Clumped isotope analysis of carbonates: comparison of two different acid digestion techniques

Research Article

Received: 4 February 2013 Revised: 25 April 2013 Accepted: 25 April 2013 Published online in Wiley Online Library

Rapid Commun. Mass Spectrom. 2013, 27, 1631–1642
(wileyonlinelibrary.com) DOI: 10.1002/rcm.6609
Clumped isotope analysis of carbonates: comparison of twodifferent acid digestion techniques

Ulrike Wacker1*, Jens Fiebig1,2 and Bernd R. Schoene31Institute of Geosciences, Goethe-University, Altenhöferallee 1, 60438 Frankfurt am Main, Germany2Biodiversity and Climate Research Center, Senckenberganlage 25, 60325 Frankfurt am Main, Germany3Department of Applied and Analytical Paleontology (INCREMENTS), Earth System Science Research Center, Institute ofGeosciences, University of Mainz, Johann-Joachim-Becherweg 21, 55128 Mainz, Germany

RATIONALE: The kinetic nature of the phosphoric acid digestion reaction enables clumped isotope analysis of carbonatesusing gas source isotope ratio mass spectrometry (IRMS). In most laboratories acid digestions are performed at 25�C insealed vessels or at 90�C in a common acid bath. Here we show that different Δ47 results are obtained depending on thedigestion technique employed.METHODS: Several replicates of a biogenic aragonite and NBS 19 were reacted with 104% H3PO4 in sealed vessels at25�C and at 90�C using a common acid bath. The sample size varied between 4 mg and 14 mg. Purification methods thatare standard for clumped isotope analyses were applied to the evolved CO2 before measuring the abundances of masses44 to 49 relative to a reference gas by IRMS.RESULTS:A systematic trend to lower and more consistent Δ47 values is observed for reactions at 25�C if the sample sizeis increased. We suggest that secondary re-equilibration of evolved CO2 or reaction intermediates with free watermolecules preferentially occurs for relatively small samples (4–7 mg), finally yielding elevated Δ47 values compared with>7 mg aliquots. In contrast, no such sample size effect on Δ47 values is observed for carbonates that are digested at 90�Cusing the common acid bath.CONCLUSIONS: The determination of Δ47 values of carbonate samples smaller than 7 mg becomes more precise andaccurate if digestions are performed at 90�C. Based on our results we propose that the difference in phosphoric acidfractionation factor between 25�C and 90�C is 0.07% for both calcite and aragonite. Copyright © 2013 JohnWiley& Sons, Ltd.

Clumped isotope analysis has recently been advanced as a newtool to reconstruct carbonate formation temperature[1] (andreferences cited therein). The carbonate clumped isotopethermometer relies on the isotope exchange reaction involvingthemost abundant isotopologue containing two heavy isotopesCa13C18O16O2:

Ca13C16O3 þ Ca12C18O16O2 ¼ Ca13C18O16O2 þ Ca12C16O3

(1)

At thermodynamic equilibrium the abundance of 13C–18Obonds in CaCO3 is a function of the equilibrium constant ofreaction (1). The equilibrium constant, in turn, largelydepends on temperature[2] and, hence, the determination ofthe abundance of 13C–18O bonds in the carbonate providesinformation about its crystallization temperature. Currently,there is no technique sensitive and precise enough to directlymeasure the abundance of isotopologues in carbonatesparticipating in reaction (1). Therefore, carbonates are digestedwith 103% H3PO4

[3] or 105% H3PO4[4] and the abundance of

13C–18O bonds within the evolved CO2 is measured instead.

* Correspondence to: U. Wacker, Institute of Geosciences,Goethe-University, Altenhöferallee 1, 60438 Frankfurt amMain, Germany.E-mail: [email protected]

Rapid Commun. Mass Spectrom. 2013, 27, 1631–1642

163

Fortunately, the acid digestion reaction is kinetically controlledsuch that the concentration of 13C–18O bonds in the evolvedCO2 remains proportional to the original abundance ofcorresponding bonds in the carbonate lattice.[3,5,6]

The temperature dependency of reaction (1) is expressed byΔ47 which quantifies the deviation of the abundances ofisotopologues of a sample gas from a theoretical randomdistribution. For this purpose, measured R47, R46 and R45

values are compared with their corresponding stochasticdistribution ratios (R47*, R46*, R45*) where Ri =mi/m44:

Δ47 ¼ R47=R47* � 1� �

� R46=R46* � 1� �

� R45*=R45* � 1� �h i

* 1000 %ð Þ (2)

A number of calibrations of the carbonate clumped isotopethermometer have been published, including those for syntheticand biogenicminerals,[3,7–10] aswell as speleothems.[11,12] Severalmodern biogenic carbonates for which growth temperatureshave been determined independently confirm the relationshipbetween Tgrowth and Δ47 of synthetic carbonates reported byGhosh et al.[3] Amongst these are mollusks, brachiopods, corals,otholiths, foraminifera and coccoliths (see[1,13] for an overview).As a consequence, most authors refer to the Ghosh et al.[3]

line when applying the Δ47-thermometer to fossil material(e.g.,[1,13,14]). However, discrepant results have also become

Copyright © 2013 John Wiley & Sons, Ltd.

1

U. Wacker, J. Fiebig and B. R. Schoene

1632

obvious. Dennis and Schrag[8] failed to reproduce thetemperature-Δ47 calibration of Ghosh et al.[3] for inorganicallyprecipitated calcite. In addition, both published experimentalcalibrations[3,8] deviate from the theoretically calculatedtemperature dependence.[5,15] Moreover, some marine speciespreferring rather cold temperatures (0–15�C), such as somebenthic foraminifera[9] and mollusks,[16] plot below the Ghoshcalibration line. The causes of these unconformities are stillunclear.[13] According to these previous calibrations,[3,8,15,16] aΔ47 variation of 0.1% corresponds to a temperature differenceof roughly 20�C or even more. As a consequence, highlyaccurate and precise measurements are required, especiallywhen applying the thermometer to the small temperaturewindow prevailing in the marine and terrestrial realms. Asstated earlier, Δ47 measurements of acid-liberated CO2 provideinformation about the temperature of carbonate crystallizationbecause of the kinetic nature of the acid digestion reaction.However, precise measurements require that variations inkinetic fractionation factors as well as secondary bondreordering in the evolved CO2 (e.g., due to oxygen isotopeexchange between water and CO2) are avoided. Kineticfractionations do not only depend on temperature, but can alsovary with relative concentrations of reagents. For example,Wendeberg et al.[17] observed a slight decrease in the d18O valueof evolved CO2 after increasing the phosphoric acidconcentration from 102% to 107%. In addition, the d18OCO2

value varied with the d18O value of the phosphoric acid ifconcentrations <102% were used. In this particular case,isotopic exchange between evolved CO2 and phosphoric acidis promoted by the presence of free water.[17] Furthermore,oxygen isotopic compositions of CO2 evolved from carbonateswere shown to vary slightly with the digestion method:[18]

systematically lower d18O values were obtained when using acommon acid bath than when using sealed McCrea-typereaction vessels.[19] This is interpreted as being the result of O-isotopic exchange between CO2 and phosphate polymers orfree water, dissolution of CO2 in the acid and incompleteremoval of CO2 during the extraction process. Swart et al.[18]

conclude that most accurate results for d18O analyses areobtained if the common acid bath technique is used.In this study we compare Δ47 signatures of CO2 obtained by

two different carbonate digestion methods which arecommonly used for clumped isotope analyses: digestions at25�C in McCrea-type reaction vessels[3] and digestions at 90�Cusing a common acid bath.[6] The main purpose is to determinewhether the accuracy and precision of the two techniques areconsistent within the range of sample size (4–14 mg) usuallyaddressed for clumped isotope measurements of almost purecarbonates.[3,4,8,14] In addition, we determine the difference inisotopic fractionations between 25�C and 90�C digestions.

EXPERIMENTAL

Samples and sample preparation

Two different standard materials were analyzed for their Δ47

signatures:

• NBS 19 (calcite, supplied by the IAEA, Vienna, Austria, andby the National Bureau of Standards, Gaithersburg, MD,USA) is a coarse-grained white marble with d18O=�2.20%

wileyonlinelibrary.com/journal/rcm Copyright © 2013 John Wi

(V-PDB) or +28.64% (V-SMOW) and d13C=1.95% (V-PDB).NBS 19 is used as an interlaboratory standard for clumpedisotope analyses.[3,20,21] For digestions at 25�C, NBS 19 wascrushed to a fine powder and homogenized using a mortarand pestle. When not treated that way, the reaction with104% H3PO4 was observed to be very slow and, sometimesincomplete, as indicated by d13C values significantly lowerthan 1.95%.

