stable carbon isotope analysis of dissolved inorganic carbon (dic) and dissolved organic carbon...

Research Article

Received: 17 March 2013 Revised: 23 May 2013 Accepted: 21 June 2013 Published online in Wiley Online Library

Rapid Commun. Mass Spectrom. 2013, 27, 2099–2107

Stable carbon isotope analysis of dissolved inorganic carbon(DIC) and dissolved organic carbon (DOC) in natural waters –Results from a worldwide proficiency test

Robert van Geldern1*, Mahendra P. Verma2, Matheus C. Carvalho3, Fausto Grassa4,Antonio Delgado-Huertas5, Gael Monvoisin6 and Johannes A. C. Barth1

1GeoZentrum Nordbayern, Applied Geosciences, Friedrich-Alexander-University Erlangen-Nuremberg, Schlossgarten 5, 91054Erlangen, Germany2Geotermia, Instituto de Investigaciones Eléctricas, Reforma 113, Col. Palmira, Cuernavaca, Mor. C.P. 62490, Mexico3Centre for Coastal Biogeochemistry Research, Southern Cross University, Lismore 2480, NSW, Australia4Istituto Nazionale di Geofisica e Vulcanologia Sezione di Palermo, Via Ugo La Malfa 153, 90146 Palermo, Italy5Laboratorio de Biogeoquímica de Isótopos Estables, Instituto Andaluz de Ciencias de la Tierra IACT(CSIC-UGR), Avda. de lasPalmeras 4, 18100 Armilla, Granada, Spain6Laboratoire Interactions et Dynamiques des Environnements de Surface, Bâtiment 504, Université Paris Sud, 91405Orsay, France

RATIONALE: Stable carbon isotope ratios of dissolved inorganic (DIC) and organic carbon (DOC) are of particularinterest in aquatic geochemistry. The precision for this type of analysis is typically reported in the range of 0.1‰ to0.5‰. However, there is no published attempt that compares δ13C measurements of DIC and DOC among differentlaboratories for natural water samples.METHODS: Five natural water samples (lake water, seawater, two geothermal waters, and petroleum well water) wereanalyzed for δ13CDIC and δ13CDOC values by five laboratories with isotope ratio mass spectrometry (IRMS) in aninternational proficiency test.RESULTS: The reported δ13CDIC values for lake water and seawater showed fairly good agreement within a range of about1‰, whereas geothermal and petroleum waters were characterized by much larger differences (up to 6.6‰ betweenlaboratories). δ13CDOC values were only comparable for seawater and showed differences of 10 to 21‰ for other samples.CONCLUSIONS: This study indicates that scatter in δ13CDIC isotope data can be in the range of several per mil forsamples from extreme environments (geothermal waters) and may not yield reliable information with respect todissolved carbon (petroleum wells). The analyses of lake water and seawater also revealed a larger than expecteddifference and researchers from various disciplines should be aware of this. Evaluation of analytical procedures of theparticipating laboratories indicated that the differences cannot be explained by analytical errors or different datanormalization procedures and must be related to specific sample characteristics or secondary effects during samplestorage and handling. Our results reveal the need for further research on sources of error and on method standardization.Copyright © 2013 John Wiley & Sons, Ltd.

(wileyonlinelibrary.com) DOI: 10.1002/rcm.6665

Dissolved carbon in natural waters is of particular interest ingeological, biological, and environmental studies. In aquaticgeochemistry the concentration measurements of carbonspecies (H2CO3, HCO3

– and CO32–) are a well-established and

widely used tool, which allows the evaluation of thebuffering capacity of water (alkalinity and acid neutralizingcapacity). These analyses are performed mainly by acid-basetitrations and provide information on the suitability of waterfor its different uses, efficiency of wastewater treatment,anthropogenic pollution and ecosystem health.[1,2] In addition

* Correspondence to: R. vanGeldern, GeoZentrumNordbayern,AppliedGeosciences, Friedrich-Alexander-University Erlangen-Nuremberg, Schlossgarten 5, 91054 Erlangen, Germany.E-mail: [email protected]

Rapid Commun. Mass Spectrom. 2013, 27, 2099–2107

209

to species distribution the source of carbon has also receivedincreasing interest in several geological and ecologicalstudies.[3–5] The stable carbon isotope ratio (δ13C value) ofdissolved inorganic (DIC) and dissolved organic carbon(DOC) provides a natural label, which indicates the originand migration pathway of the carbon sources.[6–12]

Numerous stable isotope laboratories worldwide conductδ13CDIC and δ13CDOC measurements on a routine basis. Atpresent, no soluble international isotope reference materialis available for δ13C analyses of DIC and DOC. Despite thismajor drawback, this type of analysis is considered as astandard analytical technique. This might be true as long asdry pure chemicals, which are dissolved in deionized waterby the participating laboratories, are compared during aproficiency test. The analytical results from such freshlyprepared synthetic solutions might identify some analytical

Copyright © 2013 John Wiley & Sons, Ltd.

9

R. van Geldern et al.

2100

problems but will not provide any information about otheraspects that are valid for real natural water with realbacteriological and geochemical background matrices fromdifferent environments. The samples of this study weretherefore selected to represent realistic every-day samples inenvironmental research studies.In contrast to water stable isotope ratios (δ18O and δ2H

values)[13] no published attempt exists that compares δ13Cmeasurements of DIC and DOC among different laboratories.This study presents the results of a first inter-laboratorycomparison study for the determination of carbon stableisotope ratios of DIC (δ13CDIC) and DOC (δ13CDOC). Theexercise was announced through the Isogeochem mailing liston 30 April 2012[14] and involved five laboratories fromdifferent countries that reported values for δ13CDIC; two ofthem also reported values for δ13CDOC.

