13 c analyses of calcium carbonate comparison between gb and ea

1. d13
C analyses of calcium carbonate: comparison between
the GasBench and elemental analyzer techniques
Grzegorz Skrzypek and Debajyoti Paul*
Department of Earth and Environmental Science, The University of Texas at San Antonio, One UTSA Circle, San Antonio, Texas 78249, USA
Received 1 June 2006; Revised 26 July 2006; Accepted 27 July 2006
Measurements of stable carbon isotopic composition (d13
C) of carbonates or carbonate-rich soils are
seldom performed in a continuous-flow isotope ratio mass spectrometer (IRMS) using an elemental
analyzer (EA) as an online sample preparation device. Such analyses are routinely carried out with an
external precision better than 0.1% using a GasBench II (GB) sample preparation device coupled
online with a continuous-flow IRMS. In this paper, we report and compare d13
C analyses (86 total
analyses) of calcium carbonates obtained by using both the GB and the EA. Using both techniques,
the d13
C compositions of two in-house carbonate standards (MERCK carbonate and NR calcite) and
ten selected carbonate-rich paleosol samples (of variable CaCO3 content) were analyzed, and data are
reported in the VPDB scale calibrated against international standards, NBS 18 and 19. For the in-
house standards analyzed by both techniques, a precision better than 0.08% is achieved. The
analytical errors (1s) computed from multiple analyses of the d13
C of both the MERCK and NR
obtained by the above two techniques are nearly identical. In general, the 1s (internal error) of
paleosol analyses obtained in the GB is better than 0.06%, whereas that for the analyses in the EA
(three repetitive analyses of the same sample) varies in the range 0.05–0.21%. However, for paleosols
having more than 85% CaCO3, 1s is better than 0.15% (similar to the instrument precision), and in this
case the d13
CVPDB of samples obtained by the GB is similar to that obtained by the EA. Our results
suggest that the d13
C of pure calcium carbonate samples can also be analyzed using the EA technique.
Copyright # 2006 John Wiley & Sons, Ltd.
The stable carbon isotope composition (d13
C) of calcium
carbonate (CaCO3) is widely used to infer paleoclimatic
conditions. Carbon and oxygen isotopic compositions of
carbonate-rich paleosols (ancient soils) are routinely used to
study climate/temperature variation in the past. Therefore,
d13
C analyses of calcium carbonates (henceforth referred to as
carbonates) have become an invaluable tool in climate
research.
The most commonly used technique to obtain accurate and
precise d13
C composition of carbonate utilizes a GasBench II
sample preparation/peripheral device that is coupled online
with a stable isotope ratio mass spectrometer (IRMS) in
continuous-flow mode.1,2
In the continuous-flow mode, the
sample gas is extracted by online chemistry, carried in a
stream of continuously flowing helium (He) gas and
introduced as a transient peak into the ionization chamber
of an IRMS.3
Similarly, the most common technique to obtain
bulk carbon isotopic composition utilizes an online coupling
of a combustion elemental analyzer (EA) with an IRMS via an
interface (ConFlo III), also in the continuous-flow mode.3,4
The difference between the above two techniques lies in the
mode of online chemistry and the passage of sample gases
through different analytical circuits until the analyte gases
are introduced into the IRMS.
In the Gasbench II (GB) peripheral device, sample CO2 gas
is extracted by the classical sample-acid reaction (CaCO3-
H3PO4) method. However, compared with the original
offline method reported by McCrea,5
which requires tens of
mg of sample to be reacted with a few mL of acid at 258C for
about 24 h, Revesz and Landwehr2
reported a modified
method (400 mg sample, 268C, 24–54 h reaction time) to
perform online chemistry and isotope ratio analyses (with an
external precision of 1s <0.1% for d13
C) by coupling the GB
peripheral device to the IRMS. Subsequently, Spötl and
Vennemann1
analyzed d13
C of sample CO2 gas extracted
from the reaction of 100–400 mg calcite samples with 100%
H3PO4 at a temperature of 728C for about 80 min per sample
and reported an external precision (1s) of 0.06%.
