Intercomparison Materials for Stable Isotopes of Light Elements Number: IAEA-TECDOC-825

Home , Marilyn Fogel, Stable isotope ratio

IAEA-TECDOC-825

Reference and intercomparison materialsfor stable isotopes of light elements

Proceedings consultantsa of meeting held Vienna,in December1-3 1993

INTERNATIONAL ATOMIC ENERGY AGENCY The IAEA does not normally maintain stocks of reports in this series. However, microfiche copies of these reports can be obtained from

IN IS Clearinghouse International Atomic Energy Agency Wagramerstrasse 5 P.O. Box 100 A-1400 Vienna, Austria

Orders should be accompanied by prepayment of Austrian Schillings 100,- in the form of a cheque or in the form of IAEA microfiche service coupons which may be ordered separately from the I MIS Clearinghouse. The originating Section of this publication in the IAEA was: Isotope Hydrology Section International Atomic Energy Agency Wagramerstrasse5 0 10 P.Ox Bo . A-1400 Vienna, Austria

REFERENCE AND INTERCOMPARISON MATERIALS FOR STABLE ISOTOPE LIGHF SO T ELEMENTS IAEA, VIENNA, 1995 IAEA-TECDOC-825 ISSN 1011-^4289 © IAEA, 1995 Printed by the IAEA in Austria September 1995 FOREWORD

stable Th e isotope compositio chemicaf no l elements varie naturan si l compounds as a consequence of the slightly different physico-chemical behaviour of isotopes. In particular, isotope f lighso t elements, which hav largese eth t relative mass differencd an e therefore als mose oth t different behaviour, sho widese wth t variations r instanceFo . , natural waters exhibit variation 218e e HIO/th th 1 H16f f so o O 1:2o rati t d 1:1.1 o ratit p ,an o u p ou , 16 2 1 16 due to the different evaporation-condensation rate of 'H2 O with respect to H H O and 18 'H2 O. These variation ranges become even l widenaturaal f i r l compounds containing hydrogen and/or oxygen are considered.

The possibility of measuring the stable isotope relative variations with high precision, using mass spectrometry, promoted the rise of new fields of research in geochemistr hydrologd an y y and, more recently environmentan i , l studies e steadTh .y growth of these investigations and of their practical applications has emphasized the need for high quality isotopic standards and intercomparison samples, with well determined isotopic composition intercalibratioe th r fo , analyticaf no l technique resultd san s among laboratories

Although stable isotope standards have existed for more than three decades, there was a need to re-examine the whole matter, in view of the expansion of isotope applications ancontinuoue dth s improvement refinementd san analyticae th f o s l techniques, which make it possible to detect smaller and smaller isotopic variations. Just to give an example, up to r twento n te y years ago ,precisioa f ±0.no determinatioe 1 th % n oi 13f no C/ 12C ratio relative variation generalls swa y considered acceptabl eved ean n good t wit e applicatioBu hth . f no carbon stable isotopes to study the atmospheric carbon dioxide with the aim of improving the understanding of the global carbon cycle, the long-term analytical precision had to be increased by an order of magnitude to detect the long term trend of less than 0.02%e per year, superimposed on the seasonal variations. The laboratories involved in this work had, and still havethrougo g o t , h lengthy check runintercomparisond san analyticae th f so d an l sampling techniques ordesurn e i ,b eo t r tha date th ta obtaine differenn di t place fulle sar y consistent.

pase th t r thirtFo y years Internationae th , l Atomic Energy Agency, througs hit Sectio f Isotopno e Hydrology bees ha , n activfiele preparatiof th do n ei distributiod nan f no stable isotope referenc intercomparisod ean n material determinatioe th r sfo isotopie th f no c composition of natural compounds.

The organization of the Consultants Meeting on Stable Isotope Standards and Intercomparison Materials held in Vienna from 1 to 3 December 1993, the fifth of this type (the previous meetings took place in 1966, 1976, 1983 and 1985), called for a review and a discussio e characteristicsth f no , qualit availabilitd an y e existinth f o y g standardd an s intercalibration assessmenmaterialsn a r fo materialsd w needf an ,t o ne r sfo vien i , recenf wo t developments and applications.

A large part of the discussions was devoted to the new materials prepared for sulphur isotope analysis and the analytical requirements for highly precise isotopic analysis of CO2. The papers presented at the meeting are assembled in this volume. For the first time, two institutions were represented which are actively engaged in the field of standardization and intercalibration of isotopic measurements: the Institute for Reference Material Measurementd san Commissioe th f so Europeae th f no n Communities, Geel, Belgiume Nationath d an ,l Institut r Standardfo e d S TechnologU an s e th f o y Department of Commerce, Gaithersburg, Maryland, USA. It is hoped that this marks the beginnin newa f go , fruitful collaboration betwee IAEe nth thesd Aan e institutions.

EDITORIAL NOTE

preparingIn this publication press,for IAEAthe staffof have pages madethe up from the original manuscripts submittedas authors.the viewsby The expressed necessarilynot do reflect those of the governments of the nominating Member States or of the nominating organizations. Throughout textthe names Memberof States retainedare theyas were when textthe was compiled. The use of particular designations of countries or territories does not imply any judgement by publisher,the legalthe IAEA,to the status as of such countries territories,or of their authoritiesand institutions delimitationthe of or of their boundaries. The mention of names of specific companies productsor (whether indicatednot or registered)as does implyintentionnot any infringeto proprietary rights, should construednor be it an as endorsement recommendationor partthe IAEA. ofon the The authors are responsible for having obtained the necessary permission for the IAEA to reproduce, translate or use material from sources already protected by copyrights. CONTENTS Summar meetine th f yo g ...... 7 .

Standards and intercomparison materials distributed by the International Atomic Energy Agency for stable isotope measurements ...... 13 Gonfiantini,R. Stichler,W. RozanskiK. Reportin f stablgo e carbon, hydrogen oxyged an , n isotopic abundances...... 1 3 . T.B. Coplen Audit of VSMOW distributed by the United States National Institute of Standards and Technology ...... 5 3 . T.B. Coplen, J. Hopple Sulphur isotope standards ...... 9 3 . B.W. Robinson Variation sulfue th n rsi isotope compositio ...... T CD f no 7 4 . G. Beaudoin, B.E. Toy lor,. Rumble. D . Thiemens M , HI Compariso conventional-SOe th f no laser-SFe th d 2an 6 methods: Implicationr sfo the sulfur isotope scale ...... 9 4 . G. Beaudoin, B.E. Taylor Interlaboratory compariso f referencno e material r nitrogen-isotope-ratisfo o measurements ...... 51 J.K. Böhlke, T.B. Coplen Interlaboratory compariso materialw ne f nr o carbo fo s oxyged nan n isotope ratio measurements ...... 67 W. Stichler Carbon, oxyge hydroged nan n isotopic intercompariso f fruid no an t vegetable juices ...... 5 7 . J. Koziet, F. Pichlmayer, A. Rossmann isotopie Th c compositio Groningee th f no n GS-1 GS-2d 9an 0 pur2 standardeCO s ...1 8 . H.A..]. Meiier measurementO ô e reporA th IAEe n th o tf Aso cellulose intercompariso18 n material IAEA-C3 ...... 85 W.M. Buhay, B.B. Wolfe, R.J. Elgood, T.W.D. Edwards f platinizeo e Us d magnesiu reagens ma t replacing zin hydrogen ci n isotope analysis ...... 7 8 . S. Halas, B. Jasinska Carbon isotope (14C, 12C) measurements to quantify sources of atmospheric carbon monoxid urban ei r ...... nai 3 9 . O.A. Klouda, M.V. Connolly The carbon dioxide isotopic measurement process: Progress at NIST on measurements, reduction algorithms and standards ...... Ill R.M. Verkouteren, G.A. Klouda, L.A. Currie High precision stable isotope measurements of atmospheric trace gases ...... 131 C.E. Allison, R.J. Francey Recommendations for the reporting of stable isotope measurements of carbon and oxygen in CO2 gas ...... 155 C.E. Allison, R.J. Francey, H.A. Meijer

Lis f Participanto t s ...... 3 16 .

IAEA report publicationd san stabln so e isotope standard measuremend san t intercomparison ...... 5 16 . SUMMARY OF THE MEETING

n thiI s summar e fiftth h f o yIAE A Meetin n Stablo g e Isotope Standardd an s Intercomparison Materials e majoth , r conclusions e recommendationreacheth d an d s formulated at the meeting are presented. They are submitted to the community of those intereste determinatioe th n di stablf no e isotope environmental variation variour sfo s scientific and technical purposes. Comment proposalsd san , relate conclusione t onlth dno o y t d san recommendation f thio s s specific meetin t als bu o gstablt o e isotope referencd an e intercomparison materials in general, will be most welcome and useful, and should be addressed to the IAEA, Section of Isotope Hydrology. They will be brought to the attention consultante ofth wilo lswh participat futurn ei e IAEA meeting same th en so subject wild an ,l be of help to all national and international institutions involved in the preparation and distribution of stable isotope reference materials, in formulating working programmes and defining lines of co-operation and co-ordination.

1. Sulphur isotope reference and intercomparison samples

Early investigators adopted meteoritic sulphur particulan i , r tha troilitf to e (FeS) from the Canyon Diablo meteorite, as a reference standard for sulphur isotope studies. The choice was apparene baseth n do t isotopic homogeneit f meteoritiyo c idesulphue th a d thaan rt i t could hav same eth e isotopic compositio primordiae th s na l sulphur, supporte face th t y db that its S/ S value is close to the average of terrestrial samples. 34 32 Later, however, it was recognized that the sulphur isotopic composition of Canyon Diablo Troilit s homogeneouea (CDTt no s )wa desirables sa : difference 0.25%o t p u f oo s were observed. Thi t surprisins no fac s i t geologican gi l materials, which rarely have fully homogeneous isotopic and chemical compositions. Even so, CDT continued to be used as a primary reference for sulphur isotopes, and this unfortunate continuance has hampered progres agreemend san t among laboratories.

Intercompariso f resultno se improve b shoul w dno adoptiny db r measuremengfo t calibration the sample of chemically pure Ag2S, having a homogeneous isotopic composition calibrated vs. CDT, which was prepared by B.W. Robinson and C.A.M. Brenninkmeijer (Institut f Geologicaeo Nuclead an l r Sciences, Lower HuttZealandw Ne , 1987n i ) . This sample has been tested by several laboratories, and is now distributed by IAEA with the code name IAEA-S-1 (see Appendix). CDT, for obvious reasons, is kept as a reference to express o34S results.

The following recommendations were formulated at the meeting:

(i) It has been recognized that it is practically impossible to define a very precise value of ÎAEA-S-1 versus the Canyon Diablo Troilite (CDT) reference standard, simply because CDT was not isotopically homogeneous and, in addition, has been exhausted fo lona r g time. Thus proposes i t i , adopo dt r IAEA-S-fo t e valu1th f 5eo 34 S= -0.30%o vs. a defined, hypothetical V-CDT (where V stands for Vienna, in analogy with V-SMO V-PDB)d Wan recommendes i t I . d that laboratories re-calibrate their measurements versus IAEA-S-1 in order to remove ambiguities on the zero of the 634S scal improvd ean e intercompariso resultsf no reportinn I . g their results, laboratories should specify that the reference is V-CDT, and not CDT, indicating that the measurements calibratio mads nwa e through IAEA-S-1. (ii) A new Ag S sample shall be prepared with a <5 S value close to -30%o from 1 kg of 2 34 sphalerite (ZnS) supplie . HalasS y db , Maria Curie-Sklodowska University, Lublin, Poland. B.W. Robinson Zealandw Ne principl,n i s i , e read preparo yt samplee eth , with the support of IAEA. The new sample will be called IAEA-S-3.

BaSOw ne A sampl(iii) e shal preparee b l d wit h<5a S valu e% 0 closo -3 fro g k o e t m 1 4 34 of spahlerite supplied by S. Halas, Maria Curie-Sklodowska University, Lublin, Poland. This sample would complement NBS-127, also BaSO4, with 634 S= +20.32% (averagT laboratories0 CD 1 . f eo vs ; Hut, 1987). H.R. Krouse, University of Calgary, Canada, is in principle ready to prepare the sample, with the support of IAEA. The new sample will be called IAEA-SO-4 (the code SO implies that the sample can be used for sulphur and oxygen isotopes).

(iv) The above new samples, together with IAEA-S-1 and IAEA-S-2, should be widely

distributed to practising laboratories, which should report their results to IAEA with the objective of establishing recommended S0 34 values for IAEA-S-2, IAEA-S-3 and IAEA-SO-4 vs. IAEA-S-1.

(v) The determination of the absolute isotopic ratios and abundances of IAEA-S-1 is recommended.

33 36 34 (vi) The determination of 6 S and 5 S, in addition to S, by SF6 technique and by thermal ionization mass spectrometry IAEA-S-2r fo , , IAEA-S- IAEA-SO-d 3an . 4vs IAEA-S-1, shoul encouragede db .

(vii) Consideration shoul establishmene givee db th o nt f intercomparisoo t n sampler fo s organic sulphur isotope analyses. In fact, the isotopic composition of organic sulphur

is finding increase petroleun i e dus m geochemistry, medical sciences, agriculture, forestry food an d, studies wide vien th I .ef w o rang Sf 6e o 34 value s encounteren di these variete studieth d f methodyo an s s used (including recent on-line combustion techniques) s appropriati t i , o havt e e relevant intercomparison materialse On . difficulty in selecting the chemical compounds for the intercomparison samples is that different forms of organic sulphur exist in many natural materials, which may differ considerably in isotopic composition. For instance, in some oils these forms exhibit up to 20%o difference in 6 S, attesting to their different geochemical histories. Cystine (melting point 260 °C34 ) and methionine (melting point 282 °C) have been suggested as possible materials for the establishment of these intercomparison samples.

2. Other new intercomparison samples

intercomparisow Ne n sample r hydrogenfo s , carbon d oxygean , n stable isotope determination varioun si s categorie f compoundso suggestee sar r possibldfo e consideration at future IAEA meetings. They include: ) (i a-cellulos hydrogenr efo , carbon oxyged an , n stable isotope variation woodn i s A . sample of a-cellulose is now available at IAEA with the code name IAEA-C-3, whic intercalibrationC use14 s h i r dfo proposes i t i : checo dt isotopis kit c composition d homogeneityan orden i , judgo t r e whethe s suitabli n intercomparisot i a r s a e n sample. The o13C of IAEA-C-3 was measured by a number of laboratories for a 14C intercomparison exercisee greath d t majoritan , f resulto y s ranged from -24%oo t -26%o (Rozanski, 1991). (ii) A methane having <52H at about -300 %o vs V-SMOW and 513C at about -60 %o vs V-PDB samplA . e with these characteristic available b y sma t NISTea .

(iii) Two carbon dioxide samples to replace NBS-16 and NBS-17, which have been exhauste r severadfo samplew l ne years e s Th shoul. d have approximatel same yth e isotopic composition as NBS-16 and NBS-17, the first of which had 5 C = -41.59%o and 018O = -36.11%o, and the second 013C = -4.45%o and 018O =13 -18.76%o vs. V-PDB (Hut, 1987). NIST will consider preparing these samples.

(iv) Two silver phosphate samples having respectively o O of about 0%o and +20%o vs. V-SMOW usee b r oxyge dfo o t , n isotope measurement phosphatesn si . !8

Consideration was also given to other light elements, with the following provisional indications: mtercomparisow ne o N ) (i n materials were suggeste r lithiudfo nitroged man n isotope determinations.

(ii) Two certified reference materials (both H3BO3), are available for boron isotopes, one from NIST (NIST-SRM-951, with B/ B = 4.04362 ± 0.00137), and the other from U 10 IRMM (IRMM-G11, with UB/10B - 4.0443 + 0.0052). NIST-SRM-951 is currently usereferenca s da boron i e n isotope geochemistry watea Se . r borat enriches ei n di "B by about 40 %o with respect to NIST-SRM-951.

(iii) IAEA-NO-3 (KNO3), whic s i distributeh r nitrogefo d n isotope measurements intercomparison, can possibly also serve as intercomparison sample for oxygen in nitrates oxyges It . n isotopic compositio homogeneitd nan y needcheckede b o st .

3. Determination of the absolute isotopic ratios of reference samples

The materials used for isotopic measurement calibration vs. the reference standards, whicr include mose fa th o hs r t commofo , n light elements, V-SMOW, SLAP, NBS-1d 9an IAEA-S-1 need to be fully characterized, i.e. their absolute isotopic ratios must be determined. Thi principln i s e allow checo t ) (i sk whethe isotopie th r c compositioe th f no material will change with time, and (ii) to prepare new materials with identical (within the experimental errors) or very close isotopic ratios, when the present materials are exhausted or no longer suitable, or if the storage and handling conditions prove to be inadequate for their long-term conservation.

r thesFo e purposes absolute th , e isotopic samplee ratioth f so s indicated above must be determined with an overall error which is equal or better than the best current analytical precision for relative isotopic difference determinations. Thus, in principle, the error on the absolute isotopic ratios shoul t exceedno d 0.02%o (13C/12C)for carbon, oxygen (18O/16d O)an sulphur ( S/ S), and 0.2%o for hydrogen ( H/ H). These limits have so far been approached 34 32 2 1 e absolutth onl n i y e 2H/JH ratio determinatio f V-SMOo n d SLAWan P y carrieb t ou d Hageman . (1970)al quott o ne wh ,e error 0.32%f so Q.56d oan (7x>, respectively errore Th . s quoted by Baertschi (1976) and Li et al. (1988), for the O/ O and the O/ O determination 18 16 17 16 of V-SMOW, 0.22%0 and 2.1%0 respectively, still appear large.

e absolutTh e isotopic ratio usee r correctioar s dfo f resultno s from interferencef o s isobaric ion masn si s spectrometric analysis additionan thia d s i san , l reaso r whicnfo h their determinatio materialnin s use measuremenfor d t calibratio desirablenis . Howeverthe , accuracies require thin di t s leas a ordee cas e on tf magnitud ear o r e less than those quoted above.

The Institute for Reference Materials and Measurements (IRMM) in Geel, Belgium, is currently involve Avogadre th n di o constant re-determination. Wit same hth e procedure and equipment, the overall precisions which can be attained on isotopic ratios are: 0.04%o 13 12 15 14 18 16 34 32 16 for C/ C, 0.2%0 for N/ N, O/ O and S/ S, 1%. for "O/ O. It is anticipated that these precisions will improve considerably in the near future; in particular, for 34S/32S the precision will soon reach 0.04%o IRMe principln Th i . Ms i e read measuro yt absolute eth e isotopic calibratioe ratioth f so n materials listed above determinatione Th . carriee b n sdca out in two phases: in the first phase the isotopic ratios will be determined by using the Avogadro procedur equipment d secone ean th n usiny i d b d gan , synthetic isotopic mixtures.

Calibration materials of other light elements, including Li, B, Si, Cl, whose natural isotopic variation e use ar n sgeochemistryi d , should als e considereb o r accuratfo d e (re)determination of absolute isotopic ratios. For silicon, it is suggested to determine the Si/d SSi/an i Si ratio f NBS-2o s 8 (SiO2): this materia e useb principln i dn ca l r efo calibratio30 2829 28f silicono n isotope measurements.

4. Data reduction procedures

For high precision measurements of stable isotope variations, such as those carried r atmospherifo t ou c tracintercomparisoe th er gasesfo d an , f resultno s between laboratories, there is a need to adopt common procedures of data reduction. The following is therefore recommended: date Th a reductio ) (i n procedure use modery db n automated mass spectrometers should be clearly documented and accessible to the user, in case he wants to change it. At presen software th t r datfo ea reduction incorporate computern di s drivin e masgth s spectrometer s i oftet saccessible no e ncomputatio th d an , n procedurt no s i e sufficiently documented.

(ii) The data reduction procedure should be based on a single consistent set of assumptions for: 13 !2 18 17Î6 16 e C/th C ,) (a O/d O/Oan O absolute isotopic V-PDe ratioth n i s B CO2, i.eCOe th . 2obtained from V-PD treatmeny Bb t with H3PO4 100 %t 25°C,a and (b) the relationship between 17O and 18O in terrestrial materials.

desirabls i t I agre o ecommot a n f valueeo o t thesr nse sfo e parameters which appeae th n i r correction algorithms 18e O/th 16r OFo .ratio value th , f 0.0020883eo , base Baertschi'n do s valu r V-SMOWefo , enriche 30.9%y db 0 (recommended valu f V-PDeo . SMOWBvs d an ) by another 10.25%o (recommended value for the isotopic fractionation in CO2 extraction from CaCO3), seems to be the best assumption. Values for the othei parameters are reported in papers include thesn i d e proceedings (Alliso t al.,e n Gonfiantin t al.).e r ihig Fo h precision isotopic measurements computatioe th , n procedur datr efo a reductio valuee th d snan adopted for the above parameters should be clearly documented by the authors.

futuren I , data reduction procedure gaser sfo s other than CO2, suc Ns ha 2 O, will need developede b o t .

10 Note e summarTh : prepares . Stichlerywa W y d b . Gonfiantin R , . RozanskiK d e an i W . would like to acknowledge with thanks the contribution of all the participants in the fifth IAEA Meeting on Stable Isotope Reference and Intercomparison Materials, and in particular of the Chairmen of the four working groups which elaborated the recommendations, i.e. Messrs. C.E. Allison, T. Coplen, P. De Bièvre and H.R. Krouse.

References

C.E. Allison, R.J. Francey, and H.AJ. Meijer - Recommendations for the reporting of stable isotope measurements of carbon and oxygen in CO2 gas. These proceedings. P. Ba^rtschi, 1976 - Absolute !8O content of Standard Mean Ocean Water. Earth. Plan. Sei. Letters, , 341-34431 . . GonfiantiniR . Stichler. RozanskW K , d an , - Standardi d intercomparisoan s n materials distribute e Internationath y b d l Atomic Energy Agenc r stablfo y e isotope measurements, These proceedings. R. Hagemann, G. Nief, and E. Roth, 1970 - Absolute isotopic scale for deuterium analysi natura1 s0 l waters. Absolut rati H r SMOWeD/ ofo . Tellus, , 712-71522 . G. Hut, 1987 - Consultants' Group Meeting on Stable isotope reference samples for geochemica hydrologicad lan l investigations, IAEA, Vienna, 16-18 September 1985. Report Directore th o t General, IAEA, Vienna. pp 2 4 , W.-J , B.-LLi ., D.-Q .Ni . JinQ-Ld ,an . Zhang, 198 8Measuremen- absolute th f to e abundance of oxygen-17 in V-SMOW. Kexue Tongbao (Chinese Science Bulletin), 33, 1610-1613. K. Rozanski, 199 1- Consultants ' Group Meetin C-1n go 4 reference materialr fo s radiocarbon laboratories, IAEA, Vienna, 18-20 February 1991. Report to the Director General, IAEA, Vienna. pp 4 5 ,

11 Appendix

Starting 1994 the description of the mtercomparison materials distributed by the IAEA, was changed as indicated in the following list. The new sample code indicates the directly affected isotope.

Old Sample Code New Sample Code Substance

IAEA-C1 IAEA-CO-1 calcite

IAEA-KST IAEA-CO-8 calcite

IAEA-NZCH IAEA-CO-9 carbonate

Sucr.Anu IAEA-CH-6 sucrose

PEF-1 IAEA-CH-7 polyethelene

IAEA-N1 IAEA-N-1 ammonium sulfate

IAEA-N2 IAEA-N-2 ammonium sulfate

IAEA-N3 IAEA-NO-3 potassium nitrate

IAEA-NZ1 IAEA-S-1 silver sulfide

IAEA-NZ2 IAEA-S-2 silver sulfide

Sulfur IAEA-S-4 sulfur

12 STANDARD INTERCOMPARISOD SAN N MATERIALS DISTRIBUTE INTERNATIONAE TH Y DB L ATOMIC ENERGY AGENCY FOR STABLE ISOTOPE MEASUREMENTS

R. GONFIANTINI, W. STICHLER, K. ROZANSKI Isotope Hydrology Section, International Atomic Energy Agency, Vienna

Abstract - The Agency's current programme for the preparation and distribution of standards and intercomparison materials for the calibration of the measurements of stable isotope natural variations reviewed.is need carryThe to determinations out absolute ofthe isotopic ratios standardsin discussed.is This wouldfacilitate long-termthe quality control of existing standards as well as the establishment of new standards when the present ones are exhausted.

1 - Introduction r manFo y year Internationae sth l Atomic Energy Agenc Viennn yi bees aha n dis- tributing calibration and intercomparison materials for the interlaboratory calibration of measurements of stable isotope ratio variations in natural compounds. These materials were prepared and initially tested by laboratories having the reputation of performing high precision isotopic measurements, the collaboration of which is here gratefully acknowl- edged calibratioe Th . intercomparisod nan n materials were initially n intendei e us r dfo isotope geochemistr hydrologd yan yfield- whicn si IAEe hth Aactivels i y engaget bu d- soon they were adopted also by scientists working in other fields, such as environmental studies, biology, food-stuff adulteration, etc. calibratioe Th intercomparisod nan n materials typically consis naturaf o t l minerals and compounds commonly studied in isotope geochemistry, having the desired characteristics of isotopic composition, homogeneity, chemical purity and stability. Sometimes, however, it is not possible to find suitable natural materials, and it becomes necessary to prepare synthetic compounds with the required characteristics. Standards, calibration and intercomparison materials can be grouped under the following categorie definitionsd san : (i) Primary reference standards', natura virtuar lo l materials versus which generay ,b l agreement, the relative variations of stable isotope ratios in natural compounds are expressed, usin wele gth l know °/on6 o notation: o°/oo=hr-i xiooo (D \R* ) 13 12 15 14 18 16 34 32 where Rs and RR are the isotopic ratios (WH, C/ C, N/ N, O/ 0, S/ S) respectively in the sample and in the primary reference standard. (ii) Calibration materials: natural and synthetic compounds which have been carefully calibrated versus the primary reference standards, and the calibration values have been internationally agreed and adopted. They are used to fix the o-scale "zero" and hence express the results of isotopic composition determinations versus the common primary reference standards checo ,t mase kth s spectrometer linearity ove widra e rang isotopif eo c ratio variations calibrato t d an , e laboratory standard continuoua r sfo s internal checf ko the results.

13 (iii) Intercomparison materials: natura synthetid lan c compounds which provide eth mean r laboratoriefo s periodicallo t s y chec overale kth l qualitmeasuremente th f yo s performed, includin long-tere gth m reproducibilit samplf yo e preparation fro mvarieta y of materials comparison i , n with those obtaine othey db r laboratories. Intercomparison materials cover a bioad spectrum of chemical compositions and a wide range of isotopic ratios. Their isotopic compositio computes ni averaginy db resulte gth f severaso l laboratories obtaine intercomparison di individuan i rund an s l assays, after eliminatiof no outliers using the 2a interval criterion. Updated lists of these results are available at IAEA and are provided on request. In this article, we review the status of the programme on standards and intercomparison materials distributed by IAEA for hydrogen, carbon, nitrogen, oxygen and sulphur stable isotope analyses.

2 - Primary reference standards e primarTh y reference standards use expreso dt s natural variation f isotopio s c compositio five th e f nelemento s listed abov SMOWe ear atmospherid , PDBan T ,CD c nitrogen.

SMOW- 2.1 (Standard Mean Ocean Water) SMO uses Wwa d define s sincwa t eCraii y db 196n gi referenc1s a expreso et e sth relative variation 2 f 18H/sd o O/ !Han 16 O ratio naturan si l waters. Shortly afterwards wa t si adopte referencs da expreso et variatione sth same th f e o sisotopi c ratiol naturaal n si l materials, and then also those of the 17O/16O ratio. The latter isotopic ratio is of particular interes extraterrestriar tfo l materials. Before 1961, the reference standard mostly used for oxygen isotope determinations obtaine2 CO e treatmeny th d b s froB wa mPD t with phosphoris ha . Thi2 c% acis0 CO d10 an oxygen isotopic composition very close to that of the CO2 equilibrated with SMOW at 25°C (see section 2.2) thereford ,an switco et h fro SMOo t mt i Wreferencs a relativels ewa y easy situatione Th . , however rathes ,wa r confusing, with reference standard notationd san s differen almosr fo t t each laboratory (Gonfiantini, 1981). Craig'e Th s definitio SMOf no watebasee s Wth wa n rdo standard know NBS-1s na , watea r sample fro Potomae mth c River, Washington D.C., originally depositedl al s a , samples wit same hth e letter Nationacodee th t ,a l Burea Standardsf uo , toda Nationae yth l Institute for Standards and Technology (NIST) of the U.S. Department of Commerce in Gaithersburg, Maryland. Craig gav followine eth g relationships between SMOd Wan NBS-1: 2 2 1 ( H/'H)SMOW=1.050( H/ H)NBS.1 18 15 18 16 ( 0/ 0)SMOW=1.008( 0/ 0)NBS., Craig also evaluated the absolute isotopic ratios of SMOW: 2H/LH = (158 ± 2) x 10"6, and 18O/16O = ( 1993.4 ± 2.5) x KT6. SMOW is in principle an excellent primary reference standard especially for water, becaus oceae rathea eth s nha r uniform isotopic composition, containe th f so abou% 7 9 t water present on the Earth crust, and is by far the major source and sink of all waters taking part in the hydrological cycle. For hydrogen, the ocean is also the largest reservoir on the Earth crust.

14 However majoa , r disadvantag definee th f eo d exist a SMO no s a td thas di W t i twa real water sample versus whic measuremente hth s coul directle db y calibrated additionn I . ,

isotopie th c compositio NBS-f no quits 1wa e different from tha SMOWf to thid s,an could

I8 introduce a non-negligible error in fixing the "zero" of the 8H2 and O5 scales which would had been particularly prejudicial for the measurement intercomparison of ocean water samples t least. no Las t , therbu t e were some doubts abou state th tf conservatio eo f no NBS-1. All these problems wer t lasea t overcome wit preparatioe hth f V-SMOWo n (see section 3.1), which, having an isotopic composition practically identical to that of the defined SMOW, becam factoe eprimare d th y reference standar expreso dt s hydroged nan oxygen stable isotope variations.

2.2 - PDB (Peedee Belemnite) PDB consisted of calcium carbonate from the rostrum of a Cretaceous belemnite,

Belemnitella americana, from the Peedee formation of South Carolina. The CO2 obtained treatmeny b froB mPD phosphori % t wit0 10 h initialls c ° C5 aciwa 2 t da y adoptea s da reference standar palaeotemperaturdn i e investigations base oxygen do n isotope variations in calcite and aragonite deposited by marine organisms: PDB had the same 18O/16O ratio as the calcium carbonate deposited in thermodynamic equilibrium in modern "average" marine water at 16.5°C (Epstein et al, 1953; Craig, 1965). Shortly afterwards, PDB was also adopted to express carbon isotope variations (Craig, 1953). PDB has isotopic ratios close to those of limestone of marine origin, and is considerably enriched in C with respect to organic carbon compounds. In particular, PDB's 13 13C/I2C isotopic ratio is very close to that of bicarbonate dissolved in the ocean. Ocean

bicarbonate largely controls the carbon isotopic composition of atmospheric CO2, which represent firste th .f o s Atmospherionl% ocead y2 an 2 n bicarbonatCO c e provide eth starting materiaprocessee sin e mosr th th kfo f d o t an l s involve carboe th n di n biogeo- chemical cyclecarbor fo s i : nthus B isotopePD , equivalene sth f SMOo t hydroger Wfo n and oxygen isotopes.

Craig (1957) evaluated the isoiopic ratios of the PDB-derived CO2 in order to establish the correction equations for the interference of ions of the same mass in mass spectrometric determinations. From measurements previously performed by Nier (1950) he derived the following values: !3C/!2 C11237.= 2lO"x 6,18O/16 O207= 9 KT x 1710" x d O/0 6,an 6 I6.38 O= It shoul notee db d thacarboe th t n isotop ePDB-derivee ratith f oo correspond2 dCO o st tha f PDBo t oxygeo , whiltw e neth isotope ratioisotopie th differene o sar t e c fracdu t - tionation durin phosphorie gth c acid treatment. These values were indirectly derivey db Craig from measurements performe Niey db 1950)r( section I . thif o sn7 paper other values reportee PDB'ar e th r dsfo isotopic ratios base indirecn do t determinations. 18 Craig (1965) also evaluated that PDB was enriched of 30.6 °/m in O with respect definee th o t d SMOW, and, taking into accoun isotopie th t c fractionations occurrinn gi the CO2 preparation from the CaCO3 and in CO2-water equilibration, he computed that the PDB-derive s enrichewa 2 f 0.2do dCO witm 02 equilibrate°/ h CO respec e th o t td with SMOW at 25°C. Recent evaluations have indicated slightly different values, i.e. 30.9 / 0 œ for PDB versus V-SMOW, and 0.27 °/M for the PDB-derived CO2 versus the SMOW- equilibrate (Hut2 dCO , 1987). exhaustes i B PD d from long bees ha ntimet i kept primars ,bu ta y reference standard expreso t naturae sth l variation f carboso n isotope naturan si l compound thosd san f eo oxygen isotope carbonatesn si , becaus greae eth t majorit publishee th f yo d results were referred to PDB. 15 Atmospheric- 2.3 nitrogen Atmospheric nitrogen, with a very homogeneous isotopic composition all around the world (Mariotti, 1983) excellenn a s i , t primary reference standar r nitrogedfo n stable isotope variations. Thi characteristia s si c commo l ineral o tnt gases which hav verea y long residenc eatmosphere timth n ei thereford ean wele ear l mixed. Atmosphere is the largest terrestrial reservoir of nitrogen, and it is also the main sourc sind thif ean k o s element taking par naturan i t man-controlled an l d processes (production of fertilizers).

