Extreme isotopologue disequilibrium in molecular SIMS species during SHRIMP geochronology

The current limitation in the accuracy and precision of inter-element analysis in secondary ion mass spectrometry (SIMS) is the ability to find measurable quantities that allow relative differences in ionization and transmission efficiency of secondary ions to be normalized. In uranium– thorium–lead geochronology, the ability to make these corrections, or “calibrate” the data, results in an accuracy limit of approximately 1 %. This study looks at the ionization of uranium and thorium oxide species, which are traditionally used in U–Pb calibration, to explore the conditions under which isotopologues, or molecular species whose composition differs only in the isotopic composition of one or more atoms in the molecule, remain in or deviate from equilibrium. Isotopologue deficits of up to 0.2 (200 ‰) below ideal mixing are observed in UO+2 species during SIMS gechronological analyses using the SHRIMP IIe SIMS instrument. These are identified by bombarding natural U-bearing minerals with an O2 primary beam. The large anomalies are associated with repeat analyses down a single SIMS sputtering crater (Compston et al., 1984), analysis of high-uranium, radiation-damaged zircon, and analysis of baddeleyite. Analysis of zircon under routine conditions yield UO+2 isotopologue anomalies generally within a few percent of equilibrium. The conditions under which the isotopologue anomalies are observed are also conditions in which the UOx-based corrections, or calibration, for relative U vs. Pb ionization efficiencies fail. The existence of these isotopologue anomalies suggest that failure of the various UOx species to equilibrate with each other is the reason that none of them will successfully correct the U /Pb ratio. No simple isotopologuebased correction is apparent. However, isotopologue disequilibrium appears to be a more sensitive tool for detecting high-U calibration breakdowns than Raman spectroscopy, which showed sharper peaks for ∼ 37 Ma high-uranium zircons than for reference zircons OG1 and Temora. U–Th– Sm /He ages were determined for aliquots of reference zircons OG1 (755±71 Ma) and Temora (323±43 Ma), suggesting that the broader Raman lines for the Temora reference zircons may be due to something other than accumulated radiation damage. Isotopologue abundances for UO and ThO and their energy spectra are consistent with most or all molecular species being the product of atomic recombination when the primary beam impact energy is greater than 5.7 keV. This, in addition to the large UO+2 instrumentally generated isotopologue disequilibria, suggests that any attempts to use SIMS to detect naturally occurring isotopologue deviations could be tricky.

Abstract.The current limitation in the accuracy and precision of inter-element analysis in secondary ion mass spectrometry (SIMS) is the ability to find measurable quantities that allow relative differences in ionization and transmission efficiency of secondary ions to be normalized.In uraniumthorium-lead geochronology, the ability to make these corrections, or "calibrate" the data, results in an accuracy limit of approximately 1 %.This study looks at the ionization of uranium and thorium oxide species, which are traditionally used in U-Pb calibration, to explore the conditions under which isotopologues, or molecular species whose composition differs only in the isotopic composition of one or more atoms in the molecule, remain in or deviate from equilibrium.
Isotopologue deficits of up to 0.2 (200 ‰) below ideal mixing are observed in UO + 2 species during SIMS gechronological analyses using the SHRIMP IIe SIMS instrument.These are identified by bombarding natural U-bearing minerals with an 18 O − 2 primary beam.The large anomalies are associated with repeat analyses down a single SIMS sputtering crater (Compston et al., 1984), analysis of high-uranium, radiation-damaged zircon, and analysis of baddeleyite.Analysis of zircon under routine conditions yield UO + 2 isotopologue anomalies generally within a few percent of equilibrium.The conditions under which the isotopologue anomalies are observed are also conditions in which the UO x -based corrections, or calibration, for relative U vs. Pb ionization efficiencies fail.The existence of these isotopologue anomalies suggest that failure of the various UO x species to equilibrate with each other is the reason that none of them will successfully correct the U / Pb ratio.No simple isotopologue-based correction is apparent.However, isotopologue disequilibrium appears to be a more sensitive tool for detecting high-U calibration breakdowns than Raman spectroscopy, which showed sharper peaks for ∼ 37 Ma high-uranium zircons than for reference zircons OG1 and Temora.U-Th-Sm / He ages were determined for aliquots of reference zircons OG1 (755±71 Ma) and Temora (323±43 Ma), suggesting that the broader Raman lines for the Temora reference zircons may be due to something other than accumulated radiation damage.
Isotopologue abundances for UO + and ThO + and their energy spectra are consistent with most or all molecular species being the product of atomic recombination when the primary beam impact energy is greater than 5.7 keV.This, in addition to the large UO + 2 instrumentally generated isotopologue disequilibria, suggests that any attempts to use SIMS to detect naturally occurring isotopologue deviations could be tricky.