In the combustion EA peripheral device, the sample is
combusted quantitatively at an ambient temperature of
1700–18008C to produce CO2, N2 and H2O.6
These effluent
gases (after water removal) are carried in a stream of He and
introduced into the IRMS as transient peaks. Unlike the GB
technique, which allows determination of d13
C and d18
O of
sample CO2 gas during a single run, the EA technique only
allows determination of d13
C of sample CO2 during a single
RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 2006; 20: 2915–2920
Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rcm.2688
*Correspondence to: D. Paul, Department of Earth and Environ-
mental Science, The University of Texas at San Antonio, One
UTSA Circle, San Antonio, Texas 78249, USA.
E-mail: debajyoti.paul@utsa.edu; grzegorz.skrzypek@utsa.edu
Contract/grant sponsor: UTSA Faculty Research Award.
Copyright # 2006 John Wiley & Sons, Ltd.

2. run. However, precise d18
O analyses of sample can be
performed in an EA (using CO as the analyte gas) with the
high-temperature pyrolysis method7
(quantitative carbon
reduction method) instead of the oxidative combustion (CO2
analyte gas) method described earlier. In the C-reduction
mode, however, d13
C cannot be determined simultaneously
with d18
O. d13
C analyses performed in an EA coupled with a
ThermoFinnigan DeltaPlus
XP IRMS (EA/IRMS) give an
external precision of 0.15% (as stated by the manufacturer of
the IRMS), which is about twice as high as that achieved by
carbon isotopic analyses performed in the GB/IRMS.
This difference in precision of analyses between the GB/
IRMS and the EA/IRMS may be attributed to the mode of
transfer (along with different rate of He flow) of analyte gases
and the resulting fractionations in the respective analytical
circuits. However, in principle, ‘identical treatment’ of
international standards and samples that were analyzed
along with international standards should cancel out any
fractionation effect arising in the analytical circuit or inside
the IRMS.3
Therefore, the true d-value of samples relative to
the international standards can be measured with accuracy
as long as the nature and isotopic composition of samples are
nearly identical to that of the reference international
standards. Therefore, by analyzing carbonate samples
concurrently with international carbonate standards using
both the GB/IRMS and the EA/IRMS analytical technique, it
should be possible to obtain nearly identical d13
C compo-
sition of the samples irrespective of the techniques followed.
The EA/IRMS was developed for organic samples, and
has hitherto limited application to inorganic samples for
carbon. In particular, d13
C analyses of carbonates or
carbonate-rich soils are seldom performed using continu-
ous-flow EA/IRMS and to our knowledge such analyses
have not been reported. Furthermore, while d13
C values of
calcium carbonates in GB/IRMS represent exclusively the
carbon isotopic composition of the carbon in the (CO3)2
ion,
the EA/IRMS d13
C analyses are representative of the
carbon isotopic composition of the bulk carbonate (taking
into account both calcite and dolomite) and other sources of
carbon (mostly organic). It is the purpose of this research to
compare the accuracy and precision of d13
C of carbonates
analyzed using both the GB/IRMS and the EA/IRMS
technique. Carbon isotopic analyses were performed at the
Laboratory for Stable Isotope Geochemistry, University of
Texas at San Antonio (UTSA). The systems used consist of a
ThermoFinnigan GasBench II and a combustion elemental
analyzer (EA 4010, Costech Analytical Technologies Inc.,
Valencia, CA, USA) interfaced to a ThermoFinnigan DeltaPlus
XP isotope ratio mass spectrometer (IRMS) in continuous-
flow mode. In this study, the carbon isotopic composition
of two in-house calcium carbonate standards and ten
carbonate-rich paleosol samples were determined by using
both the GB/IRMS and the EA/IRMS, and a comparison of
accuracy and precision of analyses between the above two
techniques is presented.
