Embed Size (px)
Citation preview
Chemical Geology 356 (2013) 94–108
Contents lists available at ScienceDirect
Chemical Geology
j ourna l homepage: www.e lsev ie r .com/ locate /chemgeo
Accurate Mg/Ca, Sr/Ca, and Ba/Ca ratio measurements in carbonates bySIMS and NanoSIMS and an assessment of heterogeneity in commoncalcium carbonate standards
R.I. Gabitov ⁎, A.C. Gagnon 1, Y. Guan, J.M. Eiler, J.F. AdkinsCalifornia Institute of Technology, Pasadena, CA 91125, USA
⁎ Corresponding author at: Mississippi State UniversityE-mail address: [email protected] (R.I. Gabitov
1 Currently at the University of Washington, USA.
0009-2541/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.chemgeo.2013.07.019
a b s t r a c t
a r t i c l e i n f oArticle history:Received 7 April 2012Received in revised form 20 July 2013Accepted 22 July 2013Available online 29 July 2013
Editor: U. Brand
Keywords:NanoSIMSHeterogeneityElemental ratiosCalciteAragonite
As archives of past climate variability, the micron and sub-micron scales of element:calcium (Me/Ca) variabilityin both biogenic and inorganic carbonates contain important geochemical information. Ideallyworking at smallerand smaller scales leads to higher temporal resolution of past changes, but more often it has revealed the strongoverprint of other processes on the climate signal. Therefore, the role of SIMS andNanoSIMS techniques in study-ing paleoenvironmental proxies continues to increase. We evaluate the accuracy and precision of the CAMECAims 7F-GEO and NanoSIMS-50L ion probes for measurements of Sr/Ca, Mg/Ca, and Ba/Ca ratios in carbonateminerals. Nine carbonate referencematerials were examined for their 88Sr/42Ca, 24Mg/42Ca, and 138Ba/42Ca ratiosusing a primary O− beam with spot sizes of 20–40 μm (SIMS) and 0.8–2 μm (NanoSIMS). To assess accuracy,seven of these standards were analyzed for Sr/Ca and Mg/Ca with ID-ICP-MS. Variability in the elemental ratiosarising fromdrift and changes in the tuning of the ims 7F-GEOover a ninemonth period is smaller than the chem-ical heterogeneity of themost frequently analyzed standards (OKA and Blue-CC). Across a whole crystal, Blue-CCismore homogeneous (1σ of 2.39% and 1.60% for Sr/Ca andMg/Ca respectively) than OKA, but in the bulkmatrixof OKA there is even less variability (1σ of 0.85% and 0.83% respectively). We find that carbonate samples canbe accurately normalized to carbonate standards with significantly different absolute Me/Ca ratios. NanoSIMSintensity ratios follow counting statistics better than ±1% (2σ) at any one spot, but conversion to Me/Ca ratiosincreases the uncertainty. Two factors give rise to this limitation. First, the spatial heterogeneity of nominallyhomogeneous standards can lead to accuracy offsets that are as large as the ranges quoted above. Second, theNanoSIMS generates higher instrumental mass fractionation relative to SIMS. The combination of three differentanalytical techniques demonstrates that Blue-CC and homogeneous calcite zones inOKA are promising referencematerials for precise analyses. Accuracy is crucially dependent onmaking independentmeasurements on exactlythe same crystal of standard used in the SIMS and NanoSIMS machines.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
Secondary ion mass spectrometry (SIMS) has emerged as an impor-tant technique for characterizing the chemistry of carbonate minerals(Mason, 1987; Riciputi et al., 1994; Savard et al., 1995; Allison, 1996;Hart and Cohen, 1996; Denniston et al., 1997; Cohen et al., 2001;Finch et al., 2001; Heikoop et al., 2002; Meibom et al., 2004; Sanoet al., 2005; Shirai et al., 2005; Weber et al., 2005; Cohen et al., 2006;Gaetani and Cohen, 2006; Kunioka et al., 2006; Meibom et al., 2006;Allison et al., 2007; Reynaud et al., 2007; Stolarski et al., 2007; Stollet al., 2007; Allison and Austin, 2008; Meibom et al., 2008; Shirai et al.,2008; Cohen et al., 2009; Holcomb et al., 2009; Kasemann et al., 2009;Rollion-Bard et al., 2009; Rollion-Bard and Erez, 2010; Gabitov et al.,2011). This trend is motivated by the ability of SIMS to constrain both
, USA.).
ghts reserved.
major and trace element abundances as well as isotope ratios at spatialscales of 20–30 μm (for most conventional instruments) to as small as1 μm (for the NanoSIMS; or, in some cases, less; Badro et al., 2007;Meibom et al., 2008). Thus, ion probes are capable of measuring arange of analytes similar to laser ablation ICP-MS, but with a spatial res-olution similar to the electron probe or secondary electron microscopy.And, as for all in situmethods, SIMS avoids the sample handling and po-tential contamination required for most wet chemical analysis tech-niques. Several reviews of relevant instruments and techniquesdescribe these general strengths of SIMS for geoscience applications(Werner, 1978; Shimizu and Hart, 1982; Fayek, 2009, etc.).
One complication of SIMS analyses is that instrumental fraction-ations (i.e., differences between detected ion intensity ratios and corre-sponding abundance ratios of those elements or isotopes in the sample)are functions of the composition, structure and orientation of the sam-ple. For example, Steele et al. (1981) showed that detected intensityratios of Mg+/Si+ and Fe+/Si+ varied non-linearly with changing Mg/(Mg + Fe) in olivine and low-Ca pyroxene. These analytical artifacts
95R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
arise from differential sputtering yields, differences in probabilities ofionization, and fractionations associated with ion optics and detectorefficiencies. The term ‘matrix effects’ refers to the dependence of theseanalytical artifacts on the chemical and physical properties of the sam-ple.While the family of processes governingmatrix effects is recognized(Slodzian et al., 1980), it is generally unclearwhya specific instrumentalfractionation is observed for a specific material and instrument tuningcondition. Therefore, the accuracy of SIMS measurements dependsupon empirical standardization schemes in which it is assumed that theinstrumental fractionation for an unknown sample is equal to that in astandard that is closely similar in chemistry and structure (and perhapsorientation) to the sample, analyzed under the same instrumental condi-tions as the unknown.
Standardization schemes for SIMS analyses are well established fora variety of analytes in a variety of targets, but the growing importanceof this technique for the study of carbonates has raised two relatedproblems that have not been satisfactorily resolved: (1) Natural carbon-ates are chemically and structurally complex, meaning a variety of stan-dards are needed. And, natural carbonates are frequently heterogeneous,meaning many potential standards are actually ill suited to in-situanalyses (we will show here that one of the most common carbonatestandards currently in use for trace element analyses is highly heteroge-neous). It is not clear that we yet have a suitable set of interlaboratorystandards. And, (2) many natural carbonates are compositionally zonedover very short length scales (~a few microns or less), meaning theNanoSIMS is the only class of ion probes suitable for consistently resolv-ing zonation. But the NanoSIMS differs markedly in its ion optics frommost other SIMS (e.g., its short distance between sample and extractionlens means ions are collected from a wider range of emission angles —i.e., the NanoSIMS actually analyzes populations of ions different fromthose collected on a conventional instrument). Very little data havebeen presented demonstrating the limits of accuracy and precision ofNanoSIMS measurements of carbonates, or any distinctive behaviors itmight exhibit with respect to matrix effects or instrumental fraction-ations generally.
This paper addresses these problems for SIMS and NanoSIMS mea-surements of Sr/Ca, Mg/Ca, and Ba/Ca ratios in natural carbonates. Inparticular, we examine the contributions to analytical uncertaintiesin such measurements arising from instrumental causes vs. those thatarise from heterogeneity of several widely used carbonate standards.We compare and contrast the performance of the ims 7F-GEO (a rela-tively conventional SIMS) and NanoSIMS-50L machines on a commonset of standards and ratios (to the best of our knowledge, this is thefirst detailed comparison of the analytical capabilities of these twoinstruments for any substantial set of well-characterized materials).Finally, we also document the heterogeneity of each standard usingisotope-dilution solution ICP-MS measurements (ID-ICP-MS) and high-resolution SEM imaging, using a ‘nested’ approach that lets us use mea-surements at a relatively coarse scale to calibrate the accuracy of groupsof finer-scale in situ analyses. These data offer the first detailed docu-mentation of the properties of a set of standards suitable for SIMS analy-sis of trace elements in carbonates, and demonstrates the limitationsand controls of analytical accuracy and precision for suchmeasurementswith the NanoSIMS ion probe— the highest-resolution instrument avail-able for in situ trace element analysis.
2. Methods
2.1. Standards analyzed
We analyzed Mg/Ca, Sr/Ca, and Ba/Ca ratios in nine natural calcitesand aragonites: OKA, Blue-CC, AG-1, NBS-19, 135-CC, LAS-20, RPI-CC,HUJ-AR, and UCI-CC by measuring 24Mg, 42Ca, 88Sr, and 138Ba. Eight ofthese were analyzed by Isotope Dilution ICP-MS at Caltech for Sr/Caand Mg/Ca, while AG-1 was analyzed for the same ratios at WHOIusing a Thermo Scientific Element-II ICP-MS. OKA, Blue-CC, NBS-19,
and LAS-20 are commonly used to standardize SIMS analyses of traceand minor elements in foraminifera, corals, or loculi calcites (Dennistonet al., 1997; Cohen et al., 2001; Heikoop et al., 2002; Bice et al., 2005;Bond et al., 2005; Sano et al., 2005; Allison et al., 2007). 135-CC calciteis used as an electron microprobe standard at UC-Berkeley (http://epmalab.uoregon.edu/UCB_EPMA/standards.htm#calcite). RPI-CC hasbeen used previously to standardize analyses of Sr and Pb in calcite(Cherniak, 1997). Each Me/Ca ratio (where Me stands for any metalother than Ca) spans over two orders of magnitude between all ofthe standards. Further details about each standard are presented inAppendix 1.
