A new working reference material for cassiterite oxygen isotope microanalysis

Cassiterite is the principal ore mineral for tin, and its oxygen isotope is a promising proxy to trace the origin and evolution of ore‐forming fluids, which requires precise and accurate oxygen isotopic analysis. Secondary ion mass spectrometry (SIMS) is a powerful tool for oxygen isotope analysis, especially when samples bear complicated textures, but matrix‐matched reference materials are critical for accurate microanalysis. The only available matrix‐matched reference material for cassiterite oxygen isotope analysis is Yongde‐Cst, and more reference materials are required. Here, we report Piaotang‐Cst as a potential working reference material for cassiterite oxygen isotope microanalysis. Our extensive SIMS microanalysis confirmed that Piaotang‐Cst is relatively homogeneous, with an average two standard deviations (2SD) of 0.49‰ (n = 626). The δ18OVSMOW value of Piaotang‐Cst is 5.33 ± 0.07‰ (2SD, n = 5) as determined by a conventional fluorination isotope ratio mass spectrometer. We also demonstrated that there is no measurable matrix effect caused by the variable contents of trace elements for cassiterite oxygen isotope microanalysis. We proposed that the Piaotang‐Cst can be used as a working reference material for monitoring the external reproducibility and accuracy of SIMS analysis.


| INTRODUCTION
The source and evolution of ore-forming fluids are critical to establishing and refining genetic models of hydrothermal ore deposits.Oxygen is the main component of ore-forming fluids; hence, it is not surprising that its isotope composition (δ 18 O value) is widely used as a proxy to decipher the source and evolution of ore-forming fluids.However, robust evaluations of the source and evolutionary trajectory of ore-forming fluid are challenging.This is largely due to the uncertainties in obtaining and interpreting the oxygen isotope compositions of oreforming fluids.For example, fluid inclusion assemblages (FIAs) are arguably the most direct recorders of ore-forming fluids.However, a significant portion of the δ 18 O data of FIAs was collected through bulk analysis, which possesses the risks of mixing multiple generations of fluids and the post-entrapment modification of the FIAs.An alternative method is analyzing gangue minerals (e.g., carbonate and silicate) under the assumption that gangue minerals and ore minerals were precipitated by the same fluids. 1Unfortunately, determining whether gangue and ore minerals were precipitated from the same fluid with high confidence is difficult.As such, using ore minerals to obtain the oxygen isotopic composition of ore-forming fluids could yield more robust information.
3][4][5][6] As such, its oxygen isotope composition is an ideal proxy to provide a robust and direct insight into the temperature, source, and evolution of the oreforming fluid.This physical-chemical information can then be integrated with deposit geology to yield implications for controls of mineralization. 4Given the relatively large grain sizes of cassiterite (mm to cm), when analyzed at high spatial resolution (e.g., secondary ion mass spectrometry [SIMS]), it is possible to translate texturedcontrolled information into detailed temporal patterns.This translation approach has been successfully used to tackle the dynamics of fluid evolution at the Weilasituo Sn polymetallic deposit with high temporal resolution. 7Additionally, significant δ 18 O variation from a single cassiterite crystal was reported from the Piaotang tin-tungsten (Sn-W) deposit. 4The contrasting δ 18 O values of the core (À2.14‰) and rim (2.36‰) indicate that the cassiterite was crystallized from distinct pulses of magmatic-hydrothermal fluids with discrete meteoric water involvements.These pioneering studies demonstrate the great potential of cassiterite in decoding the ore-forming process and the importance of utilizing SIMS when studying minerals with complicated zonation. 4e oxygen isotopic composition of cassiterite is traditionally measured by the BrF 5 method using an isotope ratio mass spectrometer (IRMS). 8This is a bulk analytical method that is very accurate and precise but is time-consuming and requires large samples.Additionally, its poor spatial resolution (>>mm) is a limiting factor when samples are minute or bear complicated zones or textures.In sharp contrast, the recently developed SIMS method was capable of obtaining cassiterite oxygen isotope composition at the micrometer levels with minor degradation in precision. 9,10Moreover, SIMS measurement is highly efficient in contrast to bulk analytical methods.A single-spot analysis for SIMS oxygen isotope composition can be accomplished within approximately 4 min.However, the presence of a serious matrix effect is a major drawback for SIMS analysis, which requires highquality matrix-matched reference materials to correct the instrumental mass fractionation (IMF).Currently, only one cassiterite reference material is available (Yongde-Cst, δ 18 O VSMOW = 1.36 ± 0.16‰, two standard deviations [2SD]). 11As such, developing more reference materials that have a distinct δ 18 O value from that of Yongde-Cst 11 is critical to facilitate cassiterite oxygen isotopic analysis.
Here, we report Piaotang-Cst, a potential working reference material for cassiterite oxygen isotope analysis.Its oxygen isotope composition is homogeneous at the micro-scale level, as verified by SIMS, with its δ 18 O value being determined by IRMS.We also evaluated potential matrix effects related to cassiterite composition (e.g., Fe, Nb, and Ta).