• Arctica islandica, a bivalve with an aragonitic shell, grewnear Langanes, NE Iceland, in ca. 30 m water depth at anaverage temperature of 6.0� 0.5�C. The specimen wascollected alive in August 2006. Sample material was takenfrom the outer layer of the most recently formed part of theshell. For this purpose, the shell was cut into two partsalong the direction of growth from the umbo to the ventralmarging using a Buehler low-speed precision sawequipped with a 0.4 mm thick diamond-coated saw blade.Subsequently, both shell slabs were cut perpendicularly tothe growth lines. The youngest part of the valve was usedfor clumped isotope analyses. The periostracum and theinner shell layer were physically abraded before the twoshell pieces were ground and homogenized in anoscillating disc mill. Some aliquots that were reacted at25�C were pretreated for 2 h with 1.5% H2O2 prior to theanalysis to remove organic matter, as this has been foundto preclude accurate Δ47 analysis of some biogeniccarbonates.[9]

Acid digestion

The concentration of the phosphoric acid used for digestionreactions was 104%. This corresponds to a density of 1.91 g/cm3

(for the conversion of acid concentrations into acid densitieswe use the equation determined by Ghosh et al.:[22]

y= 0.0114x+ 0.723, where y is the acid density in g/cm3 andx is the acid concentration in %). The acid was prepared byslightly modifying the method of Coplen:[23] 99% H3PO4

(Merck KGaA, Darmstadt, Germany, ≥99%) was heated to150�C, and then P4O10was slowly added to the stirred solution.After 10 h an acid aliquot was cooled to room temperature forabout 1 h. The acid density was measured gravimetricallyat 25�C, and once a concentration of 104% was achieved,the acid was heated (at 150�C) for at least three additionalhours. Otherwise, either more P4O10 or more distilled waterwas added until the concentration reached 104%. The acidwas stored for at least 2 weeks to ensure that all the waterhad reacted with P4O10 to H3PO4.

[23] Before using theacid for digestion reactions, the density was controlledgravimetrically again.

Carbonate digestions at 25�C were performed in McCrea-type reaction vessels:[19] 4 to 14 mg aliquots of carbonate werereacted with 7� 1 mL 104% H3PO4. The carbonate was placedinto the side arm and the acid was filled in the main tube of thereaction vessel, before evacuating it for 2.5 to 3.5 h to reach avacuum better than 10–3 mbar. Before starting the reaction bytipping the acid over the carbonate, the vessels were placed ina common water bath at 25.0(�0.2)�C for 1 h to ensure thermalequilibration. During the reaction, the vessels were kept in thewater bath at 25�C. The reaction timewas between 16 and 20 h.

Reactions at 90�Cwere performed using an automated acidbath whose design is very similar to those used at Caltechand Johns Hopkins University.[6] The acid bath was placed

ley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2013, 27, 1631–1642

Clumped isotope analysis of carbonates

in a copper block surrounded with a heating band to keep theacid at a constant temperature of 90.0(�0.1)�C. The carbonatesamples were filled into small Ag capsules(IVA Analysentechnik e.K., Meerbusch, Germany, art. no.184.9921.36). These were loaded into a Zero Blank autosampler(Costech, Valencia, CA, USA). The acid bath and theautosampler were pumped by a turbomolecular pump backedby a membrane pump (Pfeiffer, Aßlar, Germany). After asample had been dropped in the acid, the evolved CO2 wasfrozen in a U-trap cooled at liquid nitrogen temperature afterpassing another cold trap held at �80�C to remove traceamounts of water. By monitoring the pressure of the evolvedCO2 it was observed that the digestion of the fine grainedcarbonate powder of A. islandica took only ~10 min, while thelarger crystals of NBS 19 reacted for ~20–30 min. Aftermanometrically determining the yield, the extracted CO2

was frozen into a transportable glass finger equipped withhigh-vacuum valves (Louwers, Hapert, The Netherlands, artcode 40.200.8).

CO2 cleaning procedure

The CO2 derived from phosphoric acid digestions at both 25�Cand 90�C was cleaned following the procedure described inGhosh et al.[3] Briefly, the transportable glass finger containingthe sample CO2 was connected to a cryogenic vacuumextraction line, evacuated to<10–6 mbar using a turbomolecularpump supported by amembranepump (Pfeiffer). The CO2 yieldwas determined with a capacity manometer, and the CO2 wasthen passed twice over a trap held at�80�C before being frozenback into the volume of the glass finger. Afterwards, the samplewas passed through a gas chromatography (GC) purificationsystem to remove traces of hydrocarbons. The CO2 wasentrained into a He carrier gas flow (18 mL/min; purity:99.9999%) and purged through a 1.20 m� 2.15 mm i.d. stainlesssteel column packed with Porapak Q 80/120, kept at �20�C.Before entering and after leaving the GC column the He-CO2

gas mixture was passed through additional water traps held at�77�C. The CO2 effluent from the GC column was collectedfor a period of 30 min in a U-trap immersed in liquid nitrogen.After one sample had passed through the GC column the Heflow was switched to a backflush mode, enabling the watertraps and the column to be heated to 25�C and 150�C,respectively, for at least 15 min. In a final step, the He carriergas was pumped away and the CO2 was cryogenically purifiedone final time using the vacuum extraction line described above.For isotopic analysis, the purified CO2 was transferred to atransportable glass finger that could be connected to the dualinlet system of the mass spectrometer.

163

Measurements

Isotopic analyses were performed at Goethe University(Frankfurt, Germany) on a MAT 253 gas source isotope ratiomass spectrometer (Thermo Fisher Scientific, Bremen,Germany) equipped with a dual inlet system and six Faradaycup collectors for masses 44 to 49 (resistors: 3� 108Ω,3� 1010Ω, 1011Ω for masses 44–46, respectively, and 1012Ωfor masses 47–49). The original stainless steel capillarieswere replaced with 4 feet long electroformed nickel (EFNi)

Copyright © 2013 JRapid Commun. Mass Spectrom. 2013, 27, 1631–1642

capillaries (VICI AG, Schenkon, Switzerland; 1/32" o.d.,0.005" i.d., art. no. TEFNI.505, 122 cm� 0.127 mm i.d.).[6]

Measurements were performed using the dual inlet system,after adjusting the sample and reference gas signals for mass44 to (16000� 150) mV. Ten acquisitions, consisting of tencycles with an ion integration time of 20 s each, were usedfor all measurements, corresponding to a shot-noise limit of~0.008%.[24] Before each acquisition, peak centering,background determination and pressure adjustments to 16 Vat mass 44 were carried out. The total analysis time of onesample was about 3 h. CO2 from Oztech (Safford, AZ, USA;d18O=+25.01% vs. V-SMOW); d13C=�3.63% vs. V-PDB) wasused as the reference gas.

Data processing

We report Δ47 values on the absolute scale of Dennis et al.[20]

In comparison with the data correction procedure describedby Huntington et al.,[25] absolute scaling of raw Δ47 valueshas the advantage that the processed data becomescomparable between labs as the Δ47 composition of the CO2

reference gas (which may vary between labs) is considered.Briefly, raw data is corrected in two steps. (1) To correct fornon-linearities of the mass spectrometer, CO2 gases ofdifferent bulk isotopic compositions were measured. Prior toisotopic analysis, these were heated in quartz break-seal tubesto 1000�C for more than 2 h to reach the characteristicdistribution at this temperature. After quenching to roomtemperature, the gases were purified cryogenically and byGC like carbonate samples. (2) Δ47 values corrected for non-linearity were then converted into the absolute scale using theempirical transfer function (ETF). The ETF is determined byplotting the intercepts of linearity lines (derived frommeasurements of CO2 gases of distinct bulk isotopiccompositions equilibrated to at least two different temperatures)against the corresponding theoretically expected values.[2] Wehave determined our ETF using the intercepts of CO2 gasesheated at 1000�C and equilibrated with water at 25�C. CO2

and H2O were enclosed in glass tubes, which were placed in awaterbath for at least 3 days at 25.0� 0.2�C. After quenchingwith liquid nitrogen, the glass was held in an ethanol/dry-iceslush at �80�C to release CO2, while H2O remained frozen.Separation between water and CO2 was improved bypassing the gas at least five times over a trap held at �80�Cusing the vacuum extraction line. Afterwards, water-equilibrated gases were cleaned and measured in the sameway as the sample gases.[20]

Carbonates were reacted at 25�C and 90�C between April2011 and January 2012. The instrument non-linearity wascontinually monitored by measuring heated gases daily andcomparing the slopes of heated gas regression linesdetermined from blocks of nine consecutive heated gasanalyses. As long as the slopes of these single blocks wereidentical within the standard errors, all the analyses wereused for the determination of the heated gas slope. If theslopes of the different blocks of analyses were changingsignificantly, new correction parameters were determinedfor each day using the running average of the heated gas lineof nine consecutive analyses. [For example, for linearitycorrection of the NBS 19 sample analyzed on 04 November2011 the slope of the heated gas line (m= 0.0250) was derivedusing heated gas analyses carried out between 24 October

wileyonlinelibrary.com/journal/rcmohn Wiley & Sons, Ltd.