2.2. Inductively coupled mass spectrometry (ICP-MS)
Powders prepared fromall nine carbonate standardsweremeasuredby ICP-MS at Caltech to provide an accurate reference frame for theSIMS and NanoSIMS measurements of Me/Ca ratios. Mg/Ca and Sr/Cawere measured using an improved version of the isotope-dilutionmethod described in Fernandez et al. (2010) by adapting the methodfor a Neptune multi-collector ICP-MS (ThermoFinnigan). The externalprecision of the method was assessed by the regular analysis of adissolved deep-sea coral consistency standard with a Sr/Ca ratio of10.26 mmol/mol and a Mg/Ca ratio of 3.062 mmol/mol. For the twoanalytical sessions used in this study, the reproducibility of 8 consisten-cy standards was 0.4‰ for Sr/Ca and 5‰ for Mg/Ca (2 standard devia-tions, relative error). This external error is a somewhat better thanthe long-term reproducibility of 98 consistency standardmeasurementsover the last four years, which is 1‰ for Sr/Ca and 12‰ for Mg/Ca. Evenusing the more conservative long term reproducibility as an error esti-mate, analytical uncertainty is much smaller than most of the signalsdiscussed in this paper. The external precision for Sr/Ca using our im-proved method is similar to more time consuming ID-TIMS methods(Beck et al., 1992; DeVilliers et al., 1995) with an additional advantagethat we can simultaneously measure Mg/Ca ratios with high precision.The ultimate accuracy of this method is set by our spike calibration asdescribed in Fernandez et al. (2010).
2.3. SIMS
Before analysis in the CaltechMicroanalysis Center with Cameca ims7f-Geo, all samples were polished with SIC paper with a grain size of 15,5, and 1 μm, followed by a polishing cloth containing 1 μm diamondpaste. Samples were finished on a table mounted vibrating disk toppedwith a polishing cloth containing colloidal silica. After polishing, sam-ples were rinsed and sonnicated for 10–30 s with DI water and ethanoland dried overnight at 50 °C and coated with 20 nm of gold. In mostcases sample imaging by SEM (see below) followed the SIMS analysesand this gold coating was removed by gentle polishing with a 1 μmdiamond powder. However, for the SIMS sessions numbered 4.4, 4.5,and 4.6 (see below) samples were analyzed on the SIMS after SEMimaging, and the gold coating was applied after removing the carboncoating used in the SEM.
Over 9 months a total of 146 spot analyses were collected from thesuite of proposed analytical standards. All four analytical sessions lastedabout 1 week each and these are given separate ‘session’ numbers. Ses-sions 3 and 4 contained a few sub-sessions numbered after the decimalpoint as 3.1. For each spot analysis a 13 keV primary beam of 16O− ionswas focused to a 20–40 μm spot on the sample surface. The primarybeam current spanned 0.8 to 5.8 nA between sessions but varied by10% (2 s.d.) within a single sub-session. Each spot was pre-sputteredfor 2 min, followed by data collection at the same beam conditions asthe pre-sputtering. Positive secondary ion intensities were measuredin pulse counting mode on an ETP electron multiplier using a −84 ±5 eV energy filter and magnet peak hopping to switch between iso-topes. Count rates of 42Ca+ secondary ions generally varied between~5000 and ~40,000 cps, largely in response to variations in primary
96 R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
beam current; in general ~7000 cps of secondary ions were detectedper nA of primary beam current. Count rates of 24Mg+ secondary ions(~43,000–95,000 cps), 88Sr+ (~30,000–154,000 cps), and 138Ba+
(~200–9300 cps) varied both due to variations in the intensity of theprimary beam and differences in concentration between the nine stud-ied samples. A dead time correction was not applied to the data due tothe relatively low secondary ion beam intensities. Each analysis wasdefined to be 20 or 30 cycles long, where each cycle used magneticpeak hopping to rotate through the sequence of detected ions, 24Mg+,42Ca+, 88Sr+, and 138Ba+, and each species was integrated for 2 or 4, 1or 2, 1 or 2, and 15 or 30 s per cycle, respectively (numbers of cyclesand dwell times per cycle varied from session to session). A few analy-sesweremadewith 180 cycles to estimate the effect of acquisition timeon measured isotopic compositions.
Because of the chemical complexity of geological materials, molecu-lar interferences can be significant in SIMS analyses. It has been demon-strated that energy filtering reduces interferences from molecular ionsduringmeasurements of 24Mg+, 42Ca+, 88Sr+, and 138Ba+ from carbon-ate materials (Herzog et al., 1973; Shimizu et al., 1978; Allison, 1996;Hart and Cohen, 1996; Denniston et al., 1997; Gaetani and Cohen,2006). As an additional check for molecular interferences in this study,we performed high mass resolution scans of secondary ions including24Mg+, 42Ca+, 88Sr+ and their potential isobaric interferences in OKAcalcite (Fig. S-1A–C). Mass resolution was 5700–5780 (M/ΔM) using a5%-to-95% peak width definition. Even without energy filtering, 24Mg+
and 42Ca+ appear to be completely free of molecular ion interferences.The small peak corresponding to the calcium dimer (44Ca2+) is0.012 amu from the peak center of 88Sr (Fig. S-1C). Assuming that thecenter of 44Ca2+ is at 87.917 and the peak is symmetrical, the intensityof 44Ca2+ is expected to be negligible at the center of the 88Sr+ peak(87.905). A scan of 88Sr+ using an energy offset of −80 eV yielded aneven smaller 44Ca2+ peak relative to those obtained without energy off-set (Fig. S-2). Overall, the 44Ca2+ contribution to 88Sr+ is below 0.3% ofmeasured 88Sr+ intensities at analytical conditions used in this work.
2.4. NanoSIMS
The Caltech Cameca NanoSIMS 50-L was also run in several sessionsover 12 months. Unlike most other Cameca SIMS machines, the duo-plasmatron on the NanoSIMS is vertically oriented and the exit hole ofthe source becomes obstructed with soot or other particulate matterduring operation and many analytical sessions had be to restartedafter failure of the 16O− primary beam. As with the 7f-Geo runs, thesesub-sessions were numbered after the decimal point as 2.1, 2.2, 2.3, etc.
Where data are collected under significantly different conditionsthan those described below, we note it in the Results and Discussionsections. The sample preparation procedure was the same as for theims 7F-GEO samples with the exception that samples were kept in theNanoSIMS vessel chamber for up to a day prior to analysis to evaporatevolatiles and reach the required vacuum condition (b10−9 Torr).
When using an 16O− primary beam on the NanoSIMS, intensity andspot size are largely controlled by two apertures. Two Faraday cups(FCp), located at the end of the intermediate section and (FCo) occa-sionally positioned in place of the sample are used to regularly assessincident primary beam current. Over all our analytical sessions FCpand FCo varied from 195 to 580 nA and 3 and 23 pA respectively. Mea-surements were performed with an 8 kV primary beam.
We determined the spot size of the primary O− beam on the samplesurface following methods described in Gagnon (2009); i.e., by imaginga sharp compositional boundary and defining the beam width to liebetween 16% and 84% of that contrast. Measured beamwidths generallyvaried between ~500 and 800 nm for primary beam currents varyingbetween 0.5 and 5pA (where most data generated as part of this studywas collected using a relatively intense primary beam with a widthnearer the upper end of this range). As an undamaged gold coat is re-quired tominimize sample charging, spacing between adjacent analytical
spots was generally limited by the width of ‘burn marks’ (regions wherethe gold coat was discolored), which were generally on the order of 2–3 μm across. For single spot analyses we pre-sputtered for 2 min at thesame beam conditions used for analysis in the session #1–3. In analyticalsession 4 we rastered the beam over a 2 μm × 2 μm area and pre-sputtered for 10 min with a primary beam approximately twice the in-tensity used for analysis.
Positive secondary ion intensities for all masses were collectedsimultaneously with Hamamatsu discrete dynode electron multipliersin pulse counting mode. Count rates of 42Ca+ typically varied between~700 and ~7000 counts per second (cps), depending on the intensityof the primary ion beam. Count rates of 24Mg+ (2–3700 cps), 88Sr+
(2–12,272 cps), and 138Ba+ (0.05–790 cps) varied more widely due toboth primary beam current variations and differences in compositionamong the different standards. For single spot analyses either 200 or500 total cycles were collected, but for the 2 μm × 2 μm raster of ses-sion #4 120 cycles were used. Integration times for each cycle were3 s in analytical session #4 and 1 s in all other sessions. No dead-timecorrection was applied to the data.
The NanoSIMS analyses generally had mass resolutions in the rangefrom 1950 to 2810 under the tuning conditions used for this study.Under these conditions, mass scans did not show any molecular inter-ferences at 24Mg+, 42Ca+, and 88Sr+ (Fig. S-1D–F). However, we inferthat 88Sr+ must have suffered from isobaric interferences from the44Ca2+ dimer, which can only be separated at MRP N 9000 (Weberet al., 2005). We corrected for these inferred interferences by assumingthat the Ca dimer formation rate is similar for all carbonate samplesand standards examined in this study, and that the 44Ca2+ contributioncan be treated as a constant background present in all standards andsamples that shifts allmeasured 88Sr/42Ca ratios higher by a similar abso-lute offset (see Appendix 2 for more details). We calculate the contribu-tion of 44Ca2+ to the 88Sr+ peak using:
44Caþ242Caþ
¼40Caþ240Caþ
=42Ca40Ca
�44Caþ240Caþ2
¼ 1:1 � 10−3
where 42Ca+ is the measured 42Ca+ intensity— a known value for everycycle of every measurement. Here 40Ca2+/40Ca+ (1.7 ⋅ 10−3) is the ratioof the count rates at masses 80 and 40, 42Ca/40Ca (6.5 ⋅ 10−3) is the nat-ural abundance ratio, and 44Ca2+/40Ca2+ (4.32 ⋅ 10−3) is the ratio of thedimmer formation determined by Weber et al. (2005). This correctionratio is identical, within error, to the y-intercept in our Sr/Ca versus88Sr+/42Ca+ calibration plots (see below), giving us confidence in this in-terference correction scheme. Hydrides are another potential source ofisobaric interferences, but count rates of all of the potential species(chiefly 137BaH+ and 23NaH+) are too small to affect our reported Me/Ca ratios.
2.5. Other techniques
We used a variety of standard in situ solid imaging and analyticaltechniques to provide context for our in situ SIMS analyses of the Me/Ca ratios. Samples were mounted into aluminum cylindrical holders of1.2 and 2.5 cm in diameter using Araldite epoxy. Each holder containedup to ten different reference calcites and aragonites. All samples werecarbon coated.