2.4 - CDT (Canyon Diablo Trouite) CDT consisted of FeS (troilite) present in the iron meteorite of Canyon Diablo, Arizona. Meteoritic sulphur was taken as reference standard because its S/ S isotopic 34 32 ratio exhibits only small variations and corresponds quite well to the average isotopic ratio of terrestrial sulphur (Macnamar Thoded aan , 1950). From this primordial value, isotopic fractionations in geochemical processes started to build up the differences of sulphur isotopic composition observe terrestrian di l compounds (Aul Kulpd an t , 1959; Thodt ee al, 1962). CDT has a 32S/34S ratio of 22.22 (Thode et al, 1962), and is considerably depleted (abouS in34 Q0 ly^2 t with respec o marint t e sulphate, which isotopicall e mosth s tyi homogeneous sulphur reservoir on the earth crust. Like PDB, also CDT is exhausted, but it has been kept as a reference standard because all the result published for a long time were expressed versu. sit

Calibratio- 3 n materials Among the four primary reference standards listed above, SMOW was never physically available, while PDB and CDT were exhausted for long time. In order to make possible the measurement calibration versus these reference standards, IAEA distributes materials whic turn hi n have been carefully calibrated versu primare sth y reference stan- calibratioe dardsth d an , n values have been internationally agree adoptedd dan . These calibration materials include two water and one calcite samples, which are currently availabl distributionr efo silveo tw rd sulphid,an e samples which wil reade lb y soon. They are reported in Table 1 and described below. For oxygen-18, the interrelations between calibration materials are also shown in Fig. 1.

3.1 - V-SMOW (Vienna Standard Mean Ocean Water) and SLAP (Standard Light Antarctic Precipitation) Thes wateo etw r sample distributee sar calibratioe th r 2e dH/'Hfo th f no , i8O/!6d Oan 17O/16O variation measurements. V-SMO n isotopia s Wha c composition practically identical to the SMOW defined by Craig (1961), while SLAP is considerably depleted in heavy isotopes with respec V-SMOWo tt absolute Th . e 2H/'H, 18O/ 1716d O/Oan 16O isotopic ratios of V-SMOW and SLAP are shown in Table 1. The maximum density of V-SMOW 999.97s i 5 kg-nï3 (Girar Menachéd dan , 1972). V-SMO prepares W . Craiwa . Weis H R d gy db san (Scripp s Institutio f Oceanno - ography, La Jolla, California) by mixing distilled ocean water (collected in the Pacific Ocea Juln i y 196 t latitud7 a longitudd an ° e0 e 180°) with small amount othef so r waters until reaching the desired isotopic composition. SLAP was prepared from South Pole firn collected by E. Picciotto (Université Libre de Bruxelles) at the Amundsen-Scott Station.

16 At present, V-SMOW is the main calibration material for determinations of hydrogen and oxygen isotopic variation naturan so l compounds factoe d d als,an maie oth n primary reference standar expreso dt s these variations fac n l laboratorieI .al t V-SMOe sus r Wfo calibration purposes and assume that V-SMOW and defined SMOW are identical - even fulle b t y itrueno f thiy . sma Als paperon i s wher reference eth e standar indicateds i d only by SMOW, this should in reality be read V-SMOW.

TABLE 1

Calibration standards distributed by IAEA

Name Nature Isotopic o°/oo Reference ratio standard

V-SMOW water 2H/1H (155.761 0.05) x 10^(1) 0.00 V-SMOW (155.751 0.08) x 10^(2) (155.6010.12) x 10^ (3)

18O/16O (2005.20 1 0.45) x 10^(4) 0.00 V-SMOW

I70/I60 (379.91 0.810^(5x ) ) 0.00 V-SMOW

SLAP water 2H/'H (89.021 0.0510^(1x ) ) -428.0 (6) V-SMOW (89.12 1 0,07) x 10"* (2) ) 10-x (3 ) *(88.8 18 . 810

18O/I6O ( 1893.91 ±0. 45) x W*(T) -55.5) 0(6 V-SMOW NBS-19 calcite 13C/12C ) 1.9(8 5 V-PDB I80/160 -2.20 (8) V-PDB ) 28.(9 6 V-SMOW

IAEA-S-1 Ag2S ^S/^S -0.30 (10) CDT 34 32 IAEA-S-2 Ag2S S/ S 27.7(11) CDT

(1 Hageman)- n étal., 1970. (2) - De Wit et al, 1980. The value originally reported for V-SMOW was later modified into 155.75X10"6 (Mook, personal communication, 1983). (3)-Tseerat, 1980. (4 Baertschi)- , 1976. (5)-UetaL, 1988. internationay B - ) (6 l agreement (Gonfiantini, 1978). (7) - Computed from V-SMOW using 518O - -55.50 %,• internationay B - ) (8 l agreement (Hut, 1987). 18 Compute- ) (9 0dB V-SMO. Ousin 30.PD vs = r m 9gfo °/ W (Hut, 1987). (10) - Proposed (see in the text and Stichler and Gonfiantini, 1994). (11- Provisiona) l valu Robinsoy b e Brenninkmeijed nan r (1987, 1990). IAEA-S- s unde2i r calibration.

17 CO2from V-PDB with H3PO4 100 % Î I I -0.27 CO2 equilibrated I I with V-SMOW at 25°C 1 \ t 1 10.25 -1.93 COJromNBS-19 1 1 \ 1 with H3PO4 100 % 1 1 1 j 1 1 ! 1 1 V-PDB 412 1 1 Î 1 1 -2.20 1 1 -55.50 4 NBS-19 30.91 1 1 1 1 1 1 1 1 V-SMOW 1 1 1 1

2 equilibratedCO 1 1 -55.50 with SLAP at 25 °C 4 1 1 1 SLAP

Fig. 1 - Internationally agreed values for oxygen-18 concentration differences between calibration materials, and for some related fractionation factors (in italics). Positive values give the relative 18O enrichment of the upper compound, and negative values the relative 18O depletion lowee th f ro compound valuel °/n i 00Al .. e sAdaptear d from Hut, 1987.

isotopie Th c compositio f SLAno P versus V-SMO -428.= W H gives i 6 0 y °/onb o 2 and o'8O = -55.50 X,, values which have been internationally agreed on the basis of analyses performe selectey db d laboratories. These exactly defined isotopic differences between V-SMOW and SLAP, which encompass most of the isotopic variations of natural Sd 18an O H scale2 6 waterse sth (Gonfiantinix usee fi ar , o dt , 1978) principlen I . e th l al , 52H and S18O determinations - and in future also those of 517O when the 17O concentration of SLAP wil calibratee lb d versus tha V-SMOf o texperimentae th d Wan - l fractionation factor hydrogef so oxyged nan n isotopes shoul normalizee db V-SMOW/SLAe th n do P scale. This recommendation applies also to hydrogen and oxygen-bearing compounds other than water. Some of the normalized fractionation factors are indicated in Fig. 1.

18 NBS-19- 3.2 (calcUe) 13 1812 16 This CaCO3 sample distributes i s calibratioe th r e dfo C/th d f O/no Can O variation determinations. BeinexhausteB gPD lona r gdfo time, NBS-1 bees 9ha n indirectly calibrated versus PDB. By international agreement, the isotopic composition of NBS-19 versus a hypothetical V-PDB (Vienna-PDB), supposed identical to PDB, has been fixed to 5 C = 1.95 °/ and Ô O = -2.20 / - The absolute isotopic ratios of NBS-19 have 13 m 18 0 œ not been determined. The 5 O value of V-PDB versus V-SMOW is 30.9 %o (Hut, 1987). 18 NBS-19 was prepared by I. Friedman, J.R. O'Neil and G. Cebula, U.S. Geological Survey, Denver, Colorado, and Menlo Park, California, by crushing a white marble slab of unknown origin (Friedma al.,t ce 1982) granulometre Th . NBS-1f yo 9 ranges fro0 m20 to 300 microns.

3.3 - IAEA-S-1 andIAEA-S-2 (Silver sulphide) Thes sampleo etw f synthetio s c silver sulphide, prepare B.Wy db . Robinsod nan A.M. C . Brenninknieijer (Institut Geologicaf eo Nuclead lan r Sciences, Lower Huttw Ne , Zealand) and initially denominated NZ-1 and NZ-2 (where NZ stands for New Zealand; Robinson and Brenninkmeijer, 1987, 1990), are available for distribution although the calibration of IAEA-S2 versus IAEA-S1 has not been completed and their 534S value versus primare th y referenc fixet internationaye ey t d b standar no T dCD l agreement. However, at the Consultants' Meeting held in December 1993 in Vienna, the value of 6 S = -0.30 34 °/m versus a hypothetical V-CDT was proposed for IAEA-S-1 (Stichler and Gonfiantini, 1994). Although this value is based on a limited number of results, it appears practically impossibl obtaieo t morna e precise valuisotopicallt no es becauswa T y ehomogeneousCD . IAEA-S-2r Fo , B.W. Robinso C.A.Md nan . Brenninkmeijer reporte provisionae dth l 34 34 32 § S value of 21.7 °/m versus CDT. The absolute S/ S isotopic ratios of IAEA-SI and IAEA-S beet 2 havye nt determinedeno .

Intel-compariso- 4 n materials The intercomparison materials distributed by IAEA are listed in Table 2 and briefly described below isotopie Th . c composition value resultf Table basee so th ar n e2 do s reported by Gonfiantini (1984) and Hut (1987), and on additional results subsequently reported to IAEA by other laboratories. However, for some intercomparison materials the numbe availablf ro e result stils si l limited. Probably the most used intercomparison material is GISP (Greenland Ice Sheet Precipitation), a water sample with an isotopic composition intermediate between those of V-SMOW and SLAP, which was provided by W. Dansgaard, University of Copenhagen. Three calcium carbonate intercomparison material availablee sar : NBS-18, prepared b Friedma. yI (1982. al t ne ) fro mcarbonatita e from Fen, Norway, collecte . TaylorB y db , University of California, Davis, and crushed by H. Friedrichsen, University of Berlin; IAEA-CO-1, prepared at IAEA from a slab of Carrara marble, Italy, provided by IMEG, Viareggio IAEA-CO-8d an ; , prepare t IAEda A fro mcarbonatita e from Kaiserstuhl, Germany, provide Geologischee dth s Landesamt, Freiburg. IAEA-CO- originalls 1wa y prepare measuremenC 14 r dfo t calibration (Rozanski, 1991; Rozansk , 1992)t al i t t e i bu , appear vera e yb gooo st d intercomparison sample als stablr ofo e isotope determinations. Its isotopic composition is close to that of NBS-19.

19 TABLE 2

Intercomparison samples distribute IAEy db A

Name Nature Isotope s o, n Reference om /oo Oi°/O0

GISP water 2H -189.73 0.87 44 V-SMOW ,8Q -24.784 0.075 46 V-SMOW NBS-18 calcite 13C 18 -5.029 0.049 16 V-PDB o -23.035 0.172 17 V-PDB IAEA-CO-1 calcite I3C 2.480 0.025 10 V-PDB I80 -2.437 0.073 11 V-PDB

IAEA-CO-8 calcite 13C -5.749 0.063 12 V-PDB 18Q -22.667 0.187 13 V-PDB 13 IAEA-CO-9 BaCO3 C ^17.119 0.149 10 V-PDB 180 -15.282 0.093 10 V-PDB 13 LSVEC Li2CO3 C -46.479 0.150 11 V-PDB 18Q -26.462 0.251 10 V-PDB USGS-24 graphite 13C -15.994 0.105 8 V-PDB

NBS-22 oil ,3C -29.739 0.124 17 V-PDB IAEA-CH-7 poly- I3C -31.826 0.114 18 V-PDB ethylene 2H -100.33 2.05 6 V-SMOW

IAEA-C-6 sucrose 13C -10.431 0.126 16 V-PDB NBS-28 quartz 18Q 9.579 0.092 8 V-SMOW 18 NBS-30 biotite o 5.243 0.245 4 V-SMOW 2H -65.70 0.27 3 V-SMOW !5 IAEA-N-1* (NH4)2S04 N 0.538 0.186 11 air N2 15 IAEA-N-2* (NH4)2S04 N 20.343 0.473 11 air N2 !5 IAEA-NO-3* KNO3 N 4.613 0.191 3 air N2 NBS-123 sphalerite 34S 17.088 0.308 13 CDT 34 NBS-127 BaSO4 S 20.315 0.357 10 CDT 180 9.337 0.319 3 V-SMOW

Z = m 6 N.B- . 0, -

5m, 0! and n are the values obtained by excluding outliers (20 criterion) after two iterations. Böhlke and Coplen (this volume) report slightly different average values, based on a larger set of measurements.

20 o otheTw r carbonat contenteO 18 material d e distributedan ar s C s13 witw lo h: IAEA-CO-9, synthetic barium carbonate prepared by C.A. Brenninkmeijer, Institute of Geological and Nuclear Sciences, Lower Hutt, New Zealand; and L-SVEC, lithium carbonate prepared by H. Svec, Iowa State University. Intercomparicon materials with carbon isotopic composition in the range of natural organic compounds are: USGS-24, graphite, prepare T.By db . Coplen, U.S. Geological Survey, Reston, Virginia, from Baker® technical grade graphite; NBS-22, oil, provided . bSilvermanyS , Universit f Californiayo Diegon Sa , ; IAEA-CH-7, polyethylene foil (known befor PEF-1)s ea , provide Gersienberge. H y db . HerrmanM d ran n (1983)e th f ,o formerly Zentralinstitu Isotopenr fü t Strahlenforschungd an - , Leipzig IAEA-C-6d an ; , sucrose, which was originally prepared by H. Polach, Australian National University, Canberra measuremenC 14 r fo , t calibration. Materials of interest for the intercomparison of oxygen isotopic analyses of silicates are: NBS-28, quartz sand NBS-30d ,an , biotite, prepare Friedman. I y db . O'Nei. ,J G d lan Cebula. The latter derives from the Lakeview tonalité of the South California batholith. r nitrogeFo n isotope measurement intercomparison IAEA distributes: N-SVEC, nitroge wits nga h 15N/14N absolute rati 366f oo 310"x 6, prepare Jun y dSved b kan c (1958), Iowa State College; two ammonium sulphate samples, IAEA-N1 and IAEA-N2, both prepared by E. Salati, Centro de Energia Nuclear na Agricultura, Piracicaba, Brazil; and IAEA-N3, potassium nitrate, prepare . MariottiA y db , Université Pierr t Mariee e Curie, Paris. IAEA-N consideree b y 3ma futurn di e intercomparison material als oxyger ofo n isotopic analysi nitratesf so , whic promisina s hi g too geochemicar lfo wated lan r pollution studies (Amberger and Schmidt, 1987). It should be mentioned that recently Böhlke et al. (1993) have proposed the use of 15 three additional calibration standards with 5 N spannin °/ 0 orde00n i 18 o go rt t fro 0 m-3 6'd normaliz8Oan scaleH 2 0 5e e 15sdones th th Nwitr wa e fo scalet hi similaa s a n i ,y rwa V-SMO SLAPd Wan . Finally, for sulphur isotope ratio (34S/32S) measurement intercomparison IAEA distributes NBS-123, sphalerite (ZnS), and NBS-127, barium sulphate. The latter, prepared by J.R. O'Neil by precipitation of sulphate dissolved in sea water, is used also for intercomparison of the 18O/16O ratio determinations in natural sulphates.

5 - Distribution policy All the above discussed calibration standards and intercomparison materials are kept at, and distributed by, the International Atomic Energy Agency. The amount stored for each materials will be sufficient for several decades if a prudent distribution policy is adopted. At present every laboratory can receive a portion of any IAEA calibration standard only once every three years. In fact, these standards are meant for calibration of laboratory standards to be used for routine work, an operation for which the amounts provided by IAE sufficiene Aar t wit currene hth t analytical procedures. intercomparisor Fo n sample t seemsi s desirable that IAEA adopt futurn si distriea - bution policy more liberal tha calibratior nfo n standards principlen I . , intercomparison samples are easier to establish because they do not require an international agreement on valuee th theif so r isotopic composition.

21 6 - Need to determine the absolute isotopic ratios of calibration and intercomparison materials In principle geochemican i , l investigation mosn wels i sa othef s a o lt r applications of environmental stable isotopes necessart no s i t i ,kno o yt absolute wth e isotopic ratios of standards and other materials used to calibrate the measurements. In fact, isotopic analyse aimine sar determino gt relative eth e variation isotopif so c ratio naturan si l compounds with respect to primary reference standard, rather than the absolute isotopic ratios themselves relative Th . e variation isotopif so c compositio measuree b n nca d very precisely with well known analytical technique isotopif o e sus basee c ratith n do mass spectrometers equipped wit doublha e inlet syste doubla d m triplr an eo e collector. The main requirements to obtain results which are not only internally consistent but also directly comparable with those obtaine othey db r laboratories thae measuree ar , th t - ments are well calibrated versus the primary reference standards through the internationally accepted calibration materials lattee Th .r should hav isotopin ea c composition whics hi homogeneous at the level of milligram sample size and remain constant in time. Finally, laboratories should adopt consistent method correco st t result instrumentar sfo l factors, i.e. the same correlation between 617O and 018O and the same absolute isotopic ratios for the reference standard. Measurin absolute gth e isotope ratios wit accuracn ha t leasya t equal relativn i , e terms, to the best accuracy achievable in determinations of relative isotopic ratio variations, is the only way to fix forever and unambiguously the isotopic composition of calibration and intercomparison materials, and to make it possible to periodically check whether changes have occurred since the preparation. Belo discuse ww circumstancee sth whicn si knowledge hth f absoluteo e isotopic ratio calibratiof so n materials woul helpfule db .

Changes- 6.1 of isotopic composition during storage happey Ima t n that ther indicatione ear doubtr so s tha givea t n calibration material doe t hav expectee sno eth d isotopic composition. havThiy sema changed slightlr yfo reasons which wer t foreseen eno t adequatel no r o , y considered, whe materiae nth s wa l initially stored. The most direct way to check whether such a change occurred is re- determining the absolute isotopic ratios and compare the results with those of previous determinations. But if previous results are not available, it will be difficult to prove with any confidence and to quantify any eventual change of isotopic composition. As a consequence measuremene ,th t re-calibration will become difficul perhapd an t s uncertaind ,an the isotopic results may be poorly comparable with the previous ones. A chang isotopie th f eo c composition during storag reaa es i l risk alreadt :i y happened in the past that some calibration materials had to be abandoned because there was the suspicion that their isotopic compositio changedd nha casee . instancer th Thi s ,fo swa f ,o NBS-1 (water), with respec whico t isotopie hth c compositio SMOf no originalls Wwa y defined by Craig (1961). Another case is that of the NBS-20 (Solenhofen limestone), which was used in the past to calibrate the 13C/12C and 18O/I6O measurements versus PDB on the basis of the calibration made by Craig (1957). The reason for these changes of isotopic composition - which on the other hand were small and difficult to be definitely proved - is supposed to be poor storage conditions, allowing isotopic exchanges with

atmospheric vapour and CO2 and, for water, also evaporation from leaky containers.

22 Imperfect- 6.2 homogeneity isotopic ofthe composition y occuIma t r that some calibration materials resul r becomo t e isotopicallt yno homogeneous teste spitn th i , sf etimo made th preparationf eo t ea alsy thae ob ma t I . isotopic inhomogeneities were introduced when these materials were split into portions for storage. This may have happened, for instance, to some intercomparison materials kep IAEAt ta distributioe ,th whicf no discontinue s hwa suspecteo t e ddu d inhomogeneities in isotopic composition. Re-determinin absolute gth e isotopic ratio differenn si t portions e suspecteoth f d material would allo wdegree checa th f f homogeneitko eo e th d yan identificatio thosf no e portions with isotopic values different fro expectee mth d ones.

6.3 - Establishment of new calibration materials Although calibration materials are usually prepared in quantities which assure distributio morr nfo e tha yearn nte leastt sa time ,th e will come when they wil exhaustede lb , and the preparation of new materials will become necessary. In order to assure a continuous consistenc measuremenn yi t intercalibration isotopie th , c compositio caliw ne - e th f no bration materials with respec exhaustee th o t d ones shoul fixee db d wit bese hth t possible accuracy. Besides carryin calibratiomateriala w t gne ou e th sf noneversuo d ol se whesth n these approach exhaustion e isotopith , c compositio materialw e fixene b f n no d ca s unambiguousl independentld yan y throug measuremene hth f theio t r absolute isotopic ratios.

6.4 - Correction equations for interference of equal-mass ions 1318 0e 0d Th Can O correction equation interferencr sfo equal-masf eo 2 s CO ion r sfo mass spectrometric determinations have been reporte Craiy db 957)1 g( , Gonfiantini (1981) and Santrock et al. (1985). The correction terms are more important for the computation of 6 C from 645, because the contribution of C O O ions to the mass 45 ion peak is 13 13 16 + 12 16 17 + relatively large Cth ealmos f - Oabundance o n 2io % t7 . The 17O/16O variations cannot be directly measured and are obtained from the I8O/16O variations, i.e. from the 8I8O determinations. The correction equation usually adopted is: = C 5 ., ,„ 13 13 17 1316 12 where isotopie /?= O/ th 17 1 3= e C/ R O ar d c Cmachinratioe an th n si e standardd an , 17 18 Ô= a O/Ô O. This espressio gooa s ni d approximatio relatioe th f no n A17(5//?17W= (^i8(s/^i8(A))"> wher subscripte eth s indicate reference sample th th equatiod n i ean s ea n I8 16 (1), and RI% = O/ O. 045 is defined in the usual way as in equation (1), i.e. as the relative difference betweestandarde th samplratioe n n masmasi o e th io t nd th n 4 s5 i ,s s4 e4 an being 7?45 = A13 + 2A17. 13 n examinfactor 0w R e d Cth no t effectn an correctiose i en Le th R errorf , so a n so n equation usualls i a , y take equas na 0.5o t l , i.e. assuming tha naturan i t l processee sth 17O/16O variations are the half of 18O/16O variations. In reality, a may range from 0.50 to 0.5physico-chemican i 3 l processes averagn a d an ,e valu 0.516f eo 0.0034± bees 3ha n observe ratie th or dbetweefo n 8 176d 18OOan variation naturan si l compounds (Matsuhisa et al., 1978). Using for a this more appropriate value, an additional correction of -0.01 1 8 13 %o for a 10 °/m increase of o' o is introduced by equation (2) in the § c computation with respec usinf o t0.5 = ga . Wit - 0.53 ha correctio e th , n becomem °/ 0 1 s a -0.02r fo m 0°/ increas 0f e18o O.

23 TABLE 3

Isotopic ratios of CO2 derived from V-PDB wit 25°t ha HC% 3P00 10 4

13 12 c/ c 11237.2x10^ (1) 11194.9x10^ (2)

6 180/,6Q 2079 xlO" (1) 2088. 310"x 6 (3)

170/160 379.95 X1Q-* (1) 387.95x10"* (4)

(1) From Craig (1953) (2) Computed from the value measured by Zhang and Li (1987) for NBS-20 and assuming o13C(NBs-2o/v-FDB reportes a -01 - 6 X )= Crai y db g (1957). Compute) (3 d fro value mth e measure Baertschy db V-SMOr ifo W (1976) wit enrichmenn ha f to 41.5 %„ resulting from the from the enrichment of PDB vs. V-SMOW (30.9 °/(>0) and the fractionation factor for the CO2 extraction (10.25 °/0()) (see Fig. 1). Compute) (4 d fro value mth e reporte etal.i L y db (1988 V-SMOWr )fo , wit same hth e enrichment multiplieO 18 factor fo 0.516s y a rdb .

613C (Zhang)- -513C (Craig), o/ '00

-0.05

-0.10

'00

n 0 13 Fig Differenc- .2 e (Aô C /œ) between S C obtained fro wit5 4 m0 h equation d (2)an ,3 usin] R r gfo

value7 1 # s computed from determination Zhany sb coworkerd gan NBS-2n so V-SMOd 0an W and values reporte: Craiy D db ; -10: 0 g : B (Tabl ;C ; . X, followss 0 ea 183) 0 Oe -2 sar : A : +10; E: +20.

24 The probable error on /?B is about 0.5 %, and its effects are negligible.

More importan effecte th errorf e o sar t Rn s17o . Assumins i g tha „ erroe R th tn ro 0 , thi±2-% s3 entail additionan sa l correctio 0.02o t p u 2 eacr f °/no /o00fo 0 ho 1 variatiof no 18 13 8 the term (Ô45-«ô O), i.e., approximately, of the term (ô C-ô' O/2). Although this correctio appeay nma r small certair fo , n analytical t requirementnegligibleno e b y .ma t si Thus, in order 10 reduce it, a more accurate determination of the absolute isotopic ratios of calibration materials especialld an , 17e O/th f I6yo O ratio neededs i , .

instancer Fo obtaine2 adoptiny CO ,b isotopie e th dth r froB gfo cmPD ratio s values propose Craiy db g (1957) thosr ,o e whiccomputee b n hca d from measurement NBS-2n so 0 and V-SMOW performe Zhany db coworker d gan s (Tablobtaie w , ne3) 813C whicy hma differ significantly (Fig. 2). This difference is almost entirely due to the R{1 values, as indicated above. 18 equatioe Th transforo nt tripla int6 4 mr 5 o fo e 0, collectois O r mass spectrometer:

being /?46 = 2/?l8 + 2/?13/?17. The correction terms are small, and the influence of current errors on the coefficients a, /?13, R17 and /?18 is generally negligible. 13 Differences between 5 C values corrected with different algoritm differend san t /?13 and /?17 values used by the softwares installed in two mass spectrometers, have been discussed by Francey and Allison (in Rozanski, 1992).

Interferenc f equaeo l mass ions occurs als masn oi s spectrometric analysi f SOso 2. 32e S16Th O18O+ ionpeakn mas e resio se th represen6 ,th f ts 6 o bein % g3 tconstitute8. y db 34 16 + the S O2 ions (the contributio othef no r mas ion6 s 6 les tripls a si r s thaeFo . t 0.0%) 2 collector mass spectrometer the correction equation is:

&*C I 1 , 2/?!8 ] g ^ÜXlSp» (A\ ) 4 < = 1+ S ~~ 5 ~ 6--—— ° ° 34 34 34 32 influence Th correctioe = S/ 4 errorf th 3 eo A n S o n so n term transforo st int6 6 m0 o 34 5 S, with /?66 = /?34 + 2/?lg, is negligible. A 1 % variation of the RIS ratio of the SO2 analyzed - which may be produced during the sample preparation - results in an error of 0.01 ^ for each 10 %(, variation of the term (ö^ — 518O), i.e., approximately, of the term (Ô34S-Ô180). In conclusion, there is a recognized need of adopting a common algoritms for 6 correction f determinino d an , g wit accuracn ha y better tha isotopie currene th nth e on tc referencratioe th f so e standard calibratiod san n materials t possibl thif no e I .s th si n ei near future alternate th , e solution coul adopo t e d commob ta valuereferencf e o th t r nse sfo e standard isotopic ratios recommendes a , Consultantse th y db ' Meetin stabln go e isotope standard intercalibratiod san n (Stichle t al.re , these proceedings).

7 - Accuracy of absolute îsoiopic ratio determinations e moderTh n mass spectrometer capable ar s f measurineo g natural isotopic ratio variations in gaseous compounds of many light elements with an error in the range of 0.01-0.0 hydrogenr 2 °/Fo m. erroe th , usualls ri ordee y on magnitudf ro e greater because naturae th l 2H/'H isotopjordero tw magnitudf r so o ce ration s oi e smaller tha rare nth o e t

25 most abundant isotope rati othef oo r elements. Additional, often larger,e errorb y sma introduce sample th y db e treatment prio maso rt s spectrometric analysis. In order to give an idea of the overall accuracy achievable in measuring absolute isotope ratios, we report here some results obtained on calibration and intercomparison materials errore Th . s quoted represen standare on t d deviatio

respond relativn ,i e terms ±0.3o ,t ±0.5d 2respectivelyan m 1°/ , which compare well with the current errors obtained by the best laboratories in 52H determinations in water samples.

It shoul notee db d tha severar fo t l laboratorie °/1 00 typicao e ,t whicsth p u lo h errog n rca is acceptable in many geochemical and hydrological investigations. errore Th s quotesame th ey d b author SLAr s fo ±0.0 e Par 510" x ±0.0 d an 710"x 6 6 respectively. In this case, the corresponding relative errors are greater, ±0.56 and ±0.79 0 2 ! /œ, because of the much lower H/ H ratio. 2 18 16 Baertschi (1976) used mixture almosf so t pure H 2'Hd O2an Odetermino t e eth 18O/16O ratio of V-SMOW, for which he found the value of (2005.20 ± 0.45) x KT6. The relative erro ±0.2s ri 3 %(), i.e. considerably greater tha erroe n th 0 n ri 18 O determinations, whic usualls hi y better than ±0.10 "V^,. e O/Th O rati V-SMOf oo W determine Baertschy db i allow computatiosa e th f no 18 16 same ratio in PDB and NBS-19, and in the CO2 extracted from PDB with H3PO4100%. PDB is enriched 30.9 %, in O with respect to V-SMOW, and therefore its O/ O ratio !8 16 18 18 16 s 2067.i 210"*x ; NBS-1 s deplete9i s O/it d f 2.2do O an versum 0rati°/ B s oi PD s 18 2062.6 x 10^. The CO2 extracted from calcite is enriched in O of 10.25 °/m (Hut, 1987), and therefore its 18O/16O ratio is 2088.4 x 1(T*. 17e O/Th !6O rati V-SMOf oo determine(1988). s Wal wa t e i . L They db y compared

in a double inlet, triple collector mass spectrometer first the O2 obtained from V-SMOW with that obtained from a water sample depleted in oxygen heavy isotopes, and then the forme2 CO thesy db e oxygen samples with pure graphite witobtaine2 hCO d fro msampla e of NBS-20, the 13C/12C of which was previously determined (Zhang and Li, 1987; see below). This complex procedure allowed the determination of 17O/I6O ratio of V-SMOW, which resulted equal to (379.9 ± 0.8) x 10"6. The relative error is 2.1 0/oo, i.e. an order of magnitude higher than that of 617O determinations. I3e C/Th 12C rati NBS-2f oo 0 (Solenhofen limestone, distribution discontinueds )wa determine 1987( Zhan i y db L )d usingan g mixture bariuo tw f mso carbonates enrichee ,on d othe e calibratioe th th 13n d ri r C mase an fo , th C if snn12 o spectrometric measurements. 6 They found a value of (11183 ± 1.3) x IGT , i.e. with a relative error of ±0.11 °/m. Modern 013C determination ±0.0s a havn w 2errosca n lo ea 0 /oos claime s a a r, laboratoriey db s monitoring isotopic variation atmospherin si (se2 instancr ecCO fo e Rozanski, 1992).

1 -TheWH value originally reported by De Wit et al. (1980) was later modified into 155.75 x KT6 (Mook, 1983, personal communication).

26 It is worth noting that the value of the C/ C ratio of NBS-20 reported by Zhang 13 12 and Li is considerably lower (-3.77 °/00) but much more precise than that computed by Craig (1957), (11225. basie 10"x th f ) previou o sn 6 30 ,o 3± s absolute isotopic ratio determinations mad Niey eb r (1950). Besides, knowing tha0e th tNBS-2f Co 0 versus 13 13 12 PDB is -1.06 ± 0.04 °/m (Craig, 1957), it is possible to re-compute the C/ C ratio of PDB from that reported by Zhang and Li for NBS-20, and we obtain the value of (11194.9 ± 1.4) x IQT6: this value is probably more accurate than that computed by Craig. The absolute isotopic ratios of NBS-123 (sphalerite, ZnS) have been determined by Zhan mase Dind th gr san g1989spectrometrifo ( s ga )s usina 6 gSF c analysis. They obtained for the 34S/32S ratio the value of 45805 x 10"6 with a relative error of ±0.10 °/oo> well comparable wit experimentae hth l erro 0n r34i S determinations. However indicatioo ,n n metho e gives i th n no d mase useth r s dspectrometefo r calibration beet no nthif s i sd ha ,an done, the 34S/32S value reported may be affected by an unknown systematic error. Considerable improvement absolutn si e ratio determinatio carbof no n isotopes were achieved recentl Institute th y yReferencr b efo e Material Measurementd san Joine th t tsa Research Centre of the Commission of the European Communities in Geel, Belgium. Here, the Mass Spectrometry Group headed by P. De Bièvre has published (Valkiers etal., 1993) 6 13 12 the value of (10753.3 ±0.4) x 10" for the C/ C ratio in commercial CF4 gas. The error quoted by Valkiers et al. corresponds to ±0.04 °/m in relative terms, which well compares wit erroe hth r obtaine 8n d13i C determination naturan si l compounds. Institute Th Referencr efo e Material Measurementd san Geen si probabls li place yth e wher bese eth t determination absolutf so e ratio lighf so t element isotope carriee sar t dou at present withi framewore nth Avogadre th f ko o project, which aim mora t sa e precise determinatio Avogadre th f no o number t appearsI . , however determinatioe , th tha r fo t n of absolute ratio nitrogenf so , oxyge sulphud nan r isotopes t satissituatioe ye th , t - no s ni factorybelieves i t i t easile b bu ,d n ythaca improvet i t neae th rn di future . nitroger ai r oxyged Fo nan absolute nth e isotopic ratios obtaine Valkiery db e D d san Bièvre (1993) are: 15N/14N = (3612 ± 7) x 10"6 and 18O/16O = (2072 ± 2) x lO"6. This is probabl bese yth t result obtainewhice b n hca d without calibratin measuremente gth s with artificial isotopic mixtures. The errors associated with these measurements correspond to ±2 °/m for nitrogen and ±1 %) for oxygen, which are at least one order of magnitude higher than errors quoted in routine determinations of 515N and o'8O. Research is carried out at the Institute for Reference Materiails and Measurements in

Gee orden i l achievo t r accuracn ea ordee th f . 0.0o ryin 1 °/ 00relativn i e e termth n i s determination of carbon, nitrogen, oxygen and sulphur absolute isotopic ratios. This would allow to fix the characteristics of the isotopic calibration and intercomparison materials distributed by IAEA.

