Indianapolis, IN, USA (4) Department of Earth Science, University of California, Santa Barbara, CA, USA (5) Geological Survey of Canada, Ottawa, ON, Canada (6) Department of Earth and Planetary Sciences, University of California, Santa Cruz, CA, USA * Corresponding author. coblem@stanford.edu We document the composition of a natural zircon gemstone sourced from Madagascar, MAD-559 a new reference material for calibrating trace element mass fractions in zircon measured by SIMS. The composition of MAD-559 was quantified by calibration relative to the well-documented zircon reference material 91500, for which we compiled existing published data (Mg, Al, Y, rare earth elements, Hf, U, Th) and performed new measurements to characterise the mass fraction of less commonly measured elements (Li, Be, B, F, Na, P, K, Ca, Sc, Ti, Fe, Nb). Measurement results of SL13, CZ3 and MAD-1 zircons and NIST SRM glasses were performed as quality control materials to test measurement bias and repeatability. We show the intermediate precision for most trace element measurement results of MAD-559 to be between ± 3% and ± 5% RSD based on 139 measurements by SIMS on twenty-five individual polished zircon chips measured during a 24-h period, as well as repeat measurements performed over five separate analytical sessions. Trace element mass fractions were also measured by LA-ICP-MS in two different laboratories, and major element compositions measured by electron microprobe, to compare with results measured by SIMS. Based on laser Raman and hyperspectral cathodoluminescence spectroscopy, we show MAD-559 to have high crystal disorder due to radiation damage relative to crystalline zircon (e.g., SL13 and zircon). Although the high cumulative alpha dose of MAD-559 zircon makes it a poor reference material for geochronology, the consistency of the trace element mass fraction results measured in multiple sessions and by various measurement methods shows that it is an ideal reference material for microanalytical trace element mass fraction quantification of zircon. Keywords: zircon, MAD, MAD-559, MADDER, reference material, trace elements, SHRIMP-RG, SIMS, secondary ion mass spectrometry, SL13, CZ3. Received 01 Jan 18 Accepted 20 Aug 18 Advances in microanalytical measurement techniques and instrumentation have allowed determination of the trace element composition of minerals, including zircon, to become routine. High spatial resolution techniques such as SIMS and LA-ICP-MS are increasingly common for measuring trace element mass fractions and U-Th-Pb isotope ratios for geochronological dating. The accuracy of these data is dependent on availability of matrix-matched, homogeneous, reference materials. For zircon (ZrSiO 4 ), there are several well-characterised reference materials; however, many are no longer distributed due to small original quantity (e.g., CZ3 Ireland and Williams 2003, Cavosie et al. 2006, SL13 Harrison and Schmitt 2007, Reid et al. 2011) or contain very low trace element mass fractions, making them analytically challenging to measure with high precision (e.g., Mud Tank zircon Hoskin and Ireland 2000, Yuan et al. 2008). Choosing zircon reference materials with relatively high mass fractions of trace elements, particularly Ti, Sc, Nb and rare earth elements (REE), has become increasingly important for accurate calibration and use of the Ti-in-zircon doi: /ggr The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 1

2 thermometer (Ferry and Watson 2007), for zircon trace element discrimination diagrams (Grimes et al. 2007, 2015) and for petrogenetic modelling (e.g., Wang et al. 2014, Dilles et al. 2015). To address the need for zircon reference materials high in trace element mass fractions, homogeneous at the micrometre to centimetre scale, and in large enough quantities to distribute between SIMS laboratories, we endeavoured to obtain zircon gemstones from mineral dealers and online vendors. Based on a trial-and-error process, we found that natural green-coloured zircons sourced from pegmatites in south-eastern Madagascar tend to have the highest mass fractions of Ti, Nb, REE, Th and U compared with colourless or brown gemstones. Two natural green zircon megacrysts, MAD-1 and MAD- 559, sourced from Madagascar, have been the primary inhouse reference materials for trace element calibration of measurements performed on the SHRIMP-RG (sensitive highresolution ion microprobe with reverse geometry), operated by Stanford University and the USGS (Bacon et al. 2012). The benefit of the reverse geometry design on SHRIMP-RG is a longer beam path length and greater magnetic dispersion, which provides up to a factor of 3 greater mass resolving power (MRP) at high secondary ion transmission chondrite-normalised mass fractions Figure 1. Zircon rare earth element (REE) mass fractions normalised to chondrite (Sun and McDonough 1995) for commonly cited zircon reference materials used in this study. MAD-1 and MAD-559 zircon are enriched in REE, which makes them ideal for trace element calibration because of improved secondary ion intensities and counting statistics, relative to other gem-quality reference materials. Mass fractions shown are preferred values from this study (Tables 3 5). Data for Mud Tank zircon from Hoskin and Ireland (2000). compared with other forward geometry large format ion microprobes (e.g., SHRIMP II or CAMECA 1280). The SHRIMP-RG is a single collector instrument with a laminated electromagnet that is well suited for moving through a wide range of mass to charge ratios (m/z from 7 to 270 corresponding to 7 Li through 238 U 16 O 2 ) in minerals and other materials at high MRP and high transmission (e.g., Grimes et al. 2015, Benson et al. 2017, Schmitt and Vazquez 2017). In zircon, compositional data, particularly abundances of Ti, Y, REE and Hf, can be readily combined with U-Th-Pb isotopic ratio measurements, allowing trace element mass fractions and ages to be determined from the same nanogram volume of analyte. In this study, we characterise the major and trace element mass fractions of MAD-559 and MAD-1 zircon using EPMA, SIMS and LA-ICP-MS measurements, and evaluate crystallinity and homogeneity using laser Raman spectroscopy and hyperspectral cathodoluminescence (CL) images. MAD-559 and MAD-1 are well suited for trace element calibration because they contain relatively high trace element mass fractions compared with other widely distributed zircon reference materials (Figure 1). We calculate trace element compositions relative to the zircon (Wiedenbeck et al. 1995, 2004) in order to document compositions for MAD-559 and MAD-1 that are comparable with other published zircon data. Also, because many laboratory data reduction protocols are idiosyncratically developed, we provide detailed measurement procedures for data acquisition on SHRIMP-RG and calculate mass fractions using the same approach for both SIMS and LA- ICP-MS data sets. Sample material MAD-559 zircon Obtained in 2010, MAD-559 (also referred to informally in the literature as MADDER ) was a faceted 1.12 g (5.59 ct), pale-green zircon gemstone, free of visible inclusions or alteration (Appendix S1), sourced from Madagascar through an online mineral dealer. The original provenance is unknown, but based on the information provided by the seller and the Pan-African age (see below), it was likely a placer zircon from Antsiranana Province (Rakotondrazafy et al. 2008). Approximately 60% of the original mass has been crushed and sieved to lm. For measurements performed in this study, individual chips of MAD-559 were randomly selected and mounted on several epoxy mounts alongside MAD-1, 91500, SL-13, CZ3 and NIST SRM 611, 613 and 615 glasses (Table 1). Mounts were ground and polished to a 1 lm finish The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts

3 Table 1. List of measurement sessions, methods used, dates performed and material analysed for each of the SIMS and LA-ICP-MS measurements performed in this study Session No Laboratory SHRIMP-RG SHRIMP-RG SHRIMP-RG SHRIMP-RG SHRIMP-RG NRCan UCSB Method SIMS SIMS SIMS SIMS SIMS LA-ICP-MS LA-ICP-MS Date 03/07/ /09/ /11/ /11/ /11/ /05/ /08/2015 Zircon reference material analysed MAD-1 MAD-1 MAD-1 MAD-1 MAD-1 MAD-1 MAD-559 MAD-559 MAD-559 MAD-559 MAD-559 MAD SL13 SL13 SL13 SL13 SL13 CZ3 CZ3 Glass calibrators NIST SRM 613 NIST SRM 611 NIST SRM 613 analysed NIST SRM 615 BCR-2G Normalising m/z 28 Si 16 O + 28 Si 16 O + 28 Si 16 O + 28 Si 16 O + 30 Si 16 O + 29 Si + 28 Si + Primary intensity (na) n/a n/a MAD-1 zircon Obtained in 2004, MAD-1 (also called Madagascar Green or MAD by Barth and Wooden 2010) is a darkgreen, rough, natural zircon that had an original weight of 0.24 g. The initial composition of MAD-1 was reported by Barth and Wooden (2010) based on measurements using SHRIMP-RG that were calibrated relative to synthetically grown zircons doped with trace elements (P, Y, REE, Hf) and mass fractions determined by electron microprobe (see also Mazdab 2009). All other elements were calibrated relative to NIST SRM glasses based on measurements on SHRIMP- RG. However, given that synthesised zircons are typically small and compositionally zoned (Hanchar et al. 2001, Mazdab and Wooden 2006) and accurate electron microprobe measurement results are challenging at low mass fractions, particularly of transition metals and REE, we report updated compositional values for MAD-1. CZ3 and SL13 zircon Two well-characterised zircons, CZ3 (Ireland and Williams 2003, Cavosie et al. 2006) and SL13 (Harrison and Schmitt 2007, Hiess et al. 2008, Reid et al. 2011), were measured as quality control materials during the same analytical sessions as MAD-1 and MAD-559 (Table 1). Both zircon quality control reference materials derive from gemstones originally sourced from Sri Lanka and have similar crystallisation ages: ca. 572 Ma (Claoue-Long et al. 1995) and 564 Ma (Pidgeon et al. 1994), respectively. We use the results for SL13 and CZ3 to evaluate intermediate precision over a longer (> 24 months) time period. We broadly evaluate measurement bias relative to published values, but note that the amount of previously published trace element data for SL13 and CZ3 is relatively limited (Cavosie et al. 2006, Mattinson et al. 2006, Barth and Wooden 2010, Reid et al. 2011) and there are significant differences in reported values (up to five times) for Y, REE and Hf in CZ3 (cf., Cavosie et al. 2006, Barth and Wooden 2010). The variation in published data suggests inaccuracies between different calibration methods (discussed below), which we hope to improve upon by reporting updated compositional values for SL13 and CZ3 in this study. Measurement principles and procedures Cathodoluminescence Emission wavelength spectra were obtained using a Ocean Optics xclent III QE65000 hyperspectral CL spectrometer on the JEOL JXA-8230 electron microprobe at Stanford University. Images were collected using an accelerating voltage of 20 kv and a beam current of 50 na. CL intensities were measured over a range of nm, with a resolution of ~ 10 nm. We noted no significant increase in CL intensity with beam currents above 50 na, and thus utilised a current of 50 na to minimise beaminduced crystal lattice damage. Images were acquired at a resolution of 1 lm with acquisition times of 30 ms per pixel. The CL spectra were corrected for detector dark noise, and colour CL images were generated using xclent image processing software with blue, green and red colours corresponding to nm, nm and nm intensities, respectively. Laser Raman Laser Raman spectroscopy analyses were performed on representative unmounted zircon chips at the University of California, Santa Cruz, using a Horiba Scientific LabRAM HR Evolution equipped with a CCD detector. Measurements 2018 The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 3