| SAMPLE DESCRIPTION AND PREPARATION
Piaotang-Cst is a euhedral-subhedral cassiterite crystal collected from the Piaotang deposit in the Nanling Range, China, which is a typical W-Sn deposit with niobium (Nb) as a byproduct.Most mineralization occurs as stockworks in the Cambrian-Sinian low-grade metamorphosed sandstones. 12Stockworks in the deposit mainly consist of cassiterite, wolframite, and quartz.Piaotang-Cst is black with a size of approximately 2 Â 1.5 Â 0.5 cm 3 and a weight of approximately 10 g (Figure 1A).The sample is relatively clean, except for the disseminated presence of a few fractures and wolframite inclusions in a limited area ($4 Â 5 Â 5 mm).
Four pieces of Piaotang-Cst were cut from the crystal along its largest dimension and cast in two epoxy resin mounts (L299, L300) along with the Yongde-Cst reference material.The remaining fragments were ground into powder for IRMS measurements.The mounts were polished with a diamond abrasive to reach a flat surface.Before SIMS analysis, the polished mounts were coated with approximately 30 nm high-purity gold film under vacuum to provide <5 Ω resistance across the mount surface, and then placed in the SIMS vacuum storage for approximately 24 h to reduce the possible interference of hydrides.

| Laser-Raman spectroscopic analysis
Laser Raman spectroscopy analyses were performed using a Horiba confocal Raman Microscope Evolution at the Beijing Research Institute of Uranium Geology (BRIUG).The wavelength and spot diameter were 532 nm and approximately 1 μm, respectively.A 100Â OLYMPUS objective lens was selected for excitation and detection, and the Rayleigh light was rejected by a margin filter.A 180 grooves/mm grating was used to disperse the light with a spectral resolution of 4.0 cm À1 .The measurement range was 50-1500 cm À1 .The laser with a power of 8 mW was focused on the sample, and the acquisition time was 8 s with an accumulation of four times.

| Trace elements compositions by LA-HR-ICP-MS
Trace element composition was determined at BRIUG using a Thermo Scientific high-resolution inductively coupled plasma mass spectrometry (HR-ICP-MS) equipped with a Coherent GeoLas HD 193 nm ArF excimer laser-ablation system.During analysis, the spot diameter was 44 μm with a laser pulse rate of 6 Hz.The energy density was 6 J/cm 2 .The low mass resolution was used to maximize sensitivity.A squid smoothing device was utilized to reduce the statistical error caused by laser-ablation pulses and to improve the quality of the data.Helium was used as the carrier gas to enhance the efficiency of transportation and merged with argon (make-up gas) via a T-connector before entering the HR-ICP-MS.Each analysis included approximately 20 s of background acquisition (gas blank), followed by 50 s of data acquisition.Trace elements were calibrated against multiple reference materials (NIST SRM610, SRM612, BHVO-2G, and BIR-1G) and by normalizing all metal oxides to 100%. 13Data reduction was conducted with the ICPMSDataCal software. 14