References

Amberger A. and Schmidt H-L., 1987 - Natürliche Isotopengehalte von Nitrat als Indikatoren für dessen Herkunft. Geochim. Cosmoch. Ada, , 2699-270551 . Ault W.U. and Kulp J.L., 1959 - Isotopic geochemistry of sulphur. Geochim. Cosmoch. Acta, 16, 201-235. Baertsch , 197iP. 6Absolut- contenO e18 Standarf o t d Mean Ocean Water. Earth Plan. Sei. Letters, 31, 341-344.

27 Böhlke J.K., Gwinn C.J., and Coplen T.B., 1993 - New reference materials for nitrogen- isotope-ratio measurements. Geostandards Newsletter, , 159-16417 . Craig H., 1953 - The geochemistry of the stable carbon isotopes. Geochim. Cosmoch. Acta, 3, 53-92. Craig H., 1957 - Isotopic standards for carbon and oxygen and correction factors for mass spectrometric analysis of carbon dioxide. Geochim. Cosmoch. Acta, 12, 133-149. Craig H., 1961 - Standards for reporting concentrations of deuterium and oxygen-18 in natural waters. Science, 133, 1833-1834. Craig H., 1965 - The measurement of oxygen isotope paleotemperatures. In Stable Isotopes in Océanographie Studies and Paleotemperatures (Editor E. Tongiorgi), 161-182, CNR- Laboratori Geologii od a Nucleare, Pisa. De Wit J.C., Van der Straaten C.M., and Mook W.G., 1980 - Determination of the absolute ratiH V-SMOf oD/ o SLAPd Wan . Geostandards Newsletter, , 33-364 . Epstei , BuchsbaunS. Lowensta, mR. m H.A. Ured an y, H.C., 195 Revise- 3 d carbonate-water isotopic temperature scale. Bull. Geol. Soc. of America, , 131564 . FranceAllisod an . ny RJ C.E. , 199 Recommendation- 2 measuremene th r sfo reportind tan g of stable isotope measurement f carboso oxyged nan atmospherin ni c trace gase sAnne- f o x1 Rozanski Research1st 1992The K., - Co-ordination Meeting Isotopeon variations carbonof dioxide otherd an trace gasesatmosphere.e th n i Internal Report, IAEA, Vienna. pp 6 ,7 Friedma , O'NeinI. carbonatw l Cebul J.Rne d o an ., 198eTw a G. - stabl2 e isotope standards. Geostandard Newsletter, , 11-126 . Gerstenberge Hermand an . , 198rH nM. 3 Repor- intercomparisoe th n o t isotope th r nfo e standard Limestone KH2 and Polyethylene Foil PEF-1. ZFI Mitteilungen, 66, 67-83. Girard G. and Menaché M., 1972 - Sur le calcul de la masse volumique de l'eau. C. R. Acad. Se. Paris, 274, 377-379. Gonfiantini R., 1978 - Standards for stable isotope measurements in natural compounds. Nature, 271, 534-536. Gonfiantin ô-notatio, 198e R. imas Th e 1- th sd spectrometrinan c measurement techniques. In Stable Isotope Hydrology. Deuterium Oxygen-18d an e Waterth n i . R Cycle d an t (J.RGa . Gonfiantini, Editors), IAEA, Vienna, 35-84. Gonfiantini R., 1984 - Advisory Group Meeting on Stable isotope reference samples for geochemical and hydrological investigations, IAEA, Vienna, 19-21 September 1983. Report to Directore th General, IAEA, Vienna. pp 7 7 , Hageman Rotd , 197 an , Nieh E. , nR. 0fAbsolutG. - e isotopic scal deuteriur efo m analysis of natural waters. Absolute D/H ratio for SMOW. Tellus, 22, 712-715. Hut G., 1987 - Consultants' Group Meeting on Stable isotope reference samples for geochemica hydrologicad an l l investigations, IAEA, Vienna, 16-18 September 1985. Reporte th o t Director General, IAEA, Vienna, 42 pp. Junk G. and Svec H.J., 1958 - The absolute abundance of nitrogen isotopes in the atmosphere and compressed gases from various sources. Geochim. Cosmoch. Acta, 14, 234-243. Li Wenjun, Ni Baoling, Jin Deqiu, and Zhang Qingliang, 1988 - Measurement of the absolute abundanceof oxygen-1 V-SMOW7n i . Kexue Tongbao (Chinese Science Bulletin), 33,1610-1613. Macnamara J. and Thode H.G., 1950 - Comparison of the isotopic constitution of terrestrial and meteoritic sulfur. Phys. Rev., 778, 307-308. Mariott , 198A. i 3 Atmospheri- c nitroge reliabla s ni e standar r naturadfo abundancN l e 15 measurements. Nature, 303, 685-687. Matsuhis , GoldsmitaY. h J.R. Claytod an , n R.N., 1978 - Mechanism f hydrothermao s l crystallizatio quartf no kbar 5 t 250°za 1 d . Can Geochim. Cosmoch. Acta, , 173-18242 .

28 Nier A.O., 1950 - A redetermination of the relative abundance of the isotopes of carbon, nitrogen, oxygen, argo potassiumd nan . Phys. Rev., 789-793, 77 . Robinson B.W. and Brenninkmeijer C.A.M., 1987 - Sulphur isotope standard NZ-1. Final Report Technicale IAEAth o t r fo Contract . 4320/TC,No . pp 3 Robinson B.W Brenninkmeijed .an r C.A.M., 199 0Sulphu- r isotope standard NZ-2. Final Report to IAEA for the Technical Contract No. 5654/TC, 2 pp. RozanskiK., 1991 -Consultants' Group Meeting on C-14 reference materials for radiocarbon laboratories. Report to the Director General, IAEA, Vienna, 54 pp. Rozanski K., 1992 - The 1st Research Co-ordination Meeting on Isotope variations of carbon dioxide and other trace gases in the atmosphere. Internal Report, IAEA, Vienna, 76 pp. Rozanski K., Stichler W., Gonfiantini R., Scott E.M., Beukens R.P., Kromer B., and Van der Flicht J., 1992 - The IAEA 14C intercomparison exercise 1990. Radiocarbon, 34, 506-519. Santrock J., Studley S.A., and Hayes J.M., 1985 - Isotopic analyses based on the mass spectrum of carbon dioxide. Anal. Chem., 57, 1444-1448. Thode H.G., Monste Dunford an , rJ. d H.B., 196 Sulphu- 2 r isotope geochemistry. Geochim. Cosmoch. Acta, 25, 159-174. Tse R.S., Wong S.C., and Yuen C.P., 1980 - Determination of deuterium/hydrogen ratios in natural waters by Fourier transform nuclear magnetic resonance spectrometry. Anal. Chem., 52, 2445. Bièvre ValkierD , d 199eP. an 3. Near-Absolut- s S e (isotope) mass spectrometry. Isotopic abundances (and atomic weight) determinations of nitrogen and oxygen. Report GE/R/MS/9/93, IRMM, Commissio Europeae th f no n Communities, Geel. Valkiers S., Schaefer F., Van Nevel L., and De Bièvre P., 1993 - Near-absolute gas (isotope) mass spectrometry: Isotope abundanc atomid ean c weight determination f boroso carbond nan . Report GE/R/MS/8/93, IRMM, Commission of the European Communities, Geel. Zhang Qingliang (T.L. Chang Dind )an g Tiping, 1989 Analysi- reference th f so e material NBS-123 and the atomic weight of sulphur. Chinese Science Bulletin, 34, 1086-1089. Zhang Qingliang (T.L. Chang) and Li Wenjun, 1987 - Mass spectrometric determination of the absolute isotopic abundance of carbon in NBS-20. Proc. 2nd Beijing Conf. and Exhib. on Instrum. Analysis, 391-392.

29 REPORTIN STABLF GO E CARBON, HYDROGEN, OXYGED AN N ISOTOPIC ABUNDANCES

T.B. COPLEN US Geological Survey, Reston, Virginia, United State f Americso a

Abstract

eliminato T e possible confusio reportine th n ni isotopif go c abundance non-correspondinn so g scales Commissioe th , Atomin no c Weight Isotopid san c Abundances recommendee th t da 37th General Assembly at Lisbon, Portugal that (i) 2H/1H relative ratios in all substances be expressed relative to VSMOW (Vienna Standard Mean Ocean Water) on a scale such that iH/1}! of SLAP (Standard Light Antarctic Precipitation) is 0.572 times that of VSMOW, (ii) 13C/12C relative l substanceratioal n i s expressee b s d relativ VPDo et B (Vienna Peedee belemnite) on a scale such that 13C/12C of NBS 19 carbonate is 1.00195 times that of VPDB, (iii) 18O/16O ratiol substanceal n si expressee sb d relativ eitheo et r VSMO Wr VPDo n Bo scales such that 18O/16O of SLAP is 0.9445 times that of VSMOW, and (iv) the use of discontinuede b B SMOPD d Wan . Furthermore reportef i , d isotopic abundance mineraa f so l or compound depend upon isotopic fractionation factors, users should (i) indicate the value l sucal hf o isotopic fractionation factors r (iio , ) indicat isotopie eth c abundance obtainer dfo a reference material of the same mineral or compound.

. COMMEN1 T

Abundances of stable hydrogen, carbon, and oxygen isotopes in geochemical and environmental studies are generally expressed in parts per thousand (%o or per mil) difference from a standard. Thus, for the oxygen isotopic composition of a sample x,

r 18o

160 = ) %o n (i ô18O X - I 1000. 18o"

16Q Standard

The standard may be an actual reference material or a hypothetical material whose isotopic abundanc assigniny b t se isotopis n ei ga c compositio existinn a o nt g reference material.

Irregularities concerning the choice of the standard have arisen for hydrogen, carbon, and oxygen isotope. 2] Friedma , [1 s O'Neid an n ] poin t [1 thal ou tt some laboratoriee ar s "tied eaco "t h othe acceptancy b rvalue 0 e f certaith o s f eo n comparison materialse Th .

31 situation for the SMOW (Standard Mean Ocean Water) standard has become increasingly aggravated e SMOTh .W standar s originallwa d a yhypothetica l water sample with abundance f stablso e hydroge oxyged nan n isotopes simila thoso t r f averageo e ocean water abundances It [3]. f stablso e hydroge oxyged nan n isotopes were define1 S termn di NB f so water distributed by the U.S. National Bureau of Standards (now National Institute of Standard Technology)d san :

2 1 2 1 ( H/ H)SMO = 1.050(W H/ H)NBS1

and

18 16 18 16 ( 0/ 0)SMO = 1.008(W 0/ 0)NBS1.

Subsequently . Taylo. Epstei S H Californi,e d th f an rn o a Institut Technologf eo y (Pasadena, California) used a hypothetical standard that they also called SMOW and defined it by assigning a 0 1 OO value of +15.5%o to their laboratory reference material, a sample of Potsdam Sandstone [1]. Thus, their oxygen isotope scal defines ei y db

t f 1 ^ ^Potsdam Sandstone/SMO = W+15. 5 %o.

Additionally Internationae th , l Atomic Energy Agency (IAEA) distribute dwatea r sample they named SMOW. This sampl s neaewa r e same (buth t isotopin i )t no c compositioe th o nt original SMOW defined in terms of NBS 1 water. Thus, three independent usages of SMOW are currently observed, leading to differing 2H/1H and 18O/16O abundances with the same name. The IAEA recognized the dilemma of naming a reference water as SMOW and subsequently change e namth do VSMOt e W (Vienna Standard Mean Ocean Water). Furthermore, they recommended that abundances of hydrogen and oxygen isotopes of all materials (except marine carbonates expressee )b d using VSMOW rather than SMOW [2,4, 5, 6]. However, these recommendations have not received wide distribution and may be unknown to numerous producers of data on abundances of oxygen and stable hydrogen isotopes, particularl rapidle th n yi y expanding field f environmentao s climatd an l e studies.

A second standar uses d i reportinr dfo g abundance oxygef so n isotope f marinso e carbonates an s named i (Peede B dPD e belemnite). Becaus supple eth f thiyo s materia s exhaustedi l , some laboratories have "tied" themselves to each other by adopting 518O values of various carbonate reference materials IAEe Th A. recognize potentialle dth y serious problem that oxygen isotopic scales in different laboratories might not correspond. They recommended that abundances of oxygen isotopes of carbonates be expressed relative to VPDB (Vienna Peedee belemnite adoptiny b ) g5a O consensu scarbonat 9 valu1 S f -2.2%eo NB er ofo 18 relativ VPDo et meetina t Ba 198n gi 3Viennn i a [2]. Thus,

^ ^-*NBS19/VPD ^ B-2.2%o .

This satisfactory onlsolutiod ha y s limitenha d distributio seldos i d nman used.

32 s Becausalsi e standar th oB r carbonPD efo d a simila, r problem exist r reportinfo s g abundance carbof so n isotopes IAEe Th .A provide dsolutioa reportinno t carbof go n isotopic data on non-corresponding scales by recommending that carbon isotopic results of all materials be expressed relative to VPDB by adopting a 5 C value of +1.95%o for NBS 19 13 carbonate relativ VPDBo et : ^ CNBS 19/VPD = +1.95%oB .

This valu s adoptewa e y consensudb s [2]. Again, this viable d littlsolutioha e s ha n distributio seldos i d nman employed.

5d 18an O H value2 6 e VSMOf so rangTh e th abundancef f euppecloseo We e o li th d o ren rt s f naturallo O 18 yd occurrinan oH f2 g materials recommendiny B . 5d 18an O H value2 g0 f so referenca e materiaf o range O th 18 f abundance f l eo loweo d e clos th d an o en ert H 2 f so naturally occurring materials, the IAEA effectively normalized these scales. The IAEA selected a water sample from Antarctica and gave it the name SLAP (Standard Light Antarctic Precipitation). Three absolute isotope-ratio measurements of VSMOW and SLAP [7, 8, ] suggeste9 valuH 2 5 f -428% edo a SLAr ofo P relativ VSMOWo et 18e O/Th .16 O ratif oo SLAP has not been determined and a consensus 518O value of —55.5%o of SLAP relative to

VSMOW was adopted in 1976 [4]. Also, the hydrogen isotopic composition of SLAP relative

to VSMOW was defined by consensus to be -428%o = ; —thus 428%o., oH2 W SLAP/VSMO

resule Th f normalizatioo t thas ntwa coherence betwee O0d resultan 2H n6 s reportey db different laboratories increased dramatically [4]. In 1983 the IAEA recommended that 5H2 and 618O results of all materials be reported on normalized scales [2]. Equations for normalizatio e give nar Gonfiantiny nb Copled an ] n[2 i [6]procese Th . f normalizino s g shoul n theori d y increas e agreementh e t between experimentally determined isotopic fractionation factors of physicochemical processes. f Commissioe Th Atomin no c Weight Isotopid san c Abundance 37e th t ht sGeneraa l Assembly in Lisbon, Portugal, [10] discusse reportine dth f isotopigo c abundance recommended san d that: l valueH hydrogen-bearin2 al 0 f o s ) (1 g substance e expresseb s d relativo t e 2 VSMOW on a scale such that 0 HSLAP/VSMOW = -428%o\ (2) 013C values of all carbon-bearing substances be expressed relative to VPDB on a scale such that ^CNBS^/VPDB = +1.95%o; ) (3 518O value oxygen-bearinl al f so g material expressee sb d relativ VSMOeo t W

or relative to VPDB, defined by ô^O^s -2.2%os , on a scale B

18 19/VPD such that 0 OSLAP/VSMOW - -55.5%o; and (4) reporting of isotopic abundances relative to SMOW and PDB be discontin- ued.

Several valueisotopin a r fo s c fractionation facto physicochemicaa f o r l proces havy ma se been measured reportef I . d isotopic abundance mineraa f so compounr o l d depend upon such isotopic fractionation factors, users should:

33 (1) indicate the value of all isotopic fractionation factors employed in calculating isotopic abundances, or (2) indicate the isotopic abundance obtained for a reference material of the same mineral or compound.

readee Th s remindei r d that additiv valueô tno e ar se when converting froe scalmon o et another determinee ar t relatioe bu ,th y db n oxygenr (e.g.fo , )

5 °a/ VSMOW = ô °a/b+ 5 °b/ VSMOW + 10 5 °a/bô °b/VSMOW-

This relation should be kept in mind especially in the case of hydrogen where 5 values of several hundred occur; thus, the last term in the equation above may reach a value of several tens.

APPENDIX A; SOURCES OF REFERENCE MATERIALS

Reference materials VSMOW, SLAP, and NBS 19 may be obtained from: International Atomic Energy Agency Section of Isotope Hydrology 0 10 x P.OBo . 1400 Vienna, Austria or Standard Reference Materials Program Request RM853 VSMOWr 5fo . Room 204, Building 202 Request RM8537 for SLAP. National Institut f Standardeo Technologd san y Reques. t19 RM854S NB r 4fo Gaithersburg, Maryland 20899 USA

REFERENCES

] FRIEDMAN[1 , O'NEILL , , J.R., U.S. Geol. Surv. Prof. Pap. 440-KK (1977). ] GONFIANTINI[2 , "AdvisorR. , y Group Meetin Stabln go e Isotope Reference Samples for Geochemical and Hydrological Investigations", IAEA, Vienna (1984) 75 pp. ] CRAIG[3 , SciencH. , (19613 e13 ) 1833-1834. [4] GONFIANTINI, R., Nature 271 (1978) 534-536. [5] HUT, G., "Consultants' Group Meeting on Stable Isotope Reference Samples for Geochemical and Hydrological Investigations", IAEA, Vienna (1987) 42 pp. [6] COPLEN, T.B., Chemical Geology (Isotope Geoscience Section) 72 (1988) 293-297. ] HAGEMANN[7 , NIEFR. ,, ROTHG. , , TelluE. , (19702 s2 ) 712-715. [8] DE WIT, J.C., VAN DER STRAATEN, C.M., MOOK, W.G., Geostandards Newsletter 4 (1980) 33-36. [9] TSE, R.S., WONG, S.C., YUEN, C.P., Anal. Chem. 52 (1980) 2445. [10] IUPAC COMMISSIO ATOMIN NO C WEIGHT ISOTOPID SAN C ABUNDANCES, "Atomic Weight Elemente th f so s 1993," Pure Appl. Chem. preparationn i , .

34 AUDI VSMOF TO W DISTRIBUTE UNITEE TH Y DB STATES NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY

T.B. COPLEN, J. HOPPLE US Geological Survey, Reston, Virginia, United States of America

Abstract Bright-orange floating matter (possibly algae) has been observed in the United States supply of some ampoules and in one of two 10-L primary glass storage flasks of the isotopic reference water VSMOW (Vienna Standard Mean Ocean Water). Within experimental error, ampoules with and without this orange matter are identical in stable hydrogen and oxygen isotopic composition. Thus Norte th , h American suppl VSMOf yo beet no ns Waltereha n di isotopic abundanc remaind ean s fully functiona primara s a l y isotopic reference material.

1. INTRODUCTION

The isotopic reference water VSMOW (Vienna Standard Mean Ocean Water) serves as a primary reference materia oxyger lfo hydroged nan n relative isotope-ratio measurements [1], VSMO Ws distributei n Norti d h Americe Nationath y b a l Institut f Standardo e d an s Technology (NIST) of the United States Department of Commerce and distributed elsewhere by the International Atomic Energy Agency (IAEA), Vienna, Austria.

In 1993, it was noted that numerous ampoules of VSMOW, in the United States only, contained bright orange floating material that is possibly algae. Because this reference water is of critical importance to a large number of scientific fields, an investigation was undertaken to determine if the North American supply of VSMOW had degraded and been made worthless.

First, a visual audit of the North American supply of VSMOW and also of SLAP (Standard Light Antarctic Precipitation GISd )an P (Greenlan Sheee dIc t Precipitation undertakens )wa . Second, water from ampoules with and without orange "floaters" was analyzed for stable hydrogen and oxygen isotopic composition.

2. EXPERIMENTAL METHOD

Water was analyzed for stable hydrogen isotopic composition by equilibrating samples with gaseou susinH t catalys multiplP e ga th y tb e sample metho Coplenf do , Wildma Ched nan n

[2]. 2Water was prepared for oxygen isotopic composition by equilibrating samples with carbon dioxide using the method of Epstein and Mayeda [3], modified for multiple sample analysis by evacuating air through a capillary tube [4]. Carbon dioxide samples were analyzed on a double-focusing isotope-ratio mass spectrometer [5]. Oxygen and hydrogen

35 isotopic results are reported in per mil relative to VSMOW and normalized on scales such thae oxygeth t d hydrogean n n isotopic value f SLAo s e —55.5%oar P d —428%ean , respectively [1].

. RESULT3 S

The visual audit of isotopic reference, waters at NIST was undertaken by Robert Vocke, Jr. of NIST, Pedro Morales (visitin authorge th scientis f o se t (TyleNISTa ton d r)an Coplen) on November 10, 1993. This audit provided part of the information in Table 1. Ampoules

Table 1. Results of a visual audit of isotopic reference waters at NIST on November 10, 1993.

Labe glasn o l s Quantity Comment vessel Vienna SMOW Unopened 10-L flask, Water looks clear. Under control of 77-02-21 sealed by glassblower. Robert Vocke, Jr. (NIST) P.S. GISP Unopened 10-L flask, Water looks clear. Under control of 77-07-25 sealed by glassblower. Robert Vocke . (NISTJr , ) P.S. SLAP Unopened 10-L flask, Water looks clear. Under control of 77-04-04 sealed by glassblower. Robert Vocke . (NISTJr , ) P.S. Vienna SMOW 10-L flask opened in 1985 Orange material floating in water. 77-02-22 and abou hale conf on tfo - Under control of Robert Vocke, Jr. P.S. tents remains. (NIST) GISP 10-L flask opened in 1985 Water looks clear. Under control of 77-07-25 and abou hale f conon to f - Robert Vocke, Jr. (NIST) P.S. tents remains. SLAP 10-L flask opene 198n di 5 Water looks clear. Under controf o l 77-04-04 and about one half of con- Robert Vocke . (NISTJr , ) P.S. tents remains. RM 8535 114 ampoules; about 20 ml Water looks clear. Under control of VSMOW each. Standard Reference Materials Program. RM 8536 Did not count; probably Water looks clear. Under control of GISP more than 100. Standard Reference Materials Program. RM 8537 Did not count; probably Water looks clear. Under control of SLAP more than 100. Standard Reference Materials Program.

36 of isotopic reference waters are distributed by the Standard Reference Materials Program at MIST. Ampoule f VSMOo s e designateWar 853 M shod R 5an signo w s n d a f bright o s - orange "floaters." The storage of the primary reservoirs of VSMOW, GISP, and SLAP in North Americ undes ai controe th r Roberf o l t Vocke . EacJr , thesf ho e thre stores e i th n di dark in two 10-L sealed glass flasks. The following identical documentation was found with three of the six flasks (one VSMOW, one GISP, and one SLAP, all about one-half full):

May 24, 1985

NOTE:

THIS 198, BOTTL 22 ABOUD 5OPENEY S AN 0 EMA WA T20 N DO SAMPLE SCONTAINEE WERTH F EO S REMOVEDD RWA EN E TH . BROKEN OFF, A CLEANED TEFLON TUBE INSURTED (sic) INTO THE BOTTLE AND BOTTLED IN 20 ML AMPULES ON THE OSRM AMPULE SEALING MACHINE. EACH AMPUL FLUSHES EWA D WITH PURE ARGON IMMEDIATELY BEFORE FILLING. AFTER ABOU0 T20 AMPULES WERE FILLED, THE BOTTLE WAS FLUSHED WITH ARGON GLASSBLOWEANE DTH Y SEALEB F DOF R (JEFF ANDERSON).

EACH OF THE LARGE BOTTLES OF SMOW, SLAP AND GISP HAD INDICATION SOMF SO E TYP CONTAMINATIOF EO N BEFORE OPEN- ING. I DO NOT KNOW WHAT THE CONTAMINATION WAS.

AFTER RESEALING EACH BOTTL ABOUD EHA T ONE-HALE TH F FO ORIGINAL CONTENTS.

I.L. BARNES 5/24/85

Althoug e half-fulth h l flask f GIS o sd SLA Pan P were nearly colorless, that containing VSMOW contained bright orange-floating material (possibly algae). Thi likels si source yth e of the orange material in 20-mL ampoules that prompted this study. All three full 10-L flasks were nearly colorless. Note that none of the six 10-L flasks was opened and sampled for this study.

The stable hydrogen and oxygen isotopic composition of several ampoules of VSMOW and other reference water gives si Tabln ni . e2

. DISCUSSIO4 N

Within experimental error stable th , e hydroge oxyged nan n isotopic compositio f threno e ampoules of VSMOW were identical. This suggests that the orange matter in one of the ampoule t affecteno s isotopie sth dha c compositio watee th f rno measurably. Furthermore, it is inferred that the half-full 10-L flask of VSMOW at NIST has probably not been altered in isotopic composition. Thus alteratioo n , VSMOf no W distribute Unitee th n di d States that affect stabls it s e hydroge oxyged nan n isotopic compositio bees nha n observed.

37 Tabl . eStabl2 e hydroge oxyged nan n isotopic compositio severaf no l ampoule VSMOf so W and other reference waters.

Description 02H no. 0180 no.

in %'00 in %'00 Ampoule with "SMOd Wan 0.0+0.0 1 9 0.00+0.05 "502" written in blue (contains small amounf o t orange floating matter) Ampoul 8535M R f ,eo 0.0±0.6 12 0.00+0.05 VSMOW obtained from SRMP at NIST on Nov. 10, 1993 (sample is clear) Ampoule of VSMOW 0.0±1.0 1 3 0.00+0.05 obtained from Robert Vocke, Jr. at NIST on Nov. 10, 1993 (sampl clears ei ) Puerto Rico laboratory -1.3+0.7 23 -1.52+0.07 37 reference, W-39500 Laboratory reference, -53.5±0.7 6 W-28673 Laboratory reference, -36.0±1.2 6 -6.21±0.05 30 W-38888 Antarctic ice melt laboratory -394.5±0.8 29 -50.04±0.06 18 reference, W-35000 SLAP with "210" written ni -428.0±0.8 7 -55.5+0.03 6 blue

REFERENCES

[1] GONFIANTINI, R., Nature 271 (1978) 534-536. ] COPLEN[2 , T.B., WILDMAN, J.D., CHEN , AnalJ. , . Chem (19913 6 . ) 910-912. ] EPSTEIN[3 , MAYEDAS. , , GeochimT. , . Cosmochim. Act (1953a4 ) 213-224. [4] ROLSTON, J.H., DEN HARTOG, J., BUTLER, J.P., J. Phys. Chem. 80 (1976) 1064-1067. ] COPLEN[5 , T.B., Int . MasJ . s Spec Physn Io (1973.1 1 . ) 37-40.

38 SULPHUR ISOTOPE STANDARDS

B.W. ROBINSON Institut f Geologicaeo Nuclead an l r Sciences, Lower Hutt, New Zealand

Abstract Although the early choice of meteorite standards for sulphur isotope investigations was good from the viewpoint of representing primordial sulphur and an average of terrestrial samples, it was soon recognised that chemical impurities and homogeneity were a problem. Although chemically pure AgS standards were recommende earls da 1962s ya , continued attempto st

introduce geological2 material standards sa almosr sfo year0 t3 s considerabl bact yse k progress; agreement between laboratories has been typically poor. However, three chemically pure and homogeneous sulphur isotope standards are now available from the IAEA and NIST. They NBS12e ar 7 whic BaSOa s hi 4 from 'ion exchanged wate a whic2 *se rNZ hsulphated an 1 ;NZ are Ag2S calibration materials produced from a sphalerite close to 0 %o CDT and a gypsum hav2 respectivelyT eNZ beed CD an o n 1 % circulate 1 NZ closIAEe .+2 th o r et Ay fo db calibration. Fifteen, 2

. 1 INTRODUCTION

e earlTh y choic f meteoriteo e primara s a s y standar r sulphufo d r isotope studiet no s i s surprising. Representing primordial sulphur, they lie at the mean for terrestrial samples. When problems were eventually recognised with the meteorite standards, a mineral sphalerite was introduced but also found to be inhomogeneous. This continuance of using geological and impure materials as primary standards hampered progress and agreement between laboratorie bees sha n typically poorneee purr Th dfo . e chemical compound standards a s s became increasingly urgent.

In 1986, an IAEA technical contract was awarded to the Institute of Nuclear Sciences in New Zealan f pureproduco dt o g ,k homogeneous1 e calibratioa s a t , Agac no 2t So nea% 0 r "zere th materia t o 6 e se part34 th Sd f "scaleo an l . This standard (NZ1 bees )ha n circulated to over 30 laboratories by the IAEA and the results obtained from 17 of these are reported here 1990n I . similaa , r technical contrac produc o puref t o g ,k homogeneouse1 , Ag2S near s initiatedwa o % .1 However+2 e calibratioth , n material produced (NZ2 s onl)ha y been analysed by three laboratories to date. Reported here is a short summary of the history of sulphur isotopresulte th d esan frostandards2 m NZ laborator d productioe an th , 1 NZ y f no intercomparisons.

39 2. HISTORY OF SULPHUR ISOTOPE STANDARDS

e firsTh t sulphur isotope measurements were made against available material suc Pars ha k City pyrite (USA Mercd )an k elemental sulphur (NZ). However 195y ,b 0 meteorite beed sha n s primara adopteA yUS standardse dUSSe th botth d n Rhi an . They were thoughe b o t t isotopically homogenous (32S/34S = 22.22 ± 0.01); they represented primordial sulphur and were close to the average for terrestrial samples [1]. Sea water sulphate was also recognised s beina g homogeneou t beinbu s g 21%o enriche witS hn d i respec meteoriteo t t t no s swa initially adopted as a standard. 34

Sulphide materials wer t firsa e t typically burnt witr hfo oxyges produco t ga s 2 nga SO e isotopic measurement. BaSO4 was reduced chemically or by graphite at high temperatures to give a sulphide phase. Gradually Ag2S began to be favoured for analysis because it gives uniform burning characteristics.

By 1960, the Canyon Diablo Troilite (FeS phase from a large octahedrite iron meteorite collected around Meteor Crater, Arizona) had been adopted as the primary standard in the west but the Russians continued to use the Shikote Alin meteorite. Quite early in the history of sulphur isotope researc problem] h[2 s were, however, recognised wit Canyoe hth n Diablo Troilite standard. Different chemical treatment of meteorites yielded differences of up to Q.4%o. Sinc earle eth y assignmen s somewha 32e S/**Swa th f T o t ratiCD t r s 22.2oa fo o 2% arbitrary and, taking into accoun e homogeneitth t f differeno e yus problemte th d an s meteorites, it was proposed in 1962 that three new standards be made [3]:

Ag2S around 0 %o o Ag% 25 S-3 around BaSO4 around +20 %o

reao N l progress with this recommendatio 196e madth s 6d nwa IAE ean A Advisory Group Meetin n Stablo g e Isotope Reference Sample r Geochemicafo s d Hydrologicaan l l Investigations concentrated on the water standards V-SMOW and SLAP. However, at the 1976 meeting, severa sulphuw ne l r isotope standards were proposed BaSOa : 4 froa mse water, galena or sphalerite at -20 to -30 %o, elemental sulphur derived from natural gas and two different SO2 samples for mass spectromelric calibration [4]. Two standards, OGS: a raw precipitated BaSO4 fro watera m se Soufrd an , Lucqe ed elementan a : l sulphur, were prepared and circulated. At the 1983 IAEA Standards Meeting it was advised to abandon these standards because of very poor agreement between laboratories. In the IAEA report [5], H Nielsen documente serioue dth s problems with sulphur isotope standard followss sa :

• Troilite is not as isotopically uniform as expected. • It is not stoichiometrically pure FeS. • It contains some Co and Ni sulphides and carbides.

He suggested that:

• Ag2S standards should be prepared. watea se rA sulphat • e standard from 'ion exchanged watea se ' r shoul madee db .

Meanwhile ReeE t McMasteC ,sa r Universit Canadn yi prepared aha Ag0 d1 2S standards d analyseha d an de superio o th the % 6 y technique ms b 0 SF rwa +5 e o t H fro.0 m-3 approached to make standards at +20 and -20 %o for the IAEA but unfortunately died before

40 completing this task. The meeting in 1983 also recommended that a sample of sphalerite close to 0 %o be introduced as a new standard. The latter was quickly supplied by S Halas of Polan becamd dan e NBS122.

A a 198t 5 IAEA meetin Isotopn go e Hydrolog Geochemistrd yan f Sulphuryo , analytical techniques were discussed and it was recommended [6] that three BaSO4 and three Ag2S standards with S S values of close to 0 %o, about +20 %o and about -20 %o would ideally be 34 made available O'NeiJ . l (USGS, Menlo Park alreadd )ha y prepare 'ion da n exchangeda se ' water sulphat BaSOs ea 4 (NBS127 replaco t ) e OGS RobinsoW B . n agree produco dt e eth Ag2S standards in New Zealand.

Also, later in 1985 a IAEA meeting on Reference Samples for Geochemical and Hydrological Investigations mad followine eth g recommendation publishes a ] [7 s 1987n di :

• NBS122 sphalerite is inhomogeneous and should no longer be distributed. • NBS123, a spherite at about 17 %o was introduced for intercomparison. NBS127 BaSO4 replaces OGS. • Further circulatio standardf no hele sb d unti Age th l2 S standard producee sar w Ne n di Zealand.