ANALYTICAL PROCEDURE
All carbon isotopic compositions of samples in this study
(unless otherwise stated) are reported in the standard
d-notation in the VPDB (Vienna PeeDee Belemnite) scale
(d13
CVPDB). d13
CVPDB is defined as the relative difference, in
parts per thousand (%), between the isotope ratio of the
sample and the VPDB carbonate standard (established by the
International Atomic Energy Authority, IAEA, Vienna) as
per Eqn. (1):
d13
CVPDBðsampleÞ¼
ð13
C=12
CÞsample
ð13C=12CÞstandard;VPDB
1
!
103
% (1)
GasBench II (GB)/IRMS technique
In the Gasbench II (GB) technique, fine grained (or
powdered) carbonate samples (200–400 mg) were weighed
into 10 mL Na-glass vials (Labco Limited, High Wycombe,
Houston, TX, USA) sealed with butyl rubber septa. After
flushing of vials for 700 s with high purity He (in order to
remove air contained in the vials), samples were reacted with
about 0.2 mL of 100% orthophosphoric acid (Merck,
Darmstadt, Germany) at 728C. One hour after the initiation
of the sample-acid reaction, the extracted CO2 gas was sam-
pled by a specially designed sampling needle where the flow
of He carrier gas pushes the gas mixture though the needle
hole into the VALCO multi-injection loop system. The gas
then passes through a gas chromatography (GC) column,
an active open-split, and, finally, through a capillary it is
introduced into the ionization chamber of the IRMS. During
each analysis, reference CO2 gas is injected three times
followed by ten injections of sample gas. The raw d13
C values
of ten individual sample peaks are calculated with respect to
the d value of the second reference gas injection (d13
C ¼ 0, by
convention), and the average and standard deviation
(internal error) of these ten measurements are computed.
This procedure has been reported by Paul and Skrzypek8
and
it is similar to the one reported by Spötl and Vennemann.1
Elemental analyzer (EA)/IRMS technique
In the EA/IRMS technique, the sample is instantaneously
melted and cracked by thermal treatment, oxidized (in the
presence of O2) and converted into homogenous combustion
gases (CO2, N2 and H2O) in amounts stoichiometrically
equivalent to its elemental components in the sample.6
In this
study we used a CHNS EA, which is coupled to the DeltaPlus
XP IRMS via a universal elemental analyzer interface
(Finnigan ConFlo III) for online carbon and nitrogen isotope
ratio analyses. The analytical circuit of the EA comprises of a
combustion reactor (Cr2O3 catalyst þ Co3O4 coated with
silver), a reduction reactor (reduced copper wire, 0.7 mm
diameter) and a GC column.
About 200–300 mg of carbonate samples were weighed into
tin (Sn) capsules and placed in the zero blank autosampler
(electric), followed by purging with He for 5 min. The Sn
capsules were dropped into the combustion reactor (main-
tained at 10208C) about 1–2 s after the arrival of oxygen to
ensure that enough oxygen is available for complete
combustion. The oxidation of Sn (exothermic reaction)
accelerates the breakdown and combustion of sample by
increasing the reaction temperature from 10208C to 1700–
18008C; as a result a mixture of combustion gases (CO2, N2
and H2O) is produced. In the case of a pure carbonate there
Copyright # 2006 John Wiley Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2915–2920
DOI: 10.1002/rcm
2916 G. Skrzypek and D. Paul

3. are negligible amounts of N2 present in the sample. The gas
mixture (in a stream of He) then passes through the reduction
reactor (maintained at 6508C), where copper wires scavenge
any excess oxygen not used during combustion. A water
removal trap filled with anhydrous magnesium perchlorate
[Mg(ClO4)2] removes H2O present in the mixture. Finally, the
mixture passes through the 3 m long GC column (maintained
at 558C), where N2 and CO2 peaks are separated and carried
to the Conflo III interface. During a typical analysis
(10 min), a continuous flow of He is maintained at a rate
of 100 mL/min and a msemi (5 mL volume) oxygen dosing
loop is used which is filled with a constant O2 flow rate of
25 mL/min (at 1 bar). This 5 mL of oxygen is introduced
into the EA for sample oxidation.