Backscattered (BE) and secondary electron (SE) images were ob-tained using a 1550 VP Field Emission SEM at 15 kV, 5–8 nA of current,and a 120 μmaperture. Chemical analyses ofmajor andminor elementswere done on anOxford INCAEnergy 300X-ray EnergyDispersive Spec-trometer (EDS) system. More precise chemical analyses of major andminor elements were done by a JEOL JXA-8200 electron microprobe(EMP) at 15 kV and 10 nA current with a spot size of 2–3 μm. Phaseidentificationwas performed on a RenishawM1000micro-Raman spec-trometer and HKL Electron Back Scatter Diffraction (EBSD) system (seeMa and Rossman, 2008 for description of analytical equipment).
97R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
3. Results
Separate spot analyses from different parts of each standard crystal,or on the different grains of NBS-19, form the basis of our data set. In all,123 spots were analyzed by NanoSIMS and 146 spots were analyzedusing the 7f-Geo. These data are tabulated in Tables S-1–S-7. Our goalis to understand heterogeneity in the various standards, but we firstneed to examine how variable the reported intensity ratios are for a sin-gle spot. For the 7f-Geo we monitored the effect of acquisition time on
0.94
0.96
0.98
1
1.02
1.04
1.06
1.08
1.1
1.12
0 30 60 90 120 150 180
cycle no.
(24/
42)/
mea
n
OKA
BlueCC
NBS-19
0.85
0.9
0.95
1
1.05
1.1
1.15
0 30 60 90 120 150 180
cycle no.
(88/
42)/
mea
n
OKA
BlueCC
NBS-19
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 30 60 90 120 150 180
(138
/42)
/mea
n
OKA
BlueCC
NBS-19
(24/
42)/
mea
n
cycle no.
(88/
42)/
mea
n(1
38/4
2)/m
ean(
Blu
eCC
&13
5CC
)
A
B
C
ims 7F-GEO
Fig. 1. Examples of ims 7F-GEO (A–C) and NanoSIMS-50L (D–F) intensity ratios normalized to 4
to their means and plotted versus cycle number for single analytical spots in different standard
the behavior of intensity ratios by performing a few relatively long ac-quisitions. Fig. 1A–C show results for 60 and 180 cycles of collection,versus our normal 20 cycle method, for 88Sr/42Ca, 24Mg/42Ca, and138Ba/42Ca ratios. Here data fromOKA, Blue-CC, and NBS-19 are normal-ized to their mean values and plotted against the number of cycles col-lected (where all cycles are of a constant duration, 17 s for OKA and 30 sfor Blue-CC and NBS-19). Even after pre-sputtering for 2 min, the inten-sity ratios drift during a single spot analyses. We suggest these patternsarise from the sputtering process, rather than from true compositional
Nano-SIMS 50L
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 100 200 300 400 500
cycle no.
OKA
BlueCC
135CC
0.4
0.6
0.8
1
1.2
1.4
1.6
0 100 200 300 400 500
cycle no.
cycle no.
OKA
BlueCC
135CC
0
0.5
1
1.5
2
2.5
3
3.5
4
0 100 200 300 400 5000.8
0.9
1
1.1
1.2
1.3
(138
/42)
/mea
n (O
KA
)
BlueCC
135CC
OKAF
E
D
2Ca. Values of 24Mg/42Ca (A, D), 88Sr/42Ca (B, E), and 138Ba/42Ca (C, F) ratios are normalizeds.
0
1 2 3 4 5 6 7
-3
-2
-1
0
1 2 3 4 5 6
A
B
log(
2σ/m
ean)
log(
2σ/m
ean)
-0.5
-1
-1.5
-2
-2.5
-3
-3.5
-0.5
-1.5
-2.5
24Mg/42 Ca88Sr/42 Ca138Ba/42 Cacount. stat.
log[NX·N42/(NX+N42)
log[NX·N42/(NX+N42)
Fig. 2. Internal relative standard errors (2σ) as a function of total counts. Markers corre-spond to the whole data set for a single spot once the early drift has been removed.The line of slope −1/2 on these log–log plots is predicted from counting statistics (seetext and Appendix 3). A) Summary of all individual spots collected on the ims 7F-GEOas part of this work. B) Summary of all individual spots collected on the NanoSIMS-50Las part of this work.
98 R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
variation with depth, because spots at different locations and in differ-ent carbonate materials exhibit similar drifts. After the initial periodof rapid drift, ratios approach a steady-state with a relatively stableion-impregnation and sputtering condition that is the ‘sweet spot’for elemental ratio analysis. The longest run of 180 cycles in NBS-19also shows relatively high drift in Mg/Ca after about cycle 130, whichmight reflect deep craters and/or other damage of the sample surfacefrom this long sputtering time.
NanoSIMS primary beam intensities at the sample surface werethree orders of magnitude smaller than those of the 7f-Geo so longeracquisition times were required. In one case we extended the numberof cycles to 500 as a test of intensity ratio stability (Fig. 1D–F). 24Mg/42Ca ratios on the NanoSIMS drift with increasing cycles collected ina manner very similar to the 7f-Geo data. In contrast, the ratios of88Sr/42Ca and 138Ba/42Ca from the NanoSIMS analyses do not showsteep increases at the start of each analysis but rather drift linearly up-ward (Fig. 1E,F). However, similar to the SIMS data, the large scatter ofthe less intense 88Sr/42Ca and 138Ba/42Ca beams in Blue-CC and 135CCmay mask the drift of the ratios with time. Extending the number ofcycles to 500 did notmarkedly change the intensity ratio drift. However,elimination of the first 150 cycles from all of our normal 200 cycle sin-gle spot analyses improved the internal error by up to the factor of two.The first 40 cycles were removed from the data in session #4 for a sim-ilar reason.
We are interested in understanding how the observed variability ofan intensity ratio compares to the uncertainty in the mean of that ratioas predicted by Poisson statistics.Many SIMS studies examine this prop-erty by plotting intensities and primary ion beam fluence. Here we takean approach developed to examine data from a MC-ICP-MS (John andAdkins, 2010). Fig. 2 shows the standard error of themean for a popula-tion of intensity ratios collected for single spots versus the total numberof counts collected over the same time. The x-axis is actually the log of
‘n-effective’, logNX � N42
NX þ N42
� �where N42 is the total counts of 42Ca, to ac-
count for the fact that both themetal of interest (NX) and the 42Ca beamcan limit the Poisson statistics in some cases. For both the 7f-Geo(Fig. 2A) and the NanoSIMS (Fig. 2B), our single spot data demonstrateconsistencywith counting statistics by scattering around the theoreticalline of slope −1/2 (Appendix 3).
To better understand how these statistics evolve during the analysisof a single spot, a continuous error plot of the three intensity ratios fromthe 180 cycle NBS-19 run on the 7f-Geo is shown in Fig. 3A. Points cor-respond to successive additions of 10 cycle blocks, starting with cycles11–20, to the cumulative data set for a given intensity ratio, such thatthe last point corresponds to the data for cycles 11–180. The effect of amean shift in the 24Mg/42Ca ratio towards the end of the spot analysiscan be seen in the increase in standard error at the end of the seriesof diamonds in Fig. 3A. In general, spot analyses parallel the countingstatistics line after a few tens of cycles, improving our confidence thatchopping the first 10 cycles of data is enough to eliminate all but thestatistical sources of internal error. As most of the single spots were an-alyzed for 20 cycles only, each spot analysis in the final data representsthe average of cycles 11 through 20. Fig. 3B–D shows a similar analysisfor the NanoSIMS's behavior over the 500 cycle run. Again, data quicklybegin following counting statistics and intensity ratios after a fewcycles for elements of widely different masses. Overall the 7f-GeoandNanoSIMS data followPoisson statistics tomuch better than 1%pre-cision (2σ) for Me/Ca ratios with long acquisition times and enoughtotal counts.
4. Discussion
4.1. Heterogeneity in solid standards
Any analytical technique that directlymeasures solid samples is sub-ject to problems of standard inhomogeneity. Our approach to tackling
this problem has been to nest measurements of very high spatial reso-lution from the NanoSIMS in a grid of lower resolution measurementsfrom the SIMS.When these data are combinedwith backscattered elec-tron (BE) images of the analyzed domains (generated with the SEM),we can evaluate which of the many standards used in the commu-nity is truly homogeneous, or at least develop strategies for avoidingheterogeneity.
While the ICP-MS data are our reference frame for the accuracy ofthe SIMS analyses, they also reveal something about the variability ofMe/Ca ratios in the standards themselves. In Fig. 4 standards are orderedfrom highest to lowest Sr/Ca and Mg/Ca in each column. Data from thisstudy (circles) are compared to literature data (dashed lines; Fig. 4).There is a N10% range in Sr/Ca for all standards except UCI, RPI, and HUJcalcites. Similarly, there is a N7% range for theMg/Ca ratio in all standardsexcept UCI calcite. We interpret this variability as evidence for chemicalheterogeneity of standards, and suggest it provides a likely explanationfor the difference between our data and the literature values.
The ICP-MS data in Fig. 4 demonstrate that there are two types ofvariabilitywe need to assess: Ratios of Sr/Ca andMg/Ca aremore homo-geneous within single crystals of Blue-CC than they are in OKA. How-ever, inter-crystal variation of Blue-CC, especially in Mg/Ca, is largewhen compared to OKA. These are two of the more commonly usedSIMS CaCO3 standards; both are clearly heterogeneous yet they havedifferent styles of heterogeneity — perhaps requiring two differentstrategies for effective use as calibration standards. All standards exam-ined here with two or more crystals show both Sr/Ca andMg/Ca ranges
0
1 2 3 4 5
1 2 3 4 5 6
3 4 5
-3
-1
2 3 4 5 6 7
C
D
B
A
-0.5
-1.5
-2.5
2.5 3.5 4.5 5.5 6.5
-2
log(
2σ/m
ean)
log(
2σ/m
ean)
log(
2σ/m
ean)
log(
2σ/m
ean)
log[NX·N42/(NX+N42)
log[N24·N42/(N24+N42)
log[N88·N42/(N88+N42)
log[N138·N42/(N138+N42)
24Mg/42 Ca88Sr/42 Ca138Ba/42 Cacount. stat.