3


1634

2011 and 16 November 2011 (Supplementary Table S1, seeSupporting Information).] The d47 and Δ47 values of heatedgases were, however, highly correlated, with R2-valuesalways better than 0.99. The slopes of the heated gas lineswhich were used to correct carbonate data are listed in Table 1along with raw and corrected carbonate Δ47 data.For the determination of the ETFs we considered the

intercepts of linearity lines of heated gases analyzed over aperiod of 2 months. The intercepts for April/May andJune/July 2011 are statistically indistinguishable (SupplementaryTable S1, see Supporting Information). After source cleaning asignificant change in the heated gas intercept betweenSeptember/October 2011 and November/December 2011 wasobserved, while no change occurred between November/December 2011 and January/February 2012 (SupplementaryTable S1, see Supporting Information) Gases equilibrated at25�C were measured frequently in November/December 2010and, after source cleaning, in September 2011. In addition, somegases were analyzed occasionally (February, March, April, Julyand November 2011, Supplementary Table S1, see SupportingInformation) to test the consistency of the Δ47 composition ofthe reference gas. The intercept of the line determined by25�C water-equilibrated gases was constant at �0.03%between November 2010 and July 2011, while it was�0.06% after source cleaning. We therefore applied threeETFs to correct our carbonate data: y = 1.1094x + 0.9585between November 2010 and July 2011, y = 1.0959x + 0.9910for September 2011 and y = 1.1521x + 0.9943 from November2011 to January 2012 (with x: intercepts of the equilibrationand heated gas lines and y: theoretical equilibrium valuefor Δ47 at the corresponding equilibration temperature).For the normalization of Δ47 values to acid digestions at

25�C a difference in acid fractionation factors of +0.081%[6]

was applied to data obtained from carbonates reacted at90�C, as used by Dennis et al.[20] It should be noted that thedifference of 0.081% was determined on the internal ’Caltechscale’. If this value is projected onto the absolute referenceframe, using the slope of the secondary reference frametransfer function provided by Dennis et al.,[20] a value of0.084% is obtained instead. Considering the precision of Δ47

analysis as represented by our shot-noise limit, this value isindistinguishable from the +0.081% reported by Passey et al.[6]

RESULTS

d13C values, d18O values, d47 values, raw Δ47 values andabsolute Δ47 values, and their 1s standard errors (1 se) aswell as sample size and date of analysis, are listed in Table 1.For untreated A. islandica reacted at 25�C a mean Δ47 value of0.718% is obtained, with the total interval ranging between0.674 and 0.772% and a standard error of 0.007% (n = 20).The mean value is statistically indistinguishable from theaverage obtained for sample aliquots that were treated withH2O2 before acid digestions (0.730� 0.008)% (n = 10), whosecorresponding Δ47 values vary between 0.695 and 0.787%.Altogether, an average for A. islandica of (0.722� 0.005)%(n = 30) is obtained. For NBS 19 digested at 25�C, the Δ47

values range from 0.339 to 0.396% with a mean of(0.373� 0.004)% (n = 20). For both materials, distinctivelylower mean Δ47 values are obtained if the carbonates werereacted at 90�C, applying a correction factor of +0.081%[6]


that accounts for the difference in acid digestion fractionationfactors between reactions at 25�C and 90�C: (0.707� 0.003)%(n= 19) for A. islandica (total interval between 0.682 and0.723%) and (0.359� 0.004)% (n = 19) for NBS 19 (total intervalbetween 0.329 and 0.391%).

DISCUSSION

The precision of clumped isotope analysis of carbonatereplicates is commonly reported as the 1s standard error ofthe mean (e.g.,[3,25]). Recently, three different laboratoriesanalyzed NBS 19, and mean values of (0.373� 0.007)%(Harvard University; n = 7), (0.399� 0.005)% (Johns HopkinsUniversity; n = 12) and (0.404� 0.006)% (Yale University;n = 40) were reported for Δ47 values with the average andaccepted value being 0.392%.[20] In this study, the mean Δ47

values determined for NBS 19 at reaction temperatures of25�C and 90�C are (0.373� 0.004)% and (0.359� 0.004)%,respectively. Within errors, these values are eitherindistinguishable from (25�C) or nearly identical to (90�C)the Havard mean value. It has been postulated that thedifferences in NBS 19 between the three laboratories weredue to possible heterogeneities of different aliquots of NBS19. However, the difference observed in our data cannot beattributed to inhomogeneities as all measured aliquots werefrom the same batch.

Correlation between sample size and Δ47 values

If Δ47 values are plotted against sample size, a systematictrend is observed for carbonates that were digested at 25�C(Fig. 1). Sample aliquots <7 mg show a larger scatter in Δ47

values and higher corresponding mean values than >7 mgaliquots. A. islandica (untreated plus pretreated) <7 mgaverage to (0.733� 0.006)% (n = 22), but (0.692� 0.003)%(n = 8) for >7 mg aliquots. Similarly, the average Δ47 valuefor NBS 19 samples <7 mg is 0.378� 0.003% (n = 16) and,therefore, clearly distinguishable from the value of(0.353� 0.006)% (n = 4) determined for larger (>7 mg)aliquots (Fig. 1, Table 1).

In contrast, no dependency of Δ47 values on sample size isobserved when digesting carbonates at 90�C (Fig. 1). The Δ47

values of small samples (<7 mg) of NBS 19 average to(0.357� 0.006)% (n = 11), indistinguishable from the meanΔ47 value for samples >7 mg ((0.362� 0.005)% (n = 8)). Thesame result is observed for A. islandica: for small samples(<7 mg) a mean of (0.706� 0.003)% (n = 15) is calculated,indistinguishable from the average Δ47 value of large samples(>7 mg) with (0.712� 0.004)% (n = 4).

Potential occurrence of contaminants in sample-derivedCO2

Isobaric interferences caused by non-CO2 additions to thesample gas will affect measured Δ47 values. One source ofisobaric contaminants might be traces of organic matterinside carbonates, which might be partly volatilized duringphosphoric acid digestion.[25] There are two evidentarguments against contamination of sample gas caused byorganic matter: First, we would expect more organiccompounds to be volatilized if the reaction temperature was


Table 1. Experimental results: isotopic values reported in % deviation, sample sizes in mg.

(a) Digestions at 25�C; McCrea-type reaction vessels.