The ConFlo III interface reduces the He flow rate of
100 mL/min in the EA to 0.3 mL/min before it is
introduced into the ion source of the IRMS.4
This device
consists of two active open-split cells, one for unknown
samples and the other for reference gases. The sample open-
split cell comprises three fused-silica capillaries; one (320 mm
diameter) delivers the sample gas from the EA into the cell,
the second dilutes concentrated samples if required, and the
third (the sample delivery capillary, 100 mm diameter)
samples and transfers the analyte gas to the ionization
chamber of the IRMS at a flow rate of 0.3 mL/min. Reference
CO2 gas (from laboratory cylinders) is transferred through a
separate open-split system into the IRMS. To prevent
ambient air from contaminating the open-split cells, a
continuous flow of He is maintained in the cells at a
pressure of 0.9 bar. The pressure of CO2 (reference gas tank)
at the Conflo III is maintained at 1.5 bar which results in 4.4 V
signal for m/z 44 (12
C16
O16
O). During both the GB and EA
analyses, high-purity gases are used: 99.9999% He, 99.995%
CO2, 99.999% O2.
Carbonate standards and paleosol analyses
d13
C analyses of IAEA carbonate standards (NBS 18 and 19),
carbonate samples (two in-house standards MERCK and
NR) and carbonate-rich paleosols were performed in four
separate sequences using both the GB and the EA. NBS 18, a
carbonatite (d13
CVPDB ¼ 5.00%), and NBS 19, a limestone
(d13
CVPDB ¼ 1.95%), were analyzed as supplied. The in-house
standard MERCK (Merck Co., Whitehouse Station, NJ,
USA), SUPRAPUR CaCO3 (purity 99.95%), was analyzed in
fine powder form. The other in-house standard, NR, is a fine
powder of high-purity calcite crystals collected from the
ultramafic rock veins in Gogolow-Jordanow Massif, SW
Poland. The carbonate-rich paleosol (ancient soil) samples
were collected from a well preserved and 35 m thick paleosol
unit in the Hensel Formation of Cretaceous age located in
Gillespie and Kimble Counties, Texas, USA.9
In one set of runs, d13
C compositions of eight MERCK and
eight NR samples were obtained using the GB. In the other
set, 12 samples each from both MERCK and NR were
analyzed for d13
C using the EA. Of the 40 paleosol isotopic
analyses performed in the GB at UTSA (unpublished data),
ten samples of variable (60–100%) CaCO3 compositions
were selected for analysis in the EA. During the analyses
using the GB, ten MERCK samples were also run simul-
taneously with these paleosol samples. The paleosols (three
separate analyses of each sample) were analyzed in the EA
along with two MERCK samples. During each set of
analyses, two replicates each of both NBS 18 and 19 were
also run simultaneously. In total 86 samples were analyzed
along with 23 analyses of IAEA carbonate standards. The raw
d values (relative to the reference CO2 tank value) of
unknown samples during a particular sequence are normal-
ized against international standards NBS 18 and 19 (also
analyzed in the same sequence) and reported in the VPDB
scale by using a two-point linear calibration as suggested by
Coplen et al.10
A zero enrichment of carbon (standard on/off) test was
performed both for the GB and for the EA; the variation in
d13
C of nine on-off pulses of CO2 reference (tank) gas in both
cases is given by 1s ¼ 0.02%. A linearity test for carbon was
also performed for both techniques where the pressure of
CO2 reference (tank) gas was increased in increments during
each analysis of eight pulses of CO2. The resulting linear
regression of d13
C vs. amplitude of m/z 44 (varying from 2.3 to
9.7 V) for both techniques gives a slope of 0.04%/V. We
also performed a blank test for the EA where, prior to
analyzing real samples, two empty Sn capsules were
analyzed and the m/z 44 signal was found to be 2.3 mV.
Before analyzing carbonates in the EA, we analyzed an
organic in-house standard, acetanilide, to monitor the
performance of the EA. The average d13
CVPDB of 25
acetanilide analyses (calibrated against organic IAEA
standards: NBS 22 and USGS 24) of variable weight (200–
300 mg) was 30.25% with a reproducibility of 1s ¼ 0.11%.