EXPERIMENTAL

Water samples often originate from various environments suchas freshwater, seawater, or brines. Therefore, five natural watersamples were selected for this study instead of syntheticsamples prepared from deionized water and chemicals(Table 1). There are no reference δ13C values that are regardedas correct (or true) for the samples as the key interest of thisstudy was on the relative differences between the laboratories.To ensure correct data normalization of the isotope ratio valuesto the Vienna Pee Dee Belemnite Scale (VPDB) everyparticipant reported details on its international and in-housereference materials and calibration procedures (Table 2).For the δ13C inter-laboratory comparison, the samples

were distributed among the participating laboratoriesin 125 mL Nalgene (high-density polyethylene, HDPE)bottles. The new bottles were rinsed with deionized waterand the sample before filling. No preservatives (e.g. HgCl2)were added. Samples were filtered by a laboratory vacuumfiltration unit (pore size 0.45 μm; Merck Millipore,Billerica, MA, USA) into a large container, homogenizedand subsequently filled into the individual bottles forshipping. All samples are natural waters collected fromdifferent environments with different DIC and DOCconcentrations. These water samples were also analyzedfor their carbon species distribution during anotherinternational proficiency test on species concentrations bychemical laboratories (M. P. Verma and colleagues, 2013,personal communication).

Table 1. Natural water samples that were distributed forδ13C analyses and corresponding pH values (±1σ) asdetermined by an international proficiency test on carbonspecies distribution (M. P. Verma and colleagues, 2013,personal communication)

Sample Water type pH

IIE30 lake water 8.29±0.15IIE33 geothermal water 7.38±0.11IIE34 geothermal water 7.17±0.26IIE35 seawater 7.76±0.23IIE36 petroleum well water 6.72±0.54

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

Each laboratory reported its methodology and theinstrumentation that was used to determine δ13CDIC andδ13CDOC values and DIC concentrations (Table 2). Inprinciple, DIC is extracted from the solution by addingphosphoric acid that converts the different carbon speciesinto CO2 that is released into the headspace of the samplevessel (e.g. by agitation). Subsequently, the CO2 is analyzedfor its stable carbon isotope ratio by gas isotope ratio massspectrometry (IRMS). This method also allows the amountof released CO2 to be quantified either from the isotoperatio chromatogram[15] or by a separate detector.[16] Theconcentration of DIC in the samples can then bedetermined by a calibration with standards of knownconcentration prepared from DIC-free water and a readilysoluble carbonate (generally sodium bicarbonate).

In this study, four participating laboratories used IRMSsystems in continuous flow mode with helium carrier gas,and one used offline sample preparation and IRMS analysiswith a dual inlet technique (high vacuum sample inlet port)(Table 2). Laboratories no. 1 and 2 used a Gasbench II device(Thermo Fisher Scientific GmbH, Bremen, Germany) for theconversion of the DIC into CO2.

[17] This device uses 12 mLLabco exetainer® (Labco Ltd, Lampeter, UK) with gas-tightbutyl rubber septa that are pre-loaded with a few drops of100% phosphoric acid, capped and filled with helium. Thesamples were transferred from the Nalgene bottles to thevials by disposable syringes. A comparable transfer techniquewas used by laboratory no. 4 with the difference that theyused smaller sample vials (5.9 mL) and a 10 mL gas-tightsyringe with stopcock (Hamilton Bonaduz AG, Bonaduz,Switzerland). Laboratory no. 3 filled the samples into 40 mLglass vials that meet the standards of the US EnvironmentalProtection Agency, so-called EPA vials, which weresubsequently capped by pierceable caps with septa. The vialswere filled gently to avoid air bubbles and to the brim withno headspace. The sample were transferred from theEPA vials by an autosampler to a fully automated Aurora1030W total organic carbon (TOC) analyzer (OI Analytical,College Station, TX, USA) and reacted with phosphoric acid(5% H3PO4) in a heated reactor.[16] Laboratory no. 5 convertedthe DIC into CO2 manually in an offline vacuum line with acryogenic trap. For each replicate 100 mL was transferred intoa dry vacuum flask without air contact. The CO2 wasliberated from the sample by adding H3PO4 to the flask.Samples were stirred for 5 min with magnetic agitation andlow heating to release the CO2.

Stable carbon isotope values are reported in the standarddelta notation in per mil (‰) versus Vienna Pee DeeBelemnite (VPDB) according to:

δ ¼ Rsample=Rreference–1� �� 1000 (1)

where R is the isotope ratio of the heavy to light isotope(e.g. 13C/12C) in sample and reference. The maximum absolutedifference of reported δ-values for a sample is referred to as:

Δδ ¼ δmax – δmin (2)

where subscripts max and min denote the highest and lowestreported value for a sample, respectively.

Two laboratories also reported results for δ13CDOC. Bothused hot wet chemical oxidation by sodium persulfate toconvert organic constituents into CO2.