Date ofanalysis

d13C(V-PDB)

d18O(V-PDB)

Samplesize d47 Δ47 Δ47 abs se

Slope of heatedgas line d48 Δ48

A. islandica4/8/2011 1.53 3.05 4.6 25.38 0.530 0.761 0.010 0.0279 64.507 24.4154/26/2011 1.56 3.28 4.9 25.66 0.548 0.772 0.013 0.0279 63.752 23.2554/28/2011 1.60 3.31 7.1 25.69 0.505 0.692 0.007 0.0290 58.068 17.7285/3/2011 1.63 3.51 4.3 26.00 0.577 0.762 0.011 0.0290 64.303 23.3135/3/2011 1.60 3.34 4.3 25.75 0.529 0.717 0.009 0.0290 60.649 20.1505/3/2011 1.69 3.39 10.2 25.87 0.521 0.705 0.010 0.0290 56.466 16.0215/4/2011 1.67 3.36 8.0 25.79 0.498 0.682 0.011 0.0290 54.985 14.6725/5/2011 1.90 3.07 4.2 25.77 0.541 0.729 0.012 0.0290 59.883 19.9545/5/2011 1.55 3.25 4.3 25.61 0.531 0.724 0.010 0.0290 62.182 21.8035/6/2011 1.67 3.11 4.1 25.62 0.564 0.760 0.009 0.0290 61.023 20.9695/10/2011 1.67 3.19 7.2 25.64 0.506 0.695 0.008 0.0290 54.969 14.9805/10/2011 1.66 3.20 9.9 25.64 0.508 0.697 0.006 0.0290 55.096 15.0865/11/2011 1.57 3.31 8.8 22.22 0.498 0.687 0.005 0.0290 54.958 14.7445/31/2011 1.60 3.05 4.8 25.43 0.528 0.703 0.012 0.0298 61.002 21.0875/31/2011 1.62 3.01 4.6 25.39 0.501 0.674 0.007 0.0298 57.110 17.4165/31/2011 1.64 3.23 4.6 25.68 0.540 0.708 0.009 0.0298 62.398 22.0576/8/2011 1.60 3.08 4.7 25.51 0.561 0.735 0.011 0.0299 61.590 21.5856/8/2011 1.64 3.12 4.8 25.59 0.562 0.733 0.010 0.0299 62.782 22.6476/8/2011 1.56 3.07 4.7 25.46 0.561 0.736 0.009 0.0299 62.641 22.6116/9/2011 1.69 3.23 14.2 25.71 0.521 0.684 0.014 0.0299 58.845 18.637average 1.63 3.21 0.718se 0.007

Pretreated A. islandica5/17/2011 1.59 3.16 5.1 25.54 0.524 0.718 0.010 0.0290 60.180 20.0655/18/2011 1.51 3.16 4.1 25.46 0.511 0.701 0.008 0.0292 64.338 24.0595/19/2011 1.64 3.09 4.9 25.55 0.550 0.741 0.010 0.0292 60.383 20.3985/20/2011 1.58 3.37 11.1 25.74 0.516 0.695 0.009 0.0293 56.040 15.6575/23/2011 1.67 3.13 4.0 25.66 0.595 0.787 0.013 0.0292 67.693 27.3585/25/2011 1.62 3.26 4.8 25.70 0.553 0.728 0.010 0.0296 63.828 23.3755/26/2011 1.66 3.36 4.2 25.85 0.561 0.731 0.007 0.0296 64.594 23.8975/26/2011 1.70 2.85 4.3 25.36 0.546 0.732 0.007 0.0296 60.177 20.6865/26/2011 1.71 3.15 4.4 25.70 0.573 0.751 0.009 0.0296 62.176 22.0156/9/2011 1.67 2.90 4.5 25.32 0.494 0.717 0.008 0.0299 58.788 19.253average 1.64 3.14 0.730se 0.008

NBS 194/12/2011 1.93 �2.14 4.3 19.40 0.034 0.396 0.013 0.0279 44.274 16.3324/12/2011 1.95 �2.22 4.6 19.33 0.021 0.384 0.010 0.0279 42.901 15.1504/12/2011 2.01 �2.03 4.4 19.59 0.020 0.374 0.013 0.0279 44.327 16.1454/13/2011 1.95 �2.26 4.5 19.28 0.012 0.375 0.010 0.0279 43.497 15.8124/14/2011 1.97 �2.05 5.2 19.49 �0.006 0.349 0.014 0.0279 40.182 12.1714/15/2011 1.91 �2.24 5.0 19.27 0.025 0.389 0.011 0.0279 46.020 18.2284/18/2011 1.95 �2.18 4.1 19.37 0.025 0.387 0.012 0.0279 42.245 14.4444/19/2011 1.97 �2.17 6.2 19.39 0.011 0.369 0.009 0.0280 40.530 12.7454/19/2011 1.95 �2.16 4.4 19.39 0.020 0.378 0.012 0.0280 42.644 14.7794/20/2011 1.91 �2.25 4.4 19.26 0.023 0.373 0.008 0.0286 45.142 17.3934/26/2011 1.94 �2.25 4.8 19.31 0.045 0.396 0.011 0.0286 45.681 17.9154/27/2011 1.96 �2.11 4.6 19.47 0.039 0.375 0.008 0.0290 45.631 17.5804/28/2011 1.98 �2.17 5.4 19.42 0.039 0.377 0.013 0.0290 44.897 16.9925/4/2011 1.99 �2.08 5.6 19.54 0.052 0.387 0.008 0.0290 41.304 13.3075/10/2011 2.05 �2.51 4.9 19.15 0.029 0.374 0.012 0.0290 41.910 14.7565/11/2011 1.98 �2.25 4.6 19.31 0.020 0.359 0.009 0.0290 36.232 8.7305/13/2011 2.06 �2.44 8.6 19.22 0.025 0.368 0.009 0.0290 39.234 12.0245/16/2011 2.05 �2.51 9.3 19.11 0.011 0.356 0.006 0.0290 36.933 9.929

(Continues)


wileyonlinelibrary.com/journal/rcmCopyright © 2013 John Wiley & Sons, Ltd.Rapid Commun. Mass Spectrom. 2013, 27, 1631–1642

1635

Table 1. (Continued)

(a) Digestions at 25�C; McCrea-type reaction vessels.

Date ofanalysis

d13C(V-PDB)

d18O(V-PDB)

Samplesize d47 Δ47 Δ47 abs se

Slope of heatedgas line d48 Δ48

6/9/2011 2.02 �2.10 12.4 19.54 0.036 0.350 0.010 0.0299 41.524 13.5636/21/2011 2.06 �2.31 8.4 19.33 0.010 0.339 0.008 0.0294 33.261 5.943average 1.98 �2.22 0.373se 0.004

(b) Digestion at 90�C; common acid bath.

Date ofanalysis

d13C(V-PDB)

d18O(V-PDB)

Samplesize

d47 Δ47 Δ47 abs se Slope of heatedgas line

d48 Δ48

A. islandica

9/2/2011 1.73 3.28 9.9 23.28 0.269 0.714 0.010 0.0256 44.069 8.6849/5/2011 1.53 3.27 4.0 23.06 0.257 0.707 0.011 0.0256 46.716 11.2749/5/2011 1.61 3.13 5.0 23.00 0.263 0.715 0.006 0.0256 44.957 9.8569/6/2011 1.71 3.33 9.8 23.31 0.266 0.709 0.005 0.0256 45.600 10.0609/6/2011 1.63 3.30 5.2 23.18 0.255 0.702 0.010 0.0256 45.303 9.8449/6/2011 1.60 3.25 4.2 23.09 0.248 0.696 0.011 0.0256 49.478 13.9759/6/2011 1.63 3.38 10.0 23.27 0.258 0.702 0.008 0.0256 43.964 8.3879/7/2011 1.67 3.25 6.3 23.15 0.237 0.683 0.005 0.0256 46.321 10.9269/7/2011 1.65 3.24 4.0 23.16 0.270 0.719 0.010 0.0256 47.231 11.8319/7/2011 1.67 3.34 4.5 23.29 0.278 0.723 0.009 0.0256 45.868 10.2999/7/2011 1.60 3.26 3.9 23.10 0.242 0.689 0.008 0.0256 46.437 11.01811/7/2011 1.59 3.05 6.2 22.86 0.230 0.682 0.010 0.0250 49.285 14.18111/8/2011 1.57 3.18 4.4 23.00 0.254 0.708 0.006 0.0249 51.604 16.17311/10/2011 1.63 3.12 5.7 23.00 0.263 0.719 0.011 0.0249 46.765 11.61711/14/2011 1.62 3.05 5.1 22.89 0.227 0.697 0.009 0.0243 49.193 14.1011/9/2012 1.60 3.14 8.0 22.95 0.219 0.722 0.010 0.0229 46.410 11.2361/10/2012 1.57 3.13 4.7 22.90 0.212 0.716 0.010 0.0229 47.540 12.6901/11/2012 1.59 3.20 4.3 23.00 0.212 0.712 0.000 0.0229 48.068 12.7041/12/2012 1.51 2.96 4.5 22.67 0.209 0.718 0.006 0.0229 48.068 12.704average 1.62 3.20 0.707se 0.003