3. PREPARATION OF NZ1

The preparation of this standard was facilitated through the IAEA Technical Contract 4320/TC, March 1986-March 1987. The requirement was to produce 1 kg of pure, 34 homogeneous Ag2S with a 0 S value close to 0 %o. The starting material was 600 g of the same sphalerite whic beed hha n discontinue NBS122s da , HalassupplieS r D . y db Initia l analyses of this material indicated inclusions of chalcopyrite and a coarse darker sphalerite

O2 free N

Ag2s Kiba reagent plus mineral

: Larg1 g Fi e column reactio2 n traiNZ d productione an th use 1 r dfo NZ f no

41 which had probably contributed to the inhomogeneity problems. The sphalerite was reground, remixe divided dan d into batche reactior sfo n with Kiba reagent (Sn2"1" with strong phosphoric acid t 200°a ) larga Cn i e volume reaction trai. Thi1) n g s(se Fi traie n C consist 5 a f o s reaction vessel in a mantle heater, two 3 f wash traps and two 3 Q AgNO3 traps all connected to a supply of oxygen-free N2 gas. Initially, two batches of 80 g sphalerite with 2 C phosphoric acid and 200 g SnCl^ were run at 200°C to produce about 140 g Ag2S each. Then four batches of 100 g sphalerite each were run to produce about 200 g Ag2S per run. Each batch was reacted for six hours. The large volume of the reaction train and the double wash traps help to mix and homogenise the H2S gas which is transported by the carrier gas. All the Ag2S batches were washed with distilled water until free of silver ions (as tested by washings)e th o additiot weighte " batchee Cl Th .th f no f so s were 133» 148, 178, 6 20419 , and 188 g. A preliminary check of the batches indicated that they were homogeneous and similar to each other within error of 534S measurements (±0.2 %o). These batches were then ground and mixed togethe larga n ri e morta pestled mixee an r Th . d sampl sieves ewa d throug hnyloa n sieve . Remainin: pm siz 3 e5 g materia regrouns wa l paso dt s throug sieve hth e usin sievga e shaker. Eventuall grindine yth g produce residuda coarso et e 'metallic' balls which coult dno be sieved further and this residue (~50 g) was discarded. The bulk sample (<53 urn) was 'turned' in a plastic container held in a tilted lathe bed for one day to thoroughly mix the powder bule Th k. sampl supplies IAEee wa th Ao dt whersplis portiong wa tm t einti 0 o50 s for bottlin distributiond gan .

4. PREPARATION OF NZ2

The IAEA Technical Contract (5654/TC, September 1989-Septernber 1990) for this standard 34 producf o puret o s g ,k wa homogeneou1 e s Ag2S s startinwitA h. 8g%o S1 clos+2 o et material acquire e f gypsumw ,o g k d6 , naturally precipitated from watea presense n y i r da t evaporite ponds owned by Dominion Salt Ltd, New Zealand. The material was washed to remove halite dried an t 200°, da converCo t gypsue th t anhydritmo t e (CaSO4). Severag 1 l aliquots of the anhydrite were converted to Ag2S using strong Kiba reagent (800 g SnC^ in dehydrateC 5 2. d phosphoric acid t 280°a ) tesCo conversioe t th t n process e efficiencTh . y of conversion is about 80% and the batches varied by less than 1 %o in 834S.

largA e volume reaction train simila thao rt productioe t th use (show1 r g dfo Fi NZ f n nni o . Eleveup t se n s CaSOg batche wa 0 9 ) i f 4 o s stronwit C h3 g Kiba solutiont a wer n eru houro 280°tw produco r st C fo f Ageo abou2g S 0 eac9 t h e givinth l totaga Al f 102. o lg 4 Ag2S batches were washed with distilled water until fre silvef eo r ion testes s(a additioy db n of Cl" to the washings).

A preliminary check of the batches indicated that they were homogeneous and similar to each other within erro f measuremeno r f 0o t S. These batches were then mixe ground dan a n di 34 large morta pestled mixee an r Th . d sampl sieves ewa d throug nyloha n sieve: siz urn3 e5 , until only coarse metallic balls were left and about 70 g of this residue had to be discarded. The bulk sample (<53 um) was 'turned' in a plastic container held in a tilted lathe bed for one day to thoroughly mix the powder. The 0.95 kg was split into 500 mg samples and bottle distributior dfo IAEAe th y nb .

42 Table 1: INTER-LABORATORY COMPARISONS OF THE SULPHUR ISOTOPE STANDARD NZ1

COT ° "CDT

-0.34 -0.34 -0.34 -0.37 -0.37 -0.37 -0.1 -0.1 -0.1 -0.25 -0.25 -0.25 -0.3 -0.3 -0.3 -0.3 -0.3 -0.3 -0.13 -0.13 -0.13 -0.55 -0.55b -0.25 -0.25 -0.25 -0.35 -0.35 -0.35 -0.14 -0.14 -0.14 -0.39 -0.39 -0.39 -0.21 -0.21 -0.21 -0.33 -0.33 -0.33 0.19b -0.15 -0.15 -0.15 -0.27 -0.27 -0.275

Mean m -0.249 -0.277 -0.259 l a 0.155-0.113 -0.091 al(n-l) 0.1600.114 0.094 n 17 16 15 m+2a 0.066 -0.054 -0.076 m-2a -0.574 -0.510 -0.447 m a 0.039 0.029 0.024

e firs*Th t column contain resulte sth s reported fro laboratorie7 m1 s Outlierb s criterioa exclude2 e th nn do c finae Th l column give acceptee sth d analyse= ( m sa togethed an n r„ a wit , m h6

4. RESULTS FOR NZ1 AND NZ2

1 4. Analyse Zealanw Ne n si d

Eight 15 mg and 2.5 mg portions were taken from the bulk sample of NZ1 and analysed for their sulphur isotopic compositio MicromasG V a n o n s 1202E mass spectrometere Th . individual analyses did not vary outside the overall error of ±0.2 %o which is the estimated maximum combined error for our chemical and mass spectrometric processes. Therefore for the sample sizes taken the standard is homogeneous. For a sample size of 10 mg, AgS will

contain more than 2500 grains, whic sufficiens hi ensuro t t e isotopic homogeneitye w d an , 2 sugges purposee t th thar fo t f laboratorso y intercompariso reasonablna y larg constand ean t amount of 10 mg is taken for analysis. Our best analytical results were produced on the 15

43 mg samples (because of the greater difficulty in handling smaller samples in our inlet system) to give an average value of -0.34 ± 0.07 %o CDT (n=8) using a value of +3.3 %c CDT for our laboratory standard R2268 [8]. wer) mg e 0 take(1 Portionn2 fro NZ bule mf th so k sampl analysed ean Micromasa n do s 1202E mass spectrometer. The results for NZ2 gave an average value of +21.7 ± 0.1 %o CDT (n=8). The individual analyses do not vary outside the overall error of ±0.2 %o, indicating sample th homogeneous ei least sa thio t t s level. Agai recommene nw purposee d th tha r fo t s of inter-laboratory comparison a reasonably large and constant sample amount, say 10 mg, is taken for isotopic analysis.

2 4. Inter-laborator2 y NZ comparison d an I NZ f so

NZI was circulated by the IAEA to over 30 laboratories and to date 17 have supplied the results shown in Table 1. Fifteen of these passed the 2a screening to give a mean 834S of -0.26 %o particulaf O . r not three analyse6 ear I froeSF mNZ f sdiffereno t laboratories. They gave 034S values of -0.27, -0.30 and -0.33, again indicating that the material is homogeneous,

wit h8a 34S valu f -0.3 eo CDT o 0% . dato T e only four other S6 34 analysehav2 eNZ f beeo s n reporte IAEAe th o dt . They are: +21.42, +21.48, +21.52 and +21.58 %o; none are by SF6. Three of these laboratories and our Zealanw Ne d results show NZ2-NZ1 value f 21.38o s , 21.66, 21.7 22.0d . an 8 4%o

. 5 CONCLUSIONS

After a very protracted period during which geological and impure mineral samples were unsuccessfully use s sulphuda r isotope standards chemicallw fe a , y pur homogeneoud ean s compounds havbeew enno prepare readile ar d ydan available fro IAEe mNISth d Aan T with reserve welr sfo l oveyears0 10 r .BaSO e Theth e y4ar fro watea mse ro (NBS127tw e th d an ) silver sulphide calibration materials (NZI and NZ2).

NZI fulfils all the criteria for a calibration material. Although CDT is essentially exhausted, the value must be kept as a primary reference because most of the results published to date expressee ar d relativ CDTo et . Howeverbees ha n I thasuccessfullw NZ tno , y analysen di many laboratories it could act as the calibration material for the primary reference standard (CDT). Future results would then be expressed as 5 values relative to V-CDT with NZI set at -0.30 %o. Coherence betwee valuen0 s reported from different laboratorie improvee b n sca d adoptiny b seconga d sulphur reference standar whico dt 6e 34hth S scal thes ei n normalised. NZ2 partly fills this requirement, but in addition, there is a need for a third standard at about -30 %o to be produced and widen the span to a more useful 50 %o. Furthermore, a BaSO4 with 'light' sulphu oxyged an r n isotopic composition alss si o required.

REFERENCES

[1] MACNAMARA J., THODE H.G., "Comparison of the isotopic constitution of terrestrial and meteoritic sulfur". Phys. Rev. 78 (1950) 307-308.

[2] JENSEN M.L., NAKAI N., "Sulfur isotope meteorite standards results and recommendations Biogeochemistr: In " sulfuf yo r isotopes (Ed: M.L. Jensen) Proc. NSF Symp. Yale Univ. (1962) 30-35.

44 ] [3 AULT W.V., JENSEN M.L., "Summar sulfuf yo r isotope standards : Biogeo"In - chemistry of sulfur isotopes (Ed: M.L. Jensen) Proc. NSF Symp. Yale Univ. (1962) 16-29.

[4] GONFIANTINI R., "Standards for stable isotope measurements on natural compounds". Nature 271 (1978) 534-536.

] [5 GONFIANTIN "Advisor, IR. y group meetin stabln go e isotope reference samples for geochemical and hydrological investigations" (1984) Report to the Director General, IAEA, Vienna.

[6] GONFIANTINI R., Hut G., "Advisory group meeting on isotope hydrology and geo-chemistry of sulphur" (1985) Report to the Director General, IAEA, Vienna.

[7] HUT G., "Consultants' group meeting on stable isotope reference samples for geo-chemical and hydrological investigations" (1987) Report to the Director General, IAEA, Vienna.

] [8 ROBINSON B.W., KUSAKAB , "QuantitativEM. e preparatio sulfuf no r dioxide for S/ S analyses, from Sulfide combustioy b s n with cuprous oxide. Anal. 34 32 Chem. 47 (1975) 1179-1181.

45 VARIATION SULFUE TH N SRI ISOTOPE COMPOSITION OF CDT

G. BEAUDOIN Départemen géologie d t génie d t ee géologique, Université Lavai, Québec, Canada B.E. TAYLOR Geological Surve f Canadayo , Ottawa, Ontario, Canada D. RUMBLE, III Geophysical Laboratory, Washington, D.C., United State Americf so a M. THIEMENS Departmen f Chemistryo t , Universit f Californiyo Diegon aSa , Jollaa L , California, United State f Americso a

Abstract. sulfurThe isotope composition Canonof Diablo troilite variableis shownas by high-precision analyses of three different interlaboratoryan samples by and comparison.The 34 present range &in S values °/00.0.4 Becauseis Canon Diablo Standardtroilitea not is Reference Material shouldusedit be calibrationfor not sulfur ofthe isotope scale.

Canon Diablo troilite (CDT), although accepte reference sulfurth e s dth a r e fo isotop e Standara scalet no s ,i d Reference Material manufacture internationar dfo l distributions it d ,an sulfur isotope composition is not well characterized. The troilite forms blebs in massive octahedrit musd eseparatean e tb d from fragment Canof so n Diablo iron meteorite obtained either from museum collection commerciar so l rock dealers. Interlaboratory comparisoa f no referenc (Beaudois ga e eSF t al.ne , 1994) indicat rangea 6n e34i ScoT valuee on 0.4f s, o 1%o orde magnitudf ro e larger than analytical uncertainty (<0.03 %c, TablsmalA . e1) l standard

deviatio valueS &* 0.00 ± sf ( n o 5 %o, n=8alss i )o obtaine Geologicae th t da l Survef yo Tabl . Summar1 e f Sco8yo 34 T (%») e referencvalues th ga r g fo s SF e n a l ± x Geophysical Laboratory - 6.54 ± 0.03 12 Universit f Californio y 6.9- ± Diegn 50.0 2 Sa a1 2o _____ (From Beaudoin et al., 1994) __

Canad thir afo s referenc gase eSF . Aliquotisotopicalle ar referenc e s th ga f so o yeSF identical andifference dth mean ei n O^Scjyj- value must therefor resula e eb differenf to t calibrations relative to CDT, and thus imply that isotopically different aliquots of CDT were used to calibrate the sulfur isotope scales used in each laboratory.

47 e averagTh e 634S valu threr efo e different samplerelativT commoa CD f o et so n

referenc 6 rangeSF e s from 6.5 46.9o t 1 %o (Tabl ; Beaudoie2 t al.ne , 1994) range .Th n ei 0 S values (0.37 %o differenr )fo t sample Canof so n Diablo troilit almoss ei ordee on t f ro 34 magnitude larger than typical analytical uncertainty (0.05 %c), indicating measurable difference sulfun si r isotope compositio primare th f no y referenc sulfue th r re fo isotop e scale. Interlaboratory compariso samplf no e CDT3 yields similar 534S laboratorievalueo tw n si s 34 (Tabl . Large2) e range f averageso d 5 SCDT values %o9 (1.alse 1. ) ar 1 oo t reporter dfo sulfur isotope intercomparison sample severay sb l laboratories (HUT, 1987) significanA . t part of these variations could result from sulfur isotope variations in the samples of Canon Diablo troilite used to calibrate the sulfur isotope scale in each of these laboratories. It is noteworthy, however, that recent high-precision sulfur isotope analyse intercomparisof so n 34 materials from the I. A.E. A and China has yielded identical S SCDT values, within analytical uncertainty, to those accepted for these materials (GAO and THIEMENS, 1993).

Table 2. Summary of 634S (%c) values for samples of Canon Diablo troilite Sample Geological Survey of Canada Geophysical Laboratory X ± la n x ± la n CDT1 6.91 ± 0.05 6 — CDT2 6.73 ± 0.04 4 — CDT3 6.63 ± 0.17 12 6.5 ± 40.1 1 3 (From Beaudoin et al., 1994)

Conclusion: The sulfur isotope composition of Canon Diablo troilite is variable as shown by high-precision analyses of three different samples and by an interlaboratory comparison. The present range in S^S values is 0.4 %c, but more samples of different troilite inclusions fro Canoe mth n Diablo meteorite shoul analysede db . Because Canon Diablo troilita t no s ei Standard Reference Material it should not be used for calibration of the sulfur isotope scale.

References

Beaudoin, G., Taylor, B.E., Rumble, D. Ill and Thiemens, M.H., 1994. Variations in the sulfur isotope composition of troilite from Canon Diablo iron meteorite. Geochim. Cosmochim. Acta, 58, 4253-4255. Gao, X., and Thiemens, M.H. 1993. Isotopic composition and concentration of sulfur in carbonaceous chondrites. Geochim. Cosmochim. Acta, 57, 3159-3169 Hut . 1987,G . Consultants Group Meetin Stabln go e Isotope Reference Sample Geochemicar sfo l Hydrologicad an l Investigations. Final Report, International Atomic Energy Agency, Vienna, 42 p.

48 COMPARISO CONVENTIONAL-SOE TH F NO E TH D 2AN LASER-SF6 METHODS: IMPLICATIONE TH R SFO SULFUR ISOTOPE SCALE

. BEAUDOIG N Département de géologie et de génie géologique, Université Lavai, Québec, Canada B.E. TAYLOR Geological Surve f Canadayo , Ottawa, Ontario, Canada

34 Abstract. Comparison of & SCDT values using conventional SO2 and laser-SF6 methods for powdered sulfide samples presented.is laser-SFThe 6 method recommendedforis calibrationthe of new reference and intercomparison materials for sulfur isotope analyses.

Comparison of 8 S values obtained using the conventional-SO2 and the laser-SFg 34 methods raises questions abou accurace th tpresentle th f yo y accepted sulfur isotope scale (Beaudoin and Taylor, 1994). Sulfur isotope intercomparison material has been dominantly analyzed by the conventional-SOa method but several corrections must be applied to these dat accouno at oxyger tfo n isotope compositio memord nan y effectinlee th tn si syste e th mf o mass spectrometer (Rees, 1978). Rees (1978) showed that measurements using SF yield

correction-free isotope ratios whic linearle har y correlated with conventional-SO2 data r afo 6 f sampleso t se . Least-squares major axes regressio f sampleo f analyset no se y a sb f so conventional-SO2 and laser-SFg methods (Table 1, Beaudoin and Taylor, 1994) gives well- 34 correlated 5 SCDT values wit slopha e different from unity (Fig: 1) . 34 6 S (SF6) = (-0.18 ± 0.04) + (1.035 ± 0.002) Ô^S (SO2); & = 0.99999 (1) This linear regression is similar to that of Rees (1978): 34 2 8 S (SF (-0.1= 6) 40.07± (1.01+ ) ) 0.0039± (2 (SOS 0.9999)S^ = 2);r 6 These similar result despite sar e different sample suites, chemical extraction methodsd ,an difference masn si s spectrometers difference .Th e between conventional-SO laser-SFd 2an g data, therefore, is most likely due to a systematic error resulting from inaccurate corrections datapplie 2 oxyger afo SO o dt n isobaric interference. SO2-derived isotope ratios shoule db empirically corrected using intercomparison material analyzed with correction-free measurements using SFg. Establishmen sulfuw ne ra isotopf o tf o t ese referenca f o d ean intercomparison material consideres si Internationae th y db l Atomic Energy Agenco yt remove ambiguities on the origin of the sulfur isotope scale and to improve laboratory intercalibrations (Gonfiantin Stichlerd an i , personnal communication 1994) calibratioe .Th n of these new reference and intercomparison materials will be improved if measurements are made using

49 34 TABLE 1. COMPARISON OF ô sCDT VALUES USING CONVENTIONAL-SO2 AND LASER-SF6 METHODS FOR POWDERED SULFIDE SAMPLES SO2 SF6 x ± a n x ± a n PAF, pyrite 27.80 0.15 28.66 ± 0.09 (5) RICO, pyrite 0.19 0.21 (4)a -0.11 ± 0.05 (6) ZNS, sphalerite -3.12 ± 0.15 (5)a -3.43 ± 0.03 (6) SILVANA, galena -7.84 + 0.15 (9)a -8.25 ± 0.06 (5) KAZA, pyrite -31.6 (2)a -32.82 ± 0.04 (4) -31.9 b Universit Ottawaf yo Universit. Albertf yo a (G.M. Ross, pers. com. 1992). ______a b (From Beaudoin and Taylor, 1994)

40- r = 0.99999 30-- PAF (pyrite), m = 1.035 20-- u. CO 10-- 0-- o •lalerite) rfr-'RICO (pyrite) CO *» co -10- SILVANA (galena) «O -20--

-30-- fkAZA (pyrite)

-40 -30 -20-10 0 10 20 30 40

Figure 1. Diagram showing the correlation between 0 SCDT values for a set of samples analysed 34 by conventiona laser-SFd an 2 lSO 6 methods analysee .Th s correlated well oveO T 7 rangra 2 6 f eo for several sulfide minerals (from Beaudoi Taylord nan , 1994).

References

Beaudoin Taylord an , ,G. , B.E., 1994. High precisio spatiad nan l resolution sulfur isotope analysis using MILES laser microprobe. Geochim. Cosmochim. Acta, , 5055-506358 .

Rees, C.E. 1978. Sulfur isotope measuremen SFd t 6usinan . 2 Geochim.gSO Cosmochim. Acta, 42, 383-389.

50 INTERLABORATORY COMPARISON OF REFERENCE MATERIALS FOR NITROGEN-ISOTOPE-RATIO MEASUREMENTS

J.K. BÖHLKE, T.B. COPLEN US Geological Survey, Reston, Virginia, United States of America

Abstract Aliquots of seven different reference materials were distributed for an interlaboratory compariso stablf no e nitrogen-isotope-ratio measurements. Results fro laboratorie5 m1 s were

compiled and evaluated selectively to yield provisional values of 515N for each material, i,

with respect to atmospheric N )(oN,. 1S The N5 values reported by the different 7air laboratorie e correlatear s thay n suc i dwa t e majo soma hth f reo discrepanciee b 2 y ma s removed by normalization (/. e., by altering the length of the ô N scale for each laboratory by an amount defined by local measurements of reference materials with extreme values). The results of the intercomparison test, after elimination of 2-u outliers, before and after normalization, are as follows:

15 15 Identifi- Substance S N,7air ± lo ô Na l l7ai ± r cation o % n i o % n i (as reported) (normalized)

NSVEC N2gas -2.77 ± 0.05 -2.78 ± 0.04

IAEA-N1 (NH4)2S04 +0.4 0.03± 7 +0.4 0.03± 7

IAEA-N2 (NH4)2S04 +20.3 0.02± 9 +20.41 ± 0.12

IAEA-N3 KNO3 +4.69 + 0.09 +4.72 ± 0.13

USGS25 (NH4)2S04 -30.25 ± 0.38 -30.4 10.2± 7

USGS26 (NH4)2S04 +53.62 ± 0.25 +53.75 ± 0.24

USGS32 KNO3 + 179.2 ± 1.3 + 180

1. INTRODUCTION

primare Th y standar relativr dfo e nitrogen (N)-isotope-ratio measurement atmospheris si c nitrogen (N^ gas, which is widespread and homogeneous and, by convention, has a 51N5 value of 0%o ([1, 2, 3] and see Appendix A for notation). Despite the long history of N- isotope-ratio measurements in the natural sciences, there has never been a set of common-use secondary reference materials with widely accepted isotopic compositions. Secondary reference materials with a large range of isotopic compositions are essential for reporting or expressing isotopic compositions far from that of the primary standard on normalized scales. The secondary reference materials most commonly analyzed for abundance of N isotopes 15 in the last decade are IAEA-N1 and IAEA-N2, both (NH4)2SO4, with 5 N values near 0 and

51 +20%o, respectivel . 5] Othe , [4 yr materials that have been widely distributed include 15 15 IAEA-N3 (KNO3; 0 N unreported) and NSVEC (N2 gas; 0 N near -2.8%o [6]). Recently ] describe, Böhlk[7 . al t ee d thre referencw ene e materials (USGS25, USGS26d an , USGS32) intended to span the range of 5 N values in all but a few of the many thousands 15 of values reported in the literature for natural terrestrial substances. Here we report the results of an interlaboratory comparison test (Appendix B) for those seven secondary reference material salts)x si d ,s an whic(on s ega h cover approximatele yth range of 5 N values known in natural terrestrial substances (Figure 1). The results of this 1S stud e complementarar y thoso t y f ParClementd o e an r s [8] reporteo ,wh d resultn a f o s interlaboratory comparison of five reference materials with high 0 N values (IAEA-305A, 15 305B, 310A, 310B, and 311) intended for calibration of biological and medical materials involving substances artificially enriche (AppendiN 15 n di . xC)

2. DESCRIPTION OF REFERENCE MATERIALS

Atmospheric Nitroge 2 gas]n[N : Atmospheri primare th s i 2 yc N referenc e materiar fo l N-isotope-ratio analysis conventiony B . 5a 15s Nha valut i , 0%of eo . Mariott report] i[3 s that it is isotopically homogeneous to within ±Q.026%o. Junk and Svec [2] report several measurement (f 15so N14N)+/(14N14N)+ wit averagn ha e valu 7.35f eo 1 xlO~ concludd 3an e 14 15 that atmospheric N2 has a N/ N ratio of 272.0±0.3 (atom ratio). The International Union of Pure and Applied Chemistry (IUPAC) recommends that conversions between the 615N scale and the 14N/15N ratio for nitrogen isotopes be based on values of exactly 0 and 272,

respectively atmospherir fo , 2 (secN e Appendi [9])d an .xA

NSVEC [N2 gas]: NSVEC is a pure tank gas held in a large high-pressure cylinder now residing in Reston, Virginia, USA, at the U.S. Geological Survey (USGS). It is believed to be the same tank gas as the one designated as "Matheson pre-purified N2" by Junk and Svec r whic[2]fo , h those authors report (15N14N)+/(14N14N) 7.326xlCT= + 3. Kendald an l 15 Gri ] reporm[6 5a s t Nthaha t i valut f —eo 2.8%o relativ atmospherio et , 2 whiccN y hb IUPAC convention would yield 14N/1S N272.8= .

IAEA-N1 [(NH4)2SO4]: IAEA-N1 is a relatively coarse-grained dried salt prepared by . SalatE i (CENA, Piracicaba Pauloo Sa , , Brazil) between 197 198d 8 an , 10]3[4 . Previous compilation f datso a reporte differeny db t laboratories indicated approximate value f <5so 15N: +0.44 ± 0.39%o (n=8) [10] +0.45 ± 0.32%o (n = 10) [11]

IAEA-N2 [(NH4)2SO4]: IAEA-N2 is a dried salt prepared by E. Salati (CENA, Piracicaba Pauloo Sa , , Brazil) between 197 198d 8, an 10] 3[4 . Previous compilation datf so a reporte differeny db t laboratories indicated approximate value f oso 15 N: +20.18 ± 0.70%c (n=8) [10] +20.20 ± 0.59%c (n = 10) [11]

IAEA-N3 [KNO3]: IAEA-N3 is a dried salt prepared by A. Mariotti (Université P. and M. Curie, Tour, Paris, France) between 1983 and 1985 [5, 10]. No previous compilation of analyse availables si .

52 I I NITRATE IAEA N3 USGS32 REFERENCE MATERIAL +180 AIR (PARTICULATE) PRECIPITATION SEAWATER AND ESTUARIES GROUND WATERS SOILS (EXTRACTABLE) DESERT "CALICHES" AND PLAYA SALTS SYNTHETIC FERTILIZER REAGENTD SAN S OXIDN E GASES

N2O IN AIR (TROPOSPHERE)

SEAWATEN I O 2 N R

GROUNN I O 2 N D WATERS (NATURALR NOAI N xI ) NOx IN AIR (FOSSIL FUEL COMBUSTION) NiTROGEN GAS NSVEC REFERENCE MATERIAL i • AIR GROUND WATERS SEDIMENTARY BASINS —————— FUMAROLE SPRINGT HO D SAN COMPRESSED TANK GAS

ORGANICN PLANTS AND ANIMALS —————— MARINE PARTICULATE ORGANIC MATTER BITUMINOUS SEDIMENTS. PEAT, COAL CRUDE OIL SOILS SYNTHETIC FERTILIZER REAGENTD SAN S BIOLOGIC FERTILIZERS

•ROCK N" METAMORPHIC ROCKS IGNEOUS ROCKS — DIAMONDS — AMMONIA AIR (CLEAN?) AIR (NEAR LIVESTOCK) AMMONIUM IAEA IAEA USGS25 N1 N2 USGS26 REFERENCE MATERIAL AIR (PARTICULATE) PRECIPITATION SEAWATE ESTUARIED RAN S SOILS (EXTRACTABLE) VOLCANI CONDENSATES CGA S SYNTHETIC FERTILIZER REAGENTD SAN S BIOLOGIC FERTILIZERS ______I______i »60 -40 -20 0 20 40 60 n %c

Figure 1. Partial compilation of <515N values for some categories of substances on Earth 15 (modified from [7]). The value o NI/air is defined by Equation 1 of Appendix A. Thin lines indicate ranges of reported values; thick bars indicate typical values for some common 15» substances secondare Th . y reference materials evaluate thin di o st stud0 5 Ny( —3 i/ai = r + 180%o) allow normalization over almost the full range of reported values in natural substances on Earth, i.e., those not enriched or depleted in 15N by artificial processes.

53 USGS25 [(NH4)2SO4]: USGS2 a drie s 5i d salt prepare J.Ky b d . Böhlke (USGS, Reston, Virginia, USA) in 1992. It was prepared by dissolving and recrystallizing a mixture of normal reagent salt and 15N-depleted salt and has a 515N value of approximately —30%o [7].

USGS26 [(NH4)2SO4]: USGS2 a drie s 6i d salt prepare J.Ky b d . Böhlke (USGS, Reston, Virginia prepare1992,s n i USAwa ) t I .dissolvin y db recrystallizind gan gmixtura e of normal reagent 15sald N-enrichean t 5a 15s Ndha valusald an tf approximatel eo y +54%c [7].

USGS32 [KNO]: USGS32 is a dried salt prepared by J.K. Böhlke (USGS, Reston,

Virginia, USA) in 19923 . It was prepared by dissolving and recrystallizing a mixture of normal reagent salt and 15N-enriched salt and has a 515N value of approximately +180%o [7].

. 3 ORGANIZATIO TESE TTH F NO

laboratorieEac2 2 f ho s that agree participato dt e receive packagda e containing aliquots of seven reference materials (NSVEC, IAEA-N1, IAEA-N2, IAEA-N3, USGS25, USGS26, USGS32 latn )i e 199 2r earlo y laboratorie5 19931 e .Th correspondind san g analysts from which results were received by March, 1994 are listed in alphabetical order in Appendix B. Analyst differene th n si t laboratories prepare materiale dth theiy methodn sb ow r used san d theicalibration ow r n scheme r isotopisfo c abundance measurements.

. ANALYTICA4 L METHODS

Method r preparatiofo s f sampleno r N-isotope-ratifo s o measurements differ widely amon participatine gth g laboratories. techniquee Somth f eo e listear s Tabln di usin1 e g abbreviations given belo squarwn i e brackets. laboratoriee Mosth f o t s used some variation of a combustion procedure with Cu metal and Cu oxide (CuO or Cu2O) to buffer oxygen fugacities at high temperature (designated as [CuO] in Table 1). In principle, the buffered combustion methods convert quantitativel l nitrogeyal n specie sampla 2 gasn si N o e,t whic h is purified either by cooling the combusted sample slowly in contact with CaO [CaO] or by subsequent cryogenic separation [Cryo] or gas chromatography [GC]. In some laboratories, ammoniu reactioy b s oxidizes ga mn wa 2 witN o dhstront a g oxidant suc NaOBs ha r [BrO], while nitrat s treateewa aqueoun di s solution wit ha metalli c substance suc s Devarda'ha s alloy [Dev] to produce ammonium, which was then neutralized and distilled from the solution [Dist], collected, dried oxidized . 2 an ,Loca N o dt l calibrations f oo N values with respect 15 to atmospheric N were made using a variety of reference materials (air, salts, and tank gases) with 015N value2 s between approximatel yd +3%o —3%an o .

54 Table 1. RESULTS OF INTERLABORATORY COMPARISON TEST OF NITROGEN ISOTOPIC REFERENCE MATERIALS* , laboratorID y identification numbers assigne orden di f increasino r g o'N values reporte r IAEA-Ndfo 1 (no alphabetican i t l order); BrO, oxidation using NaOB r LiOBro r ; CaO,

purificatio 2 usinN f go n CaO; Cryo, cryogenic separation; CuO, combustio5 r Cuo n 2witO ; hCu Dev , reaction with Devarda's alloy Distd an ; , distillation.)

NSVEC IAEA-N1 IAEA-N2 IAEA-N3 USGS25 USGS26 USGS32 ID 015N 515N n Ô!5N n Ô15N n Ô15N n Ô15N n 515N n Methods 1 0.19±0.19 3 20.68±0.13 3 4.64 ±0.02 3 -30.84±0.22 3 54.44±0.10 3 183.26±0.48 3 CuO, Cryo 2 -2.50±0,05 2 0.30±0.40 12 20.30±0.20 12 4.70±0.20 12 -30.80±0.30 10 53.80±0.50 12 180.50+1.00 12 CuO, GC 3 -2.87±O.I4 3 0.37±0.10 3 20.42 ±0.07 8 4.83±0.03 2 -30.15±0.12 3 53.74±0.09 3 179.87±0.27 3 CuO, CaO 4 -2.70±0.04 6 0.38±0.14 4 20.14±0.06 6 4.14±0.12 4 -30.16±0.09 4 53.21 ±0.15 4 177.02±0.24 4 BrO, Dev, Dist 5 -2.79±0.03 6 0.38±0.02 5 20.37 ±0.02 5 4.68±0.02 4 -30.30±0.02 4 53.76±0.03 4 180.39±0.03 4 CuO, CaO 6 -2.74 ±0.05 3 0.40+0.18 9 20.38±0.13 10 4.57 ±0.09 9 -30.92±0.16 5 53.54 ±0.19 5 180.86±0.20 5 CuO, CaO 7 -2.72±0.17 4 0.42 ±0.24 6 20.34±0.26 6 4.95 ±0.56 6 -30.46±0.32 6 53.86±0.41 6 179.05±0.83 6 BrO, Dev, Dist 8 -2.79±0.02 1 0.42±0.01 3 20.45±0.09 3 4.88±0.16 5 -29.84±0.12 3 53.28±0.20 6 178.53 ±0.05 3 CuO, Cryo 9 0.44±0.05 9 20.28±0.10 11 4.70±0.16 10 -30.07±0.14 12 53.83 ±0.10 8 179.83±0.30 8 CuO 10 -2.81 ±0.02 1 0.45 ±0.03 4 20.39±0.05 5 4.68±0.02 3 -30.21 ±0.03 5 53.74 ±0.04 5 180.10±0.10 5 CuO, CaO 11 0.48±0.11 2 20.17±0.07 2 4.59±0.08 2 -29.94±0,32 2 52.94 ±0.15 2 177.77±0.11 2 CuO 12 0.53±0.14 3 19.75±0.04 3 3.97±0.14 3 -29.63±0.10 3 51.44±0.24 3 172.41 ±0.48 3 CuO, Cryo 13 -2.75 ±0.02 5 0.55 ±0.05 2 20.20±0.00 2 4.65±0.05 2 -29.80±0.00 2 53.20±0.00 2 177.10±0.10 2 BrO, Dev, Dist 14 0.62±0.08 2 20.31 ±0.21 2 5.60±0.27 4 -30.39±0.17 2 53.84±2.78 2 173.28±2.63 4 BrO, Dev, Dist 15 -2.95 ±0.02 5 1.00±0.40 4 20.40±0.10 5 -29.10±0.10 5 48.20±0.20 6 CuO, Cryo

* o N values are given with respect to atmospheric nitrogen as reported by each laboratory (based on various local calibrations). 1

Ui Ui 4 4 NSVEC 3 - x(all) =-2 76 ± 0 11 (n=10) (n=85 0 )0 ± 7 x(-2s7 2 - )= 2 l

4 IAEA-N1 3 X(all 046±0l7(n=15)= ) x(-2s 043±007(n=!2)= ) 2

0 „HL.- o o

IAEA-N2 203= ) l ±019(n=15) x(-2s 2032±009(n=13)= )

o o ° oj

T a IAEA-N3 l 3 x(all) = 468±036(n=14) x(-2s) = 469±009(n=10) 2 T

CO CO CO

USGS25 = -3017±047(n=15) = -3025±038(n^l4)

4 USGS32 3 x(all) -17857±282(n=14) x(-2s)=17918± 1 32(n=ll) 2 1

OO 00

jr, in

Figur . e2 Histogram s showing reported laboratory mean value N 5 r secondari/ai f fo so r y reference materials. The numbers in the columns identif1 y laboratories as listed in Table 1. Mean values (± la) are given for all reported laboratory means (all) and those remaining afteoutliera 2- r s were rejected (-2er).