[16] Sample transfer

y & Sons, Ltd. Rapid Commun. Mass Spectrom. 2013, 27, 2099–2107

Table

2.Metho

dsrepo

rted

forδ1

3 CDIC

(5labo

ratories,n

os.1

to5)

andδ1

3 CDOC(2

labo

ratories;n

os.1

and3)

analyses

δ13 C

DIC

δ13 C

DOC

Parameter

no.1

no.2

no.3

no.4

no.5

no.1

no.3

Storag

eafter

arriva

ldarkan

dcool

(~4°C

)darkan

dcool

(~4°C

)darkan

dcool

(~4°C

)room

tempe

rature

(~22

°C)

labshelf(21°C

)darkan

dcool

(~4°C

)darkan

dcool

(~4°C

)Pe

riph

eral

(man

ufacturer)

Gasbe

nchII

(The

rmoScientific)

Gasbe

nchII

(The

rmoScientific)

Aurora1030W

(OIAna

lytical)

'Carbo

nate

Prep

System

'(A

nalyticalP

recision

)Offlineprep

aration

withva

cuum

line

Aurora1030W

(OIAna

lytical)

Aurora1030W

(OIAna

lytical)

Sampletran

sfer

Dispo

sable

syring

eDispo

sable

syring

eFilling

of40

mL

EPA

vials

10mLga

stight

syring

e(H

amilton

)Fille

dinto

vacu

umflask

Filling

of40

mL

EPA

vials

Filling

of40

mL

EPA

vials

CO

2conv

ersion

H3PO

4in

12mL

Exetainer®

pre-fille

dwithhe

lium

H3PO

4in

12mL

Exetainer®

pre-fille

dwith

heliu

m

H3PO

4in

automatized

TIC

/TO

Can

alyz

er

H3PO

4in

5.9m

Lvialspre-fille

dwithhe

lium

H3PO

4in

vacu

umlin

e,CO

2cryo

genictrap

Na 2S 2O

8at

98°C

;system

withtrap

andpu

rgemod

ule

Na 2S 2O

8at

98°C

;system

with

cryo

trap

Masssp

ectrom

eter

(mod

e),

(man

ufacturer)

Delta

plus

XP

(CF)

a ,(The

rmoScientific)

Delta

XP

(CF),

(The

rmoScientific)

Delta

Vplus

(CF),

(The

rmoScientific)

AP2

003

(CF),

(Ana

lyticalP

recision

)

SIRA

10(D

I),

(VG

Instrumen

ts)

Delta

Vplus

(CF),

(The

rmoScientific)

Delta

Vplus

(CF),

(The

rmoScientific)

Referen

cematerialin

analysis

(δ13C

value)

NBS19

(+1.95

‰)LSV

EC

(–46.6‰

);solid

3in-hou

seNa 2CO

3(–9.50‰

;–4.9‰

;+28.59‰

);dissolved

in-hou

seNaH

CO

3(–5.1‰

);dissolved

in-hou

seCarrara

marble(solid;

+2.45‰

)an

dNa 2CO

3(dissolved

)

in-hou

seCarrara

marble(+2.60

‰)

2in-hou

sesu

crose

(–26.5‰

;–11.8‰

);dissolved

in-hou

segluc

ose

(–10.0‰

);dissolved

Internationa

lcalib

ration

materials

NBS19,L

SVEC

NBS-18,N

BS-19,

NBS-20

NBS-19

bNBS-18,N

BS-19

NBS-18,N

BS-19,

CO-1,C

O-8,C

O-9

cUSG

S-40,

IAEA-C

H-6b

IAEA-C

H-6

b

Normalization

2-po

intd

2-po

int

1-po

int

2-po

int

1-po

int

2-po

int

1-po

int

Precision(±1σ

)0.2‰

0.1‰

0.5‰

0.2‰

0.1‰

0.5‰

0.5‰

Isotop

eva

lues

arein

permil(‰

)ve

rsus

Vienn

aPe

eDee

Belem

nite

(VPD

B);EPA

:USEnv

iron

men

talP

rotectionAge

ncy

a CF:

continuo

usflow

mod

e(useshe

lium

carrierga

sforsamplega

sinlet);D

I:dua

linlet

(high-va

cuum

sampleintrod

uction

system

);seeBrand

.[33]

bCalibratedby

elem

entala

nalyzer/isotop

eratiomasssp

ectrom

etry

(EA/IRMS).

c IAEA

referenc

ematerials

weretreatedas

samples

forqu

alitycontrol.

dFo

rareview

ofdifferen

tno

rmalizationsche

mes,see

Paul

etal.[2

2]

Carbon stable isotope proficiency test of DIC and DOC

wileyonlinelibrary.com/journal/rcmCopyright © 2013 John Wiley & Sons, Ltd.Rapid Commun. Mass Spectrom. 2013, 27, 2099–2107

2101


2102

to the 40 mL EPA-vials for DOC was identical to the DICanalysis (see above). The precision based on analyses ofin-house reference materials, expressed as standarddeviation (±1σ), is reported to be better than ±0.5‰ forδ13CDOC values; for δ13CDIC values in most cases it lowerthan ±0.2‰ (Table 2).

RESULTS

Dissolved inorganic carbon

The results of the δ13CDIC analyses are summarized in Table 3and Fig. 1. Overall, individual results for each sample show arelatively large scatter. The difference between the highestand lowest reported δ-value (Δδ) ranged from 1.1‰ (IIE35)up to 6.6‰ (IIE33). Samples IIE30 (lake water) and IIE35(seawater) showed the smallest variation (1σ <0.5‰) amongthe measurements of the participating laboratories. Theδ13CDIC values for sample IIE30 range from –1.6‰ to –0.4‰with no distinct outlier. For IIE35 four of the five laboratoriesreported δ13CDIC values between –2.7‰ and –2.5‰. Thesevalues agree within 0.16‰ and are within the analyticaluncertainty. Only one laboratory reported a value of –1.5‰.A comparable pattern could be observed for sample IIE36(petroleum water). Three laboratories reported consistentlow δ13CDIC values between –11.6‰ and –11.4‰, whereastwo laboratories measured more positive values of –10.4‰and –7.0‰. For samples IIE33 and IIE34 (geothermal water)the DIC concentration was below the limit of detection(LOD) for laboratory no. 4, i.e. not enough CO2 could bereleased from the sample for a reliable stable isotope analysis.The values reported from the other laboratories showed arelatively large scatter with Δδ values of 6.6‰ and2.2‰ for IIE33 and IIE34. The observed standarddeviations are considerably larger than the reportedanalytical precision of the laboratories. None of thelaboratories reported systematically lower or highervalues with respect to the others. In addition, none ofthe samples always showed good agreement for one ormore pairs of laboratories. Instead, the consistencybetween δ13CDIC values from different laboratoriesdiffered for individual samples.