NBS 199/8/2011 1.98 �2.04 4.97 17.05 �0.240 0.331 0.011 0.0256 30.662 7.1909/8/2011 1.97 �2.02 9.57 17.09 �0.212 0.361 0.009 0.0256 29.974 6.4709/9/2011 1.97 �2.56 5.06 16.51 �0.224 0.363 0.011 0.0256 30.302 7.8959/9/2011 2.03 �2.19 8.92 16.96 �0.228 0.347 0.006 0.0256 29.447 6.5129/9/2011 2.02 �2.10 10.62 17.05 �0.219 0.353 0.011 0.0256 29.837 6.51211/2/2011 1.97 �2.30 5.53 16.82 �0.197 0.358 0.011 0.0253 33.544 10.52611/4/2011 1.94 �2.28 4.81 16.79 �0.210 0.349 0.011 0.0250 34.685 11.60411/6/2011 1.95 �2.21 5.17 16.89 �0.185 0.375 0.012 0.0250 33.593 10.40411/7/2011 1.94 �2.29 4.45 16.78 �0.216 0.344 0.003 0.0250 31.550 8.56411/9/2011 1.96 �2.30 5.13 16.79 �0.209 0.353 0.010 0.0249 34.265 11.24011/10/2011 1.98 �2.17 7.65 16.97 �0.192 0.368 0.010 0.0249 30.609 7.39711/11/2011 1.98 �2.18 5.24 16.92 �0.227 0.329 0.007 0.0249 32.683 9.43911/14/2011 1.98 �2.23 4.75 16.89 �0.214 0.356 0.008 0.0243 33.949 10.7791/3/2012 1.97 �2.17 5.63 16.94 �0.212 0.384 0.009 0.0229 33.581 10.3061/4/2012 2.00 �2.14 8.87 16.97 �0.230 0.362 0.011 0.0229 30.857 7.5931/5/2012 1.99 �2.25 9.25 16.88 �0.207 0.391 0.010 0.0229 30.586 7.3731/9/2012 1.99 �2.17 7.95 16.92 �0.239 0.354 0.008 0.0229 33.231 10.1021/16/2012 1.97 �2.24 4.26 16.85 �0.218 0.379 0.011 0.0229 30.955 7.6781/18/2012 2.04 �2.14 8.98 17.01 �0.233 0.359 0.009 0.0229 33.721 10.653average 1.98 �2.21 0.359se 0.004


1636

increased. Hence, compared with digestions at 25�C, astronger sample size effect should be observed for digestionsat 90�C. These pattern are the reverse of those displayed by


our data. Secondly, Δ47 values for A. islandica remainedunchanged regardless of whether the samples werepretreated with H2O2 prior to acid digestion or not (Table 1,


0.80

0.78

0.76

0.74

0.72

0.70

0.68

0.66

0.46

0.44

0.42

0.40

0.38

0.36

0.34

0.32

2 4 6 8 10 12 14 16

Reacted at 25°CReacted at 25°C, pre-treatedReacted at 90°C

2 4 6 8 10 12 14 16

Reacted at 25°C

Reacted at 90°C

A. islandica

NBS 19

Sample size (mg)

Sample size (mg)

Figure 1. Correlation plot of Δ47 values (%) and sample size(mg) for A. islandica (a) and NBS 19 (b), respectively. For bothcarbonates a relatively large scatter is observed for samplesbetween 4 and ~7 mg if acid digestion is carried out at 25�C.Compared with the interval described by NBS 19 a largervariation in Δ47 values is observed for the biogenic carbonate.Using the common acid bath technique and a reactiontemperature of 90�C, no correlation between Δ47 values andsample size is displayed. For further discussion, see text.


163

Fig. 1(a)). For this reason it does not seem to be necessary toremove any trace amounts of organic matter from thearagonite of A. islandica prior to isotopic analyses. For furtherdiscussion, the data sets for untreated and H2O2-treatedmaterial of A. islandica will, therefore, be combined.Huntington et al.[25] used a correlation plot of d48 against Δ48

values to control the quality of the measured Δ47 values. Asobserved for Δ47, Δ48 is also a function of the bulk isotopiccomposition. Likewise, heated gases describe a linearrelationship between the d48 and Δ48 values. As long as thepurity of the sample gas is comparable with that of the heatedgases, the sample-derived CO2 gases should plot within thelinear array defined by the heated gases. After the installationof our mass spectrometer at the end of 2009, the mass 48 signalinitially was about 200 mV, but this continuously decreased to~40 mV within 2 years. During the same period, it wasobserved that the background of the mass 48 signal drifted tomore negative values. We, therefore, applied a correction ofthe raw values of mass 48 intensity by adding a constant offset(roughly corresponding to the negative background) to themeasured intensities of both sample and reference gas.For a given sample d48 value, we computed the deviation in

Δ48 values ((Δ(Δ48) values) between sample and heated gasregression lines (y = 0.3102x + 0.905, R2 = 0.97; after source


cleaning: y = 0.2571x+ 0.4134, R2 = 0.97). These Δ(Δ48) valuesare plotted versus measured sample Δ47 values in Fig. 2. Allthe measured samples spread around Δ(Δ48) = 0%, as dothe heated gases. Moreover, absolute deviations from theΔ(Δ48) = 0% line are of the same magnitude for both thesample and the heated gases. Especially for A. islandica, forwhich the sample size effect on Δ47 values at 25�C is mostpronounced, the absolute variations in (Δ(Δ48)) valuesof samples reacted at 25�C and at 90�C are identical(Δ(Δ48) = 0� 3%). Therefore, it is unlikely that elevatedΔ47 values for 25�Cdigestions are due to poor sample gas purity.

Incomplete gas yield

Incomplete gas yield, caused either by incomplete reactions ofcarbonates or by dissolution of CO2 in the acid, can lead tokinetic isotope effects which might depend on sample size.Walters et al.[26] observed that the evolved CO2 becameenriched in 18O during the progress of the reaction betweencarbonate and phosphoric acid. Confirming their results, weobtained a d18O value of �2.65% (vs. the accepted value of�2.20%) for coarse-grained NBS 19 that had been reacted at25�C for 8 h instead of the standard 16–20 h. In addition,the d13C value was significantly lower than the acceptedvalue (1.17% vs. 1.95%) and the Δ47 value of 0.451% wassignificantly higher than those obtained for the otherreactions at 25�C (0.339%–0.396%; Table 1). If our elevatedΔ47 values for <7 mg aliquots were due to incompletereactions, correlations between the Δ47 values and the bulkoxygen and carbon isotopic compositions would be expected.Indeed, for NBS 19 a slightly negative correlation between d13Cvalues andΔ47 values is observed, but there is no obvious trendbetweenΔ47 values and d18O values (Fig. 3(b)). In addition, thed18O values, d13C values and Δ47 values of A. islandica do notcorrelate (Fig. 3(a)), although the scatter ofΔ47 values for smallsamples is much larger than for NBS 19. Therefore, we considerit unlikely that the sample size effect results from incompletereactions. On the contrary, we assume that reactions werequantitative, because no systematic differences in Δ47 valueswere observed if the CO2 was extracted after either 16 or 20 h.

Incomplete gas yields would also be observed if theevolved CO2 is not removed quantitatively from the acid.The amount of CO2 dissolved in the acid depends on thepartial pressure of CO2 above the acid (which, in turn,depends on the sample size), the geometry of the reactionvessel, acid viscosity and reaction temperature.[18] Themigration of gases in liquids can be described by diffusionlaws. Fractionations between two CO2 reservoirs that areinduced by diffusion are more pronounced in 13C/12C and18O/16O isotope ratios than in 47/44 isotopologue ratios.[27]

Therefore, we would expect that incomplete removal ofCO2 from the acid would be accompanied by correlationsbetween the clumped and the bulk isotopic compositionof the extracted CO2. Since there is no trend betweenΔ47 valuesand d18O values for NBS 19 (Fig. 3(b)) and no correlations areobserved between Δ47 values, d

18O values and d13C values forA. islandica (Fig. 3(a)), it is unlikely that the sample size effecton Δ47 values observed for samples digested at 25�C resultsfrom incomplete extraction of CO2 from the acid. In addition,we found no manometric evidence for incomplete reactionyields at 25�C.