RESULTS AND DISCUSSION
The reproducibilities of raw d13
C composition of inter-
national standards NBS 18 and 19 analyzed simultaneously
with in-house standards and paleosols using both the GB
and the EA are presented in Fig. 1. The 1s variation of
seven NBS 18 measurements in the GB (sNBS18-GB ¼ 0.05%)
is slightly less than that in the EA (sNBS18-EA ¼ 0.08%,
n ¼ 5). The 1s variation of six NBS 19 measurements in
the GB (sNBS19-GB ¼ 0.02%) is less than that in the EA
(sNBS19-EA ¼ 0.06%, n ¼ 5). These 1s variations are similar
to the previously reported2
variations using the GB [0.09%
(n ¼ 161) for NBS 18 and 0.11% (n ¼ 182) for NBS 19]. These
results suggest high homogeneity and reproducibility of the
NBS 19 standard compared with the NBS 18, and high
reproducibility for the analyses performed using the GB
compared with the EA. Furthermore, in the analyses
performed using the GB, the difference between the
(average) raw d13
C of NBS 19 and 18 is 6.99%, which is
remarkably similar to the difference between the true
d13
CVPDB of NBS 19 and 18 of 6.95%. However, in the
analyses performed using the EA, the difference between
the (average) raw d13
C of NBS 19 and 18 is 6.69%, which is
0.26% away from the difference in true values.
It is evident that the raw d13
C value of IAEA standards is
consistently higher for analyses using the GB than for those
using the EA (Fig. 1). This may be attributed to variable
fractionation arising from different conversion pathways.
However, because of the ‘identical treatment’3
of sample and
standard, the d13
CVPDB of an unknown sample obtained by
Copyright # 2006 John Wiley Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2915–2920
DOI: 10.1002/rcm
Comparison of d13
C analyses of CaCO3 2917

4. using the GB should be close (within analytical uncertainty)
to that obtained using the EA.
A comparison of d13
CVPDB analyses of both the MERCK
and NR obtained using the GB and the EA is presented in
Table 1 and shown in Fig. 2. The average d13
C of 18 MERCK
analyses (each sample analysis represents an average of ten
injections of the same sample gas) in the GB is
35.65 0.08%, which is strikingly similar to the average
(d13
CEA, MERCK ¼ 35.62 0.07%) of 14 MERCK analyses in
the EA (Table 1). Furthermore, using both the EA and the GB
no correlation (0.03%/V and R2
0.2) is observed between
the d13
C of MERCK and the signal intensity of m/z 44, though
the intensity of m/z 44 varied in the range 3.9–6.9 V (area all
18–25 V.s; area all represents sum of peak areas of m/z 44, 45
and 46) in the GB and 6.6-8.7 V (area all 152.4–199.5 V.s) in
the EA.
The average d13
CVPDB of eight NR analyses in the GB is
4.76 0.07%, whereas the average of ten NR analyses in the
EA is 4.91 0.06% (Table 1). No significant linearity
(0.05%/V) is observed between the d13
C of NR analyses
and the signal intensity of m/z 44 for either the EA or the GB,
though the intensity of m/z 44 varied in the range 3.5-6.2 V
(area all 13-24 V.s) in the GB and 5.4–6.6 V (area all 132.4–
156.9 V.s) in the EA. Unlike the MERCK analyses in the GB
and the EA, which are virtually indistinguishable, slight
differences do exist between the NR analyses in the GB and
the EA (Fig. 2). For the NR sample, the difference between the
d13
C using the GB and that of the EA (D13
CGB-EA) is 0.15%
(Table 1). However, the magnitude of 1s precision (0.07%)
associated with these measurements suggests no significant
differences between the measured carbon isotopic compo-
sition of NR using the GB and the EA. These results suggest
that pure carbonate and/or calcite analyses may also be
performed using the EA.