Introduction
Determining the timing of geologic events is a fundamental constraint for unravelling the history of our planet.In the 120 years since the discovery of radioactivity, the use of radioactive decay has been an increasingly versatile, accurate, and precise way of measuring geologic time.One of the major advances in this field was the invention of the SHRIMP (Sensitive High Resolution Ion MicroProbe), which has been used for U-Pb geochronology of zircon for the last 34 years (Froude et al., 1983).

C. W. Magee Jr. et al.: Isotopologue disequilibrium in SIMS
The SHRIMP is a large-radius, magnetic sector secondary ion mass spectrometry (SIMS) instrument (Ireland et al., 2008).It features a low-impact-energy floating primary column optimized for Köhler illumination of a projection aperture onto the surface with a mass-filtered primary beam, a low-energy (350 V mm −1 ) initial extraction field, a quadrupole triplet for matching the secondary ion emittance to the mass spectrometer acceptance, and a large (1 m turning radius) CQH mass spectrometer with second-order focal aberration correction (Matsuda, 1974).The instrument used in this experiment is equipped with a single-electron multiplier as a detector; all data are ion counting on sequential mass peaks.The SHRIMP has been used to provide the geochronological foundations of thousands of papers, usually using a methodology best described several decades ago (Williams, 1998;Compston, et al., 1984).
A key requirement for successful SHRIMP U-Pb geochronology is accounting for the differential and variable secondary ionization yields of U and Pb in the SIMS source.The current method for doing this is the use of a calibration relating the detected ion ratio of Pb + / U + to the ratio of an oxidized molecular U-bearing ion to a reduced one.The most widely used calibration is a power law relationship between Pb + / U + and UO + / U + (Claoué- Long et al., 1995), but other calibrations using various combinations of U + , UO + , and UO + 2 have been used (Stern and Amelin, 2003).These calibrations apply to both SHRIMP and other SIMS instruments used for geochronology.
There are some drawbacks to this technique.Firstly, the covariation is not exact, and after calibration there remains a residual error that is rarely better than 0.5 %.Secondly, it requires corrections for bulk composition for zircons with more than 2000 ppm U (Williams and Hergt, 2000;White and Ireland, 2012), or for minerals that exhibit complex solid solution, such as monazite (Gregory et al., 2007;Fletcher et al., 2010).Thirdly, some simple oxides, such as baddeleyite and rutile, exhibit orientation-related deviations from the calibration (Wingate and Compston, 2000;Schmitt et al., 2010;Taylor et al., 2012).
As the relationship between uranium oxide formation and Pb ionization efficiency at the SIMS sputter site is not clearly understood, an 18 O − 2 primary beam was used to determine the source of the oxygen in the UO + species used for calibration (Magee et al., 2014).Previously, the use of 18 O implants in non-oxide species has been used for two purposes.Oxygen diffusion in either oxide or semiconductor matrices has been observed using an 18 O flood or ion implant (Kilner et al., 1996;Manning et al., 1997).In addition, 18 O implants have been used for quantifying behaviour of sputtered oxygen introduced by either a natural primary beam (Sobers Jr. et al., 2004) or by oxygen flooding (Franzreb et al., 2004).This study reverses the Sobers Jr. et al. (2004) method in which the 18 O implant is replaced with the stoichiometric natural oxygen composition (99.8 % 16 O) of the natural target material, and the 16 O − primary beam used by Sobers Jr. et al. (2004) is replaced by a 18 O − 2 primary beam.It is the first labelled oxygen SIMS experiment where an untreated natural mineral target is bombarded with an isotopically labelled primary oxygen beam, and the stoichiometry defines the 16 O content of the sputtered volume.
A description of the monoxide species formed in these experiments (Magee et al., 2014) showed that for primary energy greater than 5 keV, the relationships between uranium, thorium, and their monoxides were consistent with complete atomization and recombination (in the case of molecular species) during the sputtering process.However, the behaviour of more complicated molecular species, such as the actinide dioxides, was not considered there.This paper describes the ion abundances of the actinide dioxide species and discusses the possible meaning of the observed abundances.
One prediction of the atomization and recombination process is that the isotopologues of the molecular dioxide species should occur with abundances that are consistent with the isotopic ratio of the monoxide and elemental sputtered species.For example, if (as is the case in standard zircon geochronology) U 16 O and U 18 O are sputtered in equal proportion (Magee et al., 2014), then the predicted ratio of U 16 O 2 to U 16 O 18 O to U 18 O 2 is 1 : 2 : 1.This study tests this prediction of isotopologue equilibrium, and tries to explain the observed deviations from it.The conclusions drawn are then applied to practical SIMS geochronology, in an attempt to propose methodologies that will improve analytical performance.