7

April - July 2011

September 2011 - January 2012

-6

-4

-2

0

2

4

6

0.3 0.4 0.5 0.70.6 0.8

Δ47 (‰)

NBS 19, 25°C

NBS 19, 90°C

A. islandica, 25°C

A. islandica pretreated, 25°C

A. islandica, 90°C

Figure 2. Δ47 values (%) plotted vs. the deviation in Δ48 between samples andheated gas regression line (expressed as Δ(Δ48)). The dashed lines mark theabsolute Δ48 deviation of heated gases from the heated gas regression lineobserved within periods from April to July 2011 and September 2011 to January2012. All the Δ48 values of carbonates scatter around Δ(Δ48) = 0 and plot in theinterval described by the heated gases. This indicates that the purities of sampleand heated gases are indistinguishable from each other.

2.0

2.0 3.0 4.0 0.66 0.72 0.78 0.66 0.72 0.78

1.5

1.0

2.0

1.5

1.0

4.0

3.0

2.0

2.5

-3.0 -2.2 -1.5 0.30 0.36 0.42 0.30 0.36 0.42

2.0

1.5

2.5

2.0

1.5

-1.5

-2.2

-3.0

δ¹³C

(‰

)

NBS 19, 25°CNBS 19, 90°C

A. islandica, 25°C

A. islandica pretreated, 25°C

A. islandica, 90°C

a)

b)

y = -1.8754x + 2.6777

R² = 0.38

Figure 3. Crossplotts of d18O, d13C and Δ47 values for A. islandica and NBS 19, respectively. Values for A. islandica donot correlate with each other (a), irrespective of the reaction temperature. CO2 derived from 25�C digestions of NBS 19shows a negative trend between the Δ47 and d13C values (b). For further discussion, see text.


1638

Potential variations of fractionation factors duringacid digestion

It is well known that the acid digestion reaction of carbonatesusing phosphoric acid is kinetically controlled.[3,5] Wendeberg


et al.[17] measured the carbon and oxygen isotopic compositionof CO2 evolved from phosphoric acid digestion of NBS 19using variable concentrations of phosphoric acid. They observedthat the d18O value of the evolved CO2 decreased by 0.1%when



phosphoric acid concentration was increased from 102% to107%. They concluded that phosphoric acid concentrationmightaffect oxygen isotope fractionation.In general, it is assumed that the digestion proceeds via the

following steps:[5]

CaCO3 þ 2H3PO4 ! Ca2þ þ 2 H2PO4ð Þ� þH2CO3 (3)

H2CO3 ! H2Oþ CO2 (4)

In aqueous solutions the most common mechanism ofcarbonic acid decay at room temperature is the formation of atransition state complex consisting of one water molecule andone H2CO3 molecule. Two additional water molecules functionas polarizers that lower the activation energy, and another threewater molecules are involved in the decay of H2CO3.

[28] Thedecay of the carbonic acid reaction intermediate H2CO3 requireshigher activation energies if a smaller number ofwater molecules is involved.[28–30] If the phosphoric acidconcentration is increased, free water molecules become lessavailable, reducing reaction rates and, thereby, affecting isotopicselectivity during bond cleavage of reaction intermediates.[17]

Therefore, it might be possible that the Δ47 value of evolvedCO2 is also affected by the availability of free water molecules.The amount of produced water increases with increasingsample size because the formation of one CO2 molecule alsogenerates one water molecule. In this respect, the sample sizeeffect expressed in Fig. 1 might represent a change in reactionkinetics. If the sample sizes are smaller than 7 mg, theavailability of free water molecules might be limited and CO2

formationmight start from different competing transition states,each of which is associated with a characteristic phosphoric acidreaction fractionation factor between carbonate and CO2. Oncewater becomes available, CO2 formation might instead proceedvia a unique transition state with a constant number of watermolecules being involved in the transition state. Such scenariosmight explain why the scatter of data for <7 mg samples islarger than that for >7 mg samples.The oxygen isotope fractionation factors between carbonate

and CO2 related to phosphoric acid digestion (1000lnaCO2-CaCO3)at 25�C are 10.25% for calcite and 10.57% for aragonite,respectively,[31] while Δ* =Δ47, CO2 – Δ47, CaCO3 is 0.232% forcalcite.[5] This factor has been determined on the internal’Caltech scale’. Applying the secondary reference frametransfer function of Dennis et al.,[19] this value becomes0.268% for the factor.[16] In any case, the oxygen isotopefractionation factor for carbonate digestion at 25�C is 1–2 ordersof magnitude larger than the corresponding fractionation inΔ47. It can, therefore, be expected that variations in Δ47 valuescaused by a change in the kinetic fractionation factor Δ* willbe accompanied by even more pronounced variations in d18Ovalues. However, Fig. 3 displays no such correlations. Inconclusion, it seems unlikely that the effect of sample size onΔ47 values at a reaction temperature of 25�C is caused by avariability in reaction kinetics.

163

Secondary re-equilibration

Secondary re-equilibration of evolved CO2 can be caused byheterogeneous oxygen isotope exchange with water, which


occurs relatively quickly even at room temperature. In the courseof this exchange reaction, the original 13C–18O bonds in the CO2

gaswill be broken and readjusted depending on the temperatureand the extent of exchange. If only traces of H2O are present (e.g.adsorbed on glass walls within which the CO2 is enclosed) massbalance constraints predict that a change in the Δ47 value can beobserved exclusively,whereas the change in the d18O value of theCO2 might go unnoticed. This is due to the much higherabundance of 12C16O18O relative to 13C16O18O isotopologues. Afully re-equilibrated gas will have a Δ47 value that correspondsto the temperature of exchange. The Δ47 value of CO2 gas inequilibrium at 25�C is 0.9252%.[2] In this respect, secondaryheterogeneous oxygen exchange between reaction intermediatesor evolvedCO2 andwater at 25�C can shift theΔ47 values of CO2

derived from NBS 19 or A. islandica towards higher thanoriginal values. The extent of re-equilibration may depend onseveral factors, such as time, difference between the originalΔ47 value of the sample CO2 and the Δ47 value of CO2 at there-equilibration temperature, the surface-to-volume ratio andthe CO2/H2O concentration ratio.

Generally, secondary oxygen isotope exchange can occurwithin the McCrea-type reaction vessel and/or during gaspurification, gas storage or mass spectrometric analyses.Heated gases were measured on a daily basis, passingthrough the same gas purification system as well as the samebellow/capillary/source of the mass spectrometer as thesample gases. If the elevated Δ47 values of sample CO2

extracted at 25�C are due to partial re-equilibration occurringduring purification and/or transfer to the ion source,elevated Δ47 values should also have been measured forheated gases. However, we did not notice systematicallyincreased Δ47 values for the heated gases, and the R2 valuesfor the heated gas lines were always better than 0.99.

It is worth noting that the evolved CO2 is collected intheheadspaceof the reactionvessel for 16 to 20h,until the reactionis complete.[3] During gas collection in the headspace, CO2 mightexchange oxygen with trace amounts of H2O adsorbed on theglass walls of the reactor. In addition, several authors havesuggested that traces of free water are present even at acidconcentrations >100% because of ongoing polymerization/depolymerization of phosphoric acid molecules.[32,33] Besides,according to reaction (4), the generation of one CO2 moleculeresults in the production of one H2O molecule. This water mightinteract either with the exsolving CO2, or with reactionintermediates such as H2CO3, according to Eqns. (5) and (6):

H218OþH2

16O-13C16O2 aqð Þ ¼ H216OþH2

18O-13C16O2 aqð Þ (5)

H218Oþ13C16O16O aqð Þ ¼ H2

16Oþ13C18O16O aqð Þ (6)