5.54.53.52.51.5-2.5
-2
-1.5
-1
-0.5
0
-2.5
-2
-1.5
-1
-0.5
-1.1
-1.3
-1.5
-1.7
-1.9
-2.1
-2.3
-2.55.54.53.5
1.5 2.5 3.5 4.5 5.5
Fig. 3. Internal relative standard errors (2σ) as a function of total counts. The line of slope−1/2 on these log–log plots is predicted from counting statistics. A) Ims 7F-GEO data col-lected continuously on a single spot of NBS-19 (total 180 cycles) where each point repre-sents adding 10 more cycles to the data set and recalculating the total counts and internalerror. B–D) Same as for A for the NanoSIMS-50L data collected continuously on three spotsfrom separate standards but now separating out the 24Mg/42Ca, 88Sr/42Ca and 138Ba/42Caratios into different plots.
99R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
outside the analytical errors of our ICP-MS method. There is clearspatial heterogeneity in Me/Ca ratios in most SIMS carbonate stan-dards (Table S-8).
Backscattered electron SEM images together with EDS analyses ofOKA shed some light on at least part of the reason for this standard'slargewithin crystal variability. Fig. 5 shows thatOKA contains inclusionsof (Sr,Ca)CO3, BaSO4, and (Ba,Sr)SO4. In this image, calcite has the low-est BE intensities while the sulfate inclusions are the brightest. Theinclusionswere analyzedwith EMP,Micro Raman, and EBSD techniques(Table S-9), confirming that sulfate inclusions have a structure similar tobarite or celestine, and that carbonate inclusions have an EBSD pattern
similar to aragonite or strontianite. For all other standards measuredin this paper, the backscattered electron images are relatively homoge-neouswithin an area of ~1 mm2. OKA is the only standardwith discretemineral phases other than CaCO3.
We used the 7f-Geo to map the spatial variability in several OKAcrystals. Here we compare the 88Sr/42Ca and 24Mg/42Ca data from oneanalytical session to a backscattered BE map from the SEM (Fig. 6).This subset of all our spatially located data was chosen because of allthe crystals analyzed it has the largest number of SIMSMe/Ca ratiomea-surements. All other BSE images of the OKA standard, together withcross plots of the SIMS data, are shown in Figs. S-3–S-5, and Tables S-1and S-5, but they contain essentially the same patterns described here.Three types of calcite are evident from the different intensities ofthe grayscale image of BE intensity. The greatest area of homogeneous‘matrix’ (light gray) contrasts with dark islands or lines (Fig. 6A).The small inclusions of strontianite and barite shown in Fig. 5 are thebrightest spots entirely surrounded by dark regions in Fig. 6A.
Isotopic intensity ratios from the SIMS data show that the matrixis generally enriched in Sr and Ba, but depleted in Mg, relative to thedarker SEM regions (Fig. 6B and C). In these scatter plots, the magentarectangles correspond to heterogeneous areas including holes, inclu-sions, dark islands, dark lines, and areas of variable SEM intensities.SIMS spot sizes for this analytical session are shown as numberedovals on the BSE image in Fig. 6A. Clusters of the data in Fig. 6B and Cdocument that the matrix of OKA is an excellent candidate for use as aSIMS standard. 88Sr/42Ca, 24Mg/42Ca, and 138Ba/42Ca data in this regionhave standard deviations (1 s.d.) of 0.85%, 0.81%, and 1.59% respectively;a variability that is similar to the internal errors of the individual points.The darker, lower intensity regions, on the other hand, are highly vari-able in their Me/Ca ratios (Fig. 6). There is no easily identifiable patternto these data, but they are clearly more complex than the compositionalvariations expected for mixing between two end members (e.g., matrixmaterial mixed with inclusions of fixed composition). Overall, therange of variability from the ims 7f-Geo data set is larger than that forthe ICP-MS data. This is expected due to the larger amounts of materialthat must be micro-milled for isotope dilution experiments comparedto the spot sizes of the SIMS instrument.
We checked for even higher resolution Me/Ca ratio variation in OKAby measuring ten NanoSIMS spots on OKA-II; a different crystal thanshown in Fig. 6. The data in Fig. 7 span a 90 μm long chain with 10 μmseparating each spot. Similar to the wide area spatial study with the7f-Geo, most of the data form a cluster having a mean that is high in88Sr/42Ca and 138Ba/42Ca, and low in 24Mg/42Ca relative to the three out-lying data-points. Though we lack an SEM image for the analyzed area,the data suggest that this NanoSIMS profile crossed a heterogeneous,BSE-dark calcite zone. We speculate that spot #3 (with the lowest88Sr/42Ca and 138Ba/42Ca and the highest 24Mg/42Ca) was in the centerof a BE-dark zone and that spots 2 and 4 were at the sides (Table S-5).
It is not possible to directly compare data generated with the 7f-Geo and the NanoSIMS for any one type of carbonate in the OKA sample(i.e., BSE dark vs. light) using conventional standardization schemesbecause it is obvious from Fig. 4 that this material is compromisedby inter and intra-grain heterogeneities. However, if we assume thateach fabric (e.g., the ‘matrix’ in BSE images) is uniform within and be-tween grains, we can normalize data generated on these two machinesusing the following equation:
88Sr42Ca
!Nano‐SIMS
n adjustedð Þ¼
88Sr=42Ca� �7 f‐ims
average−matrix
88Sr=42Ca� �Nano‐SIMS
average‐matrix
�88Sr42Ca
!Nano‐SIMS
n
: ð1Þ
The ratios of 24Mg/42Ca and 138Ba/42Ca were adjusted in the same way.The scale of variability seen in the 7f-Geo data set is about 2× that seenin the NanoSIMS data despite a ~100× increase in the area sampled.This implies that textures seen in the SEM images seem to be a good
100 R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
guide for the spatial variability in OKA, and that the ‘matrix’ of this stan-dard is homogeneous at essentially the level we can measure it.
Three other potential SIMS standards all have smaller overall Me/Caratio ranges compared to the bulk OKA data, but none of them are ashomogeneous as the matrix portion of OKA (Fig. 8 and Table S-2).Thus, paradoxically, the best standard available to the community maybe one of the fabrics in this complexly zoned material. In Blue-CC, thestandard deviation (1 s) of 88Sr/42Ca, 138Mg/42Ca 24/42, and 138Ba/42Caratios are 2.37%, 1.60%, and 4.91% respectively across eight data-points.We find that 88Sr/42Ca ratiosmeasured by SIMS varymorewide-ly than Sr/Ca ratios determined for single crystals by ICP-MS, suggestingthat heterogeneity in Sr/Ca increases at smaller scales in Blue-CC. NBS-19 has slightly larger heterogeneity in 88Sr/42Ca and 24Mg/42Ca, with
14
16
18
20
crys
tal 3
crys
tal 4
crys
tal 1
n=3
crys
tal 1
n=3
Sr/
Ca
(mm
ol/m
ol)
crys
tal 2
crys
tal 3
crys
tal 4
crys
tal 1
n=3
crys
tal 2
n=
2
crys
tal 1
n=3
crys
tal 2
0.44
0.48
0.52
0.56
0.60
0.48
0.52
0.56
0.60
0.64
7.0
7.5
8.0
8.5
9.0
Sr/
Ca
(mm
ol/m
ol)
Sr/
Ca
(mm
ol/m
ol)
Sr/
Ca
(mm
ol/m
ol)
literature value
literature value
literature value
LAS-20
OKA
BlueCC
HUJ
Fig. 4. ICP-MS data from seven standards and across single and multiple crystals. Dashed linesGaetani and Cohen, 2006; Blue-CC: Anne Cohen (personal communication); 135CC: Kent RoOberdorf, Austria; LAS-20 and NBS-19: Sano et al. (2005); RPI: Cherniak (1997).
standard deviations b5% in both ratios, but 138Ba/42Ca in NBS-19 is upto a factor of three more variable than in Blue-CC. While these rangesin Me/Ca ratio may seem acceptable, we have demonstrated that inter-nal errors for single spot analyses are more precise than this spatial var-iability. NanoSIMS data from Blue-CC shows similar results as thosefrom the SIMS (Fig. 9, Table S-5). While Blue-CC Sr/Ca and Mg/Ca ratiosspan a larger range than the matrix portion of OKA, this standard isoverallmore homogeneous thanOKA. It is clear that the ‘matrix’ portionof OKA is the best candidate for a communitywide SIMS standard, but itis important that each lab document where standardization spots comefrom relative to the other inhomogenieties in OKA. In addition we sug-gest that standardization based on analyses of all of these materials isonly accurate if one has independently determined Me/Ca ratios of the
crys
tal 3
crys
tal 4
crys
tal 1
n=3
crys
tal 2
crys
tal 3
crys
tal 4
crys
tal 1
n=3
crys
tal 2
n=
2cr
ysta
l 1n=
2
crys
tal 1
n=3
crys
tal 2
3
4
5
6
25
30
35
40
45
50
Mg/
Ca
(mm
ol/m
ol)
Mg/
Ca
(mm
ol/m
ol)
Mg/
Ca
(mm
ol/m
ol)
Mg/
Ca
(mm
ol/m
ol)
2.5
3.0
3.5
4.0
literature value
4
5
6
literature value
OKA
BlueCC
LAS-20
UCI
correspond to data obtained from the literature or from personal communications. OKA:ss (personal communication) specimen from Harvard Mineralogical museum (#97189)
crys
tal 3
crys
tal 1
n=2
crys
tal 2
crys
tal 1
n=3
crys
tal 2
crys
tal 3
crys
tal 1
n=2
crys
tal 1
n=3
crys
tal 2
crys
tal 1
n=3
crys
tal 2
crys
tal 1
n=3
0.12
0.15
0.18
0.21
0.24
0.27
Sr/
Ca
(mm
ol/m
ol)
Sr/
Ca
(mm
ol/m
ol)
Sr/
Ca
(mm
ol/m
ol)
0.16
0.18
0.20
0.22
0.25
0.30
0.35
0.40
0.45
0.50
Mg/
Ca
(mm
ol/m
ol)
Mg/
Ca
(mm
ol/m
ol)
Mg/
Ca
(mm
ol/m
ol)
0.020
0.024
0.028
0.032
0.036
literature value
literature value
UCI
RPI
135CC
0.05
0.06
0.07
0.08
0.09
0.10
literature value
135CC
0.007
0.014
0.021
0.028
0.035
HUJ
RPI
Fig. 4 (continued).
50µm Electron Image 1
Fig. 5. An example of sulfate and carbonate inclusions in a backscattered electron SEMimage. The SIMS spot no. 21 (session 4.2) is in the upper right corner.
101R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
particular crystal used as a standard. There can be wide variability inabsolute Me/Ca ratios between different crystals of the same nominalstandard.