Methods
The experiments were performed on the SHRIMP IIe instrument at Geoscience Australia (Stern et al., 2009).Following the first "blank" experiment (see section A, below), the tank of (isotopically natural) high-purity oxygen gas usually used to feed the duoplasmatron was replaced with a tank of 99.9 % 18 O 2 gas (Icon Isotope Services, Troy, NY).The primary beam Wein filter electrostatic voltage was then dropped to transmit mass 36, selecting the 18 O − 2 ion for sample bombardment.As the samples were all natural oxides, this instrumental setup yielded a situation where 99.7 % of the sputtered oxygen originating from the sample is 16 O, and essentially all the resputtered primary beam oxygen is 18 O.No oxygen flooding was used, yielding a simple two-component system.
The analytical run table included standard uranium lead geochronology peaks: 90 Zr 16 2 O reference peak, a background position, 204 Pb, 206 Pb, 207 Pb, 208 Pb, 232 Th, and 238 U.In addition to the standard actinide oxides used for U-Pb calibration (Williams, 1998) Aside from the additional mass stations, and the isotopically labelled primary beam, analyses were run automatically using standard Geoscience Australia procedures.Ions are extracted from the duoplasmatron ion source at 10 kV, focused through the primary column, and subjected to an additional 680 V of acceleration between the extraction plate and the sample, for a total primary impact energy of 10 680 V. Secondary ions are initially extracted off the surface of the target at 680 V before acceleration to a total of 10 kV for mass spectrometric analysis.The SHRIMP was set up with a 110 µm source slit and a 100 µm collector slit, yielding a 1 % mass resolution (M / M) of 5000 or greater for all peaks.Data were collected in six scans through the run table on a single electron multiplier.Data were reduced using SQUID 2.5 (Ludwig, 2010).Analytical and data reduction procedures are described in detail elsewhere (Magee et al., 2012).The mounts analysed were a mix of standard 25 mm mounts and 35 mm "megamounts" (Ickert et al., 2008).Baddeleyite analyses were performed using the same instrumental setup and analytical run table .The following experiments were performed.
Experiment A: analytical "blank" Before the 18 O bottle was connected to the duoplasmatron, the run table with the 18 O species was run using standard, isotopically undisturbed oxygen with both zircon and baddeleyite targets (zircon: session 110056; baddeleyite: session 110057).The primary beam Wein filter was used to select mass 32, so that the primary beam in this experiment (as is usually the case in standard SHRIMP II and SHRIMP RG analyses) was 16 O − 2 .The purpose of this was to see if any unexpected mass interferences existed that would complicate interpretations of the 18 O-bearing species.In session 110056, 25 grains of Temora-2 (Black et al., 2004) zircon were analysed, along with 22 grains of R-33 (Black et al., 2004) and 8 grains of OG1 (Stern et al., 2009).In 110057, an oriented Phalabowra baddeleyite mount from the original orientation work (Wingate and Compston, 2000) was analysed, with a dozen analyses on each of the four oriented megacrysts.

Experiment B: standard zircon
Four analytical sessions (110061,110066,110088,110106) were run with standard zircon using the 18 O − 2 primary beam.Two of these contained zircons of unknown age, the geochronological results of which are published in Laurie et al. (2016).The reference materials in those runs consisted of the following: Session 110061: 24 Temora-2 and 21R-33.

Experiment C: high-uranium zircon
The zircon mount used in a recent study of the high-U effect (White and Ireland, 2012) was repolished and reanalysed using the 18 O − 2 beam.These sessions were as follows: Session 110067: 10 Temora and 16 Red Hill (Williams and Hergt, 2000;White and Ireland, 2012) zircons.

Experiment D: baddeleyite
Baddeleyite was analysed in session 110060 and 110065.
Twelve spots on each of four oriented sections of Phalabowra were analysed in session 110060.In 110065, 20 randomly oriented Phalabowra and 20 randomly oriented Kurinelli dolerite (Claoué-Long et al., 2008) grains were analysed.