56 . TES5 T RESULTS

5.1. Summar Selectiod an y Datf no a

Results of measurements by the 15 reporting laboratories are summarized in: (1) a table of reported value histograa s ) (Tabl(2 ; eacr 1) em fo h reference materia) (3 l (Figurd an ; e2) selected correlation diagrams illustrating laboratory biases (Figure 3). Each laboratory was assigne n identificatioa d n number whic s arbitrarilwa h y chose o reflect n e relativth t e magnitud 6s 15it Nf eo valu IAEA-Nr efo 1 (Tabl . Those1) e identification numbers appear summare eacth n i f ho y figures. Fro date mTablth n ai , arithmeti1 e c mean standard san d deviations (1er) were calculate datl d al for a) wit:(1 selectiono hn l datal a) afte(2 ; r elimination of values that differ from the means by more than two standard deviations, followed by recalculatio eliminatiod nan n unti l valueal l s were withi nstandar2 d deviations (Tabl; e2 Figur. e2)

Tabl . e2 OVERAL L MEANS, SELECTED MEANS NORMALIZED AN , D VALUES 15 FOR S Ni/ai; r (n)a l [Dat%o.]n , + i givee x aar s na

As Reported Normalized to Not Normalized 015NuSGS32/air = +l«0%o ID No Selection Minus 2-ff No Selection Minuo s2- Outliers Outliers

NSVEC -2.76 ±0.1 1(10) -2.77 ±0.0) 5(8 -2.75 ±0.1) 0(9 -2.78±0.04 (8) IAEA-N1 + 0.46±0.17(15) +0.43 ±0.07 (12) +0.43±0.11(14) +0.43 ±0.07 (12) IAEA-N2 +20.31±0.19(15) +20.32±0.09 (13) +20.46 ±0.21 (14) +20.41 ±0.12 (13) IAEA-N3 +4.68±0.36(14) +4.69±0.09(10) +4.72±0.38 (14) +4.72±0.13(11) USGS25 -30. 17 ±0.47 (15) -30.25 ±0.38 (14) -30.50±0.40 (14) -30.41 ±0.27 (13) USGS26 + 53.12±1.47(15) + 53.62±0.25(11) + 53.91 ±0.61 (14) +53.75 ±0.24 (13) USGS32 178.6±2.8(14+ ) + 179.2±1.3(11) + 180 + 180

From the data in Table 1 and Figure 2, it is evident that the range of reported 515N value eacr sfo h reference materia larges li r tha precisioe nth n reporte eacr dfo h valu mosy eb t laboratories. Some possible reason thir sfo s variability might include ) loca(1 : l calibrations with respect to atmospheric N2 differ among the laboratories; (2) methods of analysis

(including potential blanks and isotopic fractionations) differ widely among the laboratories; an) differend(3 t mass spectrometer r data-reductioo s n procedure yiely sma d different N515 scales. The reported data for some of the materials with 015N values far from 0%o (e.g., USGS26 and USGS32), after elimination of 2-a outliers, are slightly skewed and bimodal, wit majoe hth r modes farthe rr thos (Figurfroo Fo me 0% . materialse2) possibls i t i , e that the larger modes farther from 0%o may be closer to the correct values because: (1) experimental blank likele sar causo yt e deviations toward zer thosr ofo e materials with 615N value r frosfa mlaboratore zeroon ) (2 ; y that reported relativel 6w I5ylo N value r thossfo e

57 55.0

g 54.0 9 2 .-•' 14 c 4.13 3 53.0 i .-»"" \aO ;i .--'' 11 52.O 0i Cfl +air Z 51.0

50.0 170 175 180 185

-29.5 -t o ^ 12 '3

.£ -30.0 .10 «3

g -30.5 14 Z

10 , -31.0 — 170 175 180 185 §15NuSGS32/ain «n %

Figur . e3 Correlatio n diagrams showin effecte gth f interlaboratoro s y variation 5n i s15 N scales. Data labels identify laboratories as listed in Table 1. Dashed lines go through the

origin ("air", representing the isotopic composition of atmospheric N2) and through the recommended normalized values (Table 2, last column). Because most laboratory reference materials and most potential contaminants have 015N values relatively near zero, variations alon dashee gth d lines coul causee db laboratory db y blanks, error valuen si s assignee th o dt laboratory reference materials, or mass spectrometric 5-scale variations.

materials (laboratory 8) reported possible incomplete recovery of N2 from a molecular sieve trap; and (3) two other laboratories (laboratories 11 and 13) that reported relatively low 515N values for those materials reported only two trials each. Thus, for USGS32, it is possible that a value near +180%o should be preferred to the mean (—2o) of +179.2%o.

5.2 Effect of Normalization

From correlation sonee sucth ss hsummarizea Figurn di t appeari , e3 s thadifferene th t t laboratories have consistently different (expande r contracteddo ) 015N scales. Therefore, some major discrepancies among the ô N values reported by different laboratories may be removed by adjusting the length of the S15N scale for each laboratory (i.e., "normalization," as described by Gonfiantini [4] and Coplen [12]). To illustrate this effect, we calculated normalized 615N values for all of the analyses in Table 1 by assuming that the correct ô N value of USGS32 relative to N2 in air is +180%o exactly and that the corrections to the 15 15 o N,7air value other sfo r material proportionae sar theio t l r o N,-/air d values—sean ] 12 , e[4 Equation 8 in Appendix A. For example,

58 15 NUSGS32/air,realô NUSGS26/air,meas Ô N USGS32/air, norm 15 0 NUSGS32/air,meas

15 e m s where ö NUSGS26/air)nor normalize* m d valu f USGS2o e 6n airi relativ2 ,N o t e ôl5NusGS32/air,reaiis +180%o, ô^Nusosze/air.meas e imeasuresth d valu USGS2f eo 6 relative § tne to N2 in air, and ô NusGS32/air,meas i measured value of USGS32 relative to N2 in air. 15 Normalized value ô N f USGS26/aiso 5d an rN USGS25/ai summarizee ar r e Figurn di Th . e4 effect of the normalization in both cases is to reduce the range and standard deviation of S15N values and to shift the mean values slightly away from 0%c (Table 2 and Figure 4). If the selection of data for USGS32 is justified, then the preferred 515N value for USGS26 with respec atmospherio t t s approximateli 2 N c y +53.8% thad r USGS2an fo ot s -30.4%o5i .

Unnormalized

Normalized 815NuSGS32/air s 180%»

Sl5NuSGS26/airc % n ,i

Unnormalized

5 ^ 19 0 4 n I 8 I is 1 12

Normalized 815NuSGS32/air = 180%,

0 3——'

Figure 4. Histograms showing the differences between reported laboratory mean values of

<>NusGS25/ai^Nusos26/ai ^r r befor afted ean r normalization (Appendi , EquatioxA . n8)

an( 15

Th1S e numbers in the columns identify laboratories as listed in Table 1. Normalization reduces the variability of the data to levels that are not much greater than many of the individual laboratory uncertainties meae Th .n valuenormalizee th f so d dat e furtheaar r from zero than the meanreportee th f so d data becaus selectee eth d valu f oeo N 32 j (+180%o) used 15 USGS /a r normalizatioe foth r s slightlni y higher tha unbiasee nth d mean (—la) value (+179.2%c).

After normalization, the remaining variations 6/inj anNo15 25/d Nj ofrom a r r a USGS

different laboratories (Figure 4) are smaller than thosUSGS2 e of the reported values (Figure 2), but still slightly larger thareportee nth d uncertainties from individual laboratories (Tabl. 1) e

From correlations like those in Figure 3, it is evident that similar results would be achieved by normalizing N515 values over the whole range from — 30 %e to +180%o.

3 Calibratio5. n With Respec Atmospherio t c Nitrogen

Precise calibratio N-isotope-ratif no o measurements with respec atmospherio t c(thN e

primary N-isotope-ratio standard) may be complicated by mass-spectrometer-specifi2 c measurement effects of argon in air, which can cause errors in N015 measurements of 0.1 %o to 0.2%o [5, 13, 14]. None of the laboratories participating in this study reported testing the effect of argon on measurements of atmospheric N2. Therefore, it is possible that

5 2 ; 4 , [ Unnormalized 5 i 6 2 | 4 9 7 I , ____ Il2 1 11 13 10 : n

5 1 4 • Normalized 1 3 3 ! 8 5NusGS32/air ; =180%. 2 1 4 ! 2 6 5 7 • 1 9 10 11 "ill 12 n ,,-,,,,..,,.

o15NlAEA-N2/airn ,i

; 5 4 ! Unnormalized 3 • 2 • 1 » ' 0

• 5 4 • Normalized 15 3 5 NuSGS32/IAEA-N1 = 179.5%. ; 2 i :

• 0

Ö15NIAEA-N2/IAEA-N1o % n i ,

Figure 5. Histograms showing the differences between reported laboratory mean values of o15NiAEA-N2/air an^ ^15NlAEA-N2/iAEA-Ni before and after normalization (Appendix A, Equations 8 and 9). The numbers in the columns identify laboratories as listed in Table 1. After normalization, the interlaboratory differences are similar in magnitude to individual laboratory precisions. Slightly larger variability in normalized values of S^NJ^A.^/^

compare oo dt NTAEA.N2/iAEA-N y indicati componenea f uncertainto t y associated with the calibratio15 n of laboratoryma standards near 0%o.

60 measurements against a secondary reference material like IAEA-N1 could yield simpler and more precise interlaboratory calibrations than measurements against atmospheric N2. That possibility canno evaluatee b t t presenda t becausparticipatine th f o e w onlfe yga laboratories reported analyses of atmospheric N2 samples. Instead, we compare the isotopic compositions reported for IAEA-N2 with respect to IAEA-N1 and for IAEA-N2 with respect to N2 in air as determined by the various local calibrations (Figure 5). Normalized 015N values were calculate IAEA-Nr dfo 2 with respec IAEA-No t 1 wit hversioa Equatiof no Appendin i n8 x A:

5-15*7 _ ^ NUSGS32/IAEA-Nl,rea]a NIAEA-N2/IAEA-Nl,meas 0 ^lAEA-NZ/IAEA-Nl.norm ~ —————————-jj-———————————————————————»

0 NUSGS32/IAEA-Nl,meas

assumes wa e +179.49% b o dt wher ec N 5base15 valuen do f +0.4so 3 %r ofo l

IAEA-N1USGS32/Nljrea and +180%c for USGS32 (see Equations 6 and 7 in Appendix A). After

normalization, the variation in O^NI^^^/IAEA-NI>norm is slightly smaller than that for iAEA-N2/air,non^15N n (Figure 5). Thus, some of the overall uncertainties in the reported 15 S Nl7air values may be attributable to errors in local calibrations; those errors are slightly larger than most laboratory precisions, but they are smaller than the effects of varying 015N scales when applied to materials with 615N values far from zero.

4 Evaluatio5. Isotopif no c Homogeneity

Only a few laboratories reported possible evidence of isotopic heterogeneity in isolated reference materials (IAEA-N1, IAEA-N3, USGS25 USGS3d ,an 2 were each mentioned once). However, the data and the comments of the analysts do not indicate a consistent problem with any of the materials for sample sizes in the range of 10 to 100 /xmol of N (see also [7] for homogeneity tests of USGS25, USGS26, and USGS32). Some laboratories reported regrinding and homogenizing the salts before analyzing them; that practice, if followed elsewhere, apparently removed measurable isotopic heterogeneity if it existed in the original materials.

6. DISTRIBUTION INFORMATION

Aliquot reference th f o s e materials teste thin di s stud e availablyar e fro e U.Smth . National Institute of Standards and Technology (NIST), Standard Reference Materials Program, Room 204, Building 202, Gaithersburg, Maryland, 20899, USA frod e an m,th International Atomic Energy Agency, Sectio Isotopf no e Hydrology, Wagramerstrass , P.Oe5 . Box 100, A-1400 Vienna, Austria.

7. CONCLUSIONS

As a result of a new interlaboratory comparison among 15 laboratories, 515N values are know secondar7 r nfo y reference material N-isotope-ratir sfo o measurements over almose tth whole range of values encountered in natural substances on Earth. Patterns in the reported data indicate that discrepancies amon valueN ô g s measure differenn i d t laboratoriee ar s

61 largely the result of (1) variations in the N515 scales of the different laboratories (either instrumenta r procedural)o l ) error(2 f calibratiod so an , n with respec atmospherio t t . cN

Discrepancies coul reducee db d significantl adoptioy yb universally-acceptef no 2 valueN dd s for secondary reference materials for normalization purposes.

ACKNOWLEDGEMENTS thane W k sincerel participante yth thin si s study , Appendixe listeth n di shareo wh , d freely data and methods. W. Stichler supplied aliquots of some of the reference materials unpublishee studth e used th yn dan i d result somf so e previous measurements . GwinC . d nan B. Davis assisted with analyses in our laboratory. We thank K. Revesz and R. Carmody for constructive comments on the manuscript. firmf o e , brandUs tradd an , e name thin si s papeidentificatior fo s ri n purposes onld yan does not constitute endorsement by the U.S. Geological Survey.

REFERENCES

[I] HOERING, T., Science 122 (1955) 1233-1234. [2] JUNK, G., SVEC, HJ., Geochim. Cosmochim. Acta 14 (1958) 234-243. [3] MARIOTTI, A., Nature 303 (1983) 685-687. [4] GONFIANTINI, R., Nature 271 (1978) 534-536. [5] HUT, G., "Consultants' group meeting on stable isotope reference samples for geochemical and hydrological investigations", IAEA, Vienna (1987) 42 pp. ] KENDALL[6 , GRIMC. , , AnalE. , . Chem (19902 6 . ) 526-529. ] BÖHLKE[7 , J.K., GWINN, C.J., COPLEN T.B., Geostandards Newslette (19937 1 r ) 159-164. ] PARR[8 , R.M., CLEMENTS, S.A., 1991, "Intercompariso enrichef no d stable isotope reference material medicar sfo biologicad an l l studies", IAEA, Vienna (1991) NAHRES-5, 31 pp. ] COPLEN[9 , T.B., KROUSE, H.R., BÖHLKE, J.K., Pure Appl. Chem (19924 6 . ) 907- 908. [10] GONFIANTINI, R., "Advisory group meeting on stable isotope reference samples for geochemica hydrologicad an l l investigations", IAEA, Vienna . (1984pp 7 )7 [II] STICHLER, W., IAEA (1993, written communication). [12] COPLEN, T.B., Chemical Geology (Isotope Geoscience Section) 72 (1988) 293-297. [13] MARIOTTI , Natur A. , (19841 e31 ) 251-252. [14] GERSTENBERGER, H., MÜHLE, K., ROHMER, D., "The influence of Argon contents in nitrogen gas samples on 515N-values measured", in réf. [5], p. 28-29. [15] AUDI, G., WAPSTRA, A.H., Nucl. Phys. A565 (1993) 991-1002. [16] IUPAC, Pure Appl. Chem. 63 (1991) 991-1002. [17] IUPAC, Pure Appl. Chem. 64 (1992) 1519-1534. [18] DANSGAARD, W., Meddelelser Om Grönland 177 (1969) 3-36.

62 Appendix A. Background Information

N Isotope Masses

14.00307400= N 14 8 (23)u [15] 15.00010897= N 15 8 (38)u [15]

N Isotopic Abundances in Atmospheric N2

14N/15 N272.3 = 0. 0± [2] 14N/15 Ns 27 2 [9] 99.633= r atom%N 14 (o 7 % fractio )n i , nof [16] 15N = 0.3663 fraction of, in % (or atom%) [16] Standard atomic weight = 14.00673 [17]

Delta Conversions

15 15 For an unknown, x, expressed relative to z or if o NJ/air is the 5 N value of an unknown , expresse/ , d relativ atmospherio et , the2 have c N followin ne w eth g equations.

14N 15 5 Nx/2 (in %o) = 1000 (1)

14N

15 ô15N, - 100 0 272 " N^ - 1 (2) 7air 14 N j

15 15 _ o N + 1000 N i7air (3) 14 272000 N i

15 15 atom% N,. _ i (4) ô N, /ai100= r 0 272 100- atom %15N,-

100 atom%15N = f 272 (5)

1000

63 (6)

1000 = 1000 -1 (7) 1000

Normalization Equations For measurements calibrated directly by measurement of atmospheric N2 (air) under the same conditions as USGS32 and the unknown (0 (equivalent to Equation 1 of [4]):

^ NUSGS32/air,real^ Ni/air,meas ' i7air,norm (8) «"N,USGS32/air,meas

where ÔNuso^^aij.^ assumes ^i d provisionall 180%o+ measurementr e b Fo .yo t s against workina g referenc calibrated an (rg)s ega measuremeny db secondara f to y reference material (srm) under the same conditions as USGS32 and the unknown (0 (numerically equivalent to [12]n i alsEquation5 e 1 ; se o d [18])f pago an 6 4 e3 s1 :

" i/air,norm NUSGS32/rg,meas

ll5*T (9) NUSGS32/air,real Njnn/air,real

15 where o NUSGS32/airtreal and o^N^^^^^ are independently calibrated normalized values (e.g. 0.4d ,3 an 180%IAEA-N1)s i % oo m isr f .

64 Appendix B. List of participants (in alphabetical order)

Greg Arehart/Neil Sturchio, CMT-205, Argonne National Laboratory, Argonne, IL 60439, USA

John Karl Böhlke/Tyler Coplen Nationa1 43 , l Center, U.S. Geological Survey, Reston, VA 22092, USA

Hervé Casablanca, CNRS Service Central d'Analyse, Autoroute Lyon-Vienne-Echangeur de Solaise , F-693922 P B , 0 Vernaison, FRANCE

Marilyn Fogel, Geophysical Laboratory, 5251 Broad Branch Rd., Washington 20015C D , - 1305A US ,

Gerhard Gebauer, Lehrstuh Pflanzenökologier fü l , Universität Bayreuth, Universitätsstr., D-8580 Bayreuth, GERMANY

Ede Hertelendi , Institute of Nuclear Research, Hungarian Academy of Sciences, PO Box , H-40051 1 Debrecen, HUNGARY

Kaplann la , Global Geochemistry Corp., 6919 Eton Ave., Canoga Park 91303-2194A C , , USA

Barbara Kornexl, Lehrstuh Allgemeinr fü l e Chemi Biochemid eun r Technischeede n Universität, München, D-8050 Friesing-Weihenstephan, GERMANY

Joseph Montoya, Biological Laboratories, Harvard University, 16 Divinity Ave., Cambridge, MA 02138, USA

Nathaniel Ostrom, Dept f Geologicao . l Sciences, Michigan State Univ., East LansingI M , 48824-1115, USA

Yury Shukolyukov, Isotopes Geochemistry Laboratory, V.I. Vernadsky Institute, Kosygin str. 19, SU-117975 Moscow, RUSSIA

Steve Suva/Carol Kendall 435S M ,, U.S. Geological Survey Middlefiel5 34 , d Rd., Menlo Park, CA 94025, USA

Daniel Snow/Roy Spalding, Institut f Agricultureo Naturad ean l Resources Natura3 10 , l Resources Hall, Universit f Nebraskayo , Lincoln 68583-0844B N , A US ,

Suzanne Voerkelius, HYDROISOTOP Gmbh, Woelkestrass , 806e9 9 Schweitenkirchen, GERMANY

Len Wassenaar, National Hydrology Research Institute, 11 Innovation Blvd., Saskatoon, Saskatchewan CANAD5 3H N S7 A,

65 Appendi . OthexC r Reference Material Nitrogen-Isotope-Ratir sfo o Measurements

NBS-14 [N2 gas]: NBS-14 has been loosely described as "atmospheric nitrogen" [10]. Kendal Grid ] reporan lm [6 8s t tha= -1.18%Nha t i t « relativ atmospherio et t Hu . cN 15 2 [5] suggests that NBS-14 should be replaced by NSVEC in common use.

IAEA-305 IAEA-305d Aan B [(NH^SOJ: IAEA-30 IAEA-3055d Aan driee Bar d salts provide . FerE y ndb (Nestle Research Centre, Vevey, Switzerland) beforn i e e us 198r fo 9 medica biologicad an l l tracer studies [8]. ParClementd an r ] repor[8 s result e n th a t f o s intercomparison test that yielded "provisional certified values" and 95% confidence intervals:

IAEA-305A +39. 80.5%+ 0 (n=23) IAEA-305B +375.3 ± 2.3 %o (n=25)

IAEA-310A and IAEA-310B [CO(NH2)2]: ÏAEA-310A and IAEA-310B are dried salts provide . FausH y db t (Central Institut Isotopf eo Radiatiod ean n Research, Leipzig, Germany) before 1989 for use in medical and biological tracer studies [8]. Parr and Clements [8] report resulte intercomparison th a f o s n test that yielded "provisional certified % values95 d an " confidence intervals:

IAEA- A 310 +47. 2+ 1.3% o (n=24) IAEA-310B +244. 60.8%± o (n=23)

IAEA-311 [(NH4)2SO4]: IAEA-311 is a dried salt provided by E. Fern (Nestle Research Centre, Vevey, Switzerland) before 1989 for use in medical and biological tracer studies [8]. Par Clementd ran repor] s[8 resulte intercomparison tth a f so n test that yielde d"provisionaa l certified value" and 95% confidence interval:

IAEA-311 +469 %7 05 (n=28 3± )

66 INTERIABORATORY COMPARISON OF NEW MATERIALS FOR CARBON AND OXYGEN ISOTOPE RATIO MEASUREMENTS

W. STICHLER Isotope Hydrology Section, International Atomic Energy Agency, Vienna

Abstract Aliquot f fivso e different intercomparison materials were distribute interlaboratorn a r dfo y comparison test of stable carbon and oxygen isotope ratio measurements. The samples were sent twenty-threo t e laboratories whicf ,o reporte3 h1 d their results additioni h , , each laboratory received an aliquo NBS1f to NBS18d 9an result l Al . s were normalized with NBS1 reported 9an d versue sth V-PDB-scale. The results of the intercomparison, after elimination of 2-cr outliers, are listed below and represent provisional 013C and 018O values of these materials:

Identifi- Substance Carbon - 1 3 Oxygen - 1 8 cation S C-13 1-6- n S 0-18 l-ön -

IAEA-CO-1 Calcite 2.480 0.025 (10) -2.437 0.073 (11)

IAEA-CO-8 Calcite -5.749 0.063 (12)-22.667 0.187 (13)

IAEA-CO-9 Barium -47.119 0.149 (10)-15.282 0.093 (10) Carbonate LSVEC Lithium -46.479 0.150 (11)-26.462 0.251 (10) Carbonate USG4 S2 Graphite -15.994 0.105 (8)

Introduction

Since PDB is virtually non-existent, a reference Vienna-PDB (V-PDB) was introduced in 1985 defined by using NBS19 as a reference material with:

18 §i3C =+i.95%0ando O = - %o versus V-PDB

hi addition, only one carbonate intercomparison material NBS18 was available. The 0- values of this material represented mean values of measurements carried out by different laboratories.

Given the growing demand for O^C- measurements at about -50 %o versus V-PDB, the Isotope Hydrology Section decide produco d t intercompariso w ne ea n material (IAEA-CO-9)o T . avoi problemy dan s wit diminishine hth g suppl NBS1f yo NBS1d 8 an additiona o 9tw l materials suitabl o replact e e abovth e e mentioned were produced (IAEA-CO- d IAEA-CO-8)an 1 . Furthermore, the National Institute of Standards and Technology (NIST) put two additional graphitw materialne f r thedisposala o ou eme n i t (UGS-24 a sOn . should an ) d replace eth exhausted NBS21 othee th , lithius ri m carbonate (LSVEC) also deplete 13n dCi .

67 These intercompariso fivw ene n materials were distribute interlaboratoie th dn i y comparison, test together with NBS18 and NBS19.

Description of aie intercomparison materials: IAEA-CO! Calcite(lg)

This materia marbls i l eIAEe useth An di recen l IntercomparisoC t n Exercise slaA f .b o freshl t Carrarycu a marbl supplies ewa EMEGy db , Vareggjo, Ital milled yan d dow powdeo nt y rb the IAEconsensue Th A s o*3C valu thir efo s material calculate analyse9 basie 5 th f o sn do s reported by radiocarbon laboratories amounts to o"3C = + 2.42 %o versus V-PDB. It is intended to usee complementara b s da y materia NBS19o t l . IAEA-CO-8 Calcit) Ig e(

This material is a natural carbonatite originating from Schelingen at the Kaiserstuhl, Germany. The material was ground from 0.09 to 0.5 mm grain size and put at our disposal by the Geologisches Landesamt, Freiburg, Germany. Preliminary measurements yielded values in the order of 013C ~ 6 %o versus V-PDB and 98O « -23 %o versus V-PDB. It is intended to use this material complementar NBS18o yt . IAEA-CO-9 Barium carbonate (lg)

This barium carbonate powde bees rwa n prepareBrenninkmeijerA C. y db , Atmospheric Division, National Institute for the Atmosphere and Hydrosphere, Lower Hutt, New Zealand, with the following procedure. CQ was produced from burning natural gas and absorbed in a NaOH solution. Barium sulfide was used to precipitate the carbonate. The Barium carbonate was thoroughly purified over periods of weeks by washing it with distilled water. Homogenization was obtained with the solid still submerged in the liquid. The carbonate was dried under vacuum, after

which further homogenization took place. The isotope measurements performed by the producer were reported as SPC = -47.23 ± 0.03%o versus V-PDB and$C8 - -15.95 ± 0.05%o versus V-PDB.

LSVEC: Lithium carbonate (0.4g)

18 16 12 This lithium carbonat intendes eO/i O useisotope d C b 13 r / an dC do fo t e ratio analysis. Furthermore lithiue th , m isotope abundanc determines ewa published dan NISy db 7.52To t 5 atom percent 6Li and 92.475 atom percent 7Li (G.D. Flesch, AR. Anderson and H.J. Svec, hit. J. Mass Spectrom 265-27, Phys.n 12 Io .l ,vo 2 (1973)prepares . Svewa t H I .cy d b (Iow a State University). The prelirninary values were given by NIST to 513C = -46.7 ± 0.9 %o versus VPDB and018O = %1 26.o0. versu8± s VPDB.

USGS24: graphite (0.4g)

This graphit s prepareewa T.By db . Coplen (U.S. Geological Survey usiny b ) g Baker technical grade graphite (96%. 325 mesh). Prior to splitting with a sample splitter, six spatially separated ~ 1 mg samples were analysed to ensure isotopic homogeneity of the material. Peak-to- peak variation was 0.11 %o . It is intended to use this material in replacement of the exhausted NBS21 preliminarA . y valu -15.= gives C 0.2eNIS± wa 9y nVPDBb SP o 5s T% a .

68 Distributio intercompaiisoe di f no n materials

The materials were sent to 15 laboratories which were selected on the basis of participating institutes in former interlaboratory tests organized by the IAEA Eight laboratories reported their results. Furthermore, three laboratories wanted to participate in the ring test, two of them sent the results to the IAEA To improve the statistics of the provisional values, a further eleven laboratories were informe asked dan participate o dt . Five laboratories gav positivea e answet rbu only three of them reported their results. Altogether, twenty three laboratories received the intercomparison materials and thirteen of them reported their results (about 55 %) to the IAEA The different laboratories prepared the samples by their own methods and used NBS19 for calibration 0-valuese th f o .

In Appendix A the participating laboratories are listed in alphabetical order.

Results

The results of the measurements reported by the 13 laboratories are summarized in table 1 an . d2 Thes tableo e sigmtw e son acontaie value0 th erro e d s reportea nth re an s th y db participants. These data were used to calculate the arithmetic mean values (Mean), and standard deviation (Std. Dev) liste . 4 Withi threo Tablet n dd i e ean non iteratio3 s n steps values were eliminated which differ from the mean values by more than two standard deviations. The table in the Abstract summarize meae th s n values wher l measurineal g data were withi standaro ntw d deviations. The number of values (n) used to calculate the mean values is indicated.

Conclusions

For carbon-13 it is clearly indicated that the uncertainty is small for the measurements of the different laboratories carried out on the materials: NBS18, IAEA-CO 1 and IAEA-CO-8. The absolut range th NBS19f valuee0 en o i e sar . Except IAEA-CO-8e seee th b same n ni n th , eca oxygen-18 results standare Th . d deviation materiale th f so s IAEA-CO-9, LSVE VGSd e Can ar G4 2 in the range of 0.09 to 0.25 %o (both isotopes), and are most of the time larger than the reported uncertainties fro individuae mth l laboratorie. s (Table2) d an 1 s

Therefore, it is strongly recommended to introduce a second reference material (e.g. IAEA- CO-9 or LSVEC) with fixed 0-values to improve the present situation.

ACKNOWLEDGEMENTS

We sincerely thank participante sth thin si s intercomparison rinalse gar o e testgratefuW . l to NIST which supplied aliquots of two intercomparison materials (LSVEC and USGS24) distributed in the ring test.