4.2. Comparing ratios across separate analytical sessions
While is it possible to characterize the variance of a single standardby examining data from within a single analytical session, we alsowish to calibrate a standard's Me/Ca ratio accurately over a long time,spanning many analytical sessions. For our data set we can use the ma-trix portion of OKA to standardize the intensity ratios of Blue-CC crystalswithin a single analytical session, and then compare these values acrossmultiple sessions. We choose these two standards for our analysisbecause they are homogeneous at nearly the level of our internal preci-sion. Fig. 10 and Table S-3 show 7f-Geo Sr/Ca, Mg/Ca, and Ba/Ca valuesfor a single crystal of Blue-CC standardized to the matrix portionof OKA across several analytical sessions, some with multiple sub-sessions. The horizontal lines are our ICP-MS determined values for Sr/Ca and Mg/Ca in this Blue-CC crystal. In general, when collected andstandardized carefully, Me/Ca ratios from the 7f-Geo are precise and re-producible, following the level of their internal statistics to better than1% (2 s), across many months of running the machine. We find scatterduring individual sub-sessions that is larger than our internal error,but we attribute this to the slight heterogeneity of Blue-CC itself. The7f-Geo follows Poisson statistics for each spot, and these spots can be
22
5
9
10
111215
16
17
1819
21
23
0.27
22
5
9
10
1112
15
16
17
18
19
21
23
2
B
C
88S
r/42
Ca
A
2.4
2.3
2.2
2.1
1.9
1.8
288S
r/42
Ca
2.4
2.3
2.2
2.1
1.9
1.8
0.05 0.07 0.09 0.11 0.13 0.15 0.17
138Ba/42Ca
24Mg/42Ca
0.250.230.210.19
Fig. 6. Large spatial analyses of OKA using the SEM and ims 7F-GEO SIMS. (A) SEM BEimage of the lower right part in the OKA-II. Numbered spots are from ims 7F-GEO session4.5. (B, C) Plots of 88Sr/42Ca versus 138Ba/42Ca and 88Sr/42Ca versus 24Mg/42Ca fromOKA-II(ims 7F-GEO, session 4.5). The ovals on SEM image correspond to the location and approx-imate size of the beam burn marks. The blue diamonds are the data collected from thehomogeneous calcite matrix. The magenta rectangles correspond to heterogeneousareas including holes, inclusions, dark islands, dark lines, and areas of variable intensities(e.g.: spot no. 12). The error bars are 2σ. (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)
B
A
ims 7F-GEO, session 3.8 ims 7F-GEO, session 4.4ims 7F-GEO, session 4.5ims 7F-GEO, session 4.6NanoSIMS, session 2.1-adjusted
X
2.5
2.0
1.5
1.00.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
88S
r/42
Ca
88S
r/42
Ca
138Ba/42Ca
24Mg/42Ca
0.280.260.240.220.200.181.0
1.5
2.0
2.5
Fig. 7. Elemental variability in OKA-II using the NanoSIMS-50L. A) 88Sr/42Ca versus 138Ba/42Ca and B) 88Sr/42Ca versus 24Mg/42Ca intensity ratios. Also shown are all the ims 7F-GEOdata from sessions 3.8, 4.4, 4.5, and 4.6. NanoSIMS-50L data froma 90 μmlong line of spots(session 2.1) were normalized to ims 7F-GEO OKA-II matrix data from session 3.8 usingEq. (1) in the text. 2σ error bars of ims 7F-GEO data appear in Fig. 5.
102 R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
normalized to each other over long time periods by running the matrixportion of OKA during all analytical sub-sessions where the duo-plasmatron is on continuously.
Using this same approach to normalize intensity ratios across runswith the NanoSIMS gives Blue-CC ratios with much larger scatter.Fig. 11 shows the run-to-run variability for the NanoSIMS data butnow with two Blue-CC crystals shown separately because of theirlarge difference in absolute Mg/Ca ratio (Table S-6). We also show theeffect of normalizing to either the matrix region in OKA or to the gran-ular NBS-19 standard. The results are much worse than for the 7f-Geo.
In many cases the variability within an analytical session is larger thanthe differences between sessions, indicating that our normalizationworks to the degree that a Blue-CC crystal is actually homogeneous.However there is evidence that Mg/Ca and Sr/Ca data are normalizeddifferently in different analytical sessions. For Blue-CC-II in session 2the Mg/Ca and Sr/Ca data are similarly offset from the accurate valuein all sub-sessions. This is not true for Blue-CC-IV where some sub-sessions have Sr/Ca data that fall within the calibrated value and Mg/Ca data that does not. Then other sub-sessions have Mg/Ca data thatseems accurate while the Sr/Ca is not.
There is an important difference in the ion optics of the NanoSIMS-50L and the 7f-Geo that could account for these results. The distance be-tween the extraction lens and the sample surface in the NanoSIMS-50Lis shorter than in the 7f-Geo by a factor of ~13 (300–400 μm against a4.7 mm). Small differences in the incidence angle and energy of the pri-mary beam affect the emission cone of the secondary ions, which canlead to substantial element-to-element fractionations in the secondarybeam, such that moving the sample holder from sample to standardcan result in a different mass bias for the same elemental ratio in theNanoSIMS where the extraction optics are very close to the sample sur-face. In this case, the relative precision of an intensity ratio within a sin-gle sample is governed by the internal error, but the overall accuracycan be as bad as ~7.5% for Sr/Ca and ~11.0% for Mg/Ca. We did not ob-serve systematic variations in instrumental mass fractionation with
A
FE
DC
B0.080
0.077
0.074
0.071
0.068
0.080
0.077
0.074
0.071
0.0684.5E-04
88S
r/42
Ca
88S
r/42
Ca
88S
r/42
Ca
138Ba/42Ca
5.3E-04 6.1E-04 6.9E-04
Blue calcite Blue calcite
0.140 0.145
24Mg/42Ca
138Ba/42Ca 24Mg/42Ca
138Ba/42Ca 24Mg/42Ca
0.150 0.155
NBS-19
135CC
NBS-19
135CC
0.027
0.026
0.025
0.024
0.0237.0E-05
0.005
0.015
0.025
0.035
1.6E-041.3E-041.0E-04
0.E+00 2.E-05 4.E-05 6.E-05
0.035
0.025
0.015
0.0053.E-03 4.E-03 5.E-03 6.E-03 7.E-03
0.02250.73
0.0235
0.0245
0.0255
0.0265
0.75 0.77 0.79
Fig. 8. Plots of 88Sr/42Ca versus 138Ba/42Ca and 88Sr/42Ca versus 24Mg/42Ca from Blue-CC-II (A, B), NBS-19 (C, D), and 135CC (E, F). Data were obtained during ims 7F-GEO session 3.8.
103R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
respect to changes in either the primary or secondary lens settings andthus favor the sample holder geometry relative to the beams as themostlikely explanation for worse run-to-run precision on the NanoSIMScompared to the 7f-Geo.
Comparing the accuracy ofmany standards, not just themost homo-geneous ones, across several analytical sessions also shows differencesbetween the 7f-Geo and the NanoSIMS. Fig. 12 shows 7f-Geo intensityratios plotted against elemental ratios from our ICP-MS measurementsfor the same standard. Regardless of the day they were run, linear fitsover nearly 2 orders of magnitude in Me/Ca ratio show very similarslopes (Table S-4). For the 7f-Geo our data shows it is possible to usesingle carbonate standards of very different Sr/Ca and Mg/Ca ratios tonormalize unknown samples with only a slight increase in uncertaintyover the propagated internal measurement statistics. Unfortunatelythis same plot is not as consistent for the NanoSIMS (Fig. 13, Table S-7).Within a single sub-session, where all the standards were in the samesample holder, the linear fits across three or more standards are alsoquite good. Across at least one order of magnitude of concentration,intensity matching of carbonate samples and standards does not seemto be necessary on the NanoSIMS. However, between analytical sessionslarge changes occur in the calibration slope. We have noticed that thesechanges are associated with the different extents of pre-sputtering be-tween sub-sessions, the probe current, and the spot size. The probe cur-rent at the sample holder was not recorded routinely, but a few valuesshow that an increase from 3.5 pA (session 4.2) to 23 pA (session 4.1)
was associated with an increase in the Sr/Ca slope from 0.06752 to0.11935 and a decrease in the Mg/Ca slope from 0.05279 to 0.02609.This is probably another manifestation of the NanoSIMS ion opticswhere slight differences in the positioning of the primary beam andthe extraction lens can lead to large elemental fractionations.
5. Conclusions
We find that making accurate and precise Me/Ca ratio measure-ments in carbonatematerials on both SIMSmachines requires some im-portant attention to the details. Due to inter and intra crystal variability,each crystal used as a standard must be separately and independentlycharacterized for its bulk Me/Ca ratio to at least confirm literaturevalues. This is especially true for Blue-CC where separate crystals canhave large differences in their elemental concentrations. In general,intensity matching of standards to sample Me/Ca ratios is not as im-portant as finding homogeneous materials for calibration at reportedMe/Ca range.
Both SIMS machines demonstrate internal variability that are domi-nated by counting statistics to better than 1% precision. However, onlycertain standards are homogeneous enough within individual crystalsto be useful precision and accuracymonitors. Despite the heterogeneityof bulk OKA, its matrix calcite is themost homogeneous material exam-ined in this study (and the best widely available standard). Ims 7F-GEOand NanoSIMS-50L analyses show that external 2σ variations are b0.5%
A
B
- ims 7F-GEO
- NanoSIMS-50L
0.080
0.078
0.076
0.074
0.072
0.070
0.0684.0E-04
88S
r/42
Ca
0.080
0.078
0.076
0.074
0.072
0.070
0.068
88S
r/42
Ca
138Ba/42Ca
24Mg/42Ca
5.0E-04 6.0E-04 7.0E-04
0.140 0.144 0.148 0.152
Fig. 9. Same as Fig. 7 but for Blue-CC-II data on the two machines. NanoSIMS-50L data(session 2.1) were normalized to ims 7F-GEO Blue-CC-II data (session 3.8) using Eq. (1).Error bars are 2σ.