The above conclusion is also supported by the results
obtained for carbonate paleosol samples. The d13
CVPDB of
paleosol samples analyzed using both the GB and EA are
presented in Table 2 and Fig. 3. The percentages of CaCO3 in
these paleosols were determined from the linear relationship
(R2
¼ 0.99) between the sample weight of pure carbonate
standard (e.g., NBS 19 or 18) and the respective signal
intensity of m/z 44 obtained during analyses using the GB.
For our purpose, these estimates seem adequate. The 1s
(internal error) of analyses using the GB varies in the range
0.04–0.06% (except for HPS21, 0.09%), which is in the
acceptable range (about 0.08 %) for analyses performed in
Figure 1. Reproducibility of raw d13
C (relative to the laboratory CO2 reference
tank value) of international carbonate standards NBS 18 and NBS 19 analyzed on
different days and with different sets of unknown samples using both the GB and
EA techniques. Dashed lines represent averages of all the analyses of the
respective standards whereas the solid lines represent 1s deviations.
Table 1. Comparison of d13
CVPDB of MERCK (carbonate) and NR (calcite) analyzed using the GasBench II (GB) and the
elemental analyzer (EA). The number of replicate analyses of the same sample is given by n, and the average and standard
deviation (1s) of these replicate analyses are reported. Difference between d13
C of the GB and EA analyses is given by D13
CGB-EA
Sample
name
GB EA
D13
CGB-EA
[%]
Wt.
[mg]
Mass 44
[mV]
d13
C
[% VPDB]
1s
[%] n
Wt.
[mg]
Mass 44
[mV]
d13
C
[% VPDB]
1s
[%] n
MERCK 200–390 3938–6788 35.65 0.08 18 200–265 6650–8724 35.62 0.07 14 0.03
NR 184–331 3518–6163 4.76 0.07 8 203-260 5370–6655 4.91 0.06 10 0.15
Copyright # 2006 John Wiley Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2915–2920
DOI: 10.1002/rcm
2918 G. Skrzypek and D. Paul

5. the GB. For analyses carried out using the EA, the reported
d13
CVPDB of a sample represents an average of three separate
analyses of the same sample. The 1s for different samples
varies in the range 0.05–0.21%. We observed that the 1s of
multiple analyses of the same sample in the EA becomes
0.1% (maximum allowable 1s ¼ 0.15% is suggested by the
equipment manufacturer), when the weights of the samples
are close to each other (within 25 mg). This is always true
for impure carbonate samples (as also observed during
analyses of the in-house acetanilide standard). However, we
have not selectively excluded any analysis. Furthermore, we
believe that during the analyses using the EA all the paleosol
samples are completely combusted. Based on the linearity
between weight and intensity of m/z 44 in all the MERCK
analyses performed in the EA, about 287 mg of MERCK
would generate a m/z 44 intensity of 9.6 V (area
all ¼ 216.8 V.s), similar to the intensity (9.8 V, and area
all ¼ 221.6 V.s) observed for the highest amount of paleosol
sample analyzed (287 mg, HPS20).
Figure 2. Comparison of d13
CVPDB of MERCK (carbonate) and NR (calcite)
analyzed using the GB and EA techniques. Closed symbols represent only
samples run separately with external standards, whereas open symbols represent
samples analyzed as unknowns simultaneously with paleosols. Respective
averages (dashed lines) and 1s external errors (solid lines) are computed from
all the data shown for respective samples.
Figure 3. Comparison of d13
CVPDB of paleosol samples
between GB and EA analyses. 1s error bars are shown for
analyses obtained by respective techniques. Each GB
analysis represents an average of 10 loop injections of the
sample CO2 gas, whereas each EA analysis represents an
average of three multiple runs of the sample. Data shown are
also presented in Table 2. Data for three samples, HPS15,
HPS19 and HPS21, are not shown in this figure.
Table 2. Comparison of d13
CVPDB of ten paleosol samples analyzed using the GasBench (GB) and the elemental analyzer (EA).
For the GB analyses, the average d13
C and standard deviation (1s) is computed from d13
C values of ten loop injections of the same
sample gas, whereas, for the EA analyses, the average and 1s for each sample are computed from three separate analyses of the
same sample
Sample
Depth
[m]
CaCO3
[%]
GB EA
D13
CGB-EA
[%]
Wt.