69 --4 O - valueS Tabl : s1 e reporte participatine th y db g laboratories. (For laboratory No 6 the values were normalized with NBS 19, the werresult* 6 d e lin n obtainesan i e6 usiny db g different mass spectromete preparatiod ran n device)

Intercomparison: Carbon-13 Results

LAB NBS 18 IAEA-CO-1 lAEA-CO-8 IAEA-CO-9 LSVEC USGS 24

S C-13 1-6 C-15 - 3 1-6 C-1S - 3 1-6 C-15 - 3 l-< C-1S r 3 1-6 C-1S - 3 1-cr

1 -4.96 0,03 2.52 0,04 -5.72 0.02 -46.82 0.04 -45.99 0.03 -15.84 0.04 2 -5.038 0.024 2.488 0.020 -5.803 0.017 -47.03 0.05 -46.471 0.03 -15.970 0.023 3 -5.07 0.02 2.44 0.05 -5.78 0.02 -47.16 0.03 -46.56 0.04 -15.91 0.01 4 -5.10 0.008 2.45 0.028 -5.81 0,030 -47.23 0,034 -46.37 0,048 -16.08 0.007 5 -4.98 0.06 2.46 0.01 -5.77 0.01 -47.28 0.06 -46.56 0.03 -16.09 0.08 6 -4.89 0.01 2.28 0.24 -5,64 0.03 -47.42 0.01 -46.67 0.02 6* -5.02 0.04 2,49 0.01 -5,69 0.12 -47.56 0.19 -46.71 0.01 7 -16.98 0.037 8 -4.99 0.05 2.51 0.05 -5.87 0,05 -47.04 0.04 -46.24 0.05 -16.09 0.05 9 -4.95 0.09 2.48 0.03 -5.68 0.08 -46,88 0.12 -46,58 0.12 10 -5.06 0.070 2.46 0.046 -5.75 0.042 -47.11 0.055 -46.41 0.034 -16.11 0.036 11 -4,98 0.02 2.50 0.01 -5.70 0.02 -47.00 0.05 -46.47 0.05 -15.86 0.05 12 -6.68 0.14 -46.3 0.25 -45.58 0.26 13 -5,06 0.014 2.34 0.012 -5,78 0.017 -47.04 0.006 -46.23 0.012 Table 2:5- values reported by the participating laboratories. (For laboratory No 6 the values were normalized with NBS 19, calculation from VSMOW scal VPDn ei B scal performeds ewa , the wer* result6 d e linn obtainesan i e6 usiny db g different mass spectrometer and preparation device)

Intercomparison: Oxygen-18 Results

LAB NBS 18 IAEA-CO-9 LSVEC

S 0-18 l-

1 -22.97 0.09 -2.36 0.03 -22.74 0.05 -16.02 0.03 2 -23.033 0.06 -2.52 0.05 -22.795 0.03 -15.30 0.05 -26.920 0.027 3 -22.85 0.03 -2.47 0.17 -22.49 0.1 -15.70 0.03 -26.86 0.05 4 -23.19 0.019 -2.50 0.028 -22.87 0.023 -15.36 0.032 -26.74 0.036 5 -23.04 0.08 -2.44 0.03 -22.86 0.09 -15.27 0.03 -26.78 0.06 6 -23.01 0.02 -2.90 0.06 -22.69 0.18 -15.39 0.04 -26.61 0.08 6* -22.98 0.01 -2.32 0.04 -22.53 0.36 -15.3 0.32 -26.35 0.06 8 -22.75 0.1 -2.40 0.05 -22.9 0.1 -15.25 0.19 -25.58 0.05 9 -22.90 0.09 -2.34 0.05 -22.67 0.17 -15.21 0.05 -26.27 0.11 10 -22.88 0.082 -2.42 0.029 -22.7 0.2 -15.08 0.072 -26.21 0.047 11 -22.99 0.02 -2.49 0.02 -22.78 0.02 -15.24 0.02 -26.81 0.06 12 -22.32 0.3 -14.85 0.26 -25.58 0.25 13 -23.05 0.014 -2.55 0.016 -22.33 0.012 -15.42 0.010 -26.83 0.011 Table 3: Evaluation of Carbon-13 Results

Intercomparison: Carbon-13 Results

LAB NBS18 IAEA-CO-1 IAEA-CO-8 IAEA-CO-9 LSVEC USGS 24

5 C-13 5 C-13 S C-13 S C-13 S C-13 S C-13

1 -4.96 2.52 -5.72 -46.82*** -45.99** -15.84 2 -5.04 2.49 -5.80 -47.03 -46.47 -15.97 3 -5.07 2.44 -5.78 -47.16 -46.56 -15.91 4 -5.10 2.45 -5.81 -47.23 -46.37 -16.08 5 -4.98 2.46 -5.77 -47.28 -46.56 -16.09 6 -4.89 2.28* -5.64 -47.42 -46.67 6* -5.02 2.49 -5.69 -47.56** -46.71 7 -16.98* 8 -4.99 2.51 -5.87 -47.04 -46.24 -16.09 9 -4.95 2.48 -5.68 -46.88 -46.58 10 -5.06 2.46 -5.75 -47.11 -46.41 -16.11 11 -4.98 2.50 -5.70 -47.00 -46.47 -15.86 12 -6.68* -46.30* -45.58* 13 -5.06 2.34** -5.78 -47.04 -46.23

Mean: -5.008 2.452 -5.821 -47.067 -46.372 -16.103 Dev.d St : 0.058 0.068 0.255 0.296 0.297 0.325

(*) 2.467 -5.749 -47.131 -46.438 -1 5.994 0.047 0.063 0.205 0.198 0.105

(**) 2.480 -47.092 -46.479 0.025 0.166 0.150

/***\ -47.119 0.149

(*), (**), (***) : Results outside of ± 2

72 Table 4: Evaluation of Oxygen-18 Results

Intercomparison: Oxygen-18 Results

LAB NBS18 iAEA-CO-1 IAEA-CO-8 IAEA-CO-9 LSVEC

0-1S 0-15 8 0-15 8S0-15 80-1 8 8

1 -22.97 -2.36 -22.74 -16.02* 2 -23.03 -2.52 -22.80 -15.30 -26.92 3 -22.85 -2.47 -22.49 -15.70** -26.86 4 -23.19* -2.50 -22.87 -15.36 -26.74 5 -23.04 -2.44 -22.86 -15.27 -26.78 6 -23.01 -2.90* -22.69 -15.39 -26.61 6* -22.98 -2.32 -22.53 -15.30 -26.35 8 -22.75* -2.40 -22.90 -15.25 -25.58* 9 -22.90 -2.34 -22.67 -15.21 -26.27 10 -22.88 -2.42 -22.70 -15.08 -26.21 11 -22.99 -2.49 -22.78 -15.24 -26.81 12 -22.32 -14.85*** -25.58* 13 -23.05 -2.55 -22.33 -15.42 -26.83

Mean: -22.970 -2.476 -22.667 -15.338 -26.462 StdDev.: 0.109 0.146 0.187 0.2710.456

) (* -22.970 -2.437 -15.281 -26.638 0.067 0.073 0.1930.251

(**) -15.243 0.153

(***) -15.282 0.093

(*), (**), (***) : Results outside of ± 2

73 APPENDIX A: List of participants (in alphabetic order)

Mr. AR Heemskerk r Plichde n t Mrva . J . Universit Waterlof yo o Centrum voor Isotopen Onderzoek Environmental Isotope Laboratory Rijksuniversiteit Groningen Waterloo,1 Ontario3G L N2 , Nijenborgh 4 CANADA NL-9747 AG Groningen NETHERLANDS

Mr. Calude Hilare-Marcel Universit Québeu éd Montréacà l Mr . BrenninkmeijeC . r GEOTOP National Institut Watef eo d ran C.P. 8888 SuccursaleA Atmosphere Research (NIWAR) Montréal (Quebec8 3P C )H3 Atmospheric Division CANADA P.O. Box 311.312 Lower Hurt NEW ZEALAND Mr. Tiping Ding Chief, Divisio f Isotopno e Geology Mr. T. Florkowski Institut Mineraf eo l Deposits Facult Physicf yo Nuclead san r Techniques Baiwanzhuang Roa6 d2 University of Mining and Metallurgy Beijing Al. Mckiewicz0 a3 PEOPLES REPUBLIC OF CHINA 30-095 Krakow POLAND

Mr . ErlenkeuseH . r Mr. U. Siegenthaler Institu Reinr fu Angewandttd eun e Universität Bern Kernphysik der Univ. Kiel Physikalisches Institut C-14 Labor Sidlerstrasse 5 Leibnizstrasse 19 CH-3012 Bern 2300 Kiel SWITZERLAND GERMANY Mr. J. Burdett Mr . TrimborP . n Stable Isotope Laboratory GSF-Institu Hydrologir fü t e Universit Michigaf yo n München-Neuherberg 100 LittlC 6C e Building D-85758 Oberschleißheim Ann Arbor, Michigan 48109-1063 GERMANY U.S.A

Mr. R Neubert . CopleB Mr . T . n Institu Umweltphysir fü t k 431 National Center der Universität Heidelberg U.S. Geological Survey Neuenheimem I r Fel6 d3 Reston, VA 22092 D-6900 Heildelberg U.S.A GERMANY o Ra Mr M . Isotope Group Bhabha Atomic Research Centre Trombay, Bomba5 08 0 y40 INDIA

74 CARBON, OXYGEN AND HYDROGEN ISOTOPIC INTERCOMPARISO FRUIF NVEGETABLO D TAN E JUICES

. KOZIEJ T Pernod-Ricard Research Centre, Créteil, France F. PICHLMAYER Österreichisches Forschungszentrum Seibersdorf, Seibersdorf, Austria A. ROSSMANN Lehrstuhl für Allgemeine Chemie und Biochemie, Technische Universität München, Freising-Weihenstephan, Germany

Abstract. Within (European frameworkCEN the the of Committee Standardization)for the Working Group 1 of the Technical Committee 174 Fruit and Vegetable Juices" is in charge of developing and validating isotope analytical methods, capable to improve the athentication of fruit juices. Here we report the results of several round robins recently carried out.

In 199 determinatioe 1th Carbone th f no conten 3 -1 fruin ti t juice sugar carries swa d oute Th . 15 european laboratories, participating in this task are listed in Table 1. The mostle yar y equipped with online combustion mass spectrometers. Sucrose from beed tan can wels ea juic s la e orange pineappld -an e juice samples deliverel al , Pernod-Ricardy db , were analyzed along with the isotope standards NBS 22, sucrose ANU and PEF 1, granted by the IAEA. appliee Th d sample pretreatmen juicee th f sto accordin Profo gt . H.L. Schmidt, TU-Munichs ,i shown in Fig. 1 . The analyses were accomplished except in one case in triplicate and the results Tablegive. e e 5 th ar o t n i s2

The 8*3c-values were normalized against NBS-22, the mean value of which was determined

by die participants as 8l3cpDB = -29,8 ± 0,2 %o.

From the statistical treatment of the data in compliance with ISO 5725 (Prof. G.J. Martin, University of Nantes) resulting in an average repeatability of 0,3 %o and an average reproducibility of 0,7 %o, can be deduced, that sample preparation errors are of no great influence.

The ring test results can be summarized as acceptable and therefore the method was recommende standarN CE s da d procedur C-1r efo 3 content determination fruin si t sugars [1].

The second and third ring test concerning the determination of the Oxygen 18 and Deuterium content in fruit juice water respectively where accomplished in 1992. attendine Th laboratorie7 g1 listee sar orang e Tabln di Th . e6 e juice appld -an e juice samples were supplied by Pernod-Ricard. isotope Th e standards SMO SLAPd Wan , obtained fro IAEAe mth , were applied. Their GISP material was used for additional intercomparison.

75 Tabl: e1 Participants of the inter-laboratory compariso theid nan r instrumentation

No Name *) Organization Location Country COi preparation Mass spectrometer 1 S. Brookes Europa Scientific Crewe UK On-line combustionC ,G Europa Tracer, separation direct inlet 2 H. Casablanca Centre nationa Réchercha l e ld e Scientifique Solaize France Microanalyser on-line Finnigan Delta S 3 ]. Fairchild/ Atomic Energy Authority/Campden Food dan Harwell Chipping UK Manual combustion , dua602 l inleG V t A. Robertson Drink Research Association Campden 4 A. FiUy Laboratoire d'Hydrologie et de Geochimie Orsay France Manual combustion VG 602 C, dual Isotopiqu l'Universite ed é Parid sSu inlet 5 P. Johnson Burea f Stabluo e isotope Analysis Brentford UK On-line combustion, SIRG V , duaAII l trapping box system inlet 6 , Kozie} t Pernod-Ricard Research Centre Creteil France Manual combustion Finnigan Delt, aS dual inlet 7 B. McGaw Rowett Res. Services Aberdeen UK McGa) 1 1 . ( al t we SIRG , duaA12 l Biomed MS 16(1988)269 inlet 8 N. Naulet Universit Nantee éd s Nantes France a) Microanalyser, Carlo Erba (on-line) b) Manual combustion 9 F. Pichlmayer Österreichisches Forschungszentrum Seibersdorf Austria On-line combustion • , Finniga251 T A nM GC separation dual inlet r Flichde n 1t0Va . J State University Groningen Netherlands Manual combustion VG SIR , duaA9 l inlet 11 F. Reniero Istituto Agrario Provinciale Michèln Sa e Italy On-line combustion, VG SIRA II, dual trapping box system inlet 12 A. Rossmann Technical University Munich Germany Manual combustion VG MM 903/602, dual inlet 13 A.L. Thelin Nestlé Lausanne Switzerland On-line combustion, Finnigan M AT 251, trappin systex gbo m dual inlet 14 C. Tisse Délégation Général Concurrencea l e ed a l e ,d Marseille France Microanalyser VG VG 602 E, Consommation et de la Répression des Fraudes ISOPREP 13 dual inlet 15 P. Trimbom Neuherberg/Institu r Hydrologitfü e Neuherberg Germany On-line combustion, Finnigan Delt, aS trappin systex gbo m dual inlet laboratoriee *)Th referencet no e sar d wit same hth e number Table. n i 5 s o sa t s2 13 13 Table 2: 5 C PDB %o results for Tabl : 5e3 C resultPDo B% r sfo orange juice sugar pineapple juice sugar

Laboratory Mean S.D. No. of Laboratory Mean S.Df .o . No replicates replicates

1 -24.73 0.05 10 1 -12.34 0.04 10 2 -24.17 0.15 3 2 -12.05 0.25 3 3 -24.46 0.12 12 3 -11.89 0.07 12 4 -24.91 0.15 9 4 -11.85 0.18 9 5 -24.46 0.06 20 5 -12.23 0.27 17 6 -24.64 0.11 3 6 -12.36 0.02 3 7 -24.43 0.07 12 7 -12.15 0.08 12 8 -24.67 0.04 8 8 -12.31 0.07 8 9 -24.42 0.02 6 9 -12.05 0.03 6 10 -24.56 0.03 4 10 -12.29 0.13 2 11 -24.99 0.20 6 11 -12.37 0.10 6 12 -24.51 0.01 3 12 -12.03 0.02 3 13 -24.64 0.05 4 13 -11.71 0.07 4 14 -24.57 0.07 5 14 -12.29 0.07 5 15 -25.16 0.11 3 15 -12.56 0.19 3

13 13 Table 4: 5 C PDB %o results for Tabl 6: e5 C PDB %o resultr sfo beet sugar cane sugar

>oratoiry Mean S.D.f o . No Laboratory Mean S.D. No. of replicates replicates

1 -25.52 0.03 5 1 -11.26 0.05 5 2 -24.56 0.11 3 2 -11.45 0.06 3 3 -25.62 0.06 6 3 -11.07 0.10 6 4 -26.05 0.13 3 4 -11.48 0.24 3 5 -25.40 0.05 4 5 -11.02 0.06 3 6 -25.73 0.06 3 6 -11.51 0.01 3 7 -25.68 0.07 8 7 -11.27 0.09 8 8 -25.52 0.03 3 8 -11.16 0.08 3 9 -25.64 0.03 3 9 -11.19 0.03 3 10 -25.72 -- 1 10 -11.34 — 1 11 -25.78 0.05 3 11 -11.10 0.09 3 12 -25.70 0.02 3 12 -11.37 0.02 3 13 -25.67 0.06 4 13 -10.76 0.22 4 14 -25.66 0.06 5 14 -11.18 0.05 5 15 -26.02 0.03 3 15 -11.30 0.06 3

77 Tabl Participant: e6 inter-laboratore th f so y compariso fruir nfo t juice water • oxyge determinatio8 n1 ) n(o - deuterium determinatio) n(d

Name*) Del. Organisation Location Country S. T. Brookes o Europe Scientific Crewe United Kingdom H. Casablanca o, d Service Central d'Analyse CNRS Vernison France J. Fairchild 0 Atomic Energy Authority Didcot United Kingdom A. Filly o, d University Lab. Hydrogeologie Orsay France l te s r HFo . o Institu Radioagronomir tfü e Midi Germany J. F. Goiffon o Laboratoire Interregional Montpellier France . HabfasK t d Finnigan MAT Bremen Germany P. Johnson o, d Burea Stablf uo e Isotopes Analysis Brendford United Kingdom . KozieJ t o, d Pernod-Ricard Research Centre Creteil France GJ. Martin o, d Université Reactivite N RM , e Nantes France Chimique E. Milne o, d Rowett Research Institute- MS , Aberdeen United Kingdom Group J. Morrison d VG Isotech Middlewich United Kingdom F. Reniero o Istituto Agrario Provinciale San Michèle Italy A. Rossmann o, d Technical University Munich Freising Germany A. Thelin o, d Nestle Ltd. Research Centre Lusanne Switzerland P. Trimborn o,d GSF-Institut für Hydrologie Neuherberg Germany r de n Va . J o, d University, Cent, for Isotop. Groningen Netherland Plicht Research partcipante *)Th listee sar d alphabeticall thin yi s Table numbere th , 8 Table n si d an s7 represent a different sequence.

Table 7: Table 8: 8 1 2 SMO 0 resulto W % GISr o sfo P water 8 H SMOW %o results for GISP water Laboratory Mean f S.Do . No . Laboratory Mean S.D. No. of replicates replicates

1 -24.55 0.23 2 1 -194.65 0.94 3 2 -24.73 0.08 3 2 -190.30 0.70 3 3 -24.94 0.10 3 3 -188.83 0.91 3 4 -24.62 0.00 2 4 -188.17 0.42 3 5 -24.62 0.09 3 5 -190.23 0.69 3 6 -24.72 0.06 3 6 7 -24.68 0.14 3 7 -192.64 1.48 3 8 -24.75 0.13 3 8 9 -24.69 0.01 3 9 -189.53 3.15 3 10 -24.72 0.01 2 10 -189.10 0.92 3 11 -24.86 0.02 3 11 -188.10 1.97 3 12 -24.61 0.01 3 12 -187.30 0.36 3 13 -24.76 0.02 3 14 -24.77 0.05 3 15 -24.72 0.11 3

78 FRUIT JUICE (15) 0ml

CENTRIFUGATION INSOLUBLE CONSTITUENTS \ t I Ca(OH) SUPERNATANT LIQUID 2 —— » (6g) ->90°C

CENTRIFUGATION PRECIPITATION OF ORGANIIC ACIDS H2SO4 SUPERNATANT LIQUID (2N)

REFRIGATION 4«Ch 12 , CaSO4

SOLUTION; containing FRUIT SUGAR

FREEZE DRYING

FRUIT SUGAR POWDER

MASS- SPECTROMETER

Figure 1: C0 - analysis of fruit juice sugars - sample preparation 13

79 sampls A e preparation method Oxygee th r Hydroged sfo nan n measurements CC>2- equilibration and water reduction, applying Uranium [2] or Zinc [3] respectively, were proposed. nowo t p , U onl result e IAEA-GISe yth th f so P materia releasee b n d ca l givedan Tablen ni s7 and 8. They agree with the mean value of ö^O = -24,79 ± 0,09 %o obtained from a previous IAEA ring test [4].

At present two collaborative studies on the determination of D/H ratio in sugar of juices by MS or by NMR are performed.

References

[1] J. Kozietet al. : Anal. Chim. Acta 271 (1993) 31-38 ] [2 l AFNOR(1992Ni 1 CEN/Tc ) Do WG , 4 C17 [3] AFNOR, Doc CEN/TC 174 WG l N17 (1992) [4] R. Gonfiantini: Report to the Director General, IAEA 1984

80 THE ISOTOPIC COMPOSITION OF THE GRONINGEN GS-1 GS-2D 9AN 0 PUR STANDARD2 ECO S H.AJ. MEIJER Centre for Isotope Research, Groningen University, Groningen, Netherlands

Abstract. This brief note describes the production and the isotopic composition of two pure CO2 isotopic standards, intended usedbe to primarily groupsthe by engaged high-precisionin isotope monitoring of atmospheric carbon dioxide.

Of all the application fields of isotope ratios, the measurements on atmospheric CÛ2 demand highese th t accuracy. Thi causefacs se i th t tha y expectede b tth d trendisotope th n si e ratios caused by anthropogenic emissions are very small (typically = -0.005%o per year for 13C), and should nevertheless be measured as accurately as possible and with a stable calibration basis over the years. Long-term observation records exist ([1],[2]), and at the present time more and more groups are setting up measurement sites of their own [3]. Since accuracy demands are so high, the intercomparison between the different groups should be of high quality too. Therefore the IAEA Hydrology Section has taken up the initiative to organise a world-wide, extensive intercomparison, dealing with all the steps from collecting the air up to the concentration and isotope ratio measurements [4]. Among the reference materials are also two pure CÛ2 isotopic standards, prepared by our laboratory. This short contribution describe productioe sth isotopie th d ncan compositio2 thesCÛ f no o etw standards. Logically, the isotope ratios of the CO2 standards should match those of air CÜ2 (being 013C around -8 %o, 518O around 0 %o against VPDB-CO2). Unfortunately, there is no CO2 commercially available that meets these specifications. Therefor decidee ew preparo dt ea mixture of two "natural" CO2's and one artificially enriched in 18O. Table 1 shows the isotope compositions of these components, as well as their percentages in GS-19 and GS-20. The enriched Rommenhöller has been prepared by letting 10 liters of natural Rommenhöller ( componen tabln i equilibrat) 1 te1 artificiallf o l em wit0 h10 y (17O, 18O) enriched water. This water has been enriched by adding 0.4151 g water highly enriched by thermodiffusion [6],

Table I The isotope composition of the three CÛ2 sources that GS-19 and GS-20 have been composed from. !36and 185are against VPDB-CÛ2, using the method described by Gonfiantini [5], (see recommendation elsewhere thisin volume).

CO2 source 135 185 GS-1%n i 9 GS-2%n i 0

1) Rommenhöller -2.93% -10.76% 81.65 77.77 natural CC>2 from a well in Germany 0 0 2) Beer-fermentation -28.90% -26.10%o 16.78 20.60 natural CC>2 from a brewery 0 3) Rommenhöller -16.45% +831.7% 1.56 1.63 artificially (18O,17O) enriched 0 0

81 with mass percentages : 16O 47.8 %, I7O 1.77% and 18O 50.4%, to 99.61 g destilled tap %c).1 - ~ wate Obviously0 18 ( r compositioe th , thif no s enriched water deviates very much from natural waters sens,e als th enrichmen O n eoi 18 thae th t mucs i t h higheO r 17 tha e nth enrichment. Theoretically equilibriue ,th m process would resul CC>n ti 2 having isotope ratios: 17 17 18 13 18 150%%„* 5 6 woul,O 87 c.* e d Enriche d-2.9b o whic = an n i , 5 0 2 O 3h% dCO enriched equivalently, such that :

17 17 Ä J 18 Rsample / RVPDB-CO2 ^ ( ^sample / RvPDB-CO2) would give 185 = 875%o ,17o ~ 370 %o, so the third component to GS-19 and GS-20 has a hug "anomaly"O e17 . Sinc algorithml eal froo measuree 18 m th d computinr sfo an o d13 e gth 45o and 460 assume some kind of relation between 17O and 18O, based on their terrestrial value0 ratio13 e thif s,th o s artificia wil2 computee b l CÛ l erroneoun a n di s e wayth n .I Gonfiantini algorithm this will leadeviatioa o dt -14.= valuo f 13 no 3e e th %(bein n oi g = 6.5% of the 17O anomaly). The measurements in table 1 confirm this: although the 13C / 12C ratio itself is not influenced by the equilibrium process, we find 13§ = -16.45 %o for the enriched Rommenholler, or equivalently a mis-measurement of-13.52%c. This is indeed what we expect from the (necessarily not very accurate) computation above. Of cours influencee eth GS-1n so GS-2 d 9an muc e 0ar h more limited, since only 1.56d %an Itä^o of component 3 is used in GS-19 and GS-20, respectively. Still, the 17O anomaly 13 cause deviatio0 sa GS-19 n -0.2f o n o 0 -0.2GS-20d n 1% o an 0 , 2% . The above does not implicate that GS-19 and GS-20 are not a good choice for intercomparison standards contrarye th n O . , althoug isotopie hth c compositio computes na e th y db Gonfiantini algorith 4(md > ofroan measuremento m^ 4 s doe resemblt sno true eth e isotopic composition of the gases, everyone measuring 4^6 and 466, and using the Gonfiantini algorithm same , th shoul eo t results t dge . Eve nabove morth , ee so example s stres face sth t that unificatio computatioe th f no essentials ni thad an t, deviations betwee resulte nth f so different groups, both with high quality preparation technique accurateld san y calibrated mass spectrometer neverthelesn sca unacceptable sb differenf o e yus highe tonle th o du ,yt algorithms! The paper by Alison et al (elsewhere in this Tecdoc) recommends the use of the Gonfiantini algorithm, and gives examples with which one can check the algorithm used (mostly an integral part of the mass spectrometer software).

Using the Gonfiantini algorithm, the CIO's best values for GS-19 and GS-20 are:

18 13 -0.19= o 3 % -7.50= o o 0 2GS-19% : 18 13 -0.99= o c 1% -8.62= o 0 2GS-20% : The absolute calibration error is estimated to be ± 0.025%o for 13o and ± 0.04%o for 185. Our calibration is based on the use of NBS19, NBS18 and two well-known local carbonate standards loca3 n wels o la , s pura l e CO2's ratie GS-2f Th .o o 0 relativ GS-1o et knows 9i n much more accurately: 13 18 GS-20 rel. to GS-19: o = (-1.128 ± 0.011) %0, o = (-0.798 ± 0 016) %o. These numbers are to a high extent calibration-insensitive. There is about 200 liters in stock of each of the gases, in stainless steel containers with stainless steel Swagelock valves, at ~ 45 Bar. Small (200 ml) stainless steel cylinders, at about 2 Bar, with stainless steel Swagelock valves1, are available for distribution. There is no indication of spread between the different cylinders.

1 Several bottles have been distributed already. Those had brass Swagelock valves.

82 References

[1] Roeloffzen, J.C., Mook, W.G., Keeling, C.D., "Trends and variations in stable carbon isotopes of atmospheric carbon dioxide (Proc. of an International Symposium on the Use f Stablo e Isotope Plann si t Nutrition, Soil Fertilit Environmentad yan l Studies, jointly organised by the IAEA and the FOA of the United Nations held in Vienna, 1 -5 October 1990), IAEA, Vienna, 601-618 (1991). [2] Francey, R.J., Goodman, H.S., Robbins, F.J.. "13C/12C and 18O/16O in baseline COz, Baseline Atmospheric Program (Australia) (ed. Wilson, S.R Grasd .an , J.L.), Depte th f o . Arts, Sport Environmente th , , Touris Territoried man CSIROd san , 1991 . 51-52,pp . ] "Isotop[3 e Variation Carbof so n Dioxid Othed ean r Trace GaseAtmosphere th n si e (ed. Rozanski, K.) t Researcs ,1 h Coordination Meeting, IAEA Sectio Isotopf no e Hydrology Report, Vienna, 1992 referencesd ,an therein. [4] Rozanski, K. Circular Letter in the framework of the IAEA coordinated Research Program "Isotope Variations of Carbon Dioxide and Other Trace Gases in the Atmosphere", 1993. [5] Gonfiantini, R. in: "Stable Isotope Hydrology", IAEA Technical Report 210, 1981, pp. 35-84. [6] Enritech Enrichment Technologies ltd., Weizmann Institute Campus, Rehovoth, Israel.

83 A REPORT ON THE 018O MEASUREMENTS OF THE IAEA CELLULOSE INTERCOMPARISON MATERIAL IAEA-C3

W.M. BUHAY, B.B. WOLFE, R.J. ELGOOD, T.W.D. EDWARDS Department of Earth Sciences and Waterloo Center for Groundwater Research, Universit f Waterlooyo , Waterloo, Ontario, Canada

Abstract. thisIn brief note reportwe results the ofb'8O analyses IAEAthe of cellulose intercomparison material IAEA-C3, performed by five laboratories.

e stablTh e isotopic compositio organif no c material from terrestria lacustrind an l e sources have provided valuable environmental information detailing various climatid can hydrologie changes that occurred durin paste gth . Investigators from aroun worle dth d employ different techniques to measure the oxygen, hydrogen and carbon isotopic composition n organii s c material from which they infer environmental information. However o procedurat e du , l intricacie techniquee th n i s s used, ther potentiaa s ei r fo l inherent variation e isotopith n i s c measurements from different laboratoriese Th . magnitud f theso e e variations theid an ,r subsequent effec n inferreo t d environmental

information casuaw subjece fe th a l s f discussiono t wa , s durin Internationan ga l Atomic Energy Agency (IAEA) meetin Viennan gi , Austria, during April 1993. 1 These discussions resulted in an agreement, among some investigators, to test the extent to which isotopic results are bias to technique and procedure while in the process establishing a cellulose standard for isotopic studies using organic materials. Here we report the initial 0 O 18 cellulose results from investigators using various techniques. materiae Th l used,cellulose intercomparison material IAEA-C. provides W . wa 3, Dr y db Stichler at the IAEA in Vienna. Aliquots of the sample were sent from the University of Waterlo o otheot r institutions equippe o measurt d e stablth e e isotope compositiof o n organic materials. They included ) GSF-Neuherberg1 : , Germany ) Universitä2 ; t Bern, Switzerland Weizmane Th ) 3 ; n Institute, Israel ) Cambridg4 ; e eTh University) 5 d an K U , Ridgk Oa e National Laboratory, USA. The measurement results are given in Table 1. At this time we reserve comment on the variability of the results pending further analysis and discussion at a future IAEA meeting dealing with isotopic technique environmentad san l change (possibl sprine th n yi g f 1995)meantime o th n I . hopes i t ei d that these preliminary results will inspire additional 018O, as well as 82H and 013C, measurements from the Cellulose IAEA-C3. We welcome participation from other institution t originallsno y involve thin di s collaborative study.

85 Table 1.

Investigators Affiliation Technique Measurements ave. 618O %o

W.M. Buhay, University of Nickel 6 n= 3 1.8 ±0.14 B.B. Wolfe, Waterloo Pyrolysis2 R.J. Elgood M. Saurer & Universität Nickel n = 7 32.2 ±0.11 . BorellS a Bern Pyrolysis3 . TrimborP n GSF Nickel 5 n= 31.3 ±0.19 Neuherberg Pyrolysis3 D. Yakir Weizmane Th n Mercuric n = 3 3 1.4 ± NA Institute Chloride4 n = 3 A N ± 7 1. 3 V.R. Switsur Cambridge Mercuric n = 4 32.7 ±0.13 University Chloride5

Note: Results from the Oak Ridge National Laboratory are not available presently.

References

1. IAEA Symposium on Isotope Techniques in the Study of Past and Current Environmental Changes Hydrospheree th Atmosphere,n e i th d an hel Viennan di , Austria, April 19-23, 1993.

2. Edwards, T.W.D, Buhay, W.M., Elgood, R.J. and Jiang, H.B. (1994). An improved nickel-tube pyrolysis metho oxyger dfo n isotope analysi organif so c matte waterd ran .

Chemical Geology, 113press)n (i , .

16 S.Thompson, P. and Gray, J. (1977). Determination of thO/Oe18 ratios in compounds containing C, H and O. Internationaljournal of Applied Radiation and Isotopes, 28, 411.

4. Dunbar, J. and Wilson, A.T. (1983). Re-evaluation of the HgCl2 pyrolysis technique for oxygen isotope analysis. International Journal of Applied Radiation and Isotopes, 34, 932.

5. Field, E.M., Switzur, V.R. and Waterhouse, J.S. (1994). An improved technique for the determination of oxygen stable isotope ratios in wood cellulose. Applied Radiation and Isotopes, 45, 177.

86 USE OF PLATINIZED MAGNESIUM AS REAGENT REPLACING ZINC IN HYDROGEN ISOTOPE ANALYSIS

S. H ALAS, B. JASINSKA Institute of Physics, Maria Curie-Sklodowska University, Lublin, Poland

Abstract

Platinized magnesiu s recentlha m y been reducinw proposene a gs a d agent for the conversion of small quantities of water to hydrogen in a flame- sealed borosilicate glass tube at 400°C for isotopic analysis. The reagent, Mg-Pt, in contrase prepareb n o zint ca tn everi dc y laborator y coatinb ya f o g magnesium granulate with a thin layer of platinum by reaction with

H2PtCl6*6H2O dissolved in acetone-ether mixture. Excellent reproducibility of the isotope ratios in hydrogen gas prepared from water samples has been obtained at proportion 4 \\L of water to 120mg of the reagent.

Introduction

Mass spectrometric analysis of hydrogen isotope ratio is recently performed in most laboratorie hydrogen so produces nga reductioy db f smalno l quantitie watef so r with zinc [1]. The reagent used in this method is AnalaR zinc shot supplied by Hopkin and Williams, or BDH Chemicals, so far. It is not clear why zinc from other suppliers, . Merce.g , produces wrong results owin appareno gt t incompletnes reactioe th f o s n [2,3j. Although the zinc method has become matured, Tanweer et al. [4] have found optimal conditions for water reaction by Zn (BDH Chemicals) to be 460°C for the conversion temperature, and 10 x the stoichiometric amount of zinc. These conditions slightly differ from those originagivee th n i l pape. r[1] meae Inth n timChineso etw e scientist Deqin Wand Ji s an u g Zhengxi] n[5 have publishe approacw ne da hydrogen hi n preparatio r isotopnfo e analysis. They applied Mg powder activated by a very thin layer of Pt on its surface. In this paper we describe this metho detailn di experiencr Ou . e confirme tremendouda s reactivite th f yo platized magnesium and thereby excellent reproducibility of isotopic composotion of hydrogen obtained with this reducing agent. It is the purpose of this report to inform the scientific community about use of platinized magnezium for hydrogen preparation, particularly because BDH Chemicals is no more able to supply zinc as suitable as it was previously [6]. Moreover, optimum conditions have been found which differ from those giveChinese th n i e pape. r[5]

87 Reagents

Two reagents are used: a comercial magnesium in form of granulate 0.5 to 2 mm and chloroplatinic acid, HjPtCl^oILjO, dissolved in waterless acetone-ether mixture whic uses h i magnesiu r dfo m coating with platinum black. Magnesium is highly base metal. It reacts with water much below 100°C, but a thin laye f magnesiuo r m oxides once formed slows down this reaction. With boiling

wate t yieldi r s rapidly magnesium hydroxide, Mg(OH) r thiFo s. reason magnesium 2 canno usee b ttotalr dfo decompositio watef no t 100°Cra . Although Mg(OH)2 coule db decomposed at about 550°C [7], however such a high temperature exclude use of borosilicate glass (pyrex) for reaction ampoules. Jin Deqiu and Wang Zhengxin [5] have found that Mg coated with platinum black (Mg-Pt) yields total decomposition of water at 400°C. Probably the Mg-Pt is so reactive that no formation of the hydroxides proceeds. Chloroplatinic acid is prepared by dissolving of platinum in aqua regia in a ceramic dish. For this purpose waste pieces of platinum (e.g. remainings of reaction vessels, piece f thermocouplo s e wire e useetc.b dn )ca afte r their purificatiod an n roastin furnaca n gi r flameo e . When platinu ms dissolvei d completly solutioe th , n ni heated continuously while small portions of concentrated HC1 and water are added alternatel remainine th o yt g syrup-like solution chloroplatinie Th . c acid being prepared is then cooled down to room temperature if a sample taken on a glass rod quickly solidifies during its cooling down. (The melting point of H^PtCl^oH^O is 60°C, while t 110°a C start o decompost s e yielding C12, FLX) d PtClan , 2). The e reagenth n s i t transferred to a tightly closed bottle made of a dark glass. The remainings are rinsed off fro ceramie mth c dis acetone-ethey hb r being mixe proportion di n 1:5. Both solvents should be distilled prior to use. To prepare 60 mL of the acetone-ether solution one should take 10 mL of

acetone, 50 mL of ether and lOOmg HPtQ*6HO. This quantity of the solution one 6 2

btains from 40m platinumf go bottle Th . e containin solutioe gth 2 HjPtCl^oH-f no n s pi tightly closed against loss of solvents and a contact with air humid (the reagent is hygroscopic).