Mg/
Ca
(mm
ol/m
ol)
Sr/
Ca
(mm
ol/m
ol)
Ba/
Ca
(mm
ol/m
ol)
B
C
A
1
1s.d.=1.90%
1s.d.=1.17%
1s.d.=3.16%
7.7E-03
0.50
0.52
0.54
0.56
0.58
0.60
0.62
0.64
session2 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 4.1 4.2
1
session2 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 4.1 4.2
1
session2 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 4.1 4.2
1.0E-02
9.7E-03
9.2E-03
8.7E-03
8.2E-03
9.0
8.8
8.6
8.4
8.2
8.0
7.8
7.6
7.4
7.2
7.0
Fig. 10. Ims7F-GEOdata fromsingle crystal of Blue-CC-III obtainedovermultiple sub-sessions(along the x-axis). Different symbols correspond to the four different analytical sessionsconducted in December 2007 (diamonds), February 2008 (triangles), April–May 2008(circles), September 2008 (asterisks). Elemental ratios were calculated from intensity ratiosusing data from the matrix areas of the single crystal OKA-III (the same holder as Blue-III)run during the same sub-session (see Eq. (4.1) in Appendix 4). Columns of data are from dif-ferent sub-sessions and the horizontal lines were calculated from Blue-CC-I using Eq. (4.2) inAppendix4. The absence of a horizontal line inC corresponds to the lackof Ba/Ca ICP-MSdata.
104 R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
for Sr/Ca and Mg/Ca, and b1% for Ba/Ca. In our study Blue-CC and AG-1are the second most homogeneous materials in Sr/Ca (2σ b 2%), andBlue-CC is the second most homogeneous material in both Mg/Ca(2σ b 2%) and Ba/Ca (2σ b 4%). UCI calcite is homogeneous within asingle crystal.
There are discrepancies between our bulk analyses and those report-ed in literature. Therefore, adjacent pieces (within a few mm distance)of each studied SIMS or NanoSIMS standard must be analyzed indepen-dently with conventional methods. Independent measurements are re-quired for most of the standards even if they are from the same batch.The possible exception is the OKA standard where Me/Ca inter-crystalvariability could be caused by a different distribution of inclusions ofbarite, strontianite, and heterogeneous (dark) calcite. Our ID-ICP-MSMg/Ca data for NBS-19 are different from those published by Sanoet al. (2005) by 17%, suggesting that Mg/Ca varies between differentbatches of NBS-19.
Acknowledgments
Wewould like to thank Daniele Cherniak, Anne Cohen, Ellen Druffel,Jonathan Erez, Glenn Gaetani, Kent Ross, and Neil Sturchio for sharingthe carbonate standards and bulk analytical data. John Ferry providedthe carbonate material used in testing the primary beam size on theNanoSIMS. We also thank Lindsey Hedges for advising on sample prep-aration. The comments of anonymous reviewers were helpful as well.
This work was supported by the following grants from NOAA — MITSeagrant Program: NA100AR4170086 “Autonomous vehicle explorationand sampling of deep-water corals” and NSF: OCE-0929272 “Collecting
7.5
7.7
7.9
8.1
8.3
8.5
8.7
8.9
0.50
0.55
0.60
0.65
0.70
0.75
2.8
3.0
3.2
3.4
3.6
3.8
4.0
0
0.01
0.02
0.03
0.04
0 .05
0
0.01
0.02
0.03
0.04
0.05
0.50
0.55
0.60
0.65
0.70
0.75
Blue-CC-II Blue-CC-IV
no Ba datain 1.1 and 1.2
Ba/
Ca
(mm
ol/m
ol)
Mg/
Ca
(mm
ol/m
ol)
Sr/
Ca
(mm
ol/m
ol)
analytical session
306 nA
581 nA
434 nA
A
C
D
E
F
B
1.1
305 nA
236 nA
438 nA
266 nA
1.2 2.1 2.2 2.3 2.4 3.1
analytical session
1.1 1.2 2.1 2.2 2.3 2.4 3.1
Fig. 11. NanoSIMS-50L data from Blue-CC obtained over multiple sub-sessions. Intensity ratios from separate crystals of Blue-CC were normalized to elemental ratios using either OKA(circle marks) or NBS-19 (x marks) run during the same sub-session. The horizontal lines were calculated from Blue-CC-I using Eq. (4.2) in Appendix 4: here blue and green lines arethe calculated values for Blue-CC-II and Blue-CC-IV. The absence of a horizontal line in C corresponds to the lack of Ba/Ca ICP-MS data in this study. (For interpretation of the referencesto color in this figure legend, the reader is referred to the web version of this article.)
105R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
deep-sea corals from off of Tasmania to measure the history of SouthernOcean radiocarbon concentrations”.
Appendix 1. Description of the standard materials
OKA calcite (OKA-C), Blue-CC calcite (#0875), and AG-1 (aragonite)were provided by Anne Cohen and Glenn Gaetani (WHOI). OKA-C isfrom the OKA carbonatite complex in Quebec, Canada. AG-1 is a smallsingle aragonite crystal from Spain purchased from Alfa Aesar; notethat trace and minor element compositions in other crystals fromthe same batch are unknown. NBS-19 (in literature also referred to asTS-Limestone) was obtained from a single slab of whitemarble, provid-ed by the US National Bureau of Standards (Friedman et al., 1982).The 135-CC specimen (Iceland Spar) from the Harvard MineralogicalMuseum (#97189) was collected in Oberdorf, Austria and was providedbyKent Ross at UC-Berkeley. LAS-20 is aU-rich calcite froma speleothemin the Vinschgau Valley of northern Italy (Spotl et al., 2002), provided byNeil Sturchio (UL-Chicago). RPI-CC is another Iceland Spar calcite fromMexico, provided byDaniele Cherniak (themineral dealerwasNortheast
Geological Supply). HUJ-AR is an aragonite provided by Jonathan Erezat the Hebrew University in Jerusalem and UCI-CC is a calcite providedby Ellen Druffel at UC-Irvine.
Appendix 2. Interferences in NanoSIMS
Unlike the ICP-MS, doubly-charged species are not typically detectedon the NanoSIMS reflecting the different energies and physics for ionformation between in these methods. However, molecular ions arepresent in the secondary ion beam, and can interfere with the speciesof interest. To this end the adjustment of the sample voltage was inves-tigated as a tool to reduce molecular isobaric interferences, but as thesample voltage offset becomes more negative, there is a disproportion-ate penalty in sensitivity (Fig. S-6) and under the analytical conditionsused for this set of experiments it is not a useful approach. But an volt-age offset of about−15 V to−25 V, corresponding to awide and stablemaximum in secondary beam current, is still necessary to ensure goodsensitivity and stability to accommodate steady-state charge build-upon the insulating CaCO3 samples. Since molecular species cannot be
0
0.05
0.1
0.15
0.2
0.25
0 1 2 3 4 5 6
holder-I, session 4.2, slope=0.04329
holder-I, session 4.3, slope=0.04341
holder-II, session 3.8, slope=0.04207
0
0.02
0.04
0.06
0.08
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
holder-I, session 4.2, slope=0.11947
holder-I, session 4.3, slope=0.11465
holder-II, session 3.8, slope=0.12341
0
1
2
3
0 10 20
OKA
HUJ
BlueCC
LAS-20
UCI
135CC
NBS19
RPI
A
OKA
BlueCC
UCI
RPI
HUJ135CC
B
0
1
2
3
0 20 40
LAS20
NBS19
ims
7F-G
EO
: 88S
r/42
Ca
(cps
/cps
)im
s 7F
-GE
O: 24
Mg/
42C
a (c
ps/c
ps)
ICP-MS: Sr/Ca (mmol/mol)
ICP-MS: Mg/Ca (mmol/mol)
Fig. 12. Calibrations of the different standards' ims 7F-GEO intensity ratios with their cor-responding ID-ICP-MS data. Fitted lines are weighted linear regressions. A) Sr/Ca data;B) Mg/Ca data. The insets show the full concentration range. Each marker correspondsto the particular calibration line of the same color. Note that different crystals of thesame standardmay have different chemical compositions. (For interpretation of the refer-ences to color in this figure legend, the reader is referred to theweb version of this article.)
00 1 2 3 4 5
holder-IV, session 1.1, slope=0.02609holder-II, session 2.2, slope=0.02775holder-IV, session 2.4, slope=0.02447holder-IV, session 3.1, slope=0.03327holder-I, session 4.1, slope=0.03619holder-I, session 4.2, slope=0.05279
00 1
holder-IV, session 1.1, slope=0.11935, FCo=23pAholder-II, session 2.2, slope=0.12497holder-IV, session 2.4, slope=0.13182holder-IV, session 3.1, slope=0.10577holder-I, session 4.1, slope=0.10077holder-I, session 4.2, slope=0.06752, FCo=3.5pA
B
0
1
2
0 10 20
0
1
2
0 10 20 30 40 50
OKA
HUJ
AG-1
A
LAS20
NBS19
Nan
oSIM
S-5
0L: 24
Mg/
42C
a (c
ps/c
ps)
Nan
oSIM
S-5
0L: 88
Sr/
42C
a (c
ps/c
ps)
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
ICP-MS: Sr/Ca (mmol/mol)
0.2 0.4 0.6 0.8 1.2 1.4
0.22
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
4.53.52.51.50.5
ICP-MS: Mg/Ca (mmol/mol)
Fig. 13. Calibrations of the different standards' NanoSIMS intensity ratios with their corre-sponding ID-ICP-MS data. Fitted lines are weighted linear regressions. A) Mg/Ca data;B) Sr/Ca data. The insets show the full concentration range. Each marker corresponds tothe particular calibration line of the same color. Note that different crystals of the samestandard may have different chemical compositions. Error bars for the Sr data are less orsimilar to the size of the symbols. (For interpretation of the references to color in this fig-ure legend, the reader is referred to the web version of this article.)
106 R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
eliminated, interferences from these species must either be resolvedusing the high mass resolving power of the NanoSIMS-50L or correctedoffline by monitoring molecular formation.
Appendix 2.1. Dimer interferences
Two approaches to correcting for dimer interferences are possible.