Table 3. δ13C results of dissolved inorganic carbon (DIC) andmean value of two or more analyses (Table 4)

δ13CDIC (‰ vs VPD

Sample no. 1 no. 2 no. 3 no. 4

IIE30 –1.26 –0.83 –0.97 –0.40IIE33 5.03 –1.56 –1.58 —a

IIE34 –3.16 –5.32 –4.73 —a

IIE35 –2.57 –2.52 –1.54 –2.60IIE36 –11.42 –11.41 –7.02 –10.37

VPDB: Vienna Pee Dee BelemniteaBelow limit of detection.


Dissolved organic carbon

The reported δ13CDOC values are only in reasonableagreement for seawater (IIE35) that was reported withisotope values of –11.0‰ and –9.3‰ from laboratories no. 1and 3 (Table 3, Fig. 2). For all other samples, the Δδ valuesamong the samples range from 10.6‰ up to 21.4‰. Herethe δ13CDOC values of laboratory no. 1 were systematicallylower that that of laboratory no. 3. There is no correlation indirection or amplitude of the δ13C values for DIC and DOCbetween laboratories no. 1 and 3. In other words, thesystematic difference between the two laboratories observedin the δ13CDOC results is not present in the reported δ13CDIC

values. It should be noted that the number of participatinglaboratories for this parameter is low and any systematicerror would be difficult to identify.

DIC concentration

Four of the participating laboratories reported the totaldissolved concentration of CO2 (Table 5). All laboratoriesmeasured low concentrations between 0.10 and 0.16 mmol/Lfor the two geothermal water samples (IIE33 and IIE34). TheDIC concentrations for the three other samples were higherwith average concentration values of 1.89 (IIE30), 2.16 (IIE35),and 2.08 mmol/L (IIE36). However, results from participatinglaboratories show quite a large scatter for sample IIE30 andparticularly for IIE36, forwhich reported concentrations rangedfrom 0.88 to 3.83mmol/L. The values reported for sample IIE35by laboratory no. 2 were notably lower than the concentrationsreported by the other laboratories for seawater.

DISCUSSION

The large scatter of δ13C values reported for DIC and DOCfrom five natural water samples is rather surprising in viewof the typically claimed analytical precision for stable carbonisotope measurements. This may be related to:

1. isotopic inhomogeneity of the samples,2. analytical errors during analysis by the participating

laboratories,3. applied data normalization procedures,

dissolved organic carbon (DOC). The reported value is the

B) δ13CDOC (‰ vs VPDB)

no. 5 Average (±1σ) no. 1 no. 3

–1.58 –1.01±0.45 –23.6 –11.11.34 0.81±3.13 –30.0 –8.6–4.33 –4.39±0.91 –22.5 –11.9–2.68 –2.38±0.47 –11.0 –9.3–11.57 –10.36±1.93 –33.3 –13.1

T4


Figure 1. S-shape plot of individual δ13CDIC results for the five natural water samples (a–e) analyzed inthis study (see also Table 3). Error bars are the precision reported by the laboratories (Table 2).

Figure 2. Comparison of δ13CDOC results from laboratoriesno. 1 and 3. Symbol size includes reported analyticalprecision (1σ = 0.5‰).


4. specific characteristics of a sample type,5. alteration of the sample during shipping and storage, and6. further currently unknown effects.

210

Inhomogeneity

Inhomogeneity is a known problem for some solid referencematerials. However, for fluid samples it is very unlikely thataliquots that are filled into individual bottles for shippingfrom a large storage container that holds a filtered andcarefully homogenized water sample are inhomogeneouswith respect to their stable isotope composition. Therefore, it

Copyright © 2013Rapid Commun. Mass Spectrom. 2013, 27, 2099–2107

can be assumed here that all individual sample bottles hadidentical initial stable carbon isotope compositions for DICand DOC.

Analytical errors

All laboratories used comparable up-to-date equipment andcalibrated their in-house reference materials againstinternational primary reference materials that are distributedby the International Atomic Energy Agency (IAEA, Vienna,Austria[18]) and the National Institute of Standards andTechnology (NIST, Gaithersburg, MD, USA[19]).

Fractionation effects between the gaseous phase in theheadspace (CO2(g)) and in the solution (CO2(aq)) as a functionof DIC concentration and volume have been discussed as apotential source of error during analysis.[15,20,21] This mayespecially account for samples with low DIC concentrationsthat might require the injection of several milliliters of sampleinto the 12 mL Labco exetainer® sample vials to liberateenough CO2 for isotope analyses with a Gasbench II device.The DIC concentrations in samples IIE33 and IIE34 were small(≤0.16mmol/L) and therefore required a larger sample volumefor analysis in laboratories no. 1 and 2, which used the above-described 12 mL vials and a Gasbench (Table 2). Laboratoryno. 1 injected 4 mL of samples IIE33 and IIE34 to generatesample peak amplitudes of 1800 mVand 2250 mV (first samplepeak; mass 44) on a Delta plus XP IRMS instrument (ThermoFisher Scientific, Bremen, Germany), respectively. Theamplitude is well above the typically required signal heightfor isotope analysis on this type of instrument and the linearityof the ion source waswithin specifications (0.03‰/V). The rawdata was corrected for linearity and instrument drift. The

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

3

Table 4. Number of replicates that were analyzed by the participating laboratories for δ13C analysis of DIC and DOC (Table 3)and concentration measurements of DIC (Table 5). Note that a higher number of replicates does not necessarily imply higheraccuracy. Dash indicates that parameter was not analyzed

Number of replicates for δ13CDIC/c(DIC)/δ13CDOC

Sample no. 1 no. 2 no. 3 no. 4 no. 5

IIE30 2/2/8 10/6/– 3/3/3 6/6/– 2/–/–IIE33 2/2/3 9/6/– 3/3/3 6/6/– 2/–/–IIE34 2/2/8 —a 3/3/3 6/6/– 2/–/–IIE35 2/2/8 —a 3/3/3 6/6/– 2/–/–IIE36 2/2/4 10/6/– 3/3/3 6/6/– 2/–/–aBelow limit of detection.