Preparatio Mg-Pf no t

The granulate is sieved to select a fraction of 0.5 to 1.0 mm. If magnesium ribbon usede ar s , the t intn cu othe e smalyar l piece s5mmx (e.gm ).1m before their roasting in order to have more convenient material to do in weighting and filling of the ampoules. The magnesiue nth m granulat roastes ei d under vacuu t m550°a C during2 hours in order to desorb water and hydrogen. The vacuum roasted magnesium granulate is then coated by a thin layer of platinum in the following way: 28g of Mg granulate is dropped into a flask up to the reactioe th en f do n (gas bubbles disappear). The solutioe nth s poureni intf e dof oth original bottl further efo r use, wherea Mg-Pe sth dries ti d gently wit streae t hth ho f mo air until complete disappearance of any smell of the organic solvents. The reagent, Mg- Pt, prepare s storei dar a hermetid y thi n di n d ki an swa c vessel. uses Prioit e o ,th t r Mg-Pt is outgased under vacuum (10"2 mbar) at 400°C during one hour.

Preparatio hydrogef no n

The further preparation steps of a small water sample (4 to 10 \\L) to yield pure isotopie hydrogeth r fo cs analysinga zinidenticae e th sar c n i metho s a l ] wite d[1 hth exception of heating condition. The ampoule containing a water sample with the Mg-Pt reagent is heated to 400°C (instead 460°C in the case of zinc reagent) during one hour. Jin Deqiu and Wang Zhengxin used 200mg of Mg-Pt for the decomposition of 15 juL of water. We reduced these amounts about three times conserving the proportion of the reagen wateo t r [8] thin .I s stud quantite yth amounwatee f yo 4|ie s th t th Lrwa f bu o t reagent varried from 20 to 200mg in order to find optimal mass ratio.

Result discussiod san n

The isotope ratios, D/H, were determined on cycloidal mass spectrometer with dual inlet and dual collector systems [9]. The results obtained for a distilled water sample are shown in Table 1 and plotted in Fig.l as permil deviations from D/H of referenc j gas H e bes .Th t accurac d reproducibilit methoan w y ne s e i dth f o y somewhat abov recene eth t precisio mase th f sn o spectrometer (ler=0.2 t mas%a o) f so the reagent of 120 to 140mg.

Table 1. The results obtained for 4fiL water sample at different mass of the reagent.

Mass (mg) SD(%o) Comments 20 t determineno d lack of reaction 40 -183.0 extremaly little fraction reacted 60 -103.0 a major fraction reacted 80 -57.5 apparently complete reaction 100 -45.0 total decomposition of water 120 -42.2 total decomposition of water 140 -41.7 total decompositio watef no r 160 -32.3 total decompositio watef no r 200 -25.3 total decompositio watef no r

89 0.0 r-

-50.0

o-lOO.O

-150.0

-200.0 l i i i i i i i 50 100 150 200 MASS OF REAGENT [mg]

Fig.l. A plot of 3D values of H2 prepared from 4/jL of water versus mass of the reagent.

We do hope that due to better reactivity of platinised magnesium than that of zinc, Pt-Mg will be more effective in the case of analysis of brines and hydrous salts. Such material e extremelar s y cumbersom isotopn ei e analysi speciad an s l methods have been developed recently: Horita described a new approach based on Hj-water

equilibration with reusabl t catalyseP t [10], whil et [11 HoritGa ] d removeaan d Ca2+

+ and Mg fro2 m natural brine n fori s f unsolublmo e carbonates prio o azeotropit r c distillatio l residuaal f no l waters whic s thehwa n prepare e standarth y db d method. Recently Tanweer [12] develope a dprocedur r brine fo ehipersalind an s e aqueous solution without remova f alkalino l e earth metal cations whicn i , h mached amountf so zinc have to be taken per analysis. Acknowlegements Thanks are due to professor H. Roy Krouse and Dr. P Staniaszek, The University of Calgary, for the translation arrangement of the Chinese paper. This study was supported by the State Committee for Scientific Research, Warsaw.

90 References

] COLEMAN[I , M.L., SHEPHERD, T.J., DURHAM, J.J., ROUSE, I.E., MOORE, G.R., Reduction of water with zinc for hydrogen isotope analysis, Anal. Chem (19824 5 . ) 993-995. [2] FLORKOWSKI, T., Sample preparation for hydrogen isotope analysis by mass spectrometry, Int . ApplJ . . Radiâ (1985t2 Isot1 6 3 .) 991-992. ] KENDALL[3 , COPLENC. , , T.B., Multisample conversio f wateno o t r hydrogen by zinc for stable isotope determination, Anal. Chem. 57 (1985) 1437-1440. [4] TANWEER, A., HUT, G., BURGMAN, J.O., Optimal conditions for the reductio watef no hydrogeo t r ziny nmasr b cfo s spectrometric analysis of the deuterium content, Chem. Geol. (Isot. Geosci. Sec.). 73 (1988) 199-203. [5] DEQIU, J., ZHENGSIN, W., Determination of the Deuterium Content in Water with Mg-Pt Reduction Method, Acta Scientiorum Naturalium Universitatis Pekinensis, 1 (1985) 42-46. ] TANWEE[6 R ,A., Investigatio zinH r hydrogefo cBD f o n n isotope analysi masy sb s spectrometry . Naturforsch,Z (1993a 48 . ) 739-740. ] BROCKHOUS[7 ChemieC AB , , Leipzig, (1965) 830p. ] HALAS [8 d WOLACEWICZan . S , , PreparatioW. , f hydrogeno r fo n isotope analysis by means of magnesium, Proc. of 5th Working Meeting "Isotope Nature"n si , Leipzig, September (1989) 847-851. ] HALAS[9 , DuaS. , l Collector Cycloidal Mass Spectromete Precisior fo r n Analysi Hydrogef so n Isotopes, Int . ApplJ . . Radiât. Isot. 36(12) (1985) 957-960. [10] HORITA , StablJ. , e isotopes fractionation factor f wateo s hydraten i r d saline minera - brinl e systems, Earth Planet. Sei. Lett (19895 9 . ) 173- 179. [II] HORITA, J., GAT, J.R., Procedure for the hydrogen isotope analysis of water from concentrated brines, Chem. Geol. (Isot. Geosci. Sec.2 7 ) (1988) 85-88. [12] TANWEER , DeterminatioA. , f Deuteriuno mBrinen i Hypersalind san e Aqueous Solution Masy sb s Spectrometry Using Zin Reducins ca g Agent, Analyst (19938 11 , ) 835-838.

91 CARBON ISOTOPE (14C, 12C) MEASUREMENTS TO QUANTIFY SOURCES OF ATMOSPHERIC CARBON MONOXIDE IN URBAN AIR*

G.A. KLOUDA Atmospheric Chemistry Group, Surface Science and Microanalysis Science Division, National Institut f Standardeo Technologyd san , Gaithersburg, Maryland M.V. CONNOLLY Air Pollution Control Division, Albuquerque Environmental Health Department, Albuquerque Mexicw Ne , o United State f Americso a

ABSTRACT. Atmospheric air samples were collected during the Winter of 1989-90 in Albuquerque, NM USA, for carbon isotope (14C, 12C) analysis of carbon monoxide (CO). An experimental sample design was prepared to target periods when the concentration of CO exceed uL/9 e Lsth (volume fraction) hou8 , r National Ambien Qualitr Ai t y Standard (NAAQS) and during periods of attainment. Sampling sites, time of day, sampling duration, and meteorology were carefully considered so that source impacts be optimal. A balanced sampling factorial design was used to yield maximum information from the constraints imposed; the number of samples was limited by the number of sample canisters available, time, and resources. Carbon isotope measurement urbaf so n air, "dean-air" background from Niwot Ridge, Colorado, average (wood) logoxygenated-gasolined san s were use3-sourca n di e mode calculato t l e eth contributio woodburninf no totae th l o gt atmospheri burdeO cC Albuquerquen i . Results show that the estimated fractional contribution of residential wood combustion (0'RWC) ranged from concentrationO C 0.3 o f t o 00 s correcte r "clean-air"dfo background r thesFo .e same samples, the respective CO concentrations attributed to woodburning range from 0 to 0.90 umol/mol (mole fraction), well below the NAAQS. In all cases, fossil CO is the predominant source of ambient CO concentrations ranging from 0.96 to 6.34 umol/mol. A final comment is made on potentiae th f fossi o lmeasurement O C l indirecn a s sa t trace f atmospheriro c benzene, relevant o exposurt e risk estimate f motoo s r vehicle emission d occupationaan s l healt d safetan h y standards.

1. INTRODUCTION

1.1. Carbon isotopes (!4C, I2C) for tracing bio- and fossil-mass combustion sources

Radiocarbon, a cosmogenically produced radionuclide with a half-life of 5730 years, has the unique capability of quantifying the relative contribution of bio- and fossil-mass combustion sources degree Th . whico et h thes puro etw e source resolvee b n sca nominalls di y 2-orderf so

magnitude, constraine 14e C/th 12y Cdb rati f contemporaroo y carbo t -1.na 10"x 4 12 (modem

14 ratie th f fossi oo d 10"x carbo an )l 2 12 10"x carbo1. , 1 includinn= ~ t na chemicaga l process moder% blan-1 f ko n carbon powee Th .discriminateo t r , e.g., carbon monoxide (CO) from woodburnin froO C m s gv moto r vehicle emissions s furthei , r limitee uncertaintth y db f yo

* Contribution of the National Institute of Standards and Technology. Not subject to copyright.

93 knowing the 14C abundance (age) distribution of the sources. Estimates of the average age of 14 logs burne determinee dar integratiny db historicae gth l recor f atmospherido c CO2 ovee th r mean calendar year f growtso hrecor C [1,2]14 e obtaines di th ; d from direc measurementC 14 t s 14 of individual tree-rings [e.g., réf. 3] and of atmospheric CO2 [e.g., réf. 4]. For C tracer studies where woodburning and motor vehicle emissions are the predominant sources of carbonaceous gase r aerosolsso simpla , e 2-source usee modeb calculato dn t ca l fractionae eth l contribution from each source given a measure of total 14C of the chemical fraction of interest. For example, this approac s beehha n applie severao dt l studies designe quantifo dt y source f aerosolo s n si urban airsheds during winter months [e.g., réf. 2,5]. , Howeverasid CO e cas eth f o efron i , m woo d motoan d r vehicle emissions that contribut wintertimo et e concentrations, ther naturaa s ei l "clean-air" background componenf to witO C relativelha concentratiow ylo n (nominall nmol/mol)0 y5 abundanc C t wit14 bu ,ha e several times greater than that of modern carbon. This "hot" CO is a result of primary I4CO produce cosmiy db interactioy cra n wit thaN h14 t produce witC s14 h rapid oxidizatio [6]O C . o nt Its 14C/12C ratio depends on the extent that this cosmogenic CO mixes with stratospheric and tropospheric CO. This background CO component, although small, can be significant in urban air since its 14C activity is greater than that of modern carbon. Therefore, a model designed to estimat impace eth woodburninf to motod gan r vehicle emission tota n concentratioso O lC n must take into account the effect of a "clean-air" CO background.

1.2. Sources of wintertime carbon monoxide emissions - Albuquerque, NM USA

Over the past decade during the winter months, the City of Albuquerque, NM, (35.05 °Ns faileha , o complt ) d106.4 ; km elevatio y6 0°W 1. wit e U.Sf th o nh . Environmental Protection Agency's (EPA) National Ambient Air Quality Standard (NAAQS) for CO of 9 fiL/L over an 8-hour period. Results from emission inventories conducted in the past have suggested thadominane th t residentiae tar sourceO C f so l wood combustion (RWC motod )an r vehicle (MV) emissions [7J. These conclusions were further supporte wintea y db r aerosol study (1984- 85) of Albuquerque using fine-particle K and Pb concentrations as surrogate tracers to estimate contributionO C emissionsV M d an s ,C fro respectivelmRW y [8]resulte .Th s indicated thaV tM emissions were responsibl -75r ambiene efo th % f concentratioo O C t n accountewhilC eRW d for the remainder. Limitations to this indirect tracer technique are discussed by Currie et al. [9]. During this same period measurementC ,14 fractionO C f so s separated fro nighttimx msi e (1600- 0200) whol sampler eai s reveale dmediaa contributioC nRW havin % 30 f grangna o e fro% 1 m1 to 63% [10,11]. These results verified that RWC was a significant source of CO to the urban airshed and that further studies were necessary to consider source contributions during both day and night, as well as factors, e.g., meteorology, that influence the accumulation of CO at ground level. attempn a n I complo t y wit NAAQSe hth , several loca federald an l controls have been institute dt unti oveno pasle s years0 198rth 1 twa t I 8. thanumbee th t f annuaro l exceedances dropped significantly due in part to three likely factors: a cleaner vehicle fleet, the mandatory f o oxygenate e us d fuel - ethanos d methyan l l tertiarybutyl ether e (MTBE)th d an , implementatio f enforceo n d designated no-burn days e latteth , r basea meteorologica n o d l forecast 24-hour advancen si implementatioe lighn th I .f o t f no-burnno days secona , d 14CO study was funded by the City of Albuquerque, Environmental Health Department, to re-evaluate emissionV M d ambienan n so C impacte concentrationsth O RW C t f so . Constraine cosy db t and resources, a factorial (sampling) design including a few replicates and controls was constructe obtaio dt mose nth t information fro ambien0 m1 t samples three-factoA . r two-level full factorial design included the following: 1) sampling location, residential vs traffic sites, 2) time-of-day, daytime vs nighttime, and 3) forecast meteorology, dynamic vs stagnant air mass, o covet r effectively time-space additionao Tw . l samples (replicates) were collected under conditions that were more likely to favor high woodburning CO emissions. For model

94 validatio qualitd nan y control purposes ageC 14 , s were also determine followine th n do g source samples: 1) an ambient underground parking garage (UPG) sample collected during the early morning (0600-0900) on a weekday, 2) log cross-sections of pinion pine and juniper wood, and composito tw ) 3 e samples eac ethanolf ho MTBE-gasolined an - .

2. EXPERIMENTAL SECTION

2.1 Field sampling apparatus

Integrate r sampledai s were collectesteeL 2 l3 canistersn di in-linn A . e diaphragm (compressor) pump supplied filtered air to a flow controller set at -200 mL-min"1 to obtain -303 kPa (abs) in the canister over 8 hours. An MSA® ' particulate filter, having an efficiency of 99% for particles 0.3 urn diameter and larger, was used to remove particles from the air stream. Outputs from a calibrated flow meter and a CO concentration monitor were recorded continuously on a strip chart recorder while sampling.

2.2 Carbon (12C) composition

Carbon monoxide concentrations were determined by scrubbing the sample of CO2 and H2O with pre-columns of Ascarite® and Aquasorb®, respectively, followed by gas chromatography (GC) usinmoleculaÀ 5 ga r sieve column methanatioa , n syste convertinr mfo g CO to CH4, and a flame ionization detector. Concentrations are reported for dry conditions 2 (H2O and CO2 removed) in |imol/mol along with the standard uncertainty for v = 2-5 degrees of freedom (Table I). The calibration curves were based on responses from NIST CO and CH4 Standard Reference Materials (SRMs). Standards of CO and CH4 at the 10 umol/mol level (exact concentration listed below) were use determino dt i catalysefficience N eth e r th fo tf o y converting CO to CH4 for flame ionization detection; responses from CH4 SRMs at the 1 and 4 umol/mol levels were used to expand the calibration curve to include the entire sample concentration range. The following standards were used to obtain sample concentrations based on dry gas: SRM 2612a CO in air, 9.70 ±0.15 umol/mol, "primary" gravimetric CO standard X-13832 airn 9i , 16.2 0.01± 2 umol/mol air n i 1658 M 4 , 0.9SR ,aCH 80.0± 1 umol/molM SR , 3 airn 1659i 4 , 9.7aCH 90.0± 8 air n umol/mol1660i M 4 , 3.8SR a CH d 80.0± an , 4 umol/mol.

Certain commercial equipment, instruments, or materials are identified in this paper to specify adequatel! y the experimental procedure. Such identification does not imply recommendation or endorsemen Nationay b t l Institut f Standardo e Technologyd an s r doe no ,t impl i s y thae th t material r equipmenso t identifie necessarile dar bese yth t availabl purposee th r fo e .

uncertaintyAll 2 estimations Tables,textthe and in unless otherwise noted, combinedare standard uncertainties (uc, estimated standard deviation) obtained by combining individual standard uncertainties (u, standard errors) using law of propagation of standard deviations according to Ku, H.H. [12] with nomenclature defined by Taylor and Kuyatt [13]. Where appropriate, degrees of freedom (v) are reported.

3 All uncertainties for gas Standard Reference Materials are reported as 95% confidence intervals.

95 TABLE I. SAMPLE VOLUME, CO CONCENTRATION, RECOVERY, AND

RADIOCARBON (fM) RESULTS

Sample Vol. Proc. CO (M,V) Rec. Vol. (u ) Chem. Rec. f cor. c fM M [L.STP] [umol/molj [joL,STP] [%] 3 87 5.31 (0.06,2) ) (6 0 49 107 0.39 0.34 6 84 2.42 (0.02,5) ) (3 3 22 110 0.42 0.33 7 79 1.96 (0.04,3) 170 (2) 110 0.45 0.38 8 82 6.31 (0.03,2) ) (6 4 55 107 0.20 0.14 9 81 1.91 (0.01,2) ) (2 6 17 115 0.60 0.52 UPG 71 12.9 (0.10,3) (103 91 ) 100 0.16 0.16 10 83 3.60 (0.03,5) ) (4 6 32 109 0.38 0.30 11 85 3.82 (0.03,5) ) (4 3 34 107 0.36 0.32 12 77 1.37 (0.01,2) 120(1) 115 0.67 0.59 13 74 1.34(0.01,2) 113(1) 115 0.74 0.67 14 76 1.40(0.01,5) 122 (1) 115 0.59 0.50

Sample - sample identification, underground parking garage (UPG) sample Vol. Proc. - volume of air (wet) processed in Liters at STP CO - total carbon monoxide concentration (dry), standard uncertainty (u), number of degrees of freedom (v)

Rec. Vol. - volume of CO2 (uL) recovered and combined standard uncertainty (MC) from the CO fraction Chem. Rec. - recovery of the CO fraction in percent with uc ranging from 2 to 5%. See text for discussion positiva f o e bias inferred from recoveries >100%. Jj^ - unconnected fraction of modern carbon defined by the 14C/'3C ratio signal of the CO fraction (as Fe-C) relative to 0.95 times the 14C/'3C ratio signal of the NIST Oxalic Acid Radiocarbon Dating Standard Reference

Material (SRM )f ±0.0Ms o i 4990Bc u 1e baseTh . weighten do d la-Poisson countinS g statisticAM e th d san . C targeg u 5 littls t0.0= 6 a blan s M r ef 1a fo f ko fractioO C e th nf M f correcteo cor M f .- blanr dfo k CO2, i.e., exces inefficiencn a s o COt e 2 du ,f separatin o y g corM f f . o rangec u atmospherie e th s Th l fro . al m#2 0.0d trapce an 2(sample0.04th o 1 t stex# e n i , Se .t )CO 14 for details regardin correctioe gth dat C thir afo s f nexceso s CO2.

Separat C analyseG e s were performed usin e samth g e syste o obtaimt 2 CO n concentrations for quality assurance of the separation process. Similarly, H2O was removed diametem m lengt m 2 2 separatio 3. 6 prior CO hx 3. Poropa o a t r n columnno ® kQ . Carbon dioxide SRM d "primary"an s standard t ambiena s t levels were use o determint d e sample concentrations . 0 monthSample-1 r fo beed s C beforsha ° n analysisstoreC 0 eG -2 t da .

2.3 Separation of CO from whole air

The approach for chemical separation of the CO from whole air is based on the work of Klouda and coworkers [10,14]. A gas separation (GS) system has been designed to separate simultaneousl followine yth g chemical fraction 3 whols m fro 1 vapoemw 0. air lo ) r1 :pressur e volatile organic compounds (LVP-VOCs) H,O higand 2) , h vapor pressure VOCs (HVP-VOCs)

96 Hl/P-VOCs/CO, LVP-1/OCs/HO S.S. Tra 2 p# Tra1 p# Canister

Figure 1: Schematic diagra NISe s th Separatio f TmGa o n Facility. Lower half o f diagram is the manifold for separating carbonaceous gases from whole air, from right to left: 1) Trap #1 was designed to collect low vapor pressure (LVP) volatile organic compound t nominalla . s O (VOCs°C 2 H 8 d y-7 an ) For this study, trap #1 was operated at -196 °C to remove the CO2 fraction. s operate) i Tra2 2 p# t -19colleco da t C 6° t high vapor pressure (HVP) VOCs and traces of CO2 that pass through trap #1 as aerosol. 3) Trap #3 is operate t -19collecda o t 2 fro 6C oxidatioe ° mCO th t ove O C rf nSchützo e Reagent. 4) Trap #4 is operated at -196 °C to collect the CH4 fraction as CO2 following its oxidation over the high temperature catalyst, Pt on alumina at -830 °C. The upper half of the system, stainless steel vacuum lines, is designed for manometric determination of the recovered gases in calibrated volumes.

and CO,, 3) CO oxidized to CO2, and 4) CH4 oxidized to CO2. A schematic diagram of this system is illustrated in Figure 1 and described in detail elsewhere [14]. The following description of the system emphasizes the separation of the CO fraction from 0.1 m3 of air. The lower half of the GS system is the sample processing section made of glass tubing with primarily glass valves. This section contains flow controllers, traps for collecting condensible gase t cryogenia s c temperatures packea , d colum f Schützno e Reagent (I2On o 5

silic t rooaa gelm CO oxidiz o )temperaturet o t O eC packea d an , d aluminn columo t P f ano 2 pellets to oxidize CH4 to CO2 at -830 °C. A reference gas mixture or ambient air sample, previously analyzef o compositions inle) o ga 12 tw r t dr e fo port o th I f s(I o adaptes i ,e on o dt the system. For most experiments, inlet II, with a 0-5 L-min"1 flow controller, is used at a flow of 0.5 L-min"1. The 0-0.2 L-min ' flow controller is for processing He when evaluating the

97 blank or for on-line dilution of NIST high concentration (umol/mol) SRMs. For experiments reported here, trap #1, normally maintained at ca. -78 °C for isolating LVP-VOCs and H2O, is operated at -196 °C to condense the bulk of the CO2. Trap #2 is operated at -196 °C to collect HVP-VOCs tha2 traced tCO an pasf so s throug haerosolss traa 1 p# . operateTras i 3 p# t -19da 6 °C to recover CO2 after the selective oxidation of CO to CO2 over Schütze Reagent at room temperature. Finally, althoug t relevanhno thio t t s study ,s maintaine i tra 4 p# t -196°da Co t collect CO2 and H2O from the high temperature catalytic oxidation of CH4 to CO2 and H2O for isotope measurements of CH4. Sample fractions as CO2 are individually cryo-transferred to the appropriate sectio uppee th f nro manifol manometryr dfo combinee Th . d standard uncertainty of the recovered volume is 1.2% based on the volume determination assuming Ideal Gas behavior. This uncertainty considers the calibration of the pressure transducer, the containment volumetemperature th d an , e probe. Sample storee sar break-sealn di awaio st preparatioe th t n analysistargetS C 14 or AM f sfo . S systeG e s mevaluatei Th y processinb d g NIST SRMs, gravimetrically prepared mixtures, and reagent He under dry conditions with and without CO2 present, to estimate the recovery of the CO fraction and blank. As a measure of recovery, the volume measurement of each fraction is compared to the expected volume determined from the concentration of CO in the sample or standard and the volume of gas processed. For samples, the mass of air processed, obtained fro canistee mth r weight befor afted ean r processing ambiene th d r an ,ai t density are used to calculate the volume of air processed and is expressed in liters of moist air at standard conditions (STP), 101.3 kPa (abs) and 273.15 K (Table I). The moist-air density of each sample is determined from density tables given the average ambient temperature, barometric pressurepoinw de t d durinan , g sampling with subsequent conversio STPo nt e .Th combined standard uncertainty for the processed volume is 1%. Any loss of HO to the walls

of the canister while at laboratory temperature (-23 °C) should be insignificant2 . The combined efficiency for oxidation of CO to CO2 and cryogenic trapping of CO2 is 101.1 ± 0.4 % (u, v=5) based on the above mentioned controls. The equivalent blank (excess CO2) of the CO fraction is 2.3 ± 0.8 uL (u, \=13) at STP. The yield estimates, for the samples afte year5 2. r f storagso e prio processingo rt reportee ar , Tabln di ewitI h individual combined standard uncertainties ranging from 2-5% uncertaintie e baseth n do totae th lf so volum 2 CO f eo recovered concentrationO C e th , volume processedr th ai d f ean ,o relativelexcess e i 2 Th . sCO y controe smalth r fo ll experiments significans i t samplee bu , th r fo t s processed, averag± 0 2 f eo 3 uL (u, v=10) at STP. This average bias from CO2 cross-contamination amounts to 10 ± 1% (u, v=10) of the recovered volume for samples with concentrations ranging from 1.3 to 12.9 umol/mol. The total carbon recovered in the CO fraction for these samples ranged from 61 to 491 ug C with an average blank (bias) of 11 ± 2 ug C (u, v-10). The !4C results are corrected for this bias; details are described in Section 2.4. The sample recoveries reported in Table I are greater than 100% possibly because of frequent major swings temperature,in 200 to e.g.,°C 90 °C, at the heated zone of trap #2; designed to optimize CO2 collection by preventing the loss of CO2 as aerosol, but in fact contributed to the inefficiency in separating the CO2 fraction completely. By replacing trap #2 with a Russian Doll trap collection efficiencies for CO2 are expected to be consistently 100% [15].

2.4 Radiocarbon measurements

fractioO OncC e selectivels enth i quantified y an oxidize 2 rnanometryy CO db o dt s i t i , catalytically reduced to graphite with hot Zn and Fe wool and fused to an Fe-C solid solution for measuremen e 14C/th 13f o Ct ratio [16]. Accelerator mass spectrometry 14C/13C ratio measurements were performed at the NSF Facility for Radioisotope Analysis, University of Arizona, Tucson. Details of the measurement process have been reported by Linick et al [17].

98 The sample 14C/13C ratios were referenced to the measured 14C/13C ratio of the NIST Standard

Reference Material (SRM) 4990B Oxalic Acid [HOx(I)] for radiocarbon dating with a 13C/12C abundance (5C)13 of -19 %o and are reported as the fraction of modern carbon (f) according

Equatioo t belon1 w (Tabl: eI) M

w Msample CO) 1 . Eq f= —————M ——-————— 0.95 x C/ C[HOx;i)]

Modern carbon activity is defined as 0.95 times the activity of HOx(I) referenced to 13 5 C=-19. whico 0% approximatels hi y equivalen activite th o t tf 189 yo 0 wood. NormallyM f , is defined in terms of I4C/12C ratios corrected to reference 613C values. However, since SI3C measurements were not obtained for these samples, 14C/13C ratios of sample and HOx(I) as stated above closely approximate fM to within a few percent of values otherwise corrected for 13 0 C. Quality control of the isotope measurement process is best determined from fM calculated fro measuremene mth SRMf o t s 4990 4990d Ban C [HOx(II)] lattee ;th alss ri o oxalic acid with an activity abou greate% 34 t r4990B thaM certifiee nSR Th . (consensusM df ) valu r thesefo e two SRMs when calculating fM, i.e., HOX[II]/(0.95*HOX[I]), is 1.3588 ± 0.0005 («) based on a serie f o observations s fro3 independen1 m t laboratorie se resultTh [18]f o .s

HOJII]/(0.95*HOJI]) measurements for this study were within 2% of the consensus value for sample sizes ranging in mass from 70 ugC to 320 ngC. The dead C=0(14 ) control, used to represent the AMS target blank, is the NIST Reference Material (RM) 21 Graphite. Target

blank termn si f i.e.,, f so RM217(0.95 *HO[I]), range fro0.00o t mC 8 0.01ug 5 0.006 4± r fo 1 X

± 0.00M1 for 170 ugC. These measured standard and blank values have combined standard uncertainties based on weighted lo-Poisson counting statistics. fractionO C resulte M Sincf th e sf eth so contai nsignificana t contribution from cross- 14 12 contamination of atmospheric sample CO2, fM measurements are corrected given the C and C abundances of this excess CO2 according to the following equation:

- fM[ÖBL K' fM (BLK)] IM cor. = ————————————————— Eq. 2 (1 - ©BLK)

where e fractiona0th BL s Ki excesamoune n i l th 2 amoun f o sCO O t f C expecte o t e th r dfo fraction. The blank CO2 is assumed to have a fraction of modern carbon [fM(BLK)] resulting fro sourceso mtw fro) 1 :m background atmospheri 2 wit cconcentratiohCO a umol/mo5 35 f no ! 2 fromCO ) pollutio2 fd Md =1.1an an ] n5[4 sources exces n makini amoune 2 th f so CO p gu f o t background concentrationd an pollutioe C assumes i Th . 2 composee RW b n CO o % dt 50 f do 50% MV emissions with bounds for uncertainty based on the extreme values; the excess being entirely (uppefromRWC r boundexcesthe vs s) being entirely fro(lowemMV r bound). (Estimated ages [fM values] for these two sources are fM=0.10 for MV emissions and fM=1.06 for RWC emissions. Details for obtaining these fM values are reported later in the text and in Table III.) Therefore, estimated fM (BLK) values would range from 1.00 to 1.11 with lower and upper bounds equal to 0.87 and 1.14, respectively. The combined standard uncertainty for fM corrected (cor.) ranges from 0.02 to 0.04 which takes into account the lower and upper bounds

for fM(BLK).

99 3. SAMPLING DESIGN

3.1 Sample conditions

e samplinTh g conditions were designed o focut n wintertimo s e woodburning contributions to CO concentrations in Albuquerque, NM. The design was an attempt to satisfy neee th colleco dt t samples during conditions favorabl non-attainmenr efo NAAQe th f o tr Sfo CO, e.g., during the night (1630-0030) when the air is likely to be cold and stagnant at a residential site (Zuni Park [2ZE], intersection of Espanola, Mesilla and Prospect Sts.). In contrast, samples during attainment were important for comparison purposes, e.g., during the day

TABL . EFACTORIAII L DESIGN SAMPLING STRATEGY: TWO-LEVEL AND THREE FACTOR (23)

X, Sampling Period: AM (-) vs PM (+) Forecast Meteorology: Dynami Stagnans v ) c(- ) (+ t X, Sampling Site: San Mateo (-), traffic site (2ZK), vs Zuni Park (+), residential site (2ZE)

Factors (Variables)/Sampling Conditions

Design # Sample FACTOR

Xâ X2 X3 1 6 - 2 13 + 3 14 + 4 3 + + 5 9 + 6 7 + - + 7 11 + + 8 8 + + +

9 10 + 10 1 12 + + +

* Design #9 and #10 are replicates of #8.

Yates Algorithm:

ß+ ;2X $,;^ + 2Y ß 3 X Predicted$13+ + 23 0.5[^X, X$,+ 2 23+ 23 v ß^ X 123= + ]

where ß's are coefficients of variables (X's), ^ is the mean response, and Y is the predicted response. Variable coefficientd san subscriptee ar s d accordin maio gt n effects, single numerical identities interactiond ,an s representing some combinatio maif no n effectsthrer o o e variablestw , .

100 (0630-1430 likels more b wa o e)yt r whedynamiai e nth traffia t ca c site (San Mateo [2ZK], intersectio Maten MenauSa d f oan no l Blvds.). Eac shoppina hf o sit withis em i k g 1 malln~ , Winrock and Coronado, respectively. Given the three factors, i.e., sampling period, forecasted meteorology d samplinan , g sitee desigth , n allowe e factoo investigatt on s r u do y r an f i e combination of factors was a dominant influence on the CO concentration. The design conditions with reference to meteorology were influenced by the local "no-burn" control strategy residentialfor wood stoves fireplacesand when forecasted meteorology (24-h advance)in suggested cold stagnant conditions that were likely concentrationsaffectto CO extentthe to of exceeding the NAAQS. An optimized sampling ,leve o fultw l l factorial design (2 constructes 3),wa includo dt e complete coverage and balance of the above mentioned factors considered most important in affecting the C response surface (see Table II, hypothetical design #1-8). Each possible I4 combinatio three th ef n o critica l factors define uniquda e sampl collectede b o et . Since eth prime objective of the study was to focus on nighttime non-attainment of NAAQS, replicates #9 and #10 were chosen to match the conditions of #8. An additional ambient sample was collecte undergrounn a n di d parking garag controa s ea l expectes thawa t shoo dt we th tha l al t originateO C d from motor vehicle exhaust. e factoriaTh l design theory use o specift d e optimath y l sampling conditions when constrained to n=8 observations has been summarized by Box et al. [19] reference textbook on experimental design. Unfortunately, natural variability dictate e ordedth r that samples were collected even though a randomized sampling is suggested to avoid any correlation with time. Usin generalizee gth d additive model Yatee baseth n sdo algorith factoria3 2 e mth (Tablr l fo ) eII (balanced) experimental design and graphical means, results indicated that no significant main factor effect of those studied, i.e., time of day, forecasted meteorology or site, was detected. Thipartialls swa y natura e expecteth o t le variabilitddu concentratioO C f yo ©'d RWnan C thas twa evident from replicates (n=3sampline th f )o g condition most favorabl r higfo e h O'RWC., i.e., nighttime-stagnant-residential. Although this samplin gt abl no desigo detect e s y wa an nt significant factor effects, it allows for comparison of extreme conditions that is essential for developing future experiment possibld an s w whic yne leay moro dt hma e effective control strategies.