| SIMS oxygen isotopic analysis
Oxygen isotopic analysis of cassiterite was conducted using a CAMECA IMS-1280HR ion microprobe at BRIUG.The detailed procedures for oxygen isotopic analysis by SIMS are described by Li et al. 11 The 133 Cs + primary ion beam was accelerated at 10 kV with an intensity of 2.6-3.0 nA.The Gaussian model with a 400 μm mass aperture forms a reduced ion source image with a focused beam density.The high transmission was achieved using a combination of transfer lens optics of 100 times magnification.The ion beam for analysis was were tuned to simultaneously achieve high transmission and high mass resolution.To compensate for sample charging during analysis, a normal incidence electron flood gun was applied with homogeneous electron density over an approximately 100 μm 2 elliptical area.A mass resolving power (MRP) of approximately 2400 (defined at 10% peak height) was used, which is capable of separating peaks from isobaric interferences and obtaining a sufficiently flat plateau.Two Faraday cup detectors were used to measure 16 O and 18 O ions simultaneously, which are located on the L'2 and H'2 trolleys equipped with a 10 10 and 10 11 Ω preamplifier, respectively.Under the analytical conditions described above, the intensity of the 16 O signal ranged from approximately 2.9 Â 10 9 counts per second (cps) to approximately 3.6 Â 10 9 cps.
A 60 s pre-analysis sputtering was used to remove the high-purity ). 15The results of 18 O/ 16 O ratios are reported in per mil notation relative to Vienna Standard Mean Ocean Water (VSMOW, 18 O/ 16 O = 0.0020052) according to Equation (1), 16 where m denotes the SIMS measured value.
The IMF (‰) was calculated following Equation ( 2), where r denotes the recommended value of the primary reference material (i.e., Yongde-Cst in this study).

| Oxygen isotopic analysis by isotope ratio mass spectrometry
The δ 18 O values of Piaotang-Cst were determined using the conventional fluorination method in the Stable Isotope Laboratory at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGG-CAS). 11,17For quality control, Chinese national reference materials GBW 04409 (quartz), 10 and 04BXL07 (garnet) 18 were analyzed during the study.Samples were fluorinated in the nickel reaction vessel, and each analysis consumed approximately 12 mg of materials.
Samples were loaded under a positive pressure of pure N 2 to prevent the adsorption of atmospheric moisture in the nickel reaction vessel.
After the initial overnight pumping and a 10-min room-temperature pre-fluorination with BrF 5 , the vessel was evacuated and again charged with an aliquot of BrF 5 .The nickel reaction vessel was heated to 750 C for 4 h using a resistance furnace.The gases generated were purified through a series of cryogenic traps, which were chilled by liquid nitrogen.The purified oxygen gas was analyzed by a MAT253 mass spectrometer.The reproducibility of the analysis is better than 0.2‰ (2SE) as evaluated by repeated analyses of GBW04409 (quartz).

| Topography measurement
The topography of craters created by SIMS analysis was measured using a Nexview NX2 3D white light interference surface profiler.The procedures are described by He et al. 19 The epoxy mounts were set in the middle of the stage and aligned under the eyepiece of the 3D surface profiler.The distance between the sample and eyepiece was

| Trace element composition
For four pieces of Piaotang-Cst, 25 domains were randomly selected for trace elements analysis by LA-HR-ICP-MS.We also analyzed the 115-137 ppm).For both samples, reproducibilities (<10% relative standard deviation, RSD) were obtained for Co and Ni, while the reproducibilities for other elements (e.g., W, U, Mn, Ti, Nb, Ta, Zr, Hf, and REEs) (Figure 2, Table S2) were poor.

| Homogeneity evaluation on oxygen isotopic composition of Piaotang-Cst by SIMS
SIMS oxygen isotope data of Piaotang-Cst are presented in Tables S3   and S4.The analysis positions and their measured values are visualized in Figures 3 and 4. Here, we have adopted a gridding sampling approach for the homogeneity test, as our previous studies have demonstrated its feasibility and convenience when samples are large in size. 11,15,28,29r mount L300, a total of 293 SIMS oxygen isotope measurements were conducted with an analytical duration of approximately  S6.
22.3 h (Figure 3A).After excluding two anomalous measurements (likely mineral inclusions seen from the unusual 16 O intensities), the measured δ 18 O values define a Gaussian distribution with a 2SD of 0.45‰ (2SD, n = 286) (Figure 3A,B) with the presence of five outliers (Figure 3A, Table S3).Yongde-Cst analyzed at the same time yielded a 2SD of 0.33‰ (n = 29).
For mount L299, a total of 349 SIMS oxygen isotope measurements were conducted on the other three pieces of Piaotang-Cst with an analytical duration of approximately 24.5 h (Figure 4A).After excluding three anomalous measurements (likely mineral inclusions) (Figure 4A, Table S4), the measured We further evaluated the accuracy of our method by using Yongde-Cst as a primary standard and Piaotang-Cst as an unknown.
While the anomaly values were caused by accidentally incorporating mineral inclusions, the cause of the extra scatters for the outliers is not immediately clear.Potential reasons could be due to a growth zone with distinct oxygen isotope compositions or the presence of unrecognized mineral inclusions.
The SIMS results of Piaotang-Cst from both mounts (Figures 3   and 4) suggest that Piaotang-Cst is relatively homogeneous at the microscale in terms of oxygen isotope composition.Furthermore, Piaotang-Cst has a different oxygen isotope composition from that of Yongde-Cst, which is ideal for quality control purposes.It should be noted that the oxygen isotope composition of Piaotang-Cst (0.49‰, 2SD) is less homogeneous than that of Yongde-Cst (0.20‰-0.50‰, 2SD), which is also reflected by the heterogeneous presence of trace elements at elevated abundances.As such, we propose that Piaotang-Cst should be used as a working reference material to monitor data quality.