Table 5. Total dissolved concentration CO2 in water (all values in mmol/L)

Isotope laboratories (mmol/L)a

Sample Water typeChemical laboratories

(mmol/L)b no. 1 no. 2 no. 3 no. 4 Average (±1σ)

IIE30 lake water 2.25±0.14 2.51 1.35 2.33 1.37 1.89±0.62IIE33 geothermal water 0.62±0.18 0.12 0.10 0.16 –c 0.13±0.03IIE34 geothermal water 0.07±0.16 0.16 0.12 0.16 –c 0.15±0.02IIE35 seawater 2.41±0.12 2.40 1.33 2.20 2.72 2.16±0.59IIE36 petroleum water – 1.23 0.88 3.83 2.39 2.08±1.33aLaboratory no. 5 did not report this parameter.bDetermined in an international proficiency test by eight laboratories (M. P. Verma and colleagues, 2013, personal communication).cBelow limit of detection; LOD = 1.2 mmol/L.


2104

magnitudes of these post-run corrections are small and in therange of ~0.1‰. This ensures that the results do not suffer fromany instrument or linearity effect that might be problematic forvery low peak intensities. On the other hand, the data oflaboratory no. 1 was not corrected for fractionation effects thatmight have influenced the samples with low DICconcentrations, which had to be injected with a larger volumeof 4 mL into the exetainers (see above).The concentrations of samples IIE33 and IIE 34 were below

the LOD for laboratory no. 4. The fractionation effect wasconsidered by laboratory no. 4, which corrected the valuesfor samples IIE30, IIE35 and IIE36 as a function of the DICcontent and the volume of injected water. However,differences attributed to isotope fractionation due to CO2(g)–CO2(aq) partitioning are generally small (below ~0.5‰) andstrongly depend on the injected sample volume.[15,20] Forsmaller volumes this effect should be virtually negligiblebut even the injection of several mL into the sample vialscannot explain the observed discrepancies of several per milin the δ13CDIC values. Such fractionation effects might notaccount for the results from laboratories no. 3 and 5 that useddifferent conversion techniques. Laboratory no. 3 used anautomated TOC analyzer that uses acidification withphosphoric acid and subsequent extraction of the liberatedCO2 by a gas flow.[16] Laboratory no. 5 prepared the samplesmanually in a traditional offline vacuum line. Both techniquesshould not suffer from isotope fractionation due to the CO2

partition between the gaseous and the aqueous phase.


Thorough examination of the applied protocols, analyticalinstrumentation, in-house and international reference materials,and data normalization procedures (Table 2) did not indicatea major source of error. For δ13CDIC and δ13CDOC analyses alllaboratories used suitable in-house reference materials thatwere calibrated against appropriate international materials.At least two of the laboratories also verified their in-housereference materials by analyses from independent, externalisotope laboratories.

Data normalization

Different normalization procedures (single-point versus two-point) and δ13C values of in-house reference materials that areused to scale the analyzed isotope values to the VPDB scalemight account for differences that are larger than the reportedanalytical uncertainty.[22] However, this could account fordifferences in the range of 0.2‰ to 0.5‰ but does not explainthe much larger differences observed in this study. The largedifferences in δ13CDOC values observed between laboratoriesno. 1 and 3 also do not seem to be related to such analyticalissues. Both laboratories used nearly identical analytical setupswith only slight differences in in-house reference materials andnormalization procedure. As outlined above, this cannotaccount for differences of up to 20‰.

In summary, analytical uncertainties, calibration of in-housereference materials and an applied data normalizationproceduremay explain differences in the range of several tenths



of a per mil, but it is very unlikely that the observed largedifferences of several per mil in δ13CDIC and δ13CDOC valuesare related to these factors.

210

Sample type

Discrepancies in isotope values might also be related tospecific characteristics of a particular sample type. For seawater(IIE35), laboratory no. 2 reported a DIC concentration of 1.33mmol/L. This value is significantly lower than the expectedliterature seawater value of 2.0 to 2.5 mmol/L[23] and alsolower than the concentrations determined by the otherlaboratories involved in this study. The δ13CDIC value of–1.5‰ that might be regarded as an outlier for sampleIIE35 (see above) was reported by laboratory no. 3 thaton the other hand determined a concentration of 2.2mmol/L; a value that is in good agreement with theexpected literature value and also within the range of theother laboratories. From the limited number of data thereis no correlation between DIC concentration and stableisotope data. In other words, an anomalous low or highconcentration value (compared with the average) doesnot correspond to a δ13C outlier, and vice versa.Despite the large scatter, the DIC concentrations

determined during isotope analyses are in general agreementwith the results of the titrations performed by the chemicallaboratories (Table 5). The agreement of values is better forlake water (IIE30) and seawater (IIE35), whereas thegeothermal waters (IIE33 and IEE34) only agree on the factthat the carbon concentration is generally low in thesesamples. This might be related to the non-carbonic alkalinitythat could account for a large fraction of the total alkalinityin geothermal waters.[24]