An important key factor for the extent of re-equilibrationmight be the residence time of CO2 in the acid and/or inthe headspace. Compared with 25�C, the reaction rates arefaster and acid viscosity is lower at 90�C. Moreover, usingthe common acid bath technique at 90�C, CO2 is continuouslyremoved from the acid by condensing it at liquid nitrogentemperature. As a consequence, the residence time of CO2 inthe acid is significantly lower at 90�C and the reaction as wellas CO2 removal from the acid is complete after 30 min. Wetested twice whether the effect of sample size on the Δ47

values observed for reactions at 25�C could be avoided if


9


1640

the exsolved CO2 was immediately removed from theheadspace by freezing. For this purpose, the McCrea-typereaction vessel was connected to our cryoextraction line andthe evolved CO2 was immediately collected for a period of12 h in a U-trap cooled with liquid nitrogen after passingthrough a water trap held at �80�C. We ensured the reactionwas complete after 12 h by continuously monitoring theamount of released CO2. For a small sample of NBS 19(4.4 mg), a relatively high Δ47 value of 0.378% was obtained(higher than any value measured for samples >7 mg),whereas for the large sample of A. islandica (8.8 mg) a lowΔ47 value of 0.687% was determined. Hence, although theresidence time of the analyte gas in the reaction vessel wasminimized, the results confirmed the trend of elevated Δ47

values for smaller sample sizes. The observation that animmediate removal of exsolved CO2 does not affect themeasured Δ47 values might imply that secondary re-equilibration of CO2 does not occur within the headspace ofthe McCrea-type reaction vessel but, rather, takes place beforethe exsolution of CO2 from the acid.The lower the abundance of free water in the phosporic

acid the lower are the reaction rates between carbonates andacid, especially at low temperatures.[29] If, as in our case,104% H3PO4 is used, comparatively small gas bubbles areformed during digestions at 25�C. If larger samples arereacted more bubbles are produced at the same time, whichcan then coalesce to larger bubbles. Due to their increasedbuoyancy, these larger CO2 bubbles can exsolve faster fromthe highly viscous H3PO4 than the smaller ones, therebyreducing the residence time of CO2 in the acid and, hence,the extent of secondary re-equilibration with water. As aconsequence, the measured Δ47 values for samples reactedat 25�C might depend on the sample size.Heterogeneous oxygen exchange between reaction

intermediates and/or dissolved CO2 and water might evenoccur at 90�C. CO2 that has fully equilibrated with water ata temperature of 90�C should be characterized by a Δ47 valueof 0.651%.[2,20] If partial re-equilibration were to occur at90�C, the Δ47 values measured for A. islandica should besystematically shifted to lower values than those obtainedfrom 25�C reactions. However, the contrary is observed: amean value of (0.692� 0.003)% (n = 8) is obtained fromreactions of >7 mg aliquots at 25�C, whereas the average of>7 mg aliquots reacted at 90�C occurs 0.020% higher at(0.712� 0.004)% (n = 4). In addition, no sample size effect isobserved at a reaction temperature of 90�C for either NBS19 or A. islandica (Fig. 1, Table 1). Therefore, we have noevidence that partial secondary re-equilibration occurs duringthe reactions at 90�C.Comparedwith reactions at 25�C, those at90�C are much faster with the largest proportion of evolvedCO2 being released and frozen within 5 min. Due to theenhanced reaction rate and lower acid viscosity at 90�C, CO2

Table 2. Differences in acid fractionation factors between 90�C

MineralogyΔ47 abs, 25�C,

4–14mgΔ47 abs, 90�C,

4–14mg

A. islandica aragonite 0.722 0.707NBS 19 calcite 0.373 0.359


easily exsolves from the continuously stirred acid. In addition,reaction intermediates decompose much faster at 90�C than at25�C. Thus, heterogeneous oxygen isotope exchange betweenCO2 or reaction intermediates and traces of free water mightno longer occur to a significant extent, resulting inindistinguishable mean Δ47 values being determined for smalland large sample sizes.

Difference in the acid fractionation factors between90�C and 25�C

In order to compare the Δ47 values from reactions at 90�Cwith those obtained at 25�C we applied the empiricallyderived difference in the fractionation factors Δ*25–90 of0.081%.[6] Our data can be used to address the difference inthe acid fractionation factors between carbonate digestionsat 25�C and 90�C independently. Applying the Δ*25–90 valueof 0.081%, our mean Δ47 values (4–14 mg) at a reactiontemperature of 25�C are still 0.015% (A. islandica, aragonite)and 0.014% (NBS 19, calcite), higher than the correspondingaverages obtained from 90�C reactions. This would implythat the Δ*25–90 values are 0.096% and 0.095% for aragoniteand calcite, respectively. Recently, an identical value of0.091% was determined for aragonite by Henkes et al.[16]

However, because <7 mg aliquots reacted at 25�C werepartly affected by secondary re-equilibration with water, itis more accurate to consider only the results obtained withsamples >7 mg for the calculation of the corresponding meanvalues (Table 2). At 90�C the complete data set can be used asno size effect is observed. In this case, reactions at 25�C givemean values that are 0.015% (aragonite) and 0.006% (calcite)lower than those obtained at 90�C. Therefore, we concludethat the Δ*25–90 value is not 0.081%, but 0.066% for aragoniteand 0.075% for calcite (Table 2). These differences infractionation factors are very close to the theoreticallypredicted value of 0.069%,[5] supporting our contention thataliquots <7 mg reacted at 25�C should not be considered forthe determination of Δ*25–90 values.

Implications for discrepant calibrations

Inorganic calcite precipitation experiments for the calibration ofthe carbonate clumped isotope (paleo)thermometer were firstcarried out by Ghosh et al.,[3] and later by Dennis and Schrag.[8]

In a plot ofΔ47 vs. 1/T2 both calibrations reveal distinct slopes,

even if the original Δ47 values are projected to theabsolute reference frame.[19] It is worth noting that theslope reported by Dennis and Schrag agrees very wellwith those theoretically predicted[5] and empirically derived(by analyzing mollusks and brachiopods with knowngrowth temperatures).[16] Henkes et al.[16] critically evaluatedmethodological, physicochemical and biological aspects, but

and 25�C; Δ47 and Δ47*25-90 in %

Δ47*25-90Δ47 abs, 25�C,

>7mgΔ47 abs, 90�C,

4–14mg Δ47*25-90

0.096 0.692 0.707 0.0660.095 0.353 0.359 0.075



found no convincing explanation why the calibration of Ghoshet al.[3] exhibits a higher temperature sensitivity than isdisplayed by their empirical calibration.Phosphoric acid digestions of carbonate samples for the

calibration datasets of Dennis and Schrag[8] and Henkeset al.[16] were performed at 90�C, whereas the reactions ofGhosh et al.[3] were carried out at 25�C. A closer inspectionof the calibration data of Dennis and Schrag[8] and Ghoshet al.[3] is provided in Dennis et al.[20] From their Fig. 3 itbecomes obvious that it is only the lowest temperatureprecipitate of the calibration of Ghosh et al.,[3] whose Δ47

value does not conform with the dataset reported by Dennisand Schrag.[8] Compared with the latter, it deviates to morepositive values. Ghosh et al.[3] stated that the sample sizesused for replicate analyses were around 5 mg. According toour study, this is exactly the critical range where the positivebias in Δ47 values appears at a digestion temperature of 25�C.Provided that the amount of precipiated calcite at low Twaslimited, replicates for clumped isotope analysis of the low Tprecipitate might have been split into smaller aliquots thanthose used for the analysis of the higher T precipitates. As aconsequence, their low T calibration data point would bebiased by a positive offset. Unfortunately, we are not able toprove whether the data used for the Ghosh et al. calibrationstudy[3] were affected by a sample size effect as they did notreport the exact sample sizes they used for replicate analyses.

CONCLUSIONS

A detailed comparison was carried out of carbonate clumpedisotope data obtained from two different acid digestiontechniques. For reactions at 25�C in McCrea-type vessels asystematic trend to lower Δ47 values with increasing samplesize is observed for both, A. islandica (aragonite) and NBS 19(calcite). In contrast, acid digestions performed at 90�C usinga common acid bath connected to an extraction line reveal nosuch trend of decreasing Δ47 values with increasing samplesize. Although we cannot unambiguously identify the sourceof elevated Δ47 values, our results imply that phosphoric acidreactions performed at 25�C can be accompanied by secondaryprocesses, partly altering the clumped isotopic composition ofthe carbonate-derived CO2. If unrecognized, such effects mightlead to erroneous results (positive bias inΔ47 values) especiallyif relatively small sample sizes are used. Therefore, laboratoriesusing small sample sizes should investigate the effect of samplesize specific for their own analytical conditions. Partial re-equilibration of CO2 extracted at a digestion temperature of25�C might help to explain the discrepancy in the slopesobtained for calibrations at reaction temperatures of 25�C[3]

and 90�C.[8]

SUPPORTING INFORMATION

Additional supporting information may be found in theonline version of this article.