Experiment G: monazite
Session 110064 included multiple analyses of three popular monazite reference materials: 1409 (Stern and Berman, 2001), 8153 (Carson et al., 2008), and 44069 (Aleinikoff et al., 2006) were each analysed in 11 separate spots (one of each had the multiple downhole analyses described above).

Experiment H: ThO 2 disequilibria
In session 130058, the run table was further expanded to include all three major ThO 2 isotopologues, with the intention of comparing UO 2 and ThO 2 isotopologic deviations.The run table containing these additional peaks was used to analyse the high-U Raumid zircons.The purpose of this experiment is to see whether the more variable charge state of U causes different deviations to that of Th.

Raman
Laser Raman spectra were recorded on a Dilor ® Super-Labram spectrometer equipped with a holographic notch filter, 600 and 1800 g mm −1 gratings, and a liquid N 2 cooled, 2000 × 450 pixel CCD detector.The samples were illuminated with 514.5 nm laser excitation from a Melles Griot 543 Series argon ion laser, using 5 mW power at the sample surface, and a single 30 s accumulation.A 100 × Olympus microscope objective was used to focus the laser beam and collect the scattered light.The focused laser spot on the samples was approximately 1 µm in diameter.Wavenumbers are accurate to ±1 cm −1 as determined by plasma and neon emission lines.

Helium dating
In order to more fully understand the Raman result of the reference zircons Temora 2 and OG1, conventional (U-Th-Sm) / He dating was performed on several whole OG1 and Temora 2 zircons using methodology described in Danišík et al. (2012).Euhedral to subhedral single crystals were degassed at ∼ 1250 • C under ultra-high vacuum using a diode laser, and 4 He was measured by isotope dilution on a Pfeiffer Prisma QMS-200 mass spectrometer.Then the crystals were spiked with 235 U and 230 Th, dissolved and analysed by isotope dilution for U and Th on solution ICP-MS.(U-Th-Sm) / He ages were corrected for alpha recoil following the procedure of Farley et al. (1996) assuming a homogeneous U-Th distribution.
The (U-Th-Sm) / He dating was performed because, although the palaeoarchean OG1 has a crystallization age more than 8 times older than the Silurian Temora 2, it is possible that for much of that time the OG1 was at an elevated tem- perature, where radiation damage could anneal out.Assuming this temperature threshold is similar to the He retention temperature of ∼ 150-220 • C (Guenthner et al., 2013), this experiment was performed to see if a (U-Th-Sm) / He age much younger than the crystallization age could explain the Raman results.

Results
The relation between total UO + / U + and the U 18 O + / U 16 O + for different minerals is shown in Fig. 1.The relation between total UO + / U + and the U 18 O + / U 16 O + based on impact energy was shown in Magee et al. (2014).
The predicted ratios of U 18 O + 2 and U 16 O 18 O + relative to U 16 O + 2 were calculated from the observed U 18 O + / U 16 O + ratio, and compared to the observed ratios.For any observed U 18 O + / U 16 O + ratio, R, the equilibrium fractions of the dioxide species are A summary of results is listed in Table 1, while the full analysis-by-analysis results are listed in the Supplement.
Experiment A: analytical blank (standard oxygen instead of 18 O) 110056: for the Temora zircons, the total UO + / U + was about 6.2 (Supplement).U 18 O + / U 16 O + ratio averaged about 9.3 × 10 −4 , about half the natural 18 O / 16 O ratio (Fig. 1).This is consistent with about half the oxygen coming from the natural abundance sample, and half from the pure 16 O 2 beam (Magee et al., 2014).The weighted mean 272 is positive, 0.090±0.032(1σ ).This indicates an interference on U 16 O 18 O + which increases its intensity by about 10 %.However, without 18 O 2 feeding the duoplasmatron, U 16 O 18 O + is a very low intensity peak (generally less than 10 cps), so this interference is, on average, less than 1 cps.As count rates on U 16 O 18 O + using the 18 O beam were 1000 times higher, this minor excess is sufficiently small to be ignored when processing the 18 O primary beam data in the experiments below, where it will contribute less than 1 ‰.
The expected U 18 O + 2 counts are less than one millionth of the U 16 O + 2 counts, and only 100 000 U 16 O + 2 counts were collected.Therefore, we expect a fraction of a U 18 O + 2 count, and the 274 is dominated by counting noise on the background.Any interfering peak present is negligible.
110057: Although the total UO + / U + ratio for baddeleyite was much lower (a mean of 3.46, but scattered from just under 3 to almost 4), the same experiment performed on baddeleyite yielded broadly similar results; the U 18 O + / U 16 O + ratio averaged about 9.4 × 10 −4 , and the weighted mean 272 is positive but somewhat smaller than for zircon, 0.044 ± 0.049 %.This puts it within error of both zero and the zircon value.Once again, 274 is dominated by counting noise on the background, as the expected value is a fraction of a count.