For petroleum water (IIE36) the situation is yet morecomplicated. It has been demonstrated for carbonic speciesanalyses that no dissolved CO2 (i.e. no inorganic carbonicspecies) is present in this type of waters.[25,26] On the otherhand, the gas chromatography of stable isotope analysesclearly showed considerable amounts of CO2 that werereleased by acid treatment from the samples. Theconcentrations reported for petroleum water by the isotopelaboratories showed the largest discrepancies of the fivenatural sample types that were analyzed for their isotopecomposition in this study. If the detected CO2 does notoriginate from dissolved carbonic species it should be theproduct of the reaction of acid with other substances fromadmixed oil in the water. In this case the δ13C value is notrepresentative for the DIC and its value may strongly dependon analytical conditions such as temperature and reactiontime. This is further indicated by the fact that the 'outlier'δ13CDIC value of –7.0‰ for sample IIE36 was determined ona different analytical setup that transfers the sample into aheated reactor for CO2 conversion with subsequent extractionby helium flow (see Table 2).

Sample alteration

Further bias was probably introduced by changes in δ13Cvalues during shipping and storage. The δ13CDIC values ofwater samples may change within days by biological activity.To prevent this, many workers filter water by 0.2 μm syringefilters and poison the water sample (e.g. with HgCl2) to


inhibit microbial life.[7,27] Another option is to convert DICinto CO2 directly in the field by injection of the water intohelium-filled vials pre-loaded with H3PO4

[17] or completeprecipitation of the DIC as solid carbonate (SrCO3 or BaCO3).Today, the majority of stable isotope laboratories do notaccept samples that are poisoned by HgCl2 due to its hightoxicity. To account for this fact and for the security of involvedpersons (e.g. parcel service, laboratory staff), the samples of thisstudywere not treated with any toxic preservative. In addition,the use of amber glass bottles with tight sealing rubber septa isrecommended to reduce photosynthetic activity and to excludepermeability such as might occur from plastic containers.Variations in the isotopic composition of the water molecule(δ2H and δ18O values) have been reported to change over timein plastic containers[28,29] due to diffusive transfer of water andwater vapor through container walls and this might also beproblematic for the exchange of DIC and atmospheric CO2.Nonetheless, it has been demonstrated that δ13CDIC values ofunpreserved waters samples changed by ~7‰ over 6 monthsirrespective of the use of glass or HDPE sample containers.[30]

The samples of this study were filtered (0.45 μm) prior toshipping. This is the standard filter pore size that is used toseparate between 'dissolved' and 'particulate' carbon in watersamples and is used in most studies including this proficiencytest. The 0.45 μm filtering does not result in a sterile solutionsince a significant variety of bacteria species are smaller thanthis filter pore size. In addition, secondary bacterial fauna cangrow in the unsterilized containers during shipping andstorage. This can result in alteration of the initial stableisotope ratio due to heterotrophic activity. Alteration byphotosynthesis may also occur since some species ofcyanobacteria are smaller than 0.45 μm. Adding poison tothe sample will prevent this microbial activity, but wasrejected for this study (see above). The study of Doctoret al.[7] that compared different toxic chemicals (HgCl2,CuSO4, and CuCl2) observed no significant differencebetween unpreserved samples and those preserved with thechemicals HgCl2 and CuSO4 after a storage time of 9 months.

Transparent HDPE sample containers with standard screwcaps without septa were used, similar to those used forshipping the samples for carbon species titration. Theshipping time from the Instituto de Investigaciones Eléctricas(Mexico) to stable isotope laboratories worldwide took up to2 weeks. The isotope samples arrived in most laboratories bythe end of June 2012. The time between arrival and analysiswas typically in the range of weeks for most laboratoriesand not within days, as generally recommended for watersamples without toxic preservatives (e.g. laboratories no. 1and 4 analyzed the samples within 3 to 4 weeks after arrival).

During sample preparation (i.e. the transfer of the watersample to analysis vials) air contact was avoided as best aspossible since that might also change the δ13CDIC value.Some of the laboratories observed changes in some of theirwater samples such as floating particles or algae growth.Laboratories no. 4 and 5 stored their sample sets underdaylight at room temperature prior to analysis. Nonetheless,these storage conditions do not seem to correlate withdeviating isotope values.

Another point is that the lake water and seawater originatedfrom oxygenated environments whereas geothermal water andpetroleum water are from anoxic environments. The potentialof any change in the stable carbon isotope composition by


5


2106

oxidation in the laboratory, during shipping and storage orthe analysis, is much higher in the latter group becausethese solutions changed their redox environment. Thismight be important regardless of the accuracy of the handlingduring analysis.Furthermore, it might be assumed that at the time of

sampling the heterotrophic activity, the photosynthesis,and the carbonate species distribution in seawater andlake water were balanced in these ecosystems. Therefore,any heterotrophic activity during transport and storagewould probably result only in small changes to the stableisotope value, because there is only a small change in theredox state of the solution. In contrast, the heterotrophicactivity in petroleum water and geothermal water mightbe restricted to anaerobic processes. Some DOC compoundscan be resistant to anaerobic degradation processes, but incase that these anoxic samples were oxygenized at somestage, the onset of a more effective aerobic degradation ofthese organic compounds might significantly alter the δ13Cvalues of the dissolved inorganic and organic carbon.[31,32]

Despite many precautions it cannot be ruled out herethat a part of the large differences in δ13CDIC and δ13CDOC

values in some of the samples is related to secondaryeffects during shipping and sample storage. This includesthe generation of CO2 from DOC by microbial activity,diffusion processes through container walls, CO2 degassingduring sample preparation, redox changes, and potentialcarbonate precipitation due to temperature changes thatinfluence calcite saturation state.