4 were performed on lm fragments of each sample placed on a glass slide to avoid characteristic Raman peaks from the mounting medium (e.g., epoxy resin). Measurements were conducted using a red 633 nm HeNe + laser operated at 17 mw power with a ~ 2 lm spot diameter. The Raman spectral frequency resolution was 1cm -1 over a measurement range of cm -1 using a grating of 1200 grooves/mm, with acquisition time of 60 s and six accumulations for MAD-1, MAD-559 and 15 s and five accumulations for 91500, SL13 and CZ3. Spectra were measured and peaks deconvoluted to yield amplitudes (Raman intensities) and detected full width half maximum (FWHM) with a combination of Lorentzian and Gaussian peaks using Horiba Labspec6 software. The Raman shift for B 1g, m 3 [SiO 4 ] peak is measured relative to Electron probe microanalysis measurements Major elements and selected trace element mass fractions were measured on a JEOL JXA-8230 electron microprobe at Stanford University. Measurements were performed during two sessions using an acceleration voltage of 20 kv, electron beam current of 150 na and a spot focused to an area ~ 5 lm in diameter. Counting times were 30 s on peak and 30 s backgrounds for major elements (Zr, Si, Hf) and 120 s on peak and 120 s backgrounds for trace elements (P, Y, U, Th). Synthetic calibrant materials were used, including ZrSiO 4 (for Zr and Si), Hf-metal (Hf), apatite (P), Y-phosphate (Y), Th-metal (Th) and U-metal (U). Data were processed using the standard ZAF correction method built into the JEOL microprobe software with mass absorption coefficients from the NIST FFAST database (Chantler et al. 2005). Measurement bias was controlled by analyses of BR266 (Stern 2001). Results in Table 2 are shown as cation proportions calculated based on four oxygens. Ion microprobe (SHRIMP-RG) trace element measurements Trace element intensities of zircon were measured on SHRIMP-RG using an O - 2 primary ion beam with an intensity varying from 0.9 to 3.3 na for all sessions (Table 1). The primary ion spot had a diameter between 14 and 22 lm and a sputtered depth of ~ 1.5 lm for analyses performed in this study. Before every analysis, the sample surface was cleaned by rastering the primary beam for 60 s, and the primary and secondary beams were auto-tuned to maximise transmission. Each measurement included acquisitions of m/ z corresponding to the following ions: 7 Li +, 9 Be +, 11 B +, 19 F +, 23 Na +, 27 Al +, 30 Si +, 31 P +, 39 K +, 40 Ca +, 45 Sc +, 48 Ti +, 49 Ti +, 56 Fe +, 89 Y +, 93 Nb +, 92 Zr 1 H +, 96 Zr +, 139 La +, 140 Ce +, 146 Nd +, Table 2. Electron microprobe microanalysis results* Compound MAD-559 MAD CZ3 Detector 1/9/2017 (n = 6) 10/28/2016 (n = 6) 1/9/2017 (n = 14) 10/28/2016 (n = 11) 1/9/2017 (n = 12) 10/28/2016 (n = 32) 1/9/2017 (n = 15) 10/28/2016 (n = 38) g/100 g 1s g/100 g 1s g/100 g 1s g/100 g 1s g/100 g 1s g/100 g 1s g/100 g 1s g/100 g 1s ZrO PET HfO LIF SiO TAP P 2 O 5 b.d. b.d. b.d. b.d. PETL Y 2 O b.d. PETL UO PETL ThO b.d b.d. b.d. PETL Total b.d. = below detection (<0.001). = not determined. * Quoted intermediate precisions are 1s The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts

5 147 Sm +, 153 Eu +, 155 Gd +, 141 PrO +, 165 Ho +, 159 Tb 16 O +, 163 Dy 16 O +, 166 Er 16 O +, 169 Tm 16 O +, 172 Yb 16 O +, 175 Lu 16 O +, 90 Zr 16 2 O +, 180 Hf 16 O +, 206 Pb +, 207 Pb +, 232 Th 16 O and 238 U 16 O +. Session #5 included measurement of 24 Mg +, 88 Sr +, 120 Sn + and 181 Ta +, whereas sessions #1 and #4 did not include Li, Be, B, F, Na, P or PrO (see Table 1). Each mass was measured on a single ETPâ discrete-dynode electron multiplier operated in pulse-counting mode with count times ranging from 1 to 15 s/mass to optimise counting statistics for each isotope. All measurements were performed by peak-hopping once through the list of m/z, which is generally sufficient for obtaining repeatability precision of ion intensities < 5% RSD for mass fractions > 1.5 mg kg -1. The background for the electron multiplier was very low (< 0.05 cps) and is statistically insignificant for the trace elements reported in this study. Measurements were performed with MRP (M/ DM) =~ (10% peak height), which eliminated interfering molecular species, particularly for 45 Sc +, 48 Ti + and REE + (Schmitt and Vazquez 2017) without a significant loss of secondary transmission. Heavy REE intensities were measured as oxides because the metal ions contain isobaric interferences that cannot be fully resolved and which are not present for the oxides at higher mass. Table 3. Mass fractions for zircon based on a compilation of new data from this study and previously published values Element Mass fraction (mg kg -1 ) 1s Measurement principle References Li SIMS This study (MAD-1) Be SIMS This study (MAD-1) B SIMS This study (MAD-1) F 15 5 SIMS This study (MAD-1) Na SIMS This study (MAD-1) Mg SIMS R11 Al 11 3 LA-ICP-MS C14 P SIMS This study (MAD-1) K SIMS This study (MAD-1) Ca SIMS This study (MAD-1) Sc SIMS This study (MAD-1) Ti ID-ICP-MS S18 Mn SIMS R11 Fe SIMS This study (MAD-1) Sr SIMS, LA-ICP-MS W4, K14, C14 Y SIMS, LA-ICP-MS W4, L10, R11, C6, S6 Nb SIMS This study (MAD-1) Sn LA-ICP-MS B14 La SIMS, LA-ICP-MS W4 Ce SIMS, LA-ICP-MS W4 Pr SIMS, LA-ICP-MS W4 Nd SIMS, LA-ICP-MS W4 Sm SIMS, LA-ICP-MS W4 Eu SIMS, LA-ICP-MS W4 Gd SIMS, LA-ICP-MS W4 Tb SIMS, LA-ICP-MS W4 Dy SIMS, LA-ICP-MS W4 Ho SIMS, LA-ICP-MS W4 Er SIMS, LA-ICP-MS W4 Tm SIMS, LA-ICP-MS W4 Yb SIMS, LA-ICP-MS W4 Lu SIMS, LA-ICP-MS W4 Hf SIMS, LA-ICP-MS, ID-TIMS W4, W5, L10, R11 Ta SIMS, LA-ICP-MS W4, Y4, L10 Pb 14 1 ID-TIMS W5 Th ID-TIMS W5 U ID-TIMS G8, W5 Reported mass fractions are the mean with 1s RSD of multiple measurements. Measurement results from this study are calculated relative to MAD-1 zircon. Reference abbreviations B14: Bruand et al. (2014); C14: Campbell et al. (2014); C6: Cavosie et al. (2006); G8: Gehrels et al. (2008); K14: Koreshkova et al. (2014); L10: Liu et al. (2010); R11: Reid et al. (2011); S6: Schmitt and Vazquez (2006); S18: Szymanowski et al. (2018); W4: Wiedenbeck et al. (2004); W5: Wiedenbeck et al. (1995); Y4: Yuan et al. (2004) The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 5

6 The most challenging element to measure is 93 Nb +, which requires a MRP of ~ to fully resolve peak overlap from 92 Zr 1 H +.AtM/DM = 10000, there is ~ 30% overlap between 92 Nb + and 92 Zr 1 H +, and therefore, the Nb + intensity was measured on the flat, low-mass shoulder to avoid the tailing from the overlapping hydride interference (Schmitt and Vazquez 2017). At these conditions, tailing from 92 Zr 1 H + peak yields a detection limit of ~ 0.05 mg kg -1 for Nb in zircon, based on measured background of cps at m/z = (~ 0.05 AMU above the 93 Nb + peak). Another isotope that requires high MRP is 45 Sc +, which has a nearby interference with 90 Zr 2+ requiring M/DM = to fully resolve. With the measurement conditions used in this study, any overlap of a 90 Zr 2+ interference with 45 Sc + is statistically insignificant when measured at the peak centre. Measured count rates were normalised to 28 Si 16 O + or 30 Si 16 O +, a stoichiometric oxide in zircon, to account for any primary current variation during the analytical session (Table 1). For each analyte, derived ratios (R) for MAD- 559 were compared with the mean of those for the zircon reference material and mass fractions were normalised to values (Table 3) measured during each session (Table 1), using the following linear expression: x unknown analyte R I analyte =I stoichiometric m=z ¼ R I analyte =I stoichiometric m=z RM x RM analyte ð1þ where the mass fraction x of any element is calculated in mg kg -1. The measured count rates normalised to a stoichiometric oxide (R; e.g., 48 Ti + / 30 Si 16 O + ) for the unknown are divided by the mean value for measurements of the same ratio ( R) for the reference material. During sessions 5 7, mass fractions for MAD-559 and MAD-1 were also calculated relative to NIST SRM 611, 613 and 615 glasses (Jochum et al. 2011); count rates on glasses were corrected for matrix-dependent relative sensitivity factors (RSFs) calculated relative to SL13 (Reid et al. 2011) and using the following expression: x unknown analyte R I analyte =I stoichiometric m=z ¼ R I analyte =I stoichiometric m=z RM x RM analyte RSF ð2þ The RSF is a correction factor needed to calculate accurate mass fractions when the major element composition of the unknown and reference materials is different (Schmitt and Vazquez 2017). We use the mass fractions calculated for zircon MAD-559 and MAD-1 relative to NIST SRM glasses for elements with few previously published values available in the literature (e.g., Li, Be, B, F, Na, K, Ca, Sc, Fe and Nb) or where there are inconsistencies (e.g., P). U-Pb geochronology and LA-ICP-MS trace element measurements U-Th-Pb isotope ratios were measured by SIMS and LA- ICP-MS to determine the crystallisation age for MAD-559 and evaluate concordance. Measurements were performed on SHRIMP-RG at Stanford University, and on a Nu Plasma high-resolution multi-collector ICP-MS at the University of California, Santa Barbara. A description of the measurement principles and procedures and a summary of the results are presented in online Appendix S1. Zircon trace element mass fractions were also determined by LA-ICP-MS in two separate laboratories (Table 1): at the University of California, Santa Barbara, and at the Geological Survey of Canada (GSC), both using an Agilent quadrupole LA-ICP-MS. These measurements were performed to evaluate the accuracy of the SIMS results and to assess measurement precision. Trace element mass fractions for MAD-1, MAD-559, SL13 and CZ3 measured by LA-ICP-MS were calculated relative to (values from Table 3) using the same data reduction procedures used to calculate mass fractions for SIMS measurement results (Equation 1). A description of the measurement procedures and a summary of the results are presented in the Appendix S1. Results Cathodoluminescence imaging The CL spectra for MAD-559 and MAD-1 yield maximum intensities of 18.3 and 16.5 counts per pixel, respectively, and are characterised by broad-band emissions defining a single peak at nm (measured at 50% peak height; Figure 2a). The maximum CL intensity of was 325 counts and, compared with MAD-559 and MAD-1, had a higher intensity and occurs over a larger range of wavelengths from 385 to 640 nm (Figure 2a), with a dominant signal between 520 and 560 nm that is characteristic of intrinsic emissions associated with crystalline zircon. A narrow-band emission was also visible at approximately 470 nm, which has been previously correlated with Dy 3+ impurities (e.g., G otze et al. 1999, Nasdala et al. 2003). Notably, the intrinsic emissions between 520 and The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts

7 (b) counts (a) Intensity (a.u.) wavelength (nm) Raman shift (cm 1 ) Figure 2. (a) Hyperspectral cathodoluminescence (CL) emission wavelength spectra for MAD-559, MAD-1 and zircon. The decreased intensity and lack of emission between 520 and 560 nm for MAD-1 and MAD-559 are diagnostic of increased crystal lattice damage. Note that the intensity for is ~ times higher than for MAD-1 and MAD-559. The inset shows colour CL for a representative area of MAD-559 zircon, with blue, green and red corresponding to nm, nm and nm, respectively. The approximately 15 banding is interpreted to reflect structural defects. (b) Raman shift for 91500, SL13, CZ3, MAD-559 and MAD-1. amplitudes (Raman intensities) and detected full width half maximum (FWHM) corrections are shown in Appendix S1. Prominent lattice and silicate peaks that reflect zircon crystallinity from Nasdala et al. (2003) are shown. The intensity of these Raman peaks decrease, and the peak positions shift to lower values with increased crystal lattice damage. Raman shift for the m 3 [SiO 4 ] mode is measured relative to the vertical blue dashed line, defined by nm observed in are deficient in the MAD-1 and MAD-559 emission spectra, likely due to radiation-induced defects (Nasdala et al. 2002). Similarly, narrow-band emissions (e.g., those corresponding to trace elements) are not resolvable due to the low intensity and decreased signal to noise of the emission spectra. The suppression of CL intensity for MAD-559 by nearly fifteen times relative to corresponds to the increased cumulative alpha dose (Appendix S1) and increased mass fraction of radioactive trace elements (U and Th), similar to previous observations (Nasdala et al. 2003). Although decreased crystallinity due to radioactivity-induced damage is likely the primary cause of decreased CL emission on MAD-1 and MAD-559, the relatively high REE content may also contribute to optical absorption processes. The CL emission yields characteristic blue broad-band wavelengths for both MAD-559 and MAD-1. There are faint 1 10 lm parallel bands visible in MAD-559 that are likely due to structural defects (Figure 2a), perhaps corresponding with original growth of the crystal. Alternatively, the banding could reflect trace element heterogeneities. However, there is no statistically significant difference between the trace element mass fractions measured in the light or dark bands (see trace element results, below), suggesting the banding is associated with the structure of the crystal lattice. The width of the banding is generally smaller than the diameter of the ion microprobe measurement spot (15 25 lm); therefore, any heterogeneity would likely be minimised. Images of four additional chips of MAD-559 crystals displayed these same textures and at different orientations, confirming that banding is not an artefact of the electron beam or detector system geometry. Laser Raman spectroscopy Crystalline zircon reference materials 91500, SL13 and CZ3 yield prominent peaks at approximately 202, 214, 225, 356, 439, 974 and 1008 cm -1 (Figure 2b), corresponding to four lattice modes and three internal modes to the silicate tetrahedra (symmetric bend m 2, the symmetric stretch m 1 and the asymmetric stretch m 3 ; Dawson et al. 1971, Nasdala et al. 2003). The intensity, FWHM and frequency shift of the Raman active modes are negatively 2018 The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 7

8 correlated with radiation-induced crystal lattice defects: the highest intensity mode m 3 [SiO 4 ] responds similar to previous studies (e.g., Nasdala et al. 1998, 2001, 2002, Geisler et al. 2001). The background intensity also increases with higher proportions of non-crystalline disorder, resulting in some peaks being unresolvable or obscured by peaks attributed to amorphous lattice. We used zircon as a reference material for comparing the Raman shift of the m 3 [SiO 4 ] peak because it is the most crystalline zircon measured in this study and yields the highest intensity and the lowest FWHM for all Raman active modes. The Raman spectra for SL13 and CZ3 are also characteristic of highly crystalline zircon but have lower intensities and higher FWHM for the lattice and silicate modes (e.g., m 3 [SiO 4 ] peak) that scale with increasing U content. CZ3 has a larger Raman shift in the Raman modes (e.g., m 3 [SiO 4 ] peak) relative to due to increase in disordered crystalline domains from radiation damage. The relative intensities of m 1 and m 3 bands in the SL13 and CZ3 spectrum are likely due to orientation of the crystal relative to the laser polarisation plane (Dawson et al. 1971). The least crystalline zircon samples are MAD-559 and MAD-1 (Figure 2b) due to high cumulative alpha doses (Appendix S1). Both samples have elevated backgrounds and broad peaks associated with amorphous zircon. In MAD-559, there are both crystalline peaks and modes associated with amorphous lattice domains (e.g., Geisler et al. 2001). The crystalline modes present in the Raman spectra for MAD-559 are notably broadened due to decreasing short-range order and are shifted towards lower frequency (e.g., m 3 [SiO 4 ] peak). This combination of amorphous and crystalline Raman modes for MAD-559 and MAD-1 indicates that there is an increased proportion of amorphous domains associated with metamictisation (Nasdala et al. 2003) compared with 91500, SL13 and CZ3. Electron probe microanalysis Electron probe microanalysis results for MAD-1 and MAD-559 are shown in Table 2, in addition to results for quality control zircon reference materials and CZ3 measured during the same measurement sessions. Major and trace element mass fractions for and CZ3 agree with previously published values (see Pidgeon et al. 1994, Wiedenbeck et al. 2004). Phosphorus mass fraction was below the limit of detection (~ 0.01 g 100 g -1 for zircon) due to the presence of a large Zr overlap in WD spectra. Mass fractions of Th in MAD-1 and Th and Y in CZ3 were also below the limits of detection. The measurement results for most analytes in the first and second sessions were indistinguishable at 1s. The Hf measurement results for 91500, expressed as HfO 2, are the lowest of the measured zircons and are within the wide range of HfO 2 values previously reported by EPMA ( g 100 g -1 ), which varies by laboratory and fragment analysed (Wiedenbeck et al. 1995, Wiedenbeck et al. 2004) and likely reflects a combination of sample heterogeneity and differences in analytical set-up. HfO 2 mass fraction is higher in CZ3, consistent with previously published values (Cavosie et al. 2006) and the mass fraction measured by SIMS in this study, affirming the HfO 2 measurement results on MAD-1 and MAD-559 (Table 2) are accurate. Compilation of trace element mass fractions for zircon zircon is the most widely distributed and characterised zircon reference material currently available and has been determined by numerous techniques, including electron microprobe (Wiedenbeck et al. 2004), SIMS (e.g., Wiedenbeck et al. 2004, Reid et al. 2011, this study), LA-ICP-MS (e.g., Yuan et al. 2004, Szymanowski et al. 2018) and isotope dilution ICP-MS and TIMS (e.g., Wiedenbeck et al. 1995, Gehrels et al. 2008). Although some heterogeneity at lm- to mm-scale has been noted (Wiedenbeck et al. 2004), the large volume of published compositional data for (e.g., Wiedenbeck et al. 1995, Liu et al. 2010, Reid et al. 2011, Campbell et al. 2014, Yuan et al. 2004, Gehrels et al. 2008) demonstrates that the measurement results are reproducible and that is a reliable zircon trace element reference material for microanalysis by SIMS. We compiled available trace element mass fractions for (Table 3), using LA-ICP-MS and SIMS data from Wiedenbeck et al. (2004) for REE mass fractions. Several subsequent studies that have reproduced these values (e.g., Whitehouse 2004, Liu et al. 2010, Reid et al. 2011) provide justification for their accuracy. Based on the large compilation of measurement results performed on over the last two decades (references above, also compiled at we have excluded individual measurement results from the compilation reported in Wiedenbeck et al. (2004) where those results are outside the ± 3s range of the mean. For example, for Ce in 91500, three measurement results by SIMS Lab 4 were excluded because they are 1.9 to 3.0 times higher than the mean of 2.6 ± 0.2 mg kg -1 (Table 3). Uranium and Th mass fractions for are based on isotope dilution measurements by Wiedenbeck et al. (1995) and Gehrels et al. (2008), and Hf is based on isotope dilution TIMS, SIMS and LA-ICP-MS (Table 3). The solution The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts

9 measurement results of U, Th and Hf are within the range of reported microanalytical results. Titanium mass fractions are based on independent measurements of multiple fragments of zircon crystals using isotope dilution ICP-MS with a calibrated 47 Ti 49 Ti double spike, yielding a mean of 4.73 ± 0.15 mg kg -1 (Szymanowski et al. 2018). This value is confirmed based on repeat SIMS measurements of in this study, yielding a mean of 4.75 ± 0.09 mg kg -1 Ti calculated relative to SL13 (6.14 ± 0.01 mg kg -1 Ti based on LA-ICP-MS measurements of by Hiess et al. 2008, see Appendix S1). Mass fractions for Li, Be, B, F, Na, P, K, Ca, Sc, Fe and Nb are calculated based on MAD-1 (Table 3) relative to NIST SRM glasses (Jochum et al. 2011). With the exception of P, the measurement results for these elements are generally < 5mgkg -1 in and other natural zircons and not routinely measured. Few reliable measurement results are available in the literature for these analytes, and therefore, calculating the composition relative to NIST SRM is currently the most reliable approach. Despite potential matrix matching issues, we prefer the mass fractions calculated relative to MAD-1 zircon due to its relatively high trace element content, measured using high MRP on SHRIMP-RG to avoid any isobaric interferences (see below). and 615 (Jochum et al. 2011) measured on SHRIMP-RG using Equation (2). The measurements were performed in two different studies, session #5 in this study (Table 1) and Barth and Wooden (2010), following similar measurement procedures. For most elements, we use the mean mass fraction calculated from both studies (see Appendix S1). We prefer the values for P, Sc and Nb from Barth and Wooden (2010) because they were carefully calibrated, rather than calculated relative to previously published values for (e.g., Wiedenbeck et al. 1995, 2004, Liu et al. 2010, Yuan et al. 2004). These analyte mass fractions are most accurately measured on SHRIMP-RG due to the ability to resolve interferences from 30 SiH +, 92 Zr 1 H + and 90 Zr 2+ (requiring MRP of ~ 3700, and ~ to fully resolve, respectively), which can be challenging by other methods. We attempted to measure Sc mass fractions by LA- ICP-MS during sessions #6 and #7 (see Appendices S1 and S2), and the results yielded 45 Sc/ 30 Si ratios that are the same within 2s measurement uncertainty (counting statistics measured on electron multiplier) for all zircon samples. These results suggest the 90 Zr 2+ dominated the measured signal at m/z = by ICP-MS, confirming the need for high- MRP techniques to eliminate isobaric interferences for accurate measurement results of Sc and Nb. The mean of 1.17 ± 0.06 mg kg -1 Sc for (Table 3) is ~ 67% lower than the mass fraction previously reported based on INAA by Wiedenbeck et al. (1995). We conclude the new measurement results on SHRIMP-RG are more accurate because other elements (U and Th) quantified using INAA by Wiedenbeck et al. (1995) are also overestimated relative to LA-ICP-MS, SIMS and TIMS (e.g., Wiedenbeck et al. 1995, 2004, Gehrels et al. 2008, this study). Niobium mass fractions are similarly difficult to measure by other measurement principles. Previously published LA-ICP-MS results by Wiedenbeck et al. (2004), Yuan et al. (2004) and Liu et al. (2010) are variable and generally underestimate the Nb mass fraction for by ~ 60% compared with SIMS results from this study where interferences are minimised on both the reference materials (MAD-1) and We prefer the value for Nb in measured on SHRIMP-RG with high MRP calculated relative to MAD-1 zircon. Trace element results for MAD-1 zircon The preferred trace element mass fractions for MAD-1 are listed in Table 5 (see Appendices S1 and S2 for new and previously published data for MAD-1 on which the preferred values are based). Values for Li, Be, B, F, Na, P, K, Ca, Sc, Fe and Nb are calculated relative to NIST SRM 613 All other analyte mass fractions (Mg, Al, Ti, Sr, Y, Sn, REE, Hf, Ta, Pb, Th and U) in MAD-1 were calculated relative to (Equation 1). The preferred composition of MAD-1 is based on the mean of the measurement results of four measurement sessions (Table 1), which generally overlap at 1s (RSD). Values are reported as the mean of multiple measurement sessions, and the intermediate precision measured over 28 months (1s). Trace element results for MAD-559 zircon Trace element mass fractions for MAD-559 based on measurements by SIMS calculated relative to (Equation 1) are reported in Table 4. The preferred value of MAD-559 is based on the mean of measurement results from four measurement sessions (Table 1), and the associated intermediate precisions expressed as 1s are typically < 6% for most analytes. The measurement results for each analyte are generally repeatable between individual measurement sessions at 1s measurement precision. Preferred values are reported as the mean of multiple measurement sessions, and the error is the intermediate precision measured over 28 months (Figure 3, Table 4). Overall, the analyses of MAD-559 are repeatable over multiple zircon chips, measured on separate grain mounts, and over multiple analytical sessions (Figures 3 and 4) The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 9

10 percent difference MAD-1 SIMS 2 SIMS 3 SIMS 4 SIMS 5 SIMS 5 (NIST) Barth & Wooden (2010) LA-ICP-MS 6 LA-ICP-MS 7 LA-ICP-MS 6 (NIST) MAD-559 SIMS 1 SIMS 3 SIMS 4 SIMS 5 SIMS 5 (NIST) LA-ICP-MS 6 LA-ICP-MS 7 LA-ICP-MS 6 (NIST) Li Sc Ti Nb Y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Th U -80 Li Sc Ti Nb Y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Th U Figure 3. Measurement repeatability for of MAD-1 and MAD-559 for petrologically important trace elements. The y-axis shows the difference between individual measurement session means and the preferred mass fractions (horizontal black line; see Tables 4 and 5). The horizontal grey bar representing ± 5% for reference. Measurement principles (SIMS or LA-ICP-MS) and session numbers are from Table 1. Mass fractions are calculated relative to 91500, as discussed in the text, except for SIMS 5 and LA-ICP-MS 6, which are calibrated with NIST SRM glasses (noted in key by parentheses). Note that the mass fractions calculated relative to glass are in most cases lower than mass fractions calculated relative to zircon. High variance values for LREE are due to the low trace element mass fractions (< 0.1 mg kg -1 ) and poor counting statistics of these measurements. Previously published values for MAD-1 by Barth and Wooden (2010) are also shown. Mass fractions for MAD-559 calculated relative to NIST SRM glasses are included in the calculated preferred value for Li, B, Be, K and Pr because these analyte mass fractions are deficient in zircon, and few measurements exist for these elements in MAD-559 zircon. All other mass fractions calculated relative to NIST SRM glasses are reported as quality control to evaluate measurement bias, and the measurement results are generally within 20% of mean SIMS values calculated relative to zircon (Figure 3; see Appendix S1). The notable exception is P, which suggests that the RSF for P between glass and zircon is inaccurately determined. Session #5, which used a defocused ~ 22 lm diameter spot, had elevated F, Al, Ca, La and Nd mass fractions consistent with improper mount cleaning prior to analysis and/or the larger spot incorporating minor amounts of surface contamination. These higher values were excluded from the preferred values listed in Appendix S1. Fluorine (F) is the least reproducible mass fraction within individual sessions and between analytical SIMS sessions (Table 4). Previous efforts to measure F + in NIST SRM glass by SIMS reveal it to have relatively low ion yields (i.e., ions detected/atoms sputtered; Hervig et al. 2006) relative to most other petrologically important analytes. The high electron affinity of F + (in addition to Cl and Br) results in relatively low secondary ion yield (Wilson 1995) and may also cause F + to interact with secondary electrons or other anions during sputtering. Therefore, the inter- and intra-session variability may be due to surface contamination or charging of the sample, leading to instability in secondary F + ionisation. To test homogeneity of MAD-559, we performed 139 measurements on twenty-five individual polished zircon chips, with four to eight spots per chip, organised in transects with ~ 100 lm spacing. The relative precisions of the mass fractions are between ± 3 and ± 5% (1s RSD; Figure 4). Mass fractions close to the detection limit (e.g., La, Sm) show high variance due to low count rates (Figure 3). We see no correlation between CL zoning (Figure 2a) and trace element mass fractions. Several chips have slightly higher mass fractions (e.g., grain number 4, 10, 20 in Figure 4), but these grains are still within 2s of the mean. These anomalous chips were avoided during the intercalibration studies (i.e., Table 1), but demonstrate that some inhomogeneity is resolvable in MAD-559 zircon within the precision of SIMS measurements results The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts

11 Hf (mg kg -1 ) ± 2.3% mass fraction U (mg kg -1 ) ± 4.1% Y (mg kg -1 ) ± 3.2% Ti (mg kg -1 ) ± 3.8% Th/U ± 3.9% MAD-559 grain number Figure 4. Measurement repeatability for Hf, U, Y, Ti and Th/U in MAD-559 based on 139 measurements on twentyfive randomly selected individual polished zircon chips (separated by dashed vertical lines). The horizontal solid line represents the mean of all points, and the grey band denotes the 1s RSD. The 1s measurement uncertainty for individual results are also shown. Trace element results for quality control materials: SL13 and CZ3 zircon SL13 was measured as a quality control material during four analytical sessions by SIMS (Table 1), and mass fractions have intermediate precisions of < 5% (1s RSD) for P, Sc, Ti, Y, middle and heavy REEs, Hf, Th and U (Appendices S1 and S2). Our preferred mass fractions for SL13 based on new measurement results reported in Table 5 overlap within the reproducibility precision (1s RSD) of published values from Reid et al. (2011), with the exception of somewhat higher mass fractions for P, Y and Th that may be due to mass fractions reported by Reid et al. being calibrated relative to NIST SRM 610 glass instead of matrix-matched zircon. The LA-ICP-MS results yielded mass fractions that were on average 2.6 times higher for Y and REE relative to measurements by SIMS, which were not included in the compiled (preferred) values for SL13 (see Appendix S1). Published trace element mass fractions of CZ3 vary by more than a factor of 5 (cf. Cavosie et al. 2006, Barth and Wooden 2010). SIMS measurement results for CZ3 (sessions 1 and 3, Table 1) and LA-ICP-MS (session 7) are in excellent agreement and, with the exception of La, overlap at 1s (see Appendix S1). Our new preferred values for CZ3 (Table 5) generally overlap at 1s with those reported by Barth and Wooden (2010), with the exception of Y, Ti and Hf. We consider the new compilation of SL13 and CZ3 in this study relative to accurate due to the good agreement between multiple measurement principles, and over several measurement sessions (see Appendix S1). These results further support the accuracy of the preferred mass fractions for MAD-1 and MAD-559 (Tables 4 and 5) proposed herein. U-Pb geochronology for MAD-559 zircon U-Pb isotope ratio results for MAD-559 are reported in Appendix S1. The U-Pb results for LA-ICP-MS and SIMS sessions overlap within stated measurement precisions. Combining data from both techniques yields a 206 Pb/ 238 U weighted mean age of 531 ± 1Ma (MSWD = 2.7, 2s standard error of the mean) and a 207 Pb/ 206 Pb age of 515 ± 1 Ma (MSWD = 3.5, 2s). The U-Pb isotope ratio results are consistently ~ 3% reversely discordant on conventional and inverse Concordia diagrams (Appendix S1). We interpret the 207 Pb/ 206 Pb age of 2018 The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 11

12 Table 4. Mass fractions for MAD-559 zircon based on measurements by SIMS (mg kg -1 ) MAD-559 Element Compiled SIMS (preferred) 1s N Session #1 Zircon n = 14 1s Session #3 Zircon n = 27 1s Session #4 Zircon n = 12 1s Session #5 Zircon n = 17 1s Session #5 NIST SRM n = 17 1s Li Be B F Na Mg Al P K Ca Sc Ti Fe Ge Sr Y Nb Sn La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U Four measurement sessions were performed with zircon as a primary reference materials, and one (session #5) using NIST SRM. The preferred mass fractions for MAD-559 is the mean of the SIMS measurement results (N is the number of measurement sessions included in the mean). Errors are the 1s intermediate precision. Mass fractions shown in grey italics are not included in the mean (see text). Results calculated relative to NIST SRM are generally not used for compiled mass fraction, unless noted in the text. Mass fractions not measured are left blank. 515 Ma to be the most accurate estimate for the crystallisation age for MAD-559. However, as a consequence of the radiation damage and discordant U-Pb results, MAD-559 should not be used as a geochronology reference material for microanalysis. Discussion and conclusions This study presents trace element mass fractions for a new reference material, MAD-559 zircon, with mass fractions calculated relative to the widely distributed and wellcharacterised zircon (Wiedenbeck et al. 1995, 2004). The relatively high trace element mass fractions of MAD-559 (Figure 1) are useful for calibration of a wide range of element mass fractions measured in zircon by SIMS, and the compiled values for MAD-559 (Table 4) and (Table 3) zircons are consistent with data generated in other laboratories and by other measurement principles. Small aliquots of MAD-559 are available to SIMS laboratories upon request to the corresponding author. Our approach differs from previous efforts because we emphasize use of SIMS results more strongly than previous compilations of zircon reference material compositions based The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts

13 Table 5. Preferred mass fractions for SL13, CZ3 and MAD-1 zircon based on measurement results by SIMS and LA- ICP-MS (mg kg -1 ) MAD-1 CZ3 SL13 Element Compiled (preferred) 1s N Compiled (preferred) 1s N Compiled (preferred) 1s N Li Be B F Na Mg Al P K Ca Sc Ti Fe Sr Y Nb Sn La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta Pb Th U N is the number of measurement sessions included in the mean and 1s is the intermediate precision. Mass fractions left blank were below detection. heavily on LA-ICP-MS studies (e.g., Jochum and Stoll 2008). We argue that this is appropriate for elements such as La, Yb (cf. Wiedenbeck et al. 2004), Sc (cf. Campbell et al. 2014) and Nb for which we observe a systematic difference between measurements made by LA-ICP-MS and SIMS. In addition, because all measurements were performed using a pulsecounting electron multiplier, the background on individual measurements is very low and measurement results are repeatable with precisions sufficient to measure mass fractions in the ~ 2 5 lg kg -1 range for low atomic number REE. In addition to providing trace element mass fractions for MAD-559 zircon, we also provide updated trace element mass fractions for MAD-1, CZ3, SL13 and based on a compilation of published and new measurements. It is important to note that the values for in Table 3 are generally within 1s for measurements or compilations reported by Whitehouse (2004), Wiedenbeck et al. (2004), Jochum and Stoll (2008) and Reid et al. (2011). Our compilation for provides a more extensive summary of zircon trace element mass fractions, including Sc, Ti and Nb, which have become important for petrogenetic interpretations (e.g., Ferry and Watson 2007, Grimes et al. 2015, Barth et al. 2017). We also measured mass fractions of Li, B, F, Na, Mg, Al, P, K, Ca and Fe that are typically used to monitor incorporation of mineral or melt inclusions, and to 2018 The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 13

14 evaluate alteration or open-system processes (e.g., Wang et al. 2014). Given the attention of Li in zircon and other petrologic studies (e.g., Benson et al. 2017, Rubin et al. 2017), having accurate mass fractions of low atomic number elements is increasingly important. Furthermore, the relatively high mass fraction of Li in MAD-559 (16 mg kg -1 Li; Table 4) makes it a reliable reference material, compared with zircon with lower mass fractions (e.g., mg kg -1 Li for SL13, CZ3 and 91500). Although MAD-1 and MAD-559 have high cumulative alpha doses (Appendix S1) and amorphous Raman modes that suggest metamictisation, the trace element mass fractions do not reflect open-system exchange or heterogeneity due to element migrations. Highly metamict zircon that has experienced open-system element mobility are typically characterised by heterogeneity and greater than two orders of magnitude enrichment in Li, Na, K, Ca, Al, Fe and low atomic number REE (e.g., Wang et al. 2014, Takehara et al. 2018). The trace element mass fraction of MAD-1 and MAD-559 is typical of igneous and metamorphic zircon (e.g., Barth and Wooden 2010, Grimes et al. 2015, Carley et al. 2017) and is homogeneously distributed (Figures 3 and 4), suggesting the high-u mass fraction and amorphous Raman modes do not reflect alteration of the primary zircon composition. We do not observe any matrix affects in natural zircon (unannealed) from the high cumulative alpha dose of MAD-559 relative to 91500, SL13 or CZ3 within the measurement precision of the trace element measurement results. Previous studies have shown radiation damage in high-u zircons can yield inaccurate U-Pb ages (e.g., White and Ireland 2012, Wang et al. 2014, this study), but we do not observe similar discrepant trace element results when analysing low and high-u zircons by SIMS. Mass fractions calculated relative to NIST SRM glasses for La and Pr measured by LA-ICP-MS are likely more accurate than values calculated relative to zircon because trying to quantify measurements at or below the limits of detection (e.g., lg kg -1 ) for unknowns and reference materials is challenging. For other elements, the mass fractions for zircon calculated relative to NIST SRM glasses measured by SIMS provide a quality control check on the composition of MAD-559. This is particularly important for analytes on which few previously published data exist (i.e., Li, B, Ti, Sc, Nb). With the exception of P and low atomic number REE, the mass fractions for MAD-1 and MAD-559 calculated relative to NIST SRM glasses were within 20% of the preferred values calculated relative to zircon (RSD; Figure 3). Despite the application of a matrixdependent correction factor to compare zircon with glass, many important mass fractions are systematically lower when calibrated against glass compared with mass fractions calculated relative to zircon reference materials (Figure 3), particularly Y and REEs. These inconsistencies between mass fractions calculated relative to zircon versus NIST SRM glasses confirm the need for compositionally matrix-matched zircon reference materials for calculating accurate trace elements by SIMS. The most petrologically significant changes to the mass fractions for MAD-1 and MAD-559 and zircon reference materials in this study relative to previous determinations (Barth and Wooden 2010) are for Ti, Hf, Lu, Th and U mass fractions (Figure 3). The revision of Ti mass fraction is particularly important due to its implications for changes required in the model zircon crystallisation temperatures (i.e., Ti-in-zircon thermometer; Watson et al. 2006, Ferry and Watson 2007). The new results from this study revise the Ti mass fraction 34% lower compared with the value reported for MAD-green (referred to here as MAD-1) and 22% lower than the value reported for CZ3, by Barth and Wooden (2010) (see Appendix S1). For MAD-559, Hf, Th and U mass fractions revised 10.5%, 14.7% and 14.7% higher, respectively, whereas values for MAD-1 were revised 4.6%, 17.1% and 3.7% higher for the same elements relative to those reported by Barth and Wooden (2010). The agreement between new Hf, Th and U mass fractions calibrated relative to by SIMS with electron microprobe measurements (Table 2) and LA-ICP-MS results (see Appendix S1) for these elements suggests this revision is justified. Updating previously published Ti mass fractions calibrated to the values from Barth and Wooden (2010) to the updated values reported in this study can be performed using a simple linear correction factor of This correction applies to any mass fraction measured using 48 Ti + or 49 Ti +, and irrespective of the normalising mass species ( 30 Si +, 30 Si 16 O +, 96 Zr +, 96 Zr 2 O +, etc.). Calculated model temperatures are approximately -25 to -50 C cooler than temperatures calculated using the previous Ti reference values for MAD-1 and MAD-559 from Barth and Wooden (2010), assuming the same inferred or calculated activities of Ti and Si for the system (Ferry and Watson 2007). Appendix S1: Table S8 contains correction factors for updating previous data for all elements referenced to previously published values for MAD-1 and MAD-559 (Barth and Wooden 2010). Although this study provides a significant revision to the mass fraction of MAD-1 and MAD-559, it ensures accurate and consistent comparison for Ti, Y, REE, Th and U mass fractions measured by SIMS, LA-ICP-MS or other measurement principles using MAD-1, MAD-559, 91500, CZ3, SL The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts

15 or other zircon reference materials correlated to these reference values (e.g., GZ7, Szymanowski et al. 2018). Acknowledgements We thank Frank Mazdab, Lutz Nasdala and John Aleinikoff for their helpful suggestions and discussion and Ariel Strickland for identifying and purchasing the original MAD-559 (aka, MADDER) zircon gemstone. NSF funding from EAR to APB was provided for LA-ICP-MS analyses at University of California, Santa Barbara, and EAR provided support for laser Raman at the University of California, Santa Cruz. Any use of trade, firm or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. We thank Richard Hinton, Mark Stelton and an anonymous reviewer for their comments to improve this manuscript, Thomas Meisel for his careful review and editorial guidance and Richard Armstrong and Allen Kennedy for providing pieces of SL13 and CZ3, respectively. References Bacon C.R., Grove M., Vazquez J.A. and Coble M.A. (2012) The Stanford-US Geological Survey SHRIMP ion microprobe A tool for micro-scale chemical and isotopic analysis. US Geological Survey Fact Sheet , 1 4. Barth A.P. and Wooden J.L. (2010) Coupled elemental and isotopic analyses of polygenetic zircons from granitic rocks by ion microprobe, with implications for melt evolution and the source of granitic magmas. Chemical Geology, 277, Barth A.P., Tani K., Meffre S., Wooden J.L., Coble M.A., Arculus R.J., Ishizuka O. and Shukle J.T. (2017) Generation of silicic melts in the early Izu-Bonin Arc recorded by detrital zircons in proximal arc volcaniclastic rocks from the Philippine Sea. Geochemistry, Geophysics, Geosystems, 18, Benson T.R., Coble M.A., Rytuba J.J. and Mahood G.A. (2017) Lithium enrichment in rhyolite magmas of intracontinental calderas. Nature Communications, 8, 270. Bruand E., Storey C.D. and Fowler M. (2014) Accessory mineral chemistry of high Ba-Sr granites from Northern Scotland: Constraints on petrogenesis and records of whole-rock signature. Journal of Petrology, 55, Carley T.L., Miller C.F., Sigmarsson O., Coble M.A., Fisher C.M., Hanchar J.M., Schmitt A.K. and Economos R.C. (2017) Detrital zircon resolve longevity and evolution of silicic magmatism in extinct volcanic centers: A case study from the East Fjords of Iceland. Geosphere, 13, Cavosie A.J., Valley J.W. and Wilde S.A. (2006) Correlated microanalysis of zircon: Trace element, d 18 O and U-Th Pb isotopic constraints on the igneous origin of complex >3900 Ma detrital grains. Geochimica et Cosmochimica Acta, 70, Chantler C.T., Olsen K., Dragoset R.A., Chang J., Kishore A.R., Kotochigova S.A. and Zucker D.S. (2005) X-ray form factor, attenuation and scattering table (version 2.1). National Institute of Standards and Technology (Gaithersburg, USA). Available Online at nist.gov/ffast. Claoue-Long J.C., Compston W., Roberts J. and Fanning C.M. (1995) Two Carboniferous ages: A comparison of SHRIMP zircon dating with conventional zircon ages and 40 Ar/ 39 Ar analysis. Geochronology, Time Scales and Global Stratigraphic Correlation, Society for Sedimentary Geology (SEPM), Special Publication, 4, Dawson P., Hargreave M.M. and Wilkinson G.R. (1971) The vibrational spectrum of zircon (ZrSiO 4 ). Journal of Physics: Condensed Matter, 4, Dilles J.H., Kent A.J., Wooden J.L., Tosdal R.M., Koleszar A., Lee R.G. and Farmer L.P. (2015) Zircon compositional evidence for sulfur-degassing from ore-forming arc magmas. Economic Geology, 110, Ferry J.M. and Watson E.B. (2007) New thermodynamic models and revised calibrations for the Ti-in-zircon and Zr-in-rutile thermometers. Contributions to Mineralogy and Petrology, 154, Gehrels G.E., Ruiz J. and Valencia V.A. (2008) Enhanced precision, accuracy, efficiency and spatial resolution of U-Pb ages by laser ablation-multicollectorinductively coupled plasma-mass spectrometry. Geochemistry Geophysics Geosystems, 9, Q Geisler T., Pidgeon R.T., van Bronswijk W. and Pleysier R. (2001) Kinetics of thermal recovery and recrystallization of partially metamict zircon. European Journal of Mineralogy, 13, G otze J., Kempe U., Habermann D., Nasdala L., Neuser R.D. and Richter D.K. (1999) High-resolution cathodoluminescence combined with SHRIMP ion probe measurements of detrital zircons. Mineralogical Magazine, 63, Campbell L.S., Compston W., Sircombe K.N. and Wilkinson C.C. (2014) Zircon from the East Orebody of the Bayan Obo Fe-Nb- REE deposit, China, and SHRIMP ages for carbonatiterelated magmatism and REE mineralization events. Contributions to Mineralogy and Petrology, 168, The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 15