| IRMS oxygen isotope results
The oxygen isotope composition of Piaotang-Cst was precisely determined using IRMS.The sample weights for each aliquot were

| The depth of craters
The whole analytical process has generated a crater with a similar depth of 1.38-1.45μm for Piaotang-Cst and 1.16-1.42μm for Yongde-Cst, as measured by a 3D surface profiler (Figure 6, Table S6).
In addition, the depths of craters were not related to Fe, Nb, and Ta contents, which indicates that the chemical composition variation of cassiterite cannot cause the difference in ionization rates using SIMS (Table S6).

| Potential matrix-effect linked to cassiterite chemistry
Matrix-dependent IMF presents serious challenges for SIMS oxygen isotope analysis, and previous studies have shown that the variable chemical composition of minerals may lead to significant IMF.[32] Variations in Fe and Mg contents in olivine and pyroxene also cause significant matrix effects. 33,34As such, correction for IMF is a prerequisite for SIMS analysis, which is routinely achieved utilizing matrixmatched reference materials.
To explore the potential matrix effect caused by chemical impurities in cassiterite, we have compiled literature data (Table S7).
Reported Fe, Nb, and Ta contents are shown in Figure 7 for cassiterite in variable environments, including granite, granite cupolas, quartzvein, greisen, pegmatite, skarn, and xenothermal-epithermal environments.These cassiterites show considerable variations in the trace elements they contain.Among all elements, the Fe, Nb, and Ta contents are generally the highest, with concentrations varying from 20 to 18,000 ppm, from 0 to 28,120 ppm, and from 0 to 61,950 ppm, respectively (Figure 7).The remaining trace elements in cassiterite are low in general.As such, these three elements (Fe, Nb, and Ta) are considered further for matrix effects during SIMS measurements.
To evaluate the potential matrix effect, we obtained the represented varied composition for spots previously analyzed for δ 18 O values of Piaotang-Cst and Yongde-Cst using SIMS.Although Piaotang-Cst and Yongde-Cst present significant composition differences of Fe, Nb, and Ta, no obvious IMF variation was observed (Figure 8A-C, Table S6), which indicates that chemical differences cannot produce a significant matrix effect during cassiterite oxygen isotopic measurement.

F
I G U R E 1 (A) A photo showing Piaotang-Cst; (B) Raman spectrum of Piaotang-Cst and Yongde-Cst.

approximately 10 Â
10 μm 2 in size.Prior to analysis, a 60 s presputtering under a raster model (25 Â 25 μm 2 ) was used to remove the gold coating and clean the sample surface.We further applied a 15 Â 15 μm 2 raster (including the analyzing area) to eliminate the depth effect.Secondary ions were extracted with an approximate 10 kV potential.The entrance silt (150 μm), field aperture (5000 μm), contrast aperture (400 μm), energy slit (50 eV), and exit silt (405 μm) gold layer, followed by an approximately 90 s secondary ion centering and mass calibration.The signals were measured in a 96 s accumulation session, which consists of 20 cycles of measurements, and each cycle contains 0.8 s of waiting time and 4 s of counting time.The total time of a single spot is approximately 4 min.The internal precision of individual analyses was generally better than 0.20‰ (two standard deviations [2SE] Chondrite normalized trace element pattern of Piaotang-Cst (blue) and Yongde-Cst (red).The chondrite values are from McDonough and Sun.24