Other effects

Analytical results indicated that further unknown processesmight have occurred within specific sample types. For DOCanalysis, all five samples were split into two aliquots bylaboratory no. 1. The first aliquot was analyzed directlyfor its δ13CDOC value, whereas the second aliquot wastreated with H3PO4 overnight to remove the DIC prior toanalysis. This procedure is common in many laboratoriesfor samples dedicated for δ13CDOC analyses only. For fourof the five samples, the isotope values of both aliquotswere identical within analytical uncertainty. Sample IIE36(petroleum water) yielded a δ13CDOC value of –33.3‰ forthe untreated sample, a value that is indicative for C3

plants as organic carbon source. As petroleum originatesfrom plant material this signature seems to be preservedin the compounds of the petroleum water. In contrast,the acid-treated samples showed a negative shift of about20‰ to very low δ13CDOC values of –55.5‰. The exactprocesses responsible for these unexpected and somewhatambiguous changes after adding phosphoric acid certainlyneed to be explored in further experiments but are beyondthe scope of this inter-laboratory comparison.

CONCLUSIONS

To the authors' knowledge this is among the first inter-laboratory tests for δ13CDIC and δ13CDOC values with naturalwater samples. The samples represent a realistic selectionof everyday samples and their shipment. The largediscrepancies found in this studywere surprising at first glance


and raised several questions about analytical errors in theparticipating laboratories. However, closer inspection ofmethods and data normalization procedures revealed that theobserved differences of several per mil could not be explainedby analytical errors or the use of unsuitable or incorrectcalibrated reference materials in the participating stableisotope laboratories.

The best explanation for the found discrepancies is thatsome sample types are much more sensitive to change thanother. Whereas lake water and seawater showed a reasonablespread around a δ13CDIC mean value, geothermal waters werecharacterized by very low DIC concentrations that aremore difficult to analyze for stable isotope compositions.In addition, these waters might also contain dissolvedsubstances that could alter the δ13C value during acidreaction. This introduces a higher analytical uncertaintyfor this sample type than is typically derived fromsynthetic quality control material. For waters from thepetroleum well it could be shown that δ13CDIC and δ13CDOC

analyses are not indicative for the inorganic or the organiccarbon species. Stable isotope values of this sample typeshould be interpreted only with great care as they usuallycontain complex mixtures of organic molecules that mightinteract with DIC and DOC.

An additional potential source of error could be secondaryalteration of the sample during shipping and storage. The useof glass sample containers with toxic preservatives andcontinuous cooling would help to reduce the alteration ofisotope values of the dissolved carbon species. For watersamples from anoxic environments, such as geothermal orpetroleum well waters, changes in the redox conditions aremore likely to occur during storage and analysis than forsamples from the ocean or lakes. This can be another reasonfor the better agreement in δ13C values of seawater and lakewater samples in this study.

This study indicated that scatter in carbon isotope data canbe in the range of several per mil for water samples fromextreme environments (geothermal waters, petroleum wells).The analyses of lake water and seawater also revealed alarger than initially expected scatter and researchers fromvarious disciplines should be aware of this fact. Theadvantage of this inter-laboratory test is that it reflectseveryday conditions of shipping from the field to laboratoriesover large distances.

We encourage the stable isotope community to participatein further inter-laboratory tests to improve our understandingon sample alteration processes and analytical uncertainties inthis import field of aquatic geochemistry.

AcknowledgementsThe authors appreciate the authorities of the Instituto deInvestigaciones Eléctricas (Mexico) for financial supportas well as members of each participating laboratory. Wethank the technical staff and students of all institutionsthat helped during the analytical work. Norma AraceliGarcía Muñoz, Erick Luna Rojero and Enrique Portugalgraciously provided the samples. The thorough commentsof three anonymous reviewers helped to improve the finalversion of this manuscript.



REFERENCES

[1] S. A. Rounds. Alkalinity and acid neutralizing capacity,(version 4.0), U.S. Geological Survey Techniques ofWater-Resources Investigations (TRWI), Book 9, Chap.A6., sec. 6.6, September, 2012. Available: http://water.usgs.gov/owq/FieldManual/Chapter6/section6.6/ (accessed10 March 2012).

[2] W. Stumm, J. J. Morgan. Aquatic Chemistry: ChemicalEquilibria and Rates in Natural Waters, (3rd edn.). John Wiley,New York, 1996.

[3] J. A. C. Barth, J. Veizer, B. Mayer. Origin of particulateorganic carbon in the upper St. Lawrence: Isotopicconstraints. Earth Planet. Sci. Lett. 1998, 162, 111.

[4] D. T. Monteith, J. L. Stoddard, C. D. Evans, H. A. de Wit,M. Forsius, T. Hogasen, A. Wilander, B. L. Skjelkvale,D. S. Jeffries, J. Vuorenmaa, B. Keller, J. Kopacek, J. Vesely.Dissolved organic carbon trends resulting from changes inatmospheric deposition chemistry. Nature 2007, 450, 537.

[5] P. A. Raymond, N.-H. Oh, R. E. Turner, W. Broussard.Anthropogenically enhanced fluxes of water and carbonfrom the Mississippi River. Nature 2008, 451, 449.

[6] P. Schulte, R. van Geldern, H. Freitag, A. Karim, P. Négrel,E. Petelet-Giraud, A. Probst, J. L. Probst, K. Telmer, J. Veizer,J. A. C. Barth. Applications of stable water and carbonisotopes in watershed research: Weathering, carbon cycling,and water balances. Earth-Sci. Rev. 2011, 109, 20.

[7] D. H. Doctor, C. Kendall, S. D. Sebestyen, J. B. Shanley,N. Ohte, E.W. Boyer. Carbon isotope fractionation of dissolvedinorganic carbon (DIC) due to outgassing of carbon dioxidefrom a headwater stream. Hydrol. Processes 2008, 22, 2410.