16 references Grimes C.B., John B.E., Kelemen P.B., Mazdab F.K., Wooden J.L., Cheadle M.J., Hanghøj K. and Schwartz J.J. (2007) Trace element chemistry of zircons from oceanic crust: A method for distinguishing detrital zircon provenance. Geology, 35, Grimes C.B., Wooden J.L., Cheadle M.J. and John B.E. (2015) Fingerprinting tectono-magmatic provenance using trace elements in igneous zircon. Contributions to Mineralogy and Petrology, 170, Hanchar J.M., Finch R.J., Hoskin P.W., Watson E.B., Cherniak D.J. and Mariano A.N. (2001) Rare earth elements in synthetic zircon: Part 1. Synthesis, and rare earth element and phosphorus doping. American Mineralogist, 86, Harrison T.M. and Schmitt A.K. (2007) High sensitivity mapping of Ti distributions in Hadean zircons. Earth and Planetary Science Letters, 261, Hervig R.L., Mazdab F.K., Williams P., Guan Y., Huss G.R. and Leshin L.A. (2006) Useful ion yields for Cameca IMS 3f and 6f SIMS: Limits on quantitative analysis. Chemical Geology, 227, Hiess J., Nutman A.P., Bennett V.C. and Holden P. (2008) Ti-in-zircon thermometry applied to contrasting Archean metamorphic and igneous systems. Chemical Geology, 247, Hoskin P.W. and Ireland T.R. (2000) Rare earth element chemistry of zircon and its use as a provenance indicator. Geology, 28, Ireland T.R. and Williams I.S. (2003) Considerations in zircon geochronology by SIMS. Reviews in Mineralogy and Geochemistry, 53, Jochum K.P. and Stoll B. (2008) Reference materials for elemental and isotopic analyses by LA-(MC)-ICP-MS: Successes and outstanding needs. In: Sylvester P. (ed.), Laser ablation ICP-MS in the earth sciences: Current practices and outstanding issues, 40, Mineralogical Association of Canada (Quebec, QC), Jochum K.P., Weis U., Stoll B., Kuzmin D., Yang Q., Raczek I., Jacob D.E., Stracke A., Birbaum K., Frick D.A. and G unther D. (2011) Determination of reference values for NIST SRM glasses following ISO guidelines. Geostandards and Geoanalytical Research, 35, Koreshkova M.Y., Downes H., Glebovitsky V.A., Rodionov N.V., Antonov A.V. and Sergeev S.A. (2014) Zircon trace element characteristics and ages in granulite xenoliths: A key to understanding the age and origin of the lower crust, Arkhangelsk kimberlite province, Russia. Contributions to Mineralogy and Petrology, 167, 973. Liu Y.-S., Gao S., Hu Z.-C., Gao C., Zong K. and Wang D. (2010) Continental and oceanic crust recycling-induced meltperidotite interactions in the Trans-North China orogen: U- Pb dating, Hf isotopes and trace elements in zircons from mantle xenoliths. Journal of Petrology, 51, Mattinson C.G., Wooden J.L., Liou J.G., Bird D.K. and Wu C.L. (2006) Age and duration of eclogite-facies metamorphism, North Qaidam HP/UHP terrane, western China. American Journal of Science, 306, Mazdab F.K. (2009) Characterization of flux-grown trace element-doped titanite using the high-mass-resolution ion microprobe (SHRIMP- RG). Canadian Mineralogist, 47, Mazdab F.K. and Wooden J.L. (2006) Trace element analysis in zircon by ion microprobe (SHRIMP-RG): Technique and applications. Geochimica et Cosmochimica Acta, 70, A405. Nasdala L., Pidgeon R.T., Wolf D. and Irmer G. (1998) Metamictisation and U-Pb isotopic discordance in single zircons: A combined Raman microprobe and SHRIMP ion probe study. Mineralogy and Petrology, 62, Nasdala L., Wenzel M., Vavra G., Irmer G., Wenzel T. and Kober B. (2001) Metamictization of natural zircon: Accumulation versus thermal annealing of radioactivity-induced damage. Contributions to Mineralogy and Petrology, 141, Nasdala L., Lengauer C.L., Hanchar J.M., Kronz A., Wirth R., Blanc P., Kennedy A.K. and Seydoux-Guillaume A.M. (2002) Annealing radiation damage and the recovery of cathodoluminescence. Chemical Geology, 191, Nasdala L., Zhang M., Kempe U., Panczer G., Gaft M., Andrut M. and Pl otze M. (2003) Spectroscopic methods applied to zircon. Mineralogy and Geochemistry, 53, Pidgeon R.T., Furfaro D., Kennedy A.K., Nemchin A.A. and Van Bronswjk W. (1994) Calibration of zircon standards for the Curtin SHRIMP II. In: Eighth International Conference on Geochronology, Cosmochronology and Isotope Geology, Berkeley, US (Abstract). U.S. Geological Survey Circular, 1107, 251. Rakotondrazafy A.F.M., Giuliani G., Ohnenstetter D., Fallick A.E., Rakotosamizanany S., Andriamamonjy A., Ralantoarison T., Razanatseheno M., Offant Y., Garnier V. and Maluski H. (2008) Gem corundum deposits of Madagascar: A review. Ore Geology Reviews, 34, Reid M.R., Vazquez J.A. and Schmitt A.K. (2011) Zircon-scale insights into the history of a supervolcano, Bishop Tuff, Long Valley, California, with implications for the Ti-in-zircon geothermometer. Contributions to Mineralogy and Petrology, 161, Rubin A.E., Cooper K.M., Till C.B., Kent A.J., Costa F., Bose M., Gravley D., Deering C. and Cole J. (2017) Rapid cooling and cold storage in a silicic magma reservoir recorded in individual crystals. Science, 356, The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts

17 references Schmitt A.K. and Vazquez J.A. (2006) Alteration and remelting of nascent oceanic crust during continental rupture: Evidence from zircon geochemistry of rhyolites and xenoliths from the Salton Trough, California. Earth and Planetary Science Letters, 252, Schmitt A.K. and Vazquez J.A. (2017) Secondary ionization mass spectrometry analysis in petrochronology. Reviews in Mineralogy and Geochemistry, 83, Stern R.A. (2001) A new isotopic and trace element standard for the ion microprobe: Preliminary thermal ionization mass spectrometry (TIMS) U-Pb and electron microprobe data. Radiogenic age and isotopic studies: Report 14, Geological Survey of Canada, Current research F1, 11pp. Sun S.-S. and McDonough W.F. (1995) The composition of the Earth. Chemical Geology, 120, Szymanowski D., Fehr M.A., Guillong M., Coble M.A., Wotzlaw J.-F., Nasdala L., Ellis B.S., Bachmann O. and Sch onb achler M. (2018) Isotope dilution anchoring of zircon reference materials for accurate Ti-in-zircon thermometry. Chemical Geology, 481, Takehara M., Horie K., Hokada T. and Kiyokawa S. (2018) New insight into disturbance of U-Pb and trace element systems in hydrothermally altered zircon via SHRIMP analyses of zircon from the Duluth Gabbro. Chemical Geology, 484, Wang X.L., Coble M.A., Valley J.W., Shu X.J., Kitajima K., Spicuzza M.J. and Sun T. (2014) Influence of radiation damage on Late Jurassic zircon from southern China: Evidence from in situ measurements of oxygen isotopes, laser Raman, U-Pb ages and trace elements. Chemical Geology, 389, Wiedenbeck M., Hanchar J.M., Peck W.H., Sylvester P., Valley J., Whitehouse M., Kronz A., Morishita Y., Nasdala L., Fiebig J. and Franchi I. (2004) Further characterisation of the zircon crystal. Geostandards and Geoanalytical Research, 28, Wilson R.G. (1995) SIMS quantification in Si, GaAs and diamond An update. International Journal of Mass Spectrometry and Ion Processes, 143, Yuan H., Gao S., Xiaohui Liu., Li H., G unther D. and Wu F. (2004) Accurate U-Pb age and trace element determinations of zircon by laser ablation-inductively coupled plasma-mass spectrometry. Geostandards and Geoanalytical Research, 28, Yuan H.-L., Gao S., Dai M.-N., Zong C.-L., G unther D., Fontaine G.H., Liu X.-M. and Diwu C. (2008) Simultaneous determinations of U-Pb age, Hf isotopes and trace element compositions of zircon by excimer laserablation quadrupole and multiple-collector ICP-MS. Chemical Geology, 247, Supporting information The following supporting information may be found in the online version of this article: Appendix S1. Supporting text, tables, and figures. Appendix S2. Measured ratios and calculated mass fractions calculated relative to zircon. This material is available from: com/doi/ /ggr.12238/abstract (This link will take you to the article abstract). Watson E.B., Wark D.A. and Thomas J.B. (2006) Crystallization thermometers for zircon and rutile. Contributions to Mineralogy and Petrology, 151, 413. White L.T. and Ireland T.R. (2012) High-uranium matrix effect in zircon and its implications for SHRIMP U-Pb age determinations. Chemical Geology, , Whitehouse M.J. (2004) Multi-collector SIMS determination of trace lanthanides in zircon. Geostandards and Geoanalytical Research, 28, Wiedenbeck M., Alle P., Corfu F., Griffin W.L., Meier M., Oberli F., von Quadt A., Roddick J.C. and Spiegel W. (1995) Three natural zircon standards for U-Th-Pb, Lu-Hf, trace element and REE analyses. Geostandards Newsletter, 19, The Authors. Geostandards and Geoanalytical Research 2018 International Association of Geoanalysts 17

18 Trace-Element Characterization of MAD-559 Zircon Reference Material for Ion Microprobe Analysis Matthew A. Coble (1)*, Jorge Vazquez (2), Andrew P. Barth (3), Joseph Wooden (1), Dale Burns (1), Andrew Kylander-Clark (4), Simon Jackson (5), Cara E. Vennari (6) Supplemental Material for Coble et al. includes the following sections: Supplemental Material Text Methods Results References Supplemental Material Figures and Tables Supplemental Figure 1. Supplemental Figure 2. Supplemental Table 1. Supplemental Table 2. Supplemental Table 3. Supplemental Table 4. Supplemental Table 5. Supplemental Table 6. Supplemental Table 7. Supplemental Table 8.