Experiment B: standard zircons
The mean results for typical reference material and unknown zircons are shown in Table 1.These zircons showed a small but consistent negative D272 deviation, with a mean value of −0.014 ± 0.005.There was also a small but consistent positive D274 deviation of 0.018 ± 0.010.The range and uncertainty of 79 Temora analyses is shown in Fig. 2. No statistically significant isotopologue disequilibrium trends related to age are observed, but only one session (130058) had OG1 grains analysed, and the extreme age of this reference material (Stern et al., 2009) dominates any age-related statistics.The geochronologic results from unknowns M750 and M736 analysed in session 110088 are given in Laurie et al., (2016) alongside ID-TIMS results for the same samples.The agreement between the SHRIMP ages and the ID-TIMS ages for these samples suggests that the 18 O − 2 primary beam introduces no statistically significant geochronological bias.

Experiment C: high-uranium zircon
Two of three high-uranium samples, the Triassic Red Hill and the Eocene Raumid, exhibited a 274 excess of up to 0.1 (Fig. 3a, b).The Pleistocene Bishop Tuff showed no such deviation (Fig. 3c).The Red Hill and Raumid also showed apparent excess radiogenic Pb (the "High U effect"; Williams and Hergt, 2000, but this effect was not evident in the Bishop Tuff samples; Chamberlain et al., 2014;Ickert et al., 2015).There is not a strong correlation between degree of 274  anomaly and degree of excess apparent radiogenic 206 Pb (Fig. 3d-f).

Experiment D: baddeleyite
The results of the baddeleyite experiments are shown in Fig. 4.There are moderate to extreme negative 272 anomalies present.In addition, although the mean 274 is close to zero (Table 1), this belies substantial variation, between −0.07 and +0.11, and many times the intra spot uncertainty (Fig. 4).There is a strong correspondence between the degree of negative 272 and sign of 274, and the overall U 18 O + / U 16 O + ratio.In the analyses of the oriented crystal fragments (Fig. 4a), there is good overall agreement between most analyses of each segment, but large differences between them.These orthogonally mounted sections (Wingate and Compston, 2000) do not show the low U 18 O + / U 16 O + and high 272 and 274 results that were revealed from analysing randomly oriented crystals (Fig. 4b). -

Experiment E: impact energy
The results of the impact energy experiments for primary energy between 3 and 15 kV are shown in Fig. 5.The 3 kV results have poor counting stats due to the low sputter yield and primary beam current.For the rest of the analyses there was a slight trend to more negative 272 and 274 with increasing impact energy (Fig. 5).

Experiment F: downhole analyses
Additional analyses in a single spot were performed at a variety of primary ion energies between 3 and 15 kV (Supplement).Acquisitions were manually started at the end of each previous analysis without moving the stage or retuning the secondary or primary beams.For energies below 10 kV, the additional measurements downhole resulted in a slight increase in 272 and a larger increase in 274 (Fig. 6a), although the poor counting stats on the 3 kV experiment reduces the statistical significance for those results.For a primary energy of 10 kV and up, the repeated downhole analyses resulted in a slight decrease in 272, and a slight increase in 274.So the repeated downhole analyses in both instances moved away from the origin, and the location of standard, well-behaved analyses, but in different directions.
Figure 6b shows that a scan-by-scan breakdown of multiple downhole analyses can detect when the analyses "fall off" the calibration, an effect known since the early days of SHRIMP (Compston et al., 1984).This fall-off corresponds with a minimum in the U  calibration analyses which have the slightly lower 272 and slightly higher 274 values, compared to the on-calibration scans.

Experiment G: monazite
All monazite analyses showed extreme 274 anomalies, which were different between monazites of different composition.As no monazite analyses were performed using a standard oxygen primary beam, a composition-related interference cannot be ruled out, so no further interpretation will be done with these data.