[8] R. van Geldern, T. Hayashi, M. E. Böttcher, M. J. Mottl, J. A.C. Barth, S. Stadler. Stable isotope geochemistry of porewaters and marine sediments from the New Jersey shelf:Methane formation and fluid origin. Geosphere 2013, 9, 96.

[9] A. Myrttinen, V. Becker, R. van Geldern, H. Würdemann, D.Morozova, M. Zimmer, H. Taubald, P. Blum, J. A. C. Barth.Carbon and oxygen isotope indications for CO2 behaviourafter injection: First results from the Ketzin site (Germany).Int. J. Greenhouse Gas Control 2010, 4, 1000.

[10] J.-F. Hélie, C. Hillaire-Marcel. Sources of particulate anddissolved organic carbon in the St Lawrence River: isotopicapproach. Hydrol. Processes 2006, 20, 1945.

[11] A. Stögbauer, H. Strauss, J. Arndt, V. Marek, F. Einsiedl,R. van Geldern. Rivers of North-Rhine Westphalia revisited:Tracing changes in river chemistry. Appl. Geochem.2008, 23, 3290.

[12] J. Barth, J. Veizer. Three-component water balance to enableprotection of ecologically important ecosystems: an examplefrom the St. Lawrence River. Appl. Geochem. 2004, 19, 1637.

[13] Final Report on Fourth Interlaboratory ComparisonExercise for δ2H and δ18O Analysis of Water Samples(WICO2011), (Eds: M. Ahmad, P. Aggarwal, M. van Duren,L. Poltenstein, L. Araguas, T. Kurttas, L. I. Wassenaar).International Atomic Energy Agency, Vienna, Austria,2012, p. 67.

[14] Available: http://isogeochem.wikispaces.com/.


[15] N. Assayag, K. Rivé, M. Ader, D. Jézéquel, P. Agrinier.Improved method for isotopic and quantitative analysis ofdissolved inorganic carbon in natural water samples. RapidCommun. Mass Spectrom. 2006, 20, 2243.

[16] G. St-Jean. Automated quantitative and isotopic (13C)analysis of dissolved inorganic carbon and dissolved organiccarbon in continuous-flow using a total organic carbonanalyser. Rapid Commun. Mass Spectrom. 2003, 17, 419.

[17] C. Spötl. A robust and fast method of sampling and analysisof δ13 C of dissolved inorganic carbon in groundwaters. Isot.Environ. Health Stud. 2005, 41, 217.

[18] Available: http://nucleus.iaea.org/rpst/.[19] Available: http://www.nist.gov/srm/.[20] G. Capasso, R. Favara, F. Grassa, S. Inguaggiato, M. Longo.

On-line technique for preparing and measuring stablecarbon isotope of total dissolved inorganic carbon in watersamples (δ13CTDIC). Ann. Geophys. 2005, 48, 159.

[21] G. G. Salata, L. A. Roelke, L. A. Cifuentes. A rapid andprecise method for measuring stable carbon isotope ratiosof dissolved inorganic carbon. Mar. Chem. 2000, 69, 153.

[22] D. Paul, G. Skrzypek, I. Fórizs. Normalization of measuredstable isotopic compositions to isotope reference scales – Areview. Rapid Commun. Mass Spectrom. 2007, 21, 3006.

[23] S. Emerson, J. Hedges. Chemical Oceanography and the MarineCarbon Cycle. Cambridge University Press, Cambridge, 2008.

[24] M. P. Verma. A revised analytical method for HCO3 andCO3

2– determinations in geothermal waters: An assessmentof IAGC and IAEA interlaboratory comparisons.Geostandards Geoanal. Res. 2004, 28, 391.

[25] M. P. Verma. Determination of carbonic species in ground,surface, geothermal, ocean and petroleum waters:Theoretical aspects of CO2 chemistry in acid-base titration.J. Anal. Chem. 2013, submitted.

[26] M. P. Verma, P. Brikle, D. Sánchez, in Proceedings WorldGeothermal Congress 2010, Bali, Indonesia, 25–29April, 2010.

[27] I. Clark, P. Fritz. Environmental Isotopes in Hydrology. Lewis,Boca Raton, 1997.

[28] J. E. Spangenberg. Caution on the storage of waters andaqueous solutions in plastic containers for hydrogen andoxygen stable isotope analysis. Rapid Commun. MassSpectrom. 2012, 26, 2627.

[29] J. E. Spangenberg, T. W. Vennemann. The stable hydrogenand oxygen isotope variation of water stored inpolyethylene terephthalate (PET) bottles. Rapid Commun.Mass Spectrom. 2008, 22, 672.

[30] S. J. Taipale, E. Sonninen. The influence of preservationmethodand time on the δ13C value of dissolved inorganic carbon inwater samples. Rapid Commun. Mass Spectrom. 2009, 23, 2507.

[31] B. M. Van Breukelen,W. F. M. Röling, J. Groen, J. Griffioen, H.W. VanVerseveld. Biogeochemistry and isotope geochemistryof a landfill leachate plume. J. Contam. Hydrol. 2003, 65, 245.

[32] P. P. Videla, A. J. Farwell, B. J. Butler, D. G. Dixon.Examining the microbial degradation of naphthenic acidsusing stable isotope analysis of carbon and nitrogen. WaterAir Soil Pollut. 2009, 197, 107.

[33] W. A. Brand, inHandbook of Stable Isotope Analytical Techniques,vol. I, (Ed: P. A. De Groot), Elsevier, Amsterdam, 2004.


2107

http://www.nist.gov/srm/

stable carbon isotope analysis of dissolved inorganic carbon (dic) and dissolved organic carbon...

Documents