(1) FollowingWeber et al. (2005), the relatively clean and resolvablecalcium dimer peak at mass 80 can be used to monitor dimerintensity. Assuming natural abundances for the calcium isotopes,the dimer contribution can then be subtracted off-line frommass88 using the 44Ca intensity and the 80/40 ratio. For an accurate
correction, this approach assumes either a similar gain in the de-tectors measuring dimmer ratio and mass 88, or a known elec-tron multiplier gain ratio between detectors. We did notroutinely monitor detector gain during the analytical sessionspresented in this paper. However, in one session, a steady40Ca+ beam that was centered into each of the seven EMsresulted in a signal that varied from 105 cps to 1.5 · 105 cpsacross the detectors. Under these conditions explicit correctionsare not accurate, but can still be used to roughly estimate the im-pact of dimer interference on uncorrected Sr/Ca ratios. The dimerformation rate (40Ca2+/40Ca+ ratio) was monitored in two of ourNanoSIMS analytical sessions (4.1 and 4.2). In session 4.2 40Ca2+/40Ca+ was 1.7 ±0.2 · 10−3 for all the standards except OKA andLAS, which both exhibit a lower dimer yield of 1.5 ± 0.2 · 10−3.There was no trend in dimmer formation rate with time. To es-tablish the importance of dimmers, we assume similar gains
107R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
between the mass 80 andmass 88 detectors; calcium dimers areexpected to account for only 1 to 2‰ of the measured 88 signalfor the high Sr/Ca standards OKA and HUJ, 3 to 8% for most ofthe remaining standards, and ~50% for the low Sr/Ca standardRPI.
(2) In this alternative approach 44Ca2 contribution to 88Sr can betreated as a constant background in all standards and samplesthat shifts all measured 88Sr/42Ca ratios higher by a similar abso-lute offset. This dimer background contribution is then opera-tionally accounted for by the intercept in a linear calibrationcurve. While dimer yield is similar for most materials, both OKAand LAS exhibit an ~15% lower 40Ca2+/40Ca+ ratio (see above).For OKA the dimer is already only a very minor correction, 15%of 1‰ is insignificant, and for LAS the error in the dimer correc-tion is b2‰. These effects are small sowe treat the dimmer inter-ference as a background correction during data analysis.
We can roughly estimate the intercept of the Sr/Ca ICP-MS vs.88Sr/42Ca NanoSIMS calibration curve, assuming no other interferencesand perfect ICP-MS accuracy. In session 4.2, 40Ca2+/40Ca+ is roughly1.7 · 10−3; and, 42Ca/40Ca is 6.5 · 10−3 consistently through the run.Assuming natural abundances, the contribution of calcium dimer atmass 88 is 4.32 · 10−3 times the 40Ca2 counts (Weber et al., 2005).Now assuming similar gains in all the detectors, the estimated y-intercept of the calibration curve should be: (1.7 · 10−3)/(6.5 ·10−3) · (4.32 · 10−3) = 1.1 · 10−3, which is within the range of themeasured y-intercept of 0.8 − 2.2 · 10−3. Only one Sr/Ca calibrationline has a slightly negative intercept of −7.6 · 10−5 (session 3.1).This analysis shows that dimers are present, that they need to beaccounted for, and that they can be corrected for during routinemulti-standard calibrations as long as samples and standards areall CaCO3 with only minor impurities and exhibit no dimer formationmatrix effect (i.e., differences among samples and/or standards inthe relative proportions of dimers).
Appendix 2.2. Hydride interferences
Strontium hydride formation was also estimated from 87SrH+/88Sr+
in a strontianite sample, with a similar formation rate of 0.5 to 1%.Typical MRPs for this study were 2000–3000. It should be noted thatthis definition ofMRP includes the flat top part of a peak, correspondingto the exit slitwidth, and therefore underestimates the ability to optical-ly resolve different masses. The peak edge MRP definition is a morerelevant minimum criterion for resolution as it represents the abilityto separately measure plateaus on the shoulder of composite peaks.The peak edge MRP, which somewhat overestimates the practical re-solving power due to mass stability limitations is typically ~10,000 inthis study (5%-to-95%). Fig. S-1D is a scanofmass 24 during routine con-ditions demonstrating the resolution of 24Mg+ from the expected loca-tion of 23NaH+. While 87SrH+ is a non resolvable interference on 88Sr+,this interference has little impact on 88Sr/42Ca ratios (≤1‰, assuming ahydride formation rate of ≤1%), as 88Sr abundance is much larger than87Sr. Furthermore, most natural materials exhibit similar 87/88 ratios(within 1%), thus any hydride contributionwill scale with Sr concentra-tion, affecting the slope of a calibration curve in a minor way but notimpacting the final accuracy.
Appendix 3. Counting statistics in SIMS
Assuming no covariance the relative standard error (1σ/mean) isdefined by the following formula (Bevington, 1969, p. 62):
σNX
N42
� �2=
NX
N42
� �2¼ σ NXð Þ
NX
� 2þ σ N42ð Þ
N42
� 2: ð3:1Þ
According to counting statistics:
σ NXð ÞNX
� 2¼
ffiffiffiffiffiffiffiNX
pNX
" #2¼ 1
NXand
σ N42ð ÞN42
� 2¼ 1
N42: ð3:2Þ
The logarithmof the combined expressions 1.1 and1.2 yields the fol-lowing equation for 2σ/mean:
log 2σNX
N42
� �=
NX
N42
� �� ¼ 1
2log 4
NX þ N42
NX � N42
� ð3:3Þ
or
log 2σNX
N42
� �=
NX
N42
� �� ¼ −0:30103−1
2log
NX � N42
NX þ N42
� �: ð3:4Þ
In Figs. 2 and 3 the logarithms of 2σ/mean are plotted against logNX � N42
NX þ N42
� �. The logarithms of 2σ/mean was calculated as:
log 2σNX
N42
� �=
NX
N42
� �� ¼ 2 � 1s:d: X=42ð Þ
X=42ð Þmean �ffiffiffiffiffiffiffiffiffiffiffiffiffin−10
p : ð3:5Þ
Here 1 s.d.(X/42) is the standard deviation of 24Mg/42Ca, 88Sr/42Ca, or138Ba/42Ca ratios between (n − 10) cycles.
Appendix 4. Calculation of Me/Ca ratios from SIMS data
Elemental ratios of Sr/Ca in Blue-CC-III presented in Figs. 10 and 11were calculated using OKA standard analyzed together with Blue-CCat each sub-session:
SrCa
� �BlueCC‐III
data‐pointð Þk¼
88Sr=42Ca� �BlueCC‐III
averageð Þk88Sr=42Ca� �OKA‐III
averageð Þk� Sr
Ca
� �OKA‐I
ICPMS
�88Sr=42Ca� �OKA‐III
averageð Þsession‐4:288Sr=42Ca� �OKA‐I
averageð Þsession‐4:2:
ð4:1Þ
The fraction on the left side of the Eq. (4.1) is the Sr/Ca in Blue-CC-III(shown as data-point in Figs. 10 and 11) at the particular sub-session k.The first fraction on the right side of the Eq. (4.1) is the SIMS ratio of88Sr/42Ca of Blue-CC-III to 88Sr/42Ca of OKA-III obtained at the samesub-session k. The second fraction is the Sr/Ca in OKA-I determinedwith ICP-MS. The third fraction is the SIMS ratio of 88Sr/42Ca of OKA-IIIto 88Sr/42Ca of OKA-I obtained at session 4 (when both OKA-III andOKA-I were analyzed with ims 7F-GEO). The ratio of Mg/Ca in Blue-CC-III were obtained in the similar way. For Ba/Ca ratio in OKA the liter-ature value of 1.61 mmol/mol was used (see Gaetani and Cohen, 2006).The true Sr/Ca ratios were determined by normalization of the SIMSdata from Blue-CC-III to those from Blue-CC-I:
SrCa
� �BlueCC‐III¼
88Sr=42Ca� �BlueCC‐III
averageð ÞSept:588Sr=42Ca� �BlueCC‐III
averageð Þsession‐4:2
� SrCa
� �BlueCC‐III
ICPMS: ð4:2Þ
The ratios of Mg/Ca were determined similarly to Sr/Ca. Please notethat the piece broken off Blue-CC-I for SIMS analyses and those dis-solved for ICP-MS analyses came from the same single crystal of Blue-CC and were located next to each other. The same is true for all other
108 R.I. Gabitov et al. / Chemical Geology 356 (2013) 94–108
standards from holder-I. Therefore we assumed that standards fromholder-I represent the true elemental ratios determined by ICP-MS.
Appendix 5. Supplementary data
Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.chemgeo.2013.07.019.
References
Allison, N., 1996. Comparative determinations of trace andminor elements in coral aragoniteby ion microprobe analysis, with preliminary results from Phuket, southern Thailand.Geochimica et Cosmochimica Acta 60, 3457–3470.
Allison, N., Austin,W.E.N., 2008. Serial Mg/Ca and Sr/Ca chronologies across single benthicforaminifera tests. Chemical Geology 253, 83–88.
Allison, N., Finch, A.A., Webster, J.M., Clague, D.A., 2007. Palaeoenvironmental recordsfrom fossil corals: the effects of submarine diagenesis on temperature and climateestimates. Geochimica et Cosmochimica Acta 71, 4693–4703.
Badro, J., Ryerson, F.J., Weber, P.K., Ricolleau, A., Fallon, S.J., Hutcheon, I.D., 2007. Chemicalimagingwith NanoSIMS: awindow into deep-Earth geochemistry. Earth and PlanetaryScience Letters 262, 543–551.
Beck, J., Edwards, R., Ito, E., Taylor, F., Recy, J., Rougerie, F., Joannot, P., Henin, C., 1992. Sea-surface temperature from coral skeletal strontium calcium ratios. Science 257 (5070),644–647.
Bevington, P.R., 1969. Data Reduction and Error Analyses for the Physical Sciences.McGraw-Hill Book Company.
Bice, K.L., Layne, G.D., Dahl, K., 2005. Application of secondary ion mass spectrometry tothe determination of Mg/Ca in rare, delicate, or altered planktonic foraminifera:examples from the Holocene, Paleogene, and Cretaceous. Geochemistry, Geophysics,Geosystems 6, Q12P07. http://dx.doi.org/10.1029/2005GC000974.
Bond, Z.A., Cohen, A.L., Smith, S.R., Jenkins,W.J., 2005. Growth and composition of high-Mgcalcite in the skeleton of a Bermudian gorgonian (Plexaurella dichotoma): potentialfor paleothermometry. Geochemistry, Geophysics, Geosystems 6, Q08010. http://dx.doi.org/10.1029/2005GC000911.
Cherniak, D.J., 1997. An experimental study of strontiumand lead diffusion in calcite, and im-plications for carbonate diagenesis and metamorphism. Geochimica et CosmochimicaActa 61, 4173–4179.
Cohen, A.L., Layne, G.D., Hart, S.R., 2001. Kinetic control of skeletal Sr/Ca in a symbioticcoral: implications for the paleotemperature proxy. Paleoceanography 16 (1), 20–26.