Experiment H: ThO 2 isotopologue disequilibrium
ThO 2 isotopologue deviations were calculated in the same way as the UO 2 isotopologue deviations.
The predicted ratios of Th 18 O 2 and Th 16 O 18 O relative to Th 16 O 2 were calculated from the observed Th 18 O + / Th 16 O + ratio, and compared to the observed ratios.For any observed Th 18 O + / Th 16 O + ratio, R, the equilibrium fractions of the dioxide species are

Raman
The possible correlation of a D274 anomaly with the breakdown of the calibration due to high U led us to further investigate the radiation dosage and damage in the Raumid zircons relative to our reference zircons OG1 and Temora.Raman measurements were made on the high-U Raumid zircons, to determine if their high actinide contents had produced enough damage to broaden the Raman peaks.so that the Raumid results could be compared to U-Pb reference material.The results are given in Table 2 and Fig. 7.The Raman peak widths were broader than OG1 and similar to Temora.

Helium
(U-Th-Sm) / He dating was performed on several Temora 2 and OG1 grains, in order to better constrain the radiation history of these reference materials, and to estimate their relative radiation damage histories.Single-grain (U-Th-Sm) / He ages for OG1 range from 677.5 ± 36.3 to 815.5 ± 44.6 Ma (n = 6; average: 755 ± 71 Ma) and are by ∼ 400 Myr older than (U-Th-Sm) / He ages of Temora 2, which range from 287.9 ± 15.3 to 370.6 ± 19.8 Ma (n = 5; average: 323 ± 43 Ma) (Table 3).These results indicate that OG1 accumulated radiation damage for about twice as long as Temora 2, despite having a crystallization age 8 times older.

Discussion
It appears that UO 2 isotopologues are in or close to equilibrium in those circumstances where the SIMS UO + x / U +based calibration successfully works in the SHRIMP.In general, the UO 2 isotopologue disequilibrium data show that in most known situations where the SIMS UO + x / U + -based calibration breaks down, the UO 2 isotopologue disequilibrium also increases.This is consistent with the long-known observation that Pb ionization is closely related to oxygen availability at the sputter site (Schuhmacher et al., 1993;Schmitt et al., 2010), and that the uranium oxide ratios are an accurate monitor of oxygen availability (Schmitt and Zack, 2012).The UO 2 isotopologue deviations show that in the calibration breakdown scenarios previously discovered (Compston et al., 1984;Williams and Hergt, 2000;Wingate and Compston, 2000), the various combinations of uranium and oxygen are, by definition, not in equilibrium with each other.Therefore it is no surprise if they fail to predict the Pb + ionization efficiency.
Similar calibration problems have been reported from smaller radius magnetic sector ion probes (Schmitt et al., 2010;Schmitt and Zack, 2012;Schaltegger et al., 2015).The use of oxygen flooding to enhance Pb sensitivity in those experiments complicates the repeat of this study in such SIMS instruments, as it adds a third oxygen source (the flood), in addition to the silicate/oxide matrix and the primary beam.However, the reduction of calibration problems in the baddeleyite matrix under flooding conditions (Schmitt et al., 2010) is consistent with the UO 2 isotopologue results presented here, which show that the oxygen from the matrix and the primary beam have vastly different behaviour from grain to grain (as evidenced by 272 and 274 deviations in the tens of percent).Sadly, no simple 272-or 274-based correction puts the baddeleyite data collected here back on www.geosci-instrum-method-data-syst.net/6/523/2017/Geosci.Instrum.Method.Data Syst., 6, 523-536, 2017 the calibration line to a useful precision, as the spot-to-spot scatter can only be reduced from ∼ 10 to ∼ 5 % by applying such a correction.The physical explanation for the observed UO 2 isotopologue disequilibrium is still unclear, but a potential candidate is discussed below.

Molecular ion escape model
It is worth a quick thought experiment to determine whether or not any simple mechanical models can explain the behaviour that we observe.If intact U 16 O + ions from the sample are escaping the sputtering site to the mass spectrometer without reacting with other species (as opposed to U 16 O + being formed from the recombination of sputtered atoms), the U 18 O + / U 16 O + measured by the mass spectrometer will be lower than that of the sputtering site.As a result, the equilibrium mixture of UO 2 species in equilibrium with the (unknown) U   8. These results suggest that if the intact molecular ion escape for matrix UO and UO 2 is responsible for the 272 and 274 deviations, the escaped unreactive ions must be present at the 5 to 20 % level.However, unreactive UO x species should result in a UO x production that is high relative to the Pb + production, not low, and baddeleyite miscalibrations can be either high or low relative to the 207 Pb/ 206 Pb age.So even at a qualitative level, this hypothesis is somewhat lacking.
The trend towards slightly positive 274 in the UO + 2 ions is not understood, although for the low-energy ions it may be consistent with a few % ejection of intact U 16 O + ions from the surface.The negative 272 of UO + 2 ions produced by the 12.5 and 15 kV primary ions is consistent with UO + 2 and UO + ions being ejected in a 2 : 1 ratio.
All UO + 2 isotopologues exhibit gas-phase (ion energy less than the extraction potential) tails not present in UO + , U + , or Pb + ion energy spectra (Fig. 9).This suggests that inferring percentage level or better U / Pb ratios using a calibration based on UO + 2 may be fraught with unforeseen com-   plications compared with the traditional UO + -based calibration.An additional complication may occur for UO + 2 -based calibrations at impact energies greater than 10 keV, if intact matrix UO + 2 ejection is occurring and masking the oxygen availability for Pb ionization enhancement.