164

AcknowledgementsThe present work was made possible through DFG grantFI-948-4/1. Sven Hofmann is thanked for technical supportand Tanja Rutz for analytical support.


REFERENCES

[1] J.M. Eiler. Paleoclimate reconstructionusing carbonate clumpedisotope thermometry. Quaterary Sci. Rev. 2011, 30, 3575.

[2] Z. Wang, E. A. Schauble, J. M. Eiler. Equilibriumthermodynamics of multiply substituted isotopologues ofmolecular gases. Geochim. Cosmochim. Acta 2004, 68, 4779.

[3] P.Ghosh, J. Adkins,H. P.Affek, B. Balta,W.Guo, E.A. Schauble,D. P. Schrag, J. M. Eiler. 13C–18O bonds in carbonate minerals: Anew kind of paleothermometer. Geochim. Cosmochim. Acta 2006,70, 1439.

[4] S. Zaarur, G. Olack, H. P. Affek. Paleo-environmentalimplication of clumped isotopes in land snail shells.Geochim. Cosmochim. Acta 2011, 75, 6859.

[5] W. Guo, J. L. Mosenfelder, W. A. Goddard III, J. M. Eiler.Isotopic fractionations associated with phosphoric aciddigestion of carbonate minerals: Insights from first-principlestheoretical modeling and clumped isotope measurements.Geochim. Cosmochim. Acta 2009, 73, 7203.

[6] B. H. Passey, N. E. Levin, T. E. Cerling, F. H. Brown, J. M. Eiler.High-temperature environments of human evolution in EastAfrica based on bond ordering in paleosol carbonates. Proc.Nat. Acad. Sci. 2010, 107, 11245.

[7] P. Ghosh, J. M. Eiler, S. E. Campana, R. F. Feeney. Calibrationof the carbonate ’clumped isotope’ paleothermometer forotoliths. Geochim. Cosmochim. Acta 2007, 71, 2736.

[8] K. J. Dennis, D. P. Schrag. Clumped isotope thermometry ofcarbonatites as an indicator of diagenetic alteration.Geochim. Cosmochim. Acta 2010, 74, 4110.

[9] A. K. Tripati, R. A. Eagle, N. Thiagarajan, A. C. Gagnon,H. Bauch, P. R. Halloran, J. M. Eiler. 13C–18O isotopesignatures and ’clumped isotope’ thermometry inforaminifera and coccoliths. Geochim. Cosmochim. Acta2010, 74, 5697.

[10] N. Thiagarajan, J. Adkins, J. M. Eiler. Carbonate clumpedisotope thermometry of deep-sea corals and implicationsfor vital effects. Geochim. Cosmochim. Acta 2011, 75, 4416.

[11] H. P. Affek, M. Bar-Matthews, A. Ayalon, A. Matthews,J. M. Eiler. Glacial/interglacial temperature variations inSoreq cave speleothems as recorded by ’clumped isotope’thermometry. Geochim. Cosmochim. Acta 2008, 72, 5351.

[12] M. Daëron, W. Guo, J. M. Eiler, D. Genty, D. Blamart,R. Boch, R. Drysdale, R. Maire, K. Wainer, G. Zanchetta.13C18O clumping in speleothems: Observations from naturalcaves and precipitation experiments. Geochim. Cosmochim.Acta 2011, 75, 3303.

[13] J. M. Eiler. ’Clumped-isotope’ geochemistry – The study ofnaturally-occurring, multiply-substituted isotopologues.Earth Planet. Sci. Lett. 2007, 262, 309.

[14] S. Finnegan, K. Bergmann, J. M. Eiler, D. S. Jones, D. A. Fike,I. Eisenman, N. C. Hughes, A. K. Tripati, W. W. Fischer. Themagnitude and duration of Late Ordovician–Early Silurianglaciation. Science 2011, 18, 903.

[15] E. A. Schauble, P. Ghosh, J. M. Eiler. Preferential formationof 13C–18O bonds in carbonate minerals, estimated usingfirst-principles lattice dynamics. Geochim. Cosmochim. Acta2006, 70, 2510.

[16] G. A.Henkes, B. H. Passey, A. D.Wanamaker Jr, E. L. Grossman,W. G. Ambrose Jr, M. L. Carroll. Carbonate clumped isotopecompositions of modern marine mollusk and brachiopod shells.Geochim. Cosmochim. Acta. 2013, in press.

[17] M. Wendeberg, J. M. Richter, M. Rothe, W. A. Brand. d18Oanchoring to VPDB: calcite digestion with 18O-adjustedortho-phosphoric acid. Rapid Commun. Mass Spectrom.2011, 25, 851.

[18] P. K. Swart, S. J. Burns, J. J. Leder. Fractionation of the stableisotopes of oxygen and carbon in carbon dioxide during the


1


1642

reaction of calcite with phosphoric acid as a function oftemperature and technique. Chem. Geol. 1990, 86, 89.

[19] J. M. McCrea. On the Isotopic Chemistry of carbonates anda paleotemperature scale. J. Chem. Phys. 1950, 18, 849.

[20] K. J. Dennis, H. P. Affek, B. H. Passey, D. P. Schrag, J. M. Eiler.Defining an absolute reference frame for ’clumped’ isotopestudies of CO2. Geochim. Cosmochim. Acta 2011, 75, 7117.

[21] T. W. Schmid, S. M. Bernasconi. An automated method for’clumped-isotope’ measurements on small carbonatesamples. Rapid Commun. Mass Spectrom. 2010, 24, 1955.

[22] P. Ghosh, M. Patecki, M. Rothe, W. A. Brand. Calcite-CO2

mixed into CO2-free air: a new CO2-in-air stable isotopereference material for the VPDB scale. Rapid Commun. MassSpectrom. 2005, 19, 1097.

[23] T. B. Coplen, C. Kendall, J. Hopple. Comparison of stableisotope reference samples. Nature 1983, 302, 236.

[24] D. A. Merritt, J. M. Hayes. Factors controlling precision andaccuracy in isotope-ratio-monitoring mass spectrometry.Anal. Chem. 1994, 66, 2336.

[25] K. W. Huntington, J. M. Eiler, H. P. Affek, W. Guo,M. Bonifacie, L. Y. Yeung, N. Thiagarajan, B. H. Passey,A. K. Tripati, M. Daëron, R. Came. Methods and limitationsof ’clumped’ CO2 isotope (Δ47) analysis by gas-sourceisotope ratio mass spectrometry. J. Mass Spectrom. 2009,44, 1318.


[26] L. J. Walters Jr, G. E. Claypool, P. W. Choquette. Reactionrates and d18O variation for the carbonate-phosphoricacid preparation method. Geochim. Cosmochim. Acta 1971,36, 129.

[27] J. M. Eiler, E. A. Schauble. 18O13C16O in Earth’s atmosphere.Geochim. Cosmochim. Acta 2004, 68, 4767.

[28] K. R. Liedl, S. Seku�sak, E. Mayer. Has the dimer of carbonicacid a lower energy than its constituents water and carbondioxide? J. Am. Chem. Soc. 1997, 119, 3782.

[29] T. Loerting, J. Bernard. Aqueous carbonic acid (H2CO3).Chem. Phys. Chem. 2010, 11, 2305.

[30] C. S. Tautermann, A. F. Voegele, T. Loerting, I. Kohl,A. Hallbrucker, E. Mayer, K. R. Liedl. Towards theexperimental decomposition rate of carbonic acid (H2CO3)in aqueous solution. Chem. Eur. J. 2002, 8, 66.

[31] S.-T Kim, A. Mucci, B. E. Taylor. Phosphoric acidfractionation factors for calcite and aragonite between 25and 75�C: Revisited. Chem. Geol. 2007, 246, 135.

[32] J.E. Such. Linear polyphosphoric acids, in Mellor’sComprehensive Treatise on Inorganic and Theoretical Chemistry,vol. VIII, suppl. III, (Eds: A. A. Eldridge, G. M. Dyson,A. J. E. Welch, D. A. Pantony). John Wiley, New York, 1971.

[33] E. A. Wachter, J. M. Hayes. Exchange of oxygen isotopes incarbon dioxide-phosphoric acid systems. Chem. Geol. 1985,52, 365.


Clumped isotope analysis of carbonates: comparison of two different acid digestion techniques

Documents

Transcript of Clumped isotope analysis of carbonates: comparison of two different acid digestion techniques