19 Supplemental Material Text Methods Ion microprobe (SHRIMP-RG) U/Pb geochronology Measurement principles and procedures for U/Pb isotope ratio measurements on SHRIMP-RG were similar to those for trace-element mass fraction measurements, except the O2 - primary ion beam was focused to a diameter of μm, an intensity varying from 3.5 to 4.0 na, and a sputtered depth of ~2 μm. MAD-559 zircon was comounted in epoxy with zircon geochronology reference materials Temora-2 (Black et al., 2004), Duluth Gabbro (FC1; Paces and Miller, 1993), and Mt. Dromedary (Schoene et al., 2006) and polished to a 1 μm finish. Two aliquots of individual grains from MAD-559 were mounted: (1) zircon thermally annealed at 900 C for 24 hours and (2) natural untreated zircons in order to examine the effect of radiation damage on U/Pb SIMS isotope ratio measurements on SHRIMP-RG. At the beginning of every measurement, the sample surface was cleaned by rastering the primary beam for 60 seconds, and the primary and secondary beams were auto-tuned to maximize transmission. The measurement included acquisitions of m/z which relate to the following ions: a high mass normalizing intensity for 90 Zr2 16 O +, followed by 180 Hf 16 O +, 204 Pb +, a background measured at mass units above the 204 Pb + peak, 206 Pb +, 207 Pb +, 208 Pb +, 232 Th +, 238 U +, 232 Th 16 O +, 238 U 16 O +, and 238 U 16 O2 +. Measurements were performed at MRP = 6,800-7,500 (10% peak height), which completely eliminated interfering molecular species near Pb + peaks (e.g., 178 Hf 28 Si + is AMU from 206 Pb + ), with 6 scans (peak-hopping cycles) from mass 196 through 270. Calculated crystallization ages for zircon were calibrated relative to Temora-2 (418.4 Ma; Mattinson, 2012) using the MS Excel add-in SQUID 2.51 (Ludwig, 2009), with 0.5% uncertainty added in quadrature to individual U/Pb ratios to account for reproducibility of Temora-2 during the analytical session. U/Pb isotopic ratios are plotted on Tera-Wasserburg inverse Concordia diagrams corrected for common-pb using 204 Pb and a model Pb composition from Stacey and Kramers (1975). Reported U-Pb ages are

20 error-weighted means or inverse Concordia ages (Ludwig, 1998) with 2s standard error of the mean; Concordia ages include decay-constant errors. Laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) U/Pb Geochronology U Th Pb isotope ratios were measured on a Nu Plasma high-resolution multicollector LA-ICP-MS at the University of California, Santa Barbara, with 206 Pb +, 207 Pb +, 208 Pb +, 232 Th +, and 238 U + measured using faraday detectors and 204 Pb + measured using a pulse-counting electron multiplier. Isotope ratio measurements were performed following methods outlined in Kylander-Clark et al. (2013) and Spencer et al. (2013), with isotopic ratios and uncertainties calculated using Iolite (Paton et al., 2010). Calculated U-Pb model ages for zircon are calibrated relative to ( ± 0.4 Ma; Wiedenbeck et al., 1995) with 2% uncertainty added in quadrature to individual ratios to account for repeatability of quality control zircon reference materials. Data are presented following the same approach as U-Pb data acquired by SIMS. Laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) traceelement measurements Trace-element mass fraction measurements of zircon were performed by LA- ICP-MS in two separate laboratories (session #6 and #7; Table 1). Measurement results from the two LA-ICP-MS sessions were collected on the same epoxy grain mounts and the same individual zircon chips as analysed by SIMS during session #5 and represent quality control measurements for evaluating the accuracy of SIMS measurement results. Zircon trace-element mass fractions and U Th Pb isotope ratios were determined at the University of California, Santa Barbara, laser ablation split stream inductively coupled plasma mass spectrometry (LASS-ICP-MS) laboratory following methods outlined in Kylander-Clark et al. (2013) and Spencer et al. (2013). Analyte material was ablated using an Atlex ArF (λ = 193 nm) wavelength excimer laser using a laser spot diameter of 50 µm, a 4 Hz repetition rate, an energy density of 2 J cm 2, and 30 s sample acquisition. Trace-element intensities were measured on an Agilent

21 7700x quadrupole ICP-MS, and U Th Pb isotope ratios measured simultaneously on a Nu Plasma high-resolution multi-collector ICP-MS. Zircon trace-elements measurements were performed at the Geological Survey of Canada (GSC) using an Analyte excimer (λ = 193 nm) laser ablation system (Photon Machines Inc.) attached to an Agilent quadrupole 7700x ICP-MS and followed methods discussed in Longerich et al. (1996). Ablation was performed using a laser spot diameter of 43 µm, a 10 Hz laser repetition rate, an energy density of 3 J cm 2, and 60 s sample acquisition. The measurement included of m/z which relate to the following ions: 7 Li +, 9 Be +, 11 B +, 23 Na +, 25 Mg +, 27 Al +, 28 Si + (UCSB) or 29 Si + (GSC), 31 P +, 45 Sc +, 49 Ti +, 51 V +, 57 Fe +, 73 Ge +, 88 Sr +, 89 Y +, 91 Zr +, 93 Nb + (UCSB), 118 Sn +, 139 La +, 140 Ce +, 141 Pr +, 146 Nd +, 147 Sm +, 151 Eu + (GSC) or 153 Eu + (UCSB), 157 Gd +, 159 Tb +, 163 Dy +, 165 Ho +, 166 Er + (UCSB) or 167 Er + (GSC), 169 Tm +, 172 Yb + (UCSB) or 173 Yb + (GSC), 175 Lu +, 178 Hf + (UCSB) or 177 Hf + (GSC), 181 Ta +, 206 Pb +, 207 Pb +, 208 Pb +, 232 Th +, and 238 U +. Analyses performed at University of California, Santa Barbara, did not include measurement of Li, Be, B, Na, Mg, Al, P, Sc, Sr, or Sn. Isotopic ratios and precisions were calculated using Iolite (Paton et al., 2010) for data from University of California, Santa Barbara, and using GLITTER (Griffin et al., 2008) for data from Geological Survey of Canada. Measured isotopic ratios are normalized to 28 Si + or 29 Si + (Table 1) and trace-element mass fractions are calculated relative to (values from Table 3) using the same method principles used to calculate concentrations for SIMS analyses (equation 1). Results Supporting trace-element mass fractions measured by Laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) Calculated trace-element mass fractions for MAD-559 based on measurements by LA-ICP-MS are reported in Supplemental Material Table 4, and for MAD-1 listed in Supplemental Material Table 5. Measurements performed in session #6 were also calculated relative to NIST-SRM-611, which is helpful for quality control assessment the accuracy of trace-element mass fractions in with low abundance or those which are poorly constrained (e.g., B, Be, Mg, V, Ge, Sr, Sn, La, and Pr).

22 Overall, the LA-ICP-MS measurement results are in good agreement (within 2s RSD) with concentrations measured by SIMS for MAD-1 and MAD-559 (Fig. 3) with the notable exception of Al, P, Nb, Sc, and La. Although Al was only measured by LA-ICP- MS at the Geological Survey of Canada, the mass fraction measured by SIMS is 3 to 4 times higher than by LA-ICP-MS. Aluminium is measured to monitor contamination by mineral inclusions or surface contamination, and we attribute the higher concentrations and increased scatter between different SIMS sessions to surface contamination. The LA-ICP-MS measurements are more variable for MAD-559 and MAD-1 measured between the two different labs, when standardized to (difference between labs ranges from 1 to 28% for middle- and heavy-ree, Hf, U, Th; Supplemental Material Tables 4, and 5). This variability is problematic, but highlights the difficulty of measuring low-concentrations zircons by SIMS and LA-ICP-MS. The low trace-element mass fractions of are emblematic of many metamorphic zircon (e.g., Hoskin and Ireland, 2000), and care must be taken to ensure measurement conditions that affect sensitivity (i.e., backgrounds and baseline corrections) are carefully monitored. Although some heterogeneity at µm- to mm-scale has been observed in 91500, this is generally <10% (e.g., Wiedenbeck et al., 2004) and cannot explain the differences between analyses of MAD-559 and MAD-1 measured between the two different LA-ICP-MS labs. Szymanowski et al. (2018) note that the small analytical volume of SIMS measurements is more sensitive to sample heterogeneity than laser ablation methods, which average more material during measurements. Hence, some of the variability noted in this study (e.g., Table 5) is likely due to differences in instrumentation and measurement procedures. Despite this variability, the overall measurement results for MAD-559 and MAD-1 are in excellent agreement between SIMS and LA-ICP-MS (Fig. 3), which is confirmation that the SIMS results in Tables 4 and 5 are accurate. We generally prefer the average of the SIMS results for MAD-559 and MAD-1 (Tables 4 and 5) due to the variability between the LA-ICP-MS results. Some inconsistencies between the LA-ICP-MS and SIMS datasets were noted in this study, particularly for measurement results collected during session 6 (Table 1) and highlighted in Figure 3. To evaluate this further, mass fractions for quality control

23 reference material SL13 analysed by LA-ICP-MS in session 6 (Table 1) were 2.6 times higher on average for Y and REE relative to measurements by SIMS (measured in sessions 2-5). These mass fractions were not included in the mean (preferred) for SL13 (see Supplemental Material Table 6). We note that analyses by SIMS measured during session 5 and LA-ICP-MS analyses in session 6 were performed on the same polished grain mount and the same chips of and SL13, thus, sample heterogeneity is unlikely to explain the different trace-element mass fractions calculated for SL13 between SIMS and LA-ICP-MS (Supplemental Material Table 6). We attribute the inconsistently higher mass fractions for SL13 by LA-ICP-MS in session 6 relative to to imprecisely determined mean values for each material (individual measurements had a reproducibility ±10 to ±25%, 1s RSD). The increased scatter during session 6 is due to low signal to noise on both SL13 ( REESL13 = 35.7 μg g -1 ) and ( REE91500 = 127 μg g -1 ), which are both reference materials with relatively low trace element mass fractions. In comparison, reference material high in traceelement mass fractions (e.g., REEMAD-559 = 335 μg g -1 ) are less sensitive to the increased scatter of during session 6. U/Pb geochronology for MAD-559 zircon U/Pb isotope ratio measurement results for MAD-559 (see Supplemental Material Table 3) measured on SHRIMP-RG yield a 206 Pb/ 238 U weighted mean age of ± 1.7 Ma (MSWD=0.62; 2s standard error of the mean). The U/Pb measurement results are consistently ~3% reversely discordant on conventional and inverse Concordia diagrams (Fig. 6) and calculated Concordia ages overlap within measurement precision with the 206 Pb/ 238 U age. The 207 Pb/ 206 Pb ages are distinctly younger, yielding a weighted mean age of ± 2.7 Ma (MSWD=0.94; 2s), which suggests Pb and/or U were redistributed in highly radiation damaged MAD-559 zircon. Incorporation of this labile Pb (and/or U) during SIMS analysis can bias measured U/Pb ratios relative to undamaged, crystalline zircon standards. A similar phenomenon was observed by Wiedenbeck (1995) for Archean zircon enriched in 204 Pb that produced reverse discordance when measured by ion microprobe. Although MAD-559 is not particularly 204 Pb-rich (typically <20 µg kg Pb), the high cumulative alpha dose (Supplemental Material Table 2) and