High-uranium effect
The ∼ 36 Ma Raumid and ∼ 180 Ma Red Hill (Williams and Hergt, 2000;White and Ireland, 2012) zircons show a 274 excess up to about 0.1.However, the 0.8 Ma Bishop Tuff zircons, which do not seem to exhibit this high-uranium effect (Chamberlain et al., 2014;Ickert et al., 2015), have only one analytical spot with a 274 above 0.025, and none above 0.05.A comparison of 274 and 268 in the Raumid zircons show that both the uranium and the thorium isotopologues show similar behaviour (Fig. 10).Thus the effect is probably unrelated to the higher potential valence states of uranium.
Raman measurements were made on the Raumid grains in order to determine whether any lattice damage parameters could be identified that may relate to the production of a 274 excess (Fig. 7).White and Ireland (2012) have shown that Jurassic high-U zircons which show an apparent U-concentration effect are metamict, according to the spectral featured described in Nasdala et al. (1995).
Depending of the degree of radiation damage, Raman spectroscopy can be used to measure the degree of metamictization in zircons (Nasdala et al., 1995).Well-ordered zircons show narrow internal and external vibrational bands in the spectral range below 1100 cm −1 (for band assignments, see Dawson et al., 1971).With increasing metamictization, all main Raman bands decrease in intensity, become increasingly broader, and shift towards lower wavenumbers.These changes are due to decreasing short-range order in the radiation-damaged zircon (Nasdala et al., 2004).In the case of natural zircons, decreased short-range order is mainly caused by radiation damage.However, the presence of several wt % U and Th may also decrease the short-range order and cause detectable effects on the Raman band width.This has been demonstrated by Podor (1995) for synthetic rareearth monazites.
The Raman results show slight 1007 cm −1 peak broadening in the high-U Raumid samples relative to the low-U samples.However, The highest U grains are barely broader than the OG-1 analysis, and none of the Raumid Raman results show band widths as wide as the Temora zircon (Fig. 11).The (U-Th-Sm) / He data show that the Palaeoarchean OG-1 standard zircons have a Neoproterozoic He age, making their cooling age about twice as old as the Temora grains, instead of 8 times older.The He content of the OG-1 zircons is, on average, about 3 times that of Temora (Table 3).So it is unlikely that radiation dose alone explains the broader Raman peak in Temora relative to OG-1 or Raumid.Despite the "high-U" effect of excess apparent 206 Pb being present in the Raumid sample, there is no evidence of metamictization.
The absence of a "high-U effect" in the Pleistocene Bishop Tuff, combined with the presence of a "high-U effect" and the 274 and 268 anomalies in the Eocene Raumid zircons with relatively low levels of Raman broadening, suggests that the 274 anomaly is a more sensitive indicator of a high-U effect than the 1005 cm −1 Raman peak width, and that the radiation dose in the zircon required to trigger the "high-U effect" is relatively small.There is some debate as to whether or not the zircon selfannealing temperature is at the ∼ 200 • C He closure temperature (Weber et al., 1997), or up in the greenschist facies around the biotite closure temperature (Pidgeon, 2014).However, for the Temora and OG1 results described here, this is irrelevant.The Biotite Ar closure time of the Middledale Gabbro (host of the Temora zircons) cannot be older than the crystallization age of 417 Ma (Black et al., 2004).The Ar-in-biotite closure temperature for the Owen's Gully diorite (OG1) cannot be younger than the He age presented here.So just like the He age, the biotite Ar age of OG1 must be significantly older than that of Temora.Further investigation of the Raman behaviour of Temora zircons would be a good idea, to ascertain whether this particular grain was an outlier, or whether there is a non-radiation component of peak broadening.
In terms of radiation damage, a Bishop Tuff zircon would have one decay chain per 1700 nm 3 , while a Raumid zircon of the same U concentration would have one per 3700 nm 3 .This is a similar dosage to a 200 ppm zircon of 1 Ga age, and such zircons show no trace of the high-uranium effect.