3. 5 |
Chemical composition analysis by SIMSTo evaluate the potential matrix effect, we have measured 56 Fe/ 112 Sn,93  Nb/ 112 Sn, and 181 Ta/ 112 Sn raw ratios of the Piaotang-Cst without calibration of the matrix effect to estimate correlations between elemental contents and isotopic composition.A CAMECA IMS-1280HR ion microprobe at BRIUG was used.The O 2 À primary ion beam was accelerated at 13 kV with an intensity of approximately 1.0 nA.The high transmission was achieved using a combination of transfer lens optics of 200 times magnification.A 10 Â 10 μm 2 raster mode was employed for pre-analysis sputtering, and the raster size was turned off during analysis.Positive secondary ions were extracted with a 10 kV positive potential.The entrance slit (120 μm), field aperture (FA, 3000 μm), energy slit (50 eV), and exit slit (245 μm) were set.A monocollection mode was adopted to measure 56 Fe, 93 Nb, 112 Sn, and 181 Ta using an electron multiplier (EM).An MRP of 5000 was used to obtain a flat plateau, which is also capable of separating the potential isobaric interference.Under the analytical conditions described above, the intensity of the 56 Fe, 93 Nb, 112 Sn, and 181 Ta signals ranged from approximately 6.6 Â 10 4 cps to approximately 7.8 Â 10 5 cps, from approximately 3.7 Â 10 3 cps to approximately 2.1 Â 10 5 cps, from approximately 8.0 Â 10 4 cps to approximately 2.7 Â 10 5 cps, and from 48 cps to approximately 1.2 Â 10 4 cps, respectively.A 60 s pre-analysis sputtering was used to remove the high-purity gold layer and clean the sample surface to avoid contamination of impurity, followed by an approximately 120 s secondary ion centering and mass calibration.The signals were measured in an approximately 50 s accumulation session, which consists of 4 cycles of measurements, and each cycle contains approximately 9 s of waiting time and approximately 4 s of counting time.The total testing time of a single spot is approximately 4 min.F I G U R E 3 (A) Secondary ion mass spectrometry (SIMS) oxygen isotope results of the Piaotang-Cst measured in session 1 (mount L300); (B) histogram of SIMS oxygen isotopic results for Piaotang-Cst in mount L300; (C) sketch map of homogeneity testing with grid for Piaotang-Cst in mount L300.Error bars represent the internal precision.For both Yongde-Cst and Piaotang-Cst, their δ 18 O values were corrected for instrumental mass fractionation (IMF).The corrected means were plotted as red solid lines, and the grey areas represent the external precisions (two standard deviations [2SD]) of SIMS analysis.

5
The δ 18 O values of Piaotang-Cst measured by isotope ratio mass spectrometer [IRMS].Also shown are 1SD (grey) and two standard deviations (2SD) (light grey) of the mean.F I G U R E 6 The typical topographies (A, C) and sections (B, D) of craters profiled by a 3D surface profiler.F I G U R E 7 A compilation of trace element data in cassiterite from the literature.Here, we have shown the contents of Fe, Nb, and Ta content for cassiterite in granite, granite cupolas, quartz-vein, greisen, pegmatite, skarn, and xenothermal-epithermal environments; also shown is the cumulative distribution function in red.For references to data sources, please refer to Table

δ 18 O
values define a Gaussian distribution with a 2SD of 0.49‰ (n = 340) with the presence of six outliers.Yongde-Cst analyzed at the same time yielded a 2SD of 0.28‰ (n = 36, Figure 4A,B).Combined with two sessions, the measured δ 18 O values yield a Gaussian distribution with an average 2SD of 0.49‰ (n = 626).
The IMF was not related to the crater depths (Figure8D, TableS6), which indicates the ionization rates cannot produce a positive linear correlation.Therefore, the δ 18 O values variation of Piaotang-Cst may be caused by complex oreforming processes.As discussed above, although the trace element (e.g., Fe, Nb, and Ta) compositions of Piaotang-Cst and Yongde-Cst are distinct, noF I G U R E 9The relationship between 56 Fe/ 112 Sn (A),93  Nb/ 112 Sn (B),181  Ta/ 112 Sn (C), and instrumental mass fractionation (IMF) for Piaotang-Cst.