[mg]
Mass 44
[mV]
d13
C
[% VPDB]
1s
[%]
Wt.
[mg]
Mass 44
[mV]
d13
C
[% VPDB]
1s
[%]
HPS13 6.1 100 252 5813 5.45 0.04 212–255 7178-7978 5.46 0.09 0.01
HPS14 6.2 96 244 5388 5.08 0.06 231–276 6975–8260 5.18 0.13 0.10
HPS15 6.3 60 207 2880 5.20 0.06 210–278 5929–8289 5.69 0.21 0.49
HPS16 6.4 85 293 5706 5.42 0.05 244–277 6572–7558 5.60 0.13 0.17
HPS17 6.5 100 244 5632 5.07 0.06 235–264 7254–8225 5.21 0.15 0.14
HPS18 6.6 97 288 6418 4.89 0.05 225–277 6842–8596 5.03 0.17 0.13
HPS19 6.71 80 221 4083 4.98 0.05 219–250 6599–7393 5.33 0.11 0.35
HPS20 6.78 98 266 6009 5.70 0.05 214–287 6950–9793 5.71 0.20 0.01
HPS21 6.81 72 234 3875 4.99 0.09 233–283 6066–7399 5.66 0.11 0.67
HPS22 6.91 77 225 3999 5.17 0.06 252–270 7806–8460 5.27 0.05 0.11
Copyright # 2006 John Wiley Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2915–2920
DOI: 10.1002/rcm
Comparison of d13
C analyses of CaCO3 2919

6. The calculated difference between the d13
CVPDB of paleosol
samples analyzed using the GB and the EA (D13
CGB-EA)
varies in the range 0.01–0.17%, excluding three samples
(HPS15, HPS19 and HPS21). The D13
CGB-EA values for
HPS15, HPS19 and HPS21 are 0.49, 0.35 and 0.67%,
respectively, all of which exceed the analytical precision of
the EA (1s ¼ 0.15%). However, all these samples also contain
CaCO3 80%, the lowest amongst the paleosols (Table 2).
Since combustion in the EA would convert any form of
carbon present in the sample, we attribute the much higher
D13
CGB-EA (beyond analytical uncertainty) observed for these
samples to contamination with mostly organic carbon. When
the percentage of CaCO3 is in the range 85–100%, the d13
C of
paleosols analyzed by the GB is similar to that analyzed by
the EA (linear regression of d13
CVPDB of the GB vs. EA results
in R2
¼ 0.96), given the 1s error associated with the analyses.
It is evident in Fig. 3 that the trend observed for d13
CGB
variation with depth is similar to that observed for d13
CEA,
suggesting that carbon isotopic data of carbonates analyzed
using the EA can be interpreted in a meaningful way.
Furthermore, the carbonate analyses performed in the GB
and the EA may also be useful for mass balance studies.
Assuming that the d13
C of a sample analyzed in the EA (dT,
which is same as dEA) is the result of mixing of organic carbon
(dOC) and inorganic carbon (dIO, which is same as dGB
assuming that all the inorganic carbon is in the form of
CaCO3 which reacts with the acid completely), the mass
balance equation (Eqn. (2)) can be given as:
foc ¼
dT dIO
dOC dIO
¼
dEA dGB
dOC dGB
(2)
where fOC is the fraction of organic carbon in the sample and
is related to the fraction of inorganic carbon (fIC) in the
sample so that fOC þ fIC ¼ 1. In Eqn. (2), however, both fOC
and dOC are unknown; therefore, one can only be computed
given the other. For the paleosol samples, if we assume that
the average d13
C of organic carbon is 27% (average for C3
plants), then, from Eqn. (2), the organic carbon content for the
three samples with lowest CaCO3 content is in the range 1.5–
3%, probably a realistic value. On the other hand, assuming
an average d13
C of organic carbon to be 13% (average for C4
plants), the organic carbon content for the same samples is
computed in the range 4.5–8.5%.