Conclusions
When using an isotopically labelled oxygen primary beam, the ratio of beam/sample oxygen in oxide species can be used to predict the ratio of dioxide species.For UO and UO 2 in minerals of geochronological interest, these predictions are close in systems where SIMS geochronology works well, but show large deviations in systems such as baddeleyite dating, high-uranium zircon, and repetitive downhole measurements, where U-Pb geochronology has poor accuracy.This is consistent with the UO x term in the U-Pb calibration being a monitor of oxygen availability for Pb + ion production, as if various UO species can be shown to be out of disequilibrium with each other, then it is unlikely that any particular one will be useful for predicting the oxygen-based enhancement of Pb ion formation.
The excess in 274 appears to correlate with the highuranium effect which causes erroneously old SIMS U-Pb ages better than crystal lattice damage, as determined by the broadening of the 1005 cm −1 Raman band.The cost of the 18 O 2 gas (∼ USD 750 per litre) might prevent routine use for analysing zircons where the SIMS high-U effect might be present, but as the duoplasmatron uses only a few tens of millilitres per day, it is not inconceivable that this technique could be used on zircons with ages intermediate between the ∼ 37 Ma Raumid and the 0.8 Ma Bishop Tuff, to ascertain when the high-U effect is initiated, and what other crystallographic features might be associated with it.
of this in-house research.Martin Danišík was supported by the AuScope NCRIS2 programme, Australian Scientific Instruments Pty Ltd., Australian Research Council (ARC) Discovery funding scheme (DP160102427), and Curtin Research Fellowship.We thank Kenji Horie and Trevor Ireland for constructive reviews.
Edited by: Lev Eppelbaum Reviewed by: Kenji Horie and Trevor Ireland

Figure 1 .
Figure 1.Plot of U 18 O + / U 16 O + vs.Total UO + (U 16 O + + U 18 O + ) / U + for various target minerals.Monazite sample identifications are given in the text.Two Kohler aperture sizes were used in baddeleyite analyses.

Figure 3 Figure 4 .
Figure 3. (a, b, c) 272 vs. 274 plots for analyses of highuranium zircons of three different ages.(d, e, f) Apparent 206 Pb / 238 U ratio vs. 274, colour coded for U content, for highuranium zircons of three different ages.

Figure 6 .
Figure 6.(a) 272 vs. 274 plots for multiple zircon analyses down the same hole under a variety of primary acceleration energies.(b) U 18 O + / U 16 O + vs. calibration constant on a scan-byscan basis for three downhole analyses.Grey ellipses are singlescan data for individual, non-repeated analyses.

Figure 7 .
Figure 7. Raman peak position vs. peak width at half height for Raumid zircons and reference zircons OG1 and Temora.

Figure 11 .
Figure 11.Comparison of 274 vs. Raman peak width for selected ∼ 37 Ma Raumid zircon grains.Standard reference zircons OG1 and Temora are shown for comparison.

Table 1 .
Summary of isotopologue deviations for analytical sessions.

Table 2 .
Raman peak positions and widths at half height.

Table 3 .
Uranium-thorium-samarium/helium dating results.Italic is a geologically implausible result excluded from interpretation; bold is the helium retention age.
* Outlier identified based on the corresponding U-Pb age.
18 O + / U 16 O + of the sputtering site will have positive 272 and 274 values relative to the observed U 18 O + / U 16 O + .
where B is the fraction of U 16 O + which escapes without reaction, and C is the fraction of U 16 O + 2 which escapes without reacting.As the U 16 O + escape curve has a steeper slope than the U 16 O + 2 escape line, combinations of these two mechanisms can produce 272 and 274 values that lie above the U 16 O + 2 escape line in the southwest quadrant of 272 by 274 space, to the left of the U 16 O + escape curve in the northeast quadrant of 272 by 274 space, or anywhere in the northwest quadrant.The southeast quadrant is inaccessible via either of these mechanisms in any combination.
Figure 9. Normalized energy plots for the UO + and UO + 2 species U 16 O + , U 18 O + , U 16 O + 2 , and U 18 O + 2 .Energy is relative to 10 kV extraction potential.Peak intensities are normalized and plotted in log scale.Note that the UO 2 ions have a population with negative initial energy not present in the UO species.
Figure10.274 vs. 268 plot for Raumid zircon.This plot shows that the dioxide disequilibrium is the same for Th and U oxides, so that the metal phase is unlikely to have much of a role in isotopologue formation.