Cohen, A.L., Gaetani, G.A., Lundalv, T., Corliss, B.H., George, R.Y., 2006. Compositionalvariability in a cold-water scleractinian, Lophelia pertusa: new insights into “vitaleffects”. Geochemistry, Geophysics, Geosystems 7, Q12004. http://dx.doi.org/10.1029/2006GC001354.
Cohen, A.L., McCorkle, D.C., de Putron, S., Gaetani, G.A., Rose, K.A., 2009. Morphologicaland compositional changes in the skeletons of new coral recruits reared in acidifiedseawater: insights into the biomineralization response to ocean acidification.Geochemistry, Geophysics, Geosystems 10 (7), Q07005. http://dx.doi.org/10.1029/2009GC002411.
Denniston, R.F., Shearer, C.K., Layne, G.D., Vaniman, D.T., 1997. SIMS analyses of minor andtrace elements distributions in fracture calcite form Yucca Mountains, Nevada, USA.Geochimica et Cosmochimica Acta 61, 1803–1818.
DeVilliers, S., Nelson, B., Chivas, A., 1995. Biological-controls on coral Sr/Ca and δ18Oreconstructions of sea-surface temperatures. Science 269 (5228), 1247–1249.
Fayek, M. (Ed.), 2009. Secondary Ion Mass Spectrometry in the Earth Sciences: Cleaningthe Big Picture From a Small Spot. Short Course Series, 41. Mineralogical Associationof Canada.
Fernandez, D.F., Gagnon, A.C., Adkins, J.F., 2010. An isotope dilution ICP-MS method forthe determination of Mg/Ca and Sr/Ca ratios in calcium carbonate. Geostandardsand Geoanalytical Research 35, 23–37.
Finch, A.A., Shaw, P.A., Weedon, G.P., Holmgren, K., 2001. Trace element variation inspeleothem aragonite: potential for palaeoenvironmental reconstruction. Earth andPlanetary Science Letters 186, 255–267.
Friedman, I., O'Neil, J., Cebula, G., 1982. Two new carbonate stable-isotope standards.Geostandards Newsletter 6, 11–12.
Gabitov, R.I., Schmitt, A.K., Rosner, M., McKeegan, K.D., Gaetani, G.A., Cohen, A.L., Watson,E.B., Harrison, T.M., 2011. In situ δ7Li, Li/Ca, and Mg/Ca analyses of synthetic arago-nites. Geochemistry, Geophysics, Geosystems 12, Q03001. http://dx.doi.org/10.1029/2010GC003322.
Gaetani, G.A., Cohen, A.L., 2006. Element partitioning during precipitation of aragonite fromseawater: a framework for understanding paleoproxies. Geochimica et CosmochimicaActa 70, 4617–4634.
Gagnon, A.C., 2009. Geochemical Mechanisms of Biomineralization from Analysis ofDeep-Sea and Laboratory Cultured Corals. (PhD Thesis) Caltech.
Hart, S.R., Cohen, A.L., 1996. An ion probe study of annual cycles of Sr/Ca and other traceelements in corals. Geochimica et Cosmochimica Acta 60, 3075–3084.
Heikoop, J.M., Hickmott, D.D., Risk, M.J., Shearer, C.K., Atudorei, V., 2002. Potential climatesignals from the deep-sea gorgonian coral Primnoa resedaeformis. Hydrobiologia 471,117–124.
Herzog, R.F.K., Poschenrieder, W.P., Satkiewicz, F.G., 1973. Observation of clusters in asputtering ion source. Radiation Effects 18, 199–205.
Holcomb, M., Cohen, A.L., Gabitov, R.I., Hutter, J.L., 2009. Compositional andmorphologicalfeatures of aragonite precipitated experimentally from seawater and biogenically bycorals. Geochimica et Cosmochimica Acta 73, 4166–4179.
John, S., Adkins, J.F., 2010. Analysis of dissolved iron isotopes in seawater. Marine Chemistry119, 65–76.
Kasemann, S.A., Schmidt, D.N., Bijma, J., Foster, G.L., 2009. In situ boron isotope analysis inmarine carbonates and its application for foraminifera andpaleo-pH. Chemical Geology260, 138–147.
Kunioka, D., Shirai, K., Takahata, N., Sano, Y., 2006. Microdistribution of Mg/Ca, Sr/Ca, andBa/Ca ratios in Pulleniatina obliquiloculata test by using a NanoSIMS: implication forthe vital effect mechanism. Geochemistry, Geophysics, Geosystems 7 (12), Q12P20.http://dx.doi.org/10.1029/2006GC001280.
Ma, C., Rossman, G.R., 2008. Barioperovskite, BaTiO3, a new mineral from the BenitoiteMine, California. American Mineralogist 93, 154–157.
Mason, R.A., 1987. Ionmicroprobe analysis of trace elements in calcitewith an application tothe cathodoluminescence zonation of limestone cements from the Lower Carboniferousof South Wales, U.K. Chemical Geology 64, 209–224.
Meibom, A., Cuif, J.-P., Hillion, F., Constantz, B.R., Juillet-Leclerc, A., Dauphin, Y., Watanabe,T., Dunbar, R.B., 2004. Distribution of magnesium in coral skeleton. GeophysicalResearch Letters 31, L23306. http://dx.doi.org/10.1029/2004GL021313.
Meibom, A., Yurimoto, H., Cuif, J.-P., Domart-Coulon, I., Houlbreque, F., Constantz, B.,Dauphin, Y., Tambutte, E., Tambutte, S., Allemand, D., Wooden, J., Dunbar, R., 2006.Vital effects in coral skeletal composition display strict three-dimensional control.Geophysical Research Letters 33, L11608. http://dx.doi.org/10.1029/2006GL025968.
Meibom, A., Cuif, J.-P., Houlbreque, F., Mostefaoui, S., Dauphin, Y., Meibom, K.L., Dunbar, R.,2008. Compositional variations at ultra-structure length scales in coral skeleton.Geochimica et Cosmochimica Acta 72, 1555–1569.
Reynaud, S., Ferrier-Page, C., Meibom, A., Mostefaoui, S., Mortlock, R., Fairbanks, R.,Allemand, D., 2007. Light and temperature effects on Sr/Ca and Mg/Ca ratios in thescleractinian coral Acropora sp. Geochimica et Cosmochimica Acta 71, 354–362.
Riciputi, L.R., McSween Jr., H.Y., Johnson, C.A., Prinz, M., 1994. Minor and trace ele-ment concentrations in carbonates of carbonaceous chondrites, and implicationsfor the compositions of coexisting fluids. Geochimica et Cosmochimica Acta 58,1343–1351.
Rollion-Bard, C., Erez, J., 2010. Intra-shell boron isotope ratios in the symbiont-bearingbenthic foraminiferan Amphistegina lobifera: implications for δ11B vital effects andpaleo-pH reconstructions. Geochimica et Cosmochimica Acta 74, 1530–1536.
Rollion-Bard, C., Vigier, N., Meibom, A., Blamart, D., Reynaud, S., Rodolfo-Metalpa, R.,Martin, S., Gattuso, J.P., 2009. Effect of environmental conditions and skeletal ultra-structure on the Li isotopic composition of scleractinian corals. Earth and PlanetaryScience Letters 286, 63–70.
Sano, Y., Shirai, K., Takahata, N., Hirata, T., Sturchio, N.C., 2005. NanoSIMS analysis of Mg,Sr, Ba and U in natural calcium carbonate. Analytical Sciences 21, 1091–1097.
Savard, M.M., Veizer, J., Hinton, R., 1995. Cathodoluminescence at low Fe and Mnconcentrations — a SIMS study zones of natural calcites. Journal of SedimentaryResearch Section A, Sedimentary Petrology and Processes 65, 208–213.
Shimizu, N., Hart, S.R., 1982. Application of the ion microprobe to geochemistry and cos-mochemistry. Annual Review of Earth and Planetary Sciences 10, 483–526.
Shimizu, N., Semet, M.P., Allegre, C.J., 1978. Geochemical applications of quantitative ion-microprobe analysis. Geochimica et Cosmochimica Acta 42, 1321–1334.
Shirai, K., Kusakabe,M., Nakai, S., Ishii, T.,Watanabe, T., Hiyagon, H., Sano, Y., 2005. Deep-seacoral geochemistry: implication for the vital effect. Chemical Geology 224, 212–222.
Shirai, K., Kawashima, T., Sowa, K., Watanabe, T., Nakamori, T., Takahata, N., Amakawa, H.,Sano, Y., 2008. Minor and trace element incorporation into branching coral Acroporanobilis skeleton. Geochimica et Cosmochimica Acta 72, 5386–5400.
Slodzian, G., Lorin, J.C., Havette, A., 1980. Isotopic effect on the ionization probabilities insecondary ion emission. Journal de Physique Lettres 41, 555–558.
Spotl, C., Unterwurzacher, M., Mangini, A., Longstaffe, F.J., 2002. Carbonate speleothemsin the dry, inneralpine VinscgauValley, northernmost Italy:witnesses of change in cli-mate and hydrology since the last glacial maximum. Journal of Sedimentary Research72, 793–808.
Steele, I.M., Hervig, R.L., Hutcheon, I.D., Smith, J.V., 1981. Ion microprobe techniques andanalyses of olivine and low-Ca pyroxene. American Mineralogist 66, 526–546.
Stolarski, J., Meibom, A., Przeniosło, R., Mazur, M., 2007. A Cretaceous scleractinian coralwith a calcitic skeleton. Science 318, 92–94.
Stoll, H., Shimizu, N., Arevalos, A., Matell, N., Banasiak, A., Zeren, S., 2007. Insights oncoccolith chemistry from a new ion probe method for analysis of individually pickedcoccoliths. Geochemistry, Geophysics, Geosystems 8 (6), Q06020. http://dx.doi.org/10.1029/2006GC001546.
Weber, P.K., Bacon, C.R., Hutcheon, I.D., Ingram, B.L., Wooden, J.L., 2005. Ion microprobemeasurement of strontium isotopes in calcium carbonate with application to salmonotoliths. Geochimica et Cosmochimica Acta 69, 1225–1239.
Werner, H.W., 1978. Introduction to secondary ion mass spectrometry (SIMS). Proc. Adv.Study Inst. Electron Ion Spectrosc. Solids. Plenum Press, New York, pp. 324–441.