24 amorphous ZrSiO4 associated with metamictisation yields discordant U/Pb ages due to Pb-mobility and/or loss. Measurement of annealed and natural (unannealed) MAD-559 grains by SIMS yield identical results (Supplemental Material Figure 2) supporting the suggestion that the reverse discordance is not a matrix affect from radiation damage. Zircon quality control reference material measured during the same SHRIMP-RG session include FC1 (FC1; ± 0.6 Ma; Paces and Miller, 1993) and Mt. Dromedary (99.12± 0.14 Ma; Schoene et al., 2006). The calculated weighed mean 206 Pb/ 238 U age for FC1 was 1102 ± 7 Ma and the 207 Pb/ 206 Pb age was 1101± 4 Ma, while the 206 Pb/ 238 U age for Mt. Dromedary was 99.1± 0.4 Ma. The concordance of the FC1 zircon U/Pb dates and their good agreement with the published TIMS ages (Paces and Miller, 1993) confirms the discordant results for MAD-559 are not an analytical artefact. The LA-ICP-MS measurement results for MAD-559 are similar to the analyses by SIMS; the 206 Pb/ 238 U weighted mean age is ± 0.7 Ma (MSWD=4.3; 2s standard error of the mean) and reversely discordant (Fig. 6). Similarly, the 207 Pb/ 206 Pb age is distinctly younger, with a weighted mean of ± 1.2 Ma (MSWD=7.0; 2s). CZ3 zircon was analysed as a quality control reference material during the LA-ICP-MS session, and yielded a weighed mean 206 Pb/ 238 U age of ± 1.0 Ma and the 207 Pb/ 206 Pb age was 562 ± 9 Ma, which are concordant and overlap with the published age of ca. 564 Ma (no error was reported for CZ3 by Pidgeon et al., 1994). References Barth, A. P., and Wooden, J.L. (2010) Coupled elemental and isotopic analyses of polygenetic zircons from granitic rocks by ion microprobe, with implications for melt evolution and the source of granitic magmas. Chemical Geology, 277, Black, L.P., Kamo, S.L., Allen, C.M., Davis, D.W., Aleinikoff, J.N., Valley, J.W., Mundil, R., Campbell, I.H., Korsch, R.J., Williams, I.S., and Foudoulis, C. (2004)

25 Improved 206 Pb/ 238 U microprobe geochronology by the monitoring of a trace-element-related matrix effect; SHRIMP, ID TIMS, ELA ICP MS and oxygen isotope documentation for a series of zircon standards. Chemical Geology. 205, Ferry J.M. and Watson E.B. (2007) New thermodynamic models and revised calibrations for the Ti-in-zircon and Zr-in-rutile thermometers. Contributions to Mineralogy and Petrology, 154, Griffin, W.L., Powell, W.J., Pearson, N.J. and O reilly, S.Y. (2008) GLITTER: data reduction software for laser ablation ICP-MS. Laser Ablation-ICP-MS in the earth sciences. Mineralogical association of Canada short course series, 40, Hoskin P.W. and Ireland T. R. (2000) Rare earth element chemistry of zircon and its use as a provenance indicator. Geology, 28, Jochum, K.P., and Stoll, B. (2008) Reference materials for elemental and isotopic analyses by LA-(MC)-ICP-MS: Successes and outstanding needs. Laser Ablation ICP-MS in the Earth Sciences: Current practices and outstanding issues, 40, Jochum, K.P., Weis, U., Stoll, B., Kuzmin, D., Yang, Q., Raczek, I., Jacob, D.E., Stracke, A., Birbaum, K., Frick, D.A. and Günther, D. (2011) Determination of reference values for NIST SRM glasses following ISO guidelines. Geostandards and Geoanalytical Research, 35, Kylander-Clark, A.R.C., Hacker, B.R., and Cottle, J.M. (2013) Laser-ablation split-stream ICP petrochronology. Chemical Geology, 245, Longerich, H.P., Jackson, S.E. and Gunther, D. (1996) Laser ablation inductively coupled plasma mass spectrometric transient signal data acquisition and analyte concentration calculation. Journal of Analytical Atomic Spectrometry, 11, Ludwig, K.R. (1998) On the treatment of concordant uranium-lead ages. Geochimica et Cosmochimica Acta, 62,

26 Ludwig, K.R. (2009) Squid 2, A user s manual. Berkeley Geochronology Center Special Publication 5, 110. Mattinson, J.M. (2012) Analysis of the relative decay constants of 235 U and 238 U by multi-step CA-TIMS measurements of closed-system natural zircon samples. Chemical Geology, 275, Paces, J.B. and Miller, J.D. (1993) Precise U Pb ages of Duluth complex and related mafic intrusions, northeastern Minnesota: Geochronological insights to physical, petrogenetic, paleomagnetic, and tectonomagmatic processes associated with the 1.1 Ga midcontinent rift system. Journal of Geophysical Research: Solid Earth, 98, Paton, C., Woodhead, J.D., Hellstrom, J.C., Hergt, J.M., Greig, A. and Maas, R. (2010) Improved laser ablation U Pb zircon geochronology through robust downhole fractionation correction. Geochemistry, Geophysics, Geosystems, 11, 3. Pidgeon, R.T., Furfaro, D., Kennedy, A.K., Nemchin, A.A., and Van Bronswjk, W. (1994) Calibration of zircon standards for the Curtin SHRIMP II. In: Eighth International Conference on Geochronology, Cosmochronology and Isotope Geology, Berkeley, US (Abstract). U.S. Geological Survey Circular, 1107, 251. Schoene, B., Crowley, J.L., Condon, D.J., Schmitz, M.D., and Bowring, S.A. (2006) Reassessing the uranium decay constants for geochronology using ID-TIMS U Pb data. Geochimica et Cosmochimica Acta, 70, Spencer, K.J., Hacker, B.R., Kylander-Clark, A.R.C., Andersen, T.B., Cottle, J.M., Stearns, M.A., Poletti, J.E., and Seward, G.G.E. (2013) Campaign-style titanite U-Pb dating by laser-ablation ICP: Implications for crustal flow, phase transformations and titanite closure. Chemical Geology, 341, Stacey, J. and Kramers, J.D. (1975) Approximation of terrestrial lead isotope evolution by a two-stage model. Earth and Planetary Science Letters, 26,

27 Szymanowski, D., Fehr, M.A., Guillong, M., Coble, M.A., Wotzlaw, J.-F., Nasdala, L., Ellis, B.S., Bachmann, O., and Schönbächler, M. (2018) Isotope dilution anchoring of zircon reference materials for accurate Ti-in-zircon thermometry. Chemical Geology, 481, Wiedenbeck M., Alle P., Corfu F., Griffin W. L., Meier M., Oberli, F.V., Quadt, A.V., Roddick, J.C. and Spiegel, W. (1995) Three natural zircon standards for U Th Pb, Lu Hf, trace element and REE analyses. Geostandards Newsletter, 19, Wiedenbeck M., Hanchar J. M., Peck W. H., Sylvester P., Valley J., Whitehouse, M., Kronz, A., Morishita, Y., Nasdala, L., Fiebig, J. and Franchi, I. (2004) Further characterisation of the zircon crystal. Geostandards and Geoanalytical Research, 28, 9-39.

28 Supplemental Material Figures Supplemental Figure 1. Image of a fragment of the pale-green MAD-559 zircon. Original faceted faces are visible at the top of the image. Note the absence of visible inclusions or alteration. The volume pictured represents ~25% of the original stone. Supplemental Figure 2. U/Pb isotope ratio results for MAD-559 zircon measured by (A) LA-ICP-MS and (B) SIMS on SHRIMP-RG. Tera-Wasserburg inverse Concordia diagrams show individual measurements with 2s measurement precision corrected for common-pb using 204 Pb. The grey-filled ellipse is the error-weighted Concordia age (Ludwig, 1998) and 2s standard error of the mean including uncertainty in decay-constants. For SIMS measurements in (B), the grey lines represent analyses of thermally annealed MAD- 559, and black lines are natural, non-annealed zircon.

29 Supplemental Material Tables Supplemental Table 1. Mass fractions for zircon for Li, Be, B, F, Na, P, K, Ca, Sc, Fe, and Nb calculated relative to MAD-1 zircon (values from Barth and Wooden, 2010) based on new measurements collected in this study. The Ti mass fraction for is based on the isotope dilution (ID) ICP-MS measurements by Szymanowski et al. (2018), but is also reported relative to SL13 zircon for comparison. Element mass fractions for which there are no data reported are based on the compilation of previously published values in Table 1. Supplemental Table 2. Laser Raman measurement results for 91500, SL13, CZ3, MAD-559, and MAD-1 zircon. The peak position (cm -1 ), Raman shift, intensity, and FWHM (full width at half maximum) are calculated for the v3 [SiO4] mode. Mass fractions are from Tables 3, 4, and 5. Cumulative alpha dose Dα T ( /mg) was calculated assuming no annealing following Nasdala et al. (2001) and Palenik et al. (2003). Supplemental Table 3. U-Pb isotope ratio results for MAD-559 zircon measurements by SIMS and LA-ICP-MS. For SIMS data, both thermally annealed and natural zircon were measured; annealed grains are noted in the sample name. Calculated 206 Pb/ 238 U and 207 Pb/ 206 Pb ages are reported with 1s errors for individual spots and calculated weighted mean ages and 2s standard error of the mean for both methods. We also report ratios for plotting inverse concordia and conventional concordia diagrams plots. For LA-ICP-MS analyses, 2% uncertainty was added in quadrature to individual ratios to account for reproducibility of quality control reference materials. Supplemental Table 4. Mass fractions for MAD-559 zircon based on measurements by LA-ICP-MS are reported as the session mean and 1s intermediate precision. Elements not measured

30 are left blank. Mass fractions in grey are inconsistent with the preferred value measured by SIMS (typically these values are calculated relative to NIST-SRM; see Table 4). Supplemental Table 5. Mass fractions for MAD-1 zircon based on measurements by LA-ICP-MS. Mass fractions not measured are left blank. Data presentation is the same as in Supplemental Material Table 4. Supplemental Table 6. Mass fractions for SL13 and CZ3 zircon measured as quality control reference materials during this study. We report measurements by SIMS and by LA-ICP-MS, and the recommended (preferred) mass fraction is the mean and 1s intermediate precision of multiple sessions (n). Mass fractions shown in grey are not included in the mean due to surface contamination (see text). Mass fractions calculated relative to NIST-SRM for SL13 are generally not included in the mean because they are inconsistent with SIMS measurement results and previously published results, unless noted in the text. Mass fractions not measured are left blank. Supplemental Table 7. Mass fractions for MAD-1 zircon based on measurements by SIMS. Results are reported as sample means and 1s intermediate precision, as in Supplemental Table 4. Mass fractions shown in grey are not included in the compiled mean due to surface contamination (see manuscript text). Mass fractions for Li, Be, B, F, Na, P, Ca, Sc, Fe, Ge, and Nb are based on values from Barth and Wooden (2010) and are not reported for sessions #2 - #5 (see text). Other mass fractions not measured are left blank. Supplemental Table 8. Correction factors to update zircon trace-element mass fractions (unknown samples) calculated relative to previously published values for MAD-1 (Barth and Wooden, 2010) and MAD-559 (MADDER). Multiply previously published mass fractions by the listed

31 correction factors to update mass fractions relative to the new results presented herein. Mass fractions that are not changed or were not previously measured are left black.

32 Supplemental Material Figure μm