CONCLUSIONS
Stable carbon isotopic analyses (d13
C) of carbonates are
routinely performed using an online Gasbench (GB)
sample preparation device coupled to a continuous-flow
isotope ratio mass spectrometer (IRMS). Such carbonate
analyses are rarely performed using an elemental analyzer
(EA) sample preparation device that is also coupled to a
continuous-flow IRMS. The GB technique relies on the
extraction of sample CO2 gas (that carries the d13
C signature)
from the carbonate sample by using the acid-sample
reaction method, whereas the EA technique relies on
quantitative combustion of the sample to produce sample
CO2 gas. In this paper, we compared d13
C analyses (86 total
analyses performed comprising international standards
NBS 18 and 19, in-house standards MERCK carbonate
and Calcite NR, and ten carbonate-rich paleosol samples)
of carbonates obtained by using both the GB and the EA.
The reproducibility (1s of multiple analyses of the same
sample) of d13
C of in-house carbonate standards (MERCK
and NR) analyzed using both the GB and the EA is better
than 0.08%, which is in the range of instrument precision
suggested by the manufacturer. The carbon isotopic
composition of these two in-house standards obtained by
using the GB is nearly identical to that obtained using the EA.
This suggests that with proper sample handling (‘identical
treatment’ of unknown and international carbonate stan-
dards) and repetitive analyses of pure carbonate samples, it
is possible to perform d13
C analyses of carbonates using the
EA, and that accuracy and precision similar to those obtained
using the GB can be achieved.
However, the reproducibility (1s of three repetitive
analyses of the same sample) of d13
C analyses of paleosols
analyzed using the EA varied with the percentage of CaCO3
in the samples. In general, for samples having 80% CaCO3,
the differences between d13
C from analyses using the GB and
the EA (D13
CGB-EA) and the 1s deviations are better than
0.15%. However, D13
CGB-EA values for samples with lower
percentage (80%) of CaCO3 are much higher, beyond the
analytical uncertainties. We attribute these higher values to
contamination of the isotopic ratios of inorganic carbon with
that of the organic carbon present in these samples. If the
paleosol samples are expected to contain higher amounts of
organic carbon, then in order to determine the d13
C of
calcium carbonate present in paleosol samples using the EA,
an additional analytical (addition of an oxidant, e.g., H2O2,
to oxidize the organic carbon) step may be performed
to remove organic carbon contained in the sample prior to
isotope ratio analyses.
Acknowledgements
This research was supported by a UTSA Faculty Research
Award granted to D. Paul. Thanks are due to Zaneta Martinez
for help during weighing of samples. We are grateful
to Chuck Douthitt of ThermoElectron and two anonymous
reviewers for their insightful suggestions to improve the
clarity of this manuscript.
REFERENCES
1. Spötl C, Vennemann TW. Rapid Commun. Mass Spectrom.
2003; 17: 1004.
2. Revesz KM, Landwehr JM. Rapid Commun. Mass Spectrom.
2002; 16: 2102.
3. Werner RA, Brand WA. Rapid Commun. Mass Spectrom. 2001;
15: 501.
4. Werner RA, Bruch BA, Brand WA. Rapid Commun. Mass
Spectrom. 1999; 13: 1237.
5. McCrea JM. J. Chem. Phys. 1950; 18: 849.
6. Pella E. Am. Lab. 1990; 22: 116.
7. Gehre M, Strauch G. Rapid Commun. Mass Spectrom. 2003; 17:
1497.
8. Paul D, Skrzypek G. Rapid Commun. Mass Spectrom. 2006; 20:
2033.
9. Jones JO, Kullman JA. Bull. South Texas Geol. Soc. 1999; 60:
13.
10. Coplen TB, Brand WA, Gehre M, Groning M, Meijer HAJ,
Toman B, Verkouteren RM. Anal. Chem. 2006; 78: 2439.
Copyright # 2006 John Wiley Sons, Ltd. Rapid Commun. Mass Spectrom. 2006; 20: 2915–2920
DOI: 10.1002/rcm
2920 G. Skrzypek and D. Paul