3.2 Source materials for quality control and model calculations

A few representative samples of gasoline were collected from area gas stations for 14C measurements. Albuquerqu thes enwa require ethanod ad o dMTBr t gasolino le th Eo t meeo et t a minimum 2% (by wt.) oxygen content for reducing CO emissions during the winter months. Data fro mmarkea t surve spoa d t yaudian t reveale distributioe dth thesf no e fuelsMTBE% 36 , - and 64% ethanol-gasoline [20], so that the average 14C signature for gasoline could be calculated from direct measurements of gasoline samples. Two composite samples were prepared from 14 f MTBEo eac t ethanol-gasolinehd se an - determinatione ag C r combustiod sfo an 2 . CO o nt In addition, samples of two representative logs were collected to estimate the typical age of logs burned in Albuquerque. Cross-sections of these wood (log) samples were taken and the 14 cellulos determinatione ag e pluC r s lignifo 2 burnes .nwa CO o dt

MODE. 4 L ESTIMATES FRACTIO- RESIDENTIAF NO L WOOD COMBUSTION (0'RWC)

Direc measurementC 14 t source th n eso material ambiend san t samples were usea n di 3-source model (Equatio calculato t ) n3 fractionae eth l contributio emissionC e RW th f no o st tota concentration.aO C l s follows:

fM(MEAS) = [0RWC- fM(RWQ] + [0MV- fM(MV)] + [BBKG- fM(BKG)] Eq. 3

101 where 0; is the fractional contribution of source "i" given by

0i = [COy[CO],oul Eq. 4

14 If fM(MEAS) is corrected for "clean-air" background using CO measurements by Tyler and Kloud concentratioO aC [21d an ] n measurement Novelly b s. [22 al t t Ridge] e ifro wo i mN , Colorado, then fM(MEAS) correcte backgrounO C r dfo d [fM'(MEAS) expressee b n ca ] : das

fM(MEAS) - 0BKG • fM(BKG) fM'(MEAS s )———————————————————— 5 . Eq — (1 - ©BKG)

A reduced 2-source model can then be expressed by the following equation

fM'(MEAS [0'= ) 6 RWC. Eq - fM(RWQ [0'+ ] MV- fM(MV)]

anc are now where 0'RWc ^ Ö'MV relative to background corrected CO concentration. Finally, substitutin - g0' RW1 0' MC= V into Equatio solvind an r ©'n6 gfo RW C yield e followinth s g expression:

[fM'(MEAS)/fM(MV)] - 1 0'RWC = ———————————————— Eq. 7 [fM(RWC)/fM(MV)1 - ]

The controls used to evaluate accuracy and precision of 14C measurements are NIST SRMs 4990B [HOx(I)] and 4990C [HOx(II)] for radiocarbon dating and Reference Material (RM) 21 Graphit blanC 14 r k efo (contamination ) determination.

. RESULT5 DISCUSSIOD SAN N

Tabl I summarizeeII s measure resultM df sourcf so e material published san d background concentrationO C se abov useth n ei d equation o calculatt s e fractionath e l contributiof o n residential wood combustion bese Th t. estimat average th uncertainte f e o valuth M ef d ean r yfo 14 typical logs burned in Albuquerque is equivalent to fM = 1.06 ± 0.02 supported by the C measurements of log samples and estimates derived from tree-ring models [1,2] assuming on averag yea0 logsd average e20 ol r Th . valuM ef f gasolineo e use Albuquerqun di e durine gth Winter of 1989-1990 is equivalent to fM = 0.10 ± 0.01. This value for gasoline is estimated from direc measurementC t14 MTBEf so ethanol-gasolind -an distributioe th d ean thesf no e fuels base markea n do t surve spod yan t audit during this period [20].

102 14 TABLE III. MEASURED C SOURCE SIGNATURES (fM): INPUT TO MODEL FOR ESTIMATES OF THE FRACTIONAL CONTRIBUTION OF RESIDENTIAL WOOD COMBUSTIO CONCENTRATIONO C O NT S

f . ResidentiaI l Wood Combustio (RWCnM f (RWC : 1.0= ) CO 60.02± )

- Pinion Pine*: fM = 1.033 ± 0.008 - Juniper (200 year old)*: fM = 1.031 ± 0.007 - Equal Mass and Equal Width Tree-Ring Models (200 year old logs): f1.0M= 80.0± 4 (median, v=7)

t Based on measured fM values of logs and Tree-Ring Model [1,2] estimates mentioned above. - The uc is based on weighted lo-Poisson counting statistics.

. MotoII r Vehicle (MVM f s 0.1 (MV= : ) 00.01± CO ) *

- Underground Parking Garage CO Fraction*: fM = 0.16 ± 0.02 [fM = 0.13 ± 0.02 when corrected for "clean-air" CO background (see text)] - :/TBE-gasoline* (2.0% oxyge wt)y 0.02nb = M :f 0.00 6± 1 - Ethanol-gasoline* (2.3% oxygen by wt.): fM = 0.148 ± 0.002

valueM f t e Basesth measuren do additive-gasoliner dfo s (abovee th d an ) distributio f fuelno s ethanol-gasoline soldMTBE% % 64 36 , d an - , fro mmarkea t surve spod yan t audit [20]. - The uc is based on weighted la-Poisson counting statistics.

III. Average "Clean-Air" CO Background: fM (BKG) = 3.1 ± 0.5 (uf

t Baseaveragn a n do e Northern Hemispheri (14O cCC , 12C) abundanc f 12.eo ± 7 1.7 (u, v=3) 14CO molecules-cm"3 (STP) determined from clean air at Niwot Ridge, Colorado. 14C measurements were made on just one day of the following month/year: 18 December 1990/9 January 1991 (composite), 27 February 1991, 6 December 1991, and 3 January [21]. The median 12C abundance from Niwot Ridge, correspondin timn gi e wit urbae hth n samples estimates i , t 0.14da 0 umol/mol with a range from 0.115 to 0.160 umol/mol [22]. The effect of this CO

backgroun sampln do rangeM ef s from 0.0 30.30o t .

103 Albuquerqu ConcentrationO eC s

14 (UPGl O 12-

"3 10_ NAAOS 0 S - 3- 8 e 6 .*•2* ~ •i CB ( i u « j A l ' — A 5 £ •» 1 ' W v . ' u •—o, ë ^^ 2 ™ *• ' \ ^ O ^^*" *» ^b 1 ^ *^P s^% * 0 •v ° [UPG] ^•.-•--O u \n _ «^ /a a • —a — • — B-^n 0- B M T "* NB i — ! 1 — 1 -2 —— —^ _ ——r ——— , ——— , ——— , ——— , ——— , ——— , ——— , ——— , ——— , — Ç-jzzzZzcQfflflocn à Q^^5^^££££ £ mOo — 01 — vgp»r»— • ^ r-t — — — — 'mOOOf** d Date 1989 1990 WoodbummQ gMotoO (RWC r O Vehicl)C e (MVO )C (soli dnight= , ope dayn= )

Figur : e2 Ambien concentrationO C t s durin (0630-1430y gda nighd )an t (1630-0030) time periods attributed to residential wood combustion and motor vehicle emission 0 ambien1 r fo st samples. Early morning (0600-0900O C ) concentrations fro undergrounn ma d parking garag designatee ear UPGy db . Days identified by NB are periods designated as "no-burn. " See Table IV for sampling conditions and estimates of the fractional contribution to residential wood combustion.

The ambient underground parking garage fresult of 0.16 ± 0.02 is fairly consistent M wit estimatee hth d average gasolin valuM ef e considering that "clean-air" background an O dC potentiall ysmala significant bu l t contribution from wood combustion would increas signae eth l expected from motor vehicles alone. bese Th t estimat backgrounr efo d correctio f "cleanno air" comeO C s fro (S4O mC,C i2C) measurements at Ni wot Ridge, Colorado [41.03 °N, 105.53 °W, elevation of 3.15 km]. An averag abundancC eI4 December-Februarr efo y months [n(samples)=4 measures ]wa 12.e b o d7t

±1. 14) C7(u O molecules-cm" mediaa [21 P d concentratioST ]an O t nC 3a f 0.14no 0 |amol/mol

with a range from 0.115 to 0.160 umol/mol [22]. The equivalent fvalue for thiCs 14 M

104 background is 3.Î ± 0.5 (M) assuming the median CO concentration noted above. The effect of this background correctio proportionas ni fractionae th o t l l amoun backgroune th f o t d relative to the total CO concentration for each sample. Therefore, the effect of this correction when

calculatin grange 0' s from 0.03 to 0.30. See Table III for additional information. C

FigureRW 2 is a time-ordered plot of the CO concentration attributed to residential wood combustion and CO concentration attributed to motor vehicle emissions for each sample measured determines wooe i Th O . dC d fro produce mth backgroune Q'f th to RWd Can d corrected concentrationO C concentratioe Th . attributeO C f no motoo dt r vehicle simpls si quantite yth y O-O'RWC) time backgroune sth d correcte concentrationO dC . Withou measurementC I4 t e th f so fractionO C s separated from these samples, quantitative source impact s seea s Figurn ni e2 would not be possible. The plot illustrates that the motor vehicle contribution is in all cases greater than the residential wood combustion contribution. Tabl summarizeV I e e samplinth s g condition r eacfo s h ambien r sampleai t e th , designated mandatory control ambiene r th no-bur, o , bur ) nt(B n(NB) temperatur e taken from

TABLE IV. SAMPLING CONDITIONS, CO CONCENTRATIONS AND ESTIMATED FRACTIONAL CONTRIBUTIO RESIDENTIAF NO L WOOD COMBUSTION (0'RWC)

Dsgn# Smpl Date ToD FMet Site Ctrl Temp CO (M, v) Ö'RW) C(«c 4 3 23 DEC N Stg Traf NB 0.2 (0.06,21 5.3 ) 0.17 (0.04) 1 6 10 JAN D Dyn Traf B 9.8 2.42 (0.02,5) 0.07 (0.07)

6 7 10 JAN N Dyn Res B 6.4 (0.04,36 1.9 ) 0.08 (0.09) 8 8 N 11JA N Stg Res NB 7.2 (0.03,21 6.3 ) -0.03 (0.03) 5 9 12 JAN D Dyn Res B 4.9 1.91 (0.01,2) (0.083 0.2 ) - UPGN JA 1 3 EM - Dwnt B 3.2 12.9 (0.10,3) 0.03 (0.02) 9 10E FE 6 0 N Stg Res NB 2.9 3.60 (0.03,5) 0.09 (0.05)

7 11 07 FEE D Stg Res B 5.0 3.82 (0.03,5) 0.12 (0.05) 10 12 07 FEE N Stg Res B 7.2 1.37 (0.01,2) 0.22 (0.11) 2 13 21 FEE N Dyn Traf B 2.8 1.34 (0.01,2) 0.30(0.11) 3 14 24 FEE D Stg Traf B 7.4 1.40(0.01,5) 0.12(0.11)

Dsgn # - expérimental design identification number (Table II) Smpl - sample identification, underground parking garage (UPG) Date - calendar date: all sampled in 1990 except for #3 sampled in 1989 : nigh ) 1630-0030tim- day (N tf 0630-1430) D eo (D To y da , , early mornin 0600-090) g(EM 0 FMe - forecastet d meteorology: stagnanr dynamio ) ) (Stg tc(Dyn Sit e- residentia l (Zuni , traffiParkMateo n , downtow(Sa 2ZE) c, 2ZK) , n (Dwnt) Ctr - mandatorl y control r bur) o ) n(B buro n :n (NB Temp - average ambient temperature in degrees Celsius CO - total carbon monoxide concentration (dry) and standard uncertainty (u) in |jumol/mol, number of degrees of freedom (v) Q'Bw - estimater d fractionae valueth r fo s l contributio f residentiano l wood combustion correcte averagn a r dfo e "clean-air" CO background. See text for discussion of background correction. The uc is based on error propagation f variableo ] [12 s specifie Equationn di f texto 7 . 1- s

105 NOAA 3-hour averages and averaged over the sampling period, the total integrated CO concentration (dry) and the fractional contribution of residential wood combustion )(©'

calculated fro correspondine mth g measured sampl othee f valuth d r valueean modee th O si l RWC

describes a Tabln di . eHI M

5,1 Relationship of fossil CO and benzene - inferences relevant to potential changes to the US Occupational Health and Safety Standard

Aside from obtaining quantitative CO source information from this study, it also gave e opportunitth s u o investigatt y e contentioth e n thaCt I4 O measurement e usefub y l ma s information in apportioning sources of atmospheric benzene [23]. It is well known that benzene emission isa f enginno e exhaus unburned an t d gasoline Americae studA th . y yb n Petroleum Institute on tail-pipe emissions has shown that the emission factor of benzene relative to CO for auto exhaust is -0.0033 for late model (1980's) automobiles [24]. In addition, studies have shown that benzene is a product of wood burning having an emission factor for wood stoves of -0.002 [25]. Estimates of benzene emissions for 1988 in the U.S. indicate that industrial sourceresponsible b y sma onlr efo y -15 totae th % lf benzeno e emitted annually while biomass combustion, - 50%, and auto emissions, -35%, seem to account for the bulk of the benzene [24]. Given that the levels of benzene emissions from wood burning may be comparable to that of auto gasoline, total hydrocarbon speciatio same th carries en o canisten wa t dou r samples prior separatioO C o t determino nt benzene eth e concentrations.

1.0 Current OSHA: 1 n,mol/mol C6H6 Parking Garage o al I Proposed ACGIH. o 0.1|J.molAnolC6H6

0.01 e u Ambient PQ Samples NAAQS umol/mol CO

0.001 0.1 1 10 100 Fossil CO (jimol/mol)

Figure 3: Scatter plot of wintertime benzene and fossil CO for Albuquerque, NM for the same samples plotted in Figure 2. Ambient results are shown with current and proposed CO standards: U.S. National Ambient Air Quality Standard, NAAQS (defined as volume fraction, here represented as mole fraction), benzene Occupational Safet Healtd an y h Administration, OSHA, Americae th d an n Conferenc f Governmentaeo l Industrial Hygiene, ACGIH. lowea f I r benzene standarenforcede b o t s di , direc datinC 14 t benzenr gfo e may be required to help discriminate between ambient and industrial contributions.

106 A scatter plot of the Albuquerque benzene vs fossil CO concentrations in log-log space shows i parkine Figurn ni Th . e3 g garage sampl includes i ) e(« illustrato dt poinea t regarding e Occupationath l Safet d Healtan y h Administration (OSHA) Standar r benzenfo d 1 t a e umol/mol. If the OSHA standard is reduced to 0.! umol/mol, as suggested by some organizations, background benzene concentrations from sources other than industrialC , e.g.,RW and MV emissions as reported here, may also require controls for compliance with the standard. Ambient level f benzeneo s s showa , e significannb herey d variablma , an t e enougo t h occasionally cause industrial levels to exceed a proposed 0.1 umol/mol OSHA standard. Therefore e existinth f i , g OSHA standar r "benzenedfo exposure"e reducedb o t s i , isotope measurements made directly on benzene separated from whole air samples may be the only way to identify source contributions from the ambient urban air vs industrial environments.

6. CONCLUSIONS

From 14C results of 10 ambient samples, one ambient underground parking garage sample, and source signatures for logs, gasoline, and "clean-air" CO background (I4C, I2C) used in the 3-source model, estimations of the fraction of the total CO concentration due to residential wood combustion (O'RWC) ranged from 0 to 0.30. For these same samples, the respective CO concentrations attribute o woodburnint d g range 0.9o frot m0 0 umol/mol, well beloe wth NAAQS. In all cases, fossil CO is the predominant source of ambient CO concentrations ranging from 0.96 to 6.34 umol/mol. See Table IV for sample characteristics and 0'RWC results. modee Th evaluates i l compariny db ambiene gth t underground parking garage control, wher fractioe eth f residentiao n l wood combustio estimates nwa 0.0e b 3o dt 0.02± , witn ha expected value of nearly zero. The assumption that this sample is dominated by motor vehicle exhaust is reasonable given the sampling site, the high CO concentration [12.9 ±0.1 («) umol/mol)], the dominant motor vehicle source, and the CO concentration pattern consistent with time-space activitie f o motos r vehicle a downtow n i s n metropolitane areaTh . measured/modeled resultant for this sample is statistically no different than zero. This comparison also gives validity to the use of Niwot CO data for the background correction. However, futureany studies should include examples of regional background measurementsCO (chemical isotopic)and variabilityand morefor precise measures fraction ofthe residentialof wood combustion. e "clean-air"Th backgrounO C d correctio singlee th s nmostwa significant correction applied to the 14C results with an effect on the fraction of residential wood combustion ranging from 0.03 for the sample with the highest CO concentration to 0.30 for the sample wit lowese hth concentrationO C t . Therefore samplee th r fo ,s that approac NAAQe hth S , thesfoCO r e results will hav smallese eth t correction applied. ambien0 1 e r threth Fo tf eo sample s that were collected during mandatory no-bum periods, which exhibited the higher CO concentrations, estimated values for the fraction of residential wood combustion range particulae d0.17o frot On .0 m0. r sampl interestins ewa g for it was collected two nights before Christmas (23 Dec 1989) during a mandatory no-burn condition. This sample exhibited the coldest period (0.2 °C), showed one of the highest measure concentrationO dC s (5.31 estimateumol/mol)n a d ha d d an fractio, f residentiano l wood combustion equal to 0.17 ± 0.04. The low fraction of woodburning for this sample indicates thatno-burnthe control strategy appearseffectivebe to (see Figure2). Conversely, the fraction of CO from motor vehicles for this particular sample was 0.83 ± 0.04quasi-bimodaA . l patterconcentrationO C f no thir sfo s evening (sampl ) showee#3 d an early evening peak fro . 1700-200mca 0 followe vallea y db y centere t arounda d 2030 that was followed by another increase from ca. 2100-0030; both peaks being broad with frequent spikes of relatively low magnitude riding on top. This pattern may represent the

107 evening shopping pattern due to numerous cold starts that would be associated with shopping mall activities. This event seems to be atypical of the set studied. Meteorological factors, 3-hour averages, combined with the 12CO patterns for these sample y givma se valuable insight regarding what condition y favo ma sa potentiar l exceedanc thin ei s particular area [26].

? ACKNOWLEDGEMENTS

e authorTh s would lik thano et k D.J, Donahue, A.J.T. lull, D.B. Klinedinst, L.J. Toolin, A.L. Hatheway, T. Lange and D.L. Biddulph for their assistance in the measurement of 14C by AMS. Also, the authors wish to thank G. Mattingly and P. Baumgarten of the Fluid Flow Grou t NISpa their Tfo r assistanc calibratinn ei g flow controllers.

REFERENCES ] [1 CURRffi, L.A., KLOUDA, G.A., GERLACH, R.W., "Radiocarbon: nature's tracer for carbonaceous pollutants," Residential Solid Fuels: Environmental Impactd san Solutions (COOPER, J.A., MALEK, D., Eds.), Oregon Graduate Center Publ., Beaverton (1981A US , ) 365-385OR , . [2] KLOUDA, G.A., BARRACLOUGH, D., CURRIE, L.A., ZWEIDINGER, R.B., LEWIS , STEVENSW. C , , R.K., "Source apportionmen f wintertimo t e organic aerosol Boisechemican sy i b D I isotopi,d an l c (14C) methods," proceedinge th f so 84th Air and Waste Management Association Annual Meeting and Exhibition, June 16-21, Vancouver, BC, Human Health and Environmental Effects ISA (1991) 91- 131.2. [3] OLSSON, LU., POSSNERT, G., "14C activity in different sections and chemical fractions of oak tree rings, AD 1938-1981," Radiocarbon 34 3 (1992) 757-767. [4] LEVIN, I., KROMER, B., SCHOCK-FISCHER, H., BRUNS, M., MUNNICH, M., BERDAU, D., VOGEL, J.C., MUNNICH, K.O., "25 years of tropospheric 14C observation centran si l Europe," Radiocarbo (19851 7 n2 ) 1-19. ] [5 SHEFFIELD, A.E., CURRIE, L.A., KLOUDA, G.A., DONAHUE, D.J., LINICK, T.W., JULL, A.J.T., "Accelerator mass spectrometric determination of carbon-14 in the low-polarity organic fractio f atmospherio n c particles," Anal. Chem (19902 6 . ) 2098-2102. [6] VOLTZ, A., EHHALT, D.H., "Seasonal and latitudinal variation of 14CO and the tropospheric concentration of OH radicals," Jour of Geophy Res86C6 (1981) 5163- 5171. ] [7 CIT ALBUQUERQUF YO E ENVIRONMENTAL HEALTH DEPARTMENTR ,AI POLLUTION CONTROL DIVISION, Albuquerque/Bernalillo County Emissions Inventory from 1982 (revised 1984 and 1990). ] [8 EINFELD , IVEYW. , , M.D., HOMANN, P.S.,"The Albuquerque carbon monoxide source apportionment study," (1988) Sandia National Laboratories Report No. Sand88-0792/13. [9] CURRIE, L.A., SHEFFIELD, A.E., RffiDERER, G.E., GORDON, G.E., "Improved atmospheric understanding through exploratory data analysi complementard an s y modeling: The urban K-Pb-C System," Atm. Environ. (1993), in press. [10] KLOUDA, G.A., CURRIE, L.A., DONAHUE, D.J., JULL, A.J.T., NAYLOR, M.H., !4 14 "Urban atmospheric CO and CH4 measurements by accelerator mass spectrometry," Radiocarbo (1986A 2 8 n2 ) 625-633. 108 [11] KLOUDA, G.A., CURRIE, L.A., VERKOUTEREN, R.M., EINFELD , B.DW. , . ZAK, "Advances in microradiocarbon dating and the direct tracing of environmental carbon," Journal of Radioanalytical and Nuclear Chemistry, Articles, 123 1 (1988) 191-197. , H.H. propagatiof o KU [12e ,us ] "Notee th errof nn o so r formulas," Journa f Researco l h of the National Bureau of Standards - C. Engineering and Instrumentation, 70C 4 (1966) Oct-Dec. [13] TAYLOR, B.N., KUYATT, C.E., "Guideline r evaluatinfo s expressind gan e gth uncertainty of NIST measurement results," NIST Technical Note 1297 (1993). [14] KLOUDA, G.A., NORRIS, J.E., CURRIE, L.A., RHODERICK, G.G., SAMS, R.L., DORKO, W.D., LEWIS, C.W., LONNEMAN, W.A., SEILA, R.L., STEVENS, R.K., "A method for separating volatile organic carbon from 0.1 m3 of air to identify source f ozono s e precursor isotopa vi s e (14C) measurements," proceedinge th f o s 1993 U.S. EPA/A&WMA International Symposium Measurement of Toxic and Related r Pollutants,Ai Wast& r eAi Management Association (1993)y Ma , . [15] BRENNINKMEIJER, C.A.M., "Robust, high-efficiency, high-capacity cryogenic trap," Anal. Chem. 63 (1991) 1182-1183. [16] VERKOUTEREN, R.M., KLOUDA, G.A., CURRIE, L.A., DONAHUE, D.J., JULL, A.J.T., LINICK, T.W., "Preparatio f micrograo n m sample n iroo s n woor fo l radiocarbon analysi acceleratoa svi r mass spectrometry closed-systea : m approach," Nucl. Inst. and Meth., B29 (1987) 41-44. [17] LINICK. T.W., JULL, A.J.T., TOOLIN, L.J., DONAHUE, D.J., "Operation of the NSF-Arizona accelerator facilit radioisotopr yfo e analysi resultd san s from selected collaborative research projects," Radiocarbon, 28 2A (1986) 522-533. [18] DONAHUE, D.J., LINICK, T.W., JULL, A.J.T., "Isotope-ratio and background corrections for accelerator mass spectrometry radiocarbon measurements," Radiocarbo (19902 2 n3 ) 135-142. [19] BOX, G.E.P., HUNTER, W.G., HUNTER, J.S., "Statistic Experimenterr sfo n A s- Introductio Designo nt , Data Analysis Moded an , l Building," John Wile Sony& s Publ., New York (1978). [20] STAPP, S., Albuquerque Environmental Health Department, Albuquerque, N.M., personal communication, (1993). [21] TYLER, S.C., KLOUDA, G.A., unpublished data (1993). [22] NOVELLI, P.C., STEELE , TANSP. , , P.P., "Mixing ratio f carboso n monoxidn ei the troposphere," Journal of Geophysical Research 97 D18 (1992) 20,731-20,750. [23] OLLISON, W., American Petroleum Institute, personal communication (1992). [24] AMERICAN PETROLEUM INSTITUTE, Exhaust benzene emissions from late model vehicles, No. 841-44700, Washington, D.C. (1988). - [25] KLEINDIENST, T.E., SHEPSON, P.B., EDNEY, E.O., CLAXTON, L.D., CUPITT, L.T., "Wood smoke: measuremen e rnutagenith f o t c activitie s gasit d f -an o s particulate-phase photoxidation products," Environ. Sei. Technol. 20 (1986) 493. [26] KLOUDA, G.A., CONNOLLY, M.V., "Radiocarbon measurement f wintertimo s e atmospheric carbon monoxide in Albuquerque, NM: contributions of residential wood combustion," to be submitted to the Journal of Geophysical Research (1994).

109 CARBOE TH N DIOXIDE ISOTOPIC MEASUREMENT PROCESS: PROGRES NIST MEASUREMENTSSA N TO , REDUCTION ALGORITHMS AND STANDARDS

R.M. VERKOUTEREN, G.A. KLOUDA, L.A. CURRIE Atmospheric Chemistry Group, Surface and Microanalysis Science Division, National Institut f Standardeo Technologyd san , Gaithersburg, Maryland, United States of America

ABSTRACT repore W : t progres evaluatior ou n s o developmen d nan carboe th f o tn dioxide isotope measurement process, with special emphasis on highly reproducible measurements essentia globar lfo l atmospheric programs addressing climate chang humad ean n health. Three specific aspects of the measurement process are treated: 1) The propagation of uncertainty

through algorithms, and assumptions used to convert conventional measurements to <513C values; we apply Monte Carlo methods for the purpose of illustratingC 5 13 distributions from 47 several model algorithms ) Utilizin2 . g non-conventional 5 CO2 measurement o removt s e reliance on certain limiting assumptions, we describe the data reduction algorithm for calculation of S13C. Monte Carlo methods are used to define measurement reproducibiiity requirements for this method, and special procedures are used to obtain highly repeatable measurements e describW ) 3 .ea metho r productiodfo f isotopino s standardcga s having reproducibiiity essential for global programs. Feedback is invited concerning the standards and reference materials neede adequatr dfo e modeling, calibratio qualitd nan y assuranc specifir efo c

applications.

13 1. Uncertainty C0n i fro1 m Natural Variation Oxygee th n si n Isotope Relationship

Moder isotops nga e ratio mass spectrometers have improve thao ds t single-instrument measurement repeatabilit easily can betteybe r tha approacn 0.0can 1 and %oC 5h,13 0.00for 2 %o 2 under certain conditions. Expert interlaboratory measurement reproducibiiity of pure CO2 reference materials, however, has been observed as 0.05-0.06%o [2], although preparation reproducibiiity for these reference materials was 0.02-0.03%o [3]; the source of this discrepancy has been unclear. Inconsistencies from the effects of uncontrolled isotopic fractionatio suspectede nar t anothe,bu presenre b effec y ma tt (natural variation oxygee th n si n

isotope relationship) describes a , followinn di g sections. This effec accoun n pooe ca tth r fo t reproducibiiity observed in CS 13 measurements, and can also be largely avoided by following the procedures recommended in these proceedings (Annex 1) for t.he calibration and use of working standards.

^Uncertaintie measurementn si expressee sar accordancn di e with recommendatione th f so International Organization for Standardization [1].

2)e "pe513th Cs ri mill" (%o) relative differenc 13n i eC/ 12C ratios (defined her 13s ea a R f )o

1S 17 16 sample from referencthaa f o t e standard definee ar O5d ; O5an d17 analogously usinO/gO (17R) and 18O/16O (1SR) ratios, respectively.

Ill 1.1 Conventional Isotopic CO2 Measurement

Isotopic composition typicalle sar y determine differentiay db l measuremen bean io mf to ratios between a sample and reference standard using an isotope ratio mass spectrometer with multipl collectorsn eio r COFo 2. , molecular isotopic composition measuree sar expressed dan d m using the delta (5) notation, e.g. Eq.l, where R values are ratios of CO2 ions of m/z m (m = 45 or 46) versus m/z 44 ions (the CO2 base peak) for a sample and standard. The numerical 45 46 13 18 methods used to convert measured S CO2 and S CO2 values to basic 5 C and 5 O compositions were developed by Craig [4] and enhanced subsequently [5,6,7,8]. In 1981,

D IB _ _ p cn m Ksam Kstd 5 C02 1000 ED Kstd

the International Atomic Energy Agency (IAEA) recommended the general use of an explicit reduction algorith precisr mfo e intercomparabilit measurementf yo three s solv o [9]th T r e. efo 13 18 17 45 46 unknowns S C, (implicitly5d Oan ) 5 O from onlmeasuremento ytw s (5 CO Sd 2COan 2), this algorithm assumes an exact relationship among oxygen isotopes between any sample and working standard, Eq.2, where the fractionation constant a = 0.5.

17t 18In sam IVsam (2) 17 18 R.std R•std

Developments have prompte dreassessmena validite th IAEAutilite f d o tth yan f yo - 1981 algorithm valuee Th absolut.r sfo e ratios (13R, 17R, 18 referencRn i ) e standards have been redetermined through new measurements [10,11,12] and recalculated using new assumptions [13]. These studies suggest thaconventionalle th t y accepted value absolutr sfo e ratio referencn si e standard biasee b , y translatin mucs da n sma 8% i s o ha 5% biaseo gt o t p su S13C. Santroc . [10al ]t ke note d thaoxygee th t n isotope fractionation assumptio5 0. = a f no was probably inaccurate for two reasons: first, based on theoretical aspects [14] and empirical evidence [15,16], the fractionation coefficient a varies slightly among terrestrial chemical systems so that no fixed value of a could be generally applied in a reduction algorithm. Secondly, unless the true value of a was known for a particular differential measurement, a = 0.516 shoulassumee th e db d value sinc representt ei average sth e fractionation observed among 35 terrestrial silicate minerals and natural waters [15]. Later, Robert et al. [17] measured a = 0.512-0.525 in terrestrial cherts, basalts, and industrial oxygen gas (vs Vienna Standard Mean Ocean Water - VSMOW), and Thiemens et al. [18] reported a = 0.517 in tropospheric CO2 and veiy unusual values (a = 0.6-0.7) in stratospheric CO2. Mass-independent isotopic fractionations may play an important role in many natural physicochemical subsystems [19]. addreso T s concerns tha IAEA-198e th t 1 invali n algoritha e b dy approacmma dato ht a reduction, we have compiled several alternative approaches for the calculation of S13C and 518O values. An all-important perspective is the uncertainty of these calculated values, so we propagat uncertaintiee eth s from measurement repeatabilitie naturad san l e variatioTh . a n ni resulting distributions in 513C and 618O depict the true differences and limitations of conventional algorithms, and clarify several vital areas for development and implementation.

2 Mont1. e Carlo Error Propagation

e MontTh e Carlo e propagatiomethoth r fo d f uncertainto n y through complex mathematical expressions offers advantages of conceptual simplicity, choice of input error

112 distributions, accurate response to covariance, and the facility to inspect visually (and statistically resultine )th g distribution outpuf so t values [20]. The data reduction approach of Santrock et al. [10] was used to calculate, through the Monte Carlo method, distributions of values for 513C and 518O from random normal 45 46 distributions of 5 CO2, 5 CO2, a^ (sample vs working standard), astd (working standard vs 17 VSMOW) thred an , e discrete valuef sRo VsMow (Eqs.7-9). This approach required thae th t 13R, 17R and 18R values in the working standard be defined versus a reference standard.3 The input parameters to five Monte Carlo simulations are given in Table I. The effects of isobaric interferences, isotopic fractionation, and instrumental effects, which may also influence actual 6 SI3d 18COan distribution practicen si , hav beet eno n treated here. These effect expectee ar s d negligible tob e under controlled conditions.

TABLEI SIMULATION INPUT PARAMETERS

45 46 S C0(%o)u ± 2 5 CO(%o)u ± 2 «sa" m± ustd ± « Figure vs workind gst vs workind gst -30.000 ± 0.000 -30.000 ± 0.000 0.51 0.02± 6 0 0.516 ± 0.020 1 -30.000 ± 0.010 -30.00 0.030± 0 2 0.516 ± 0.000 3 0.51 0.00± 6 0 0.516 ± 0.020 4 0.00 0.010± 0 0.000 ± 0.030 5

Standard uncertaintie value0 f o hypotheticae ) sar s(w l analytical repeatabilities. Standard uncertainty (u) of a values is from [15], where uncertainty of mean is reported as 0.5164 ± 0.0033 (standard error, n=35).

13 Working Standard is defined as: R0.011259= 1 (3) 18 R0.002083= 8 (4) I7p _ I7p . rlSrj/lS K — KVSMOW L K' (5) 18RvsMow = 0.0020052 [21] (6) 17RvsMow = 0.0003732 via [4] (7) - 0.0003799 [11] (8) = 0.0004023 [10] (9)

1.3 Results of Simulations

Results of all simulations are summarized in Table II and presented in Figs.1-5. Each 17 r plot contain point0 s90 s eac(30r fo 0h valuf eRo VsMow) referenc;f° plotsl al IAEA e n ei th , - 1981 valu alss ei o presented discussionn I . s below tere th ,m "unrelated uses "i o dt describe CO2 samples or working standards that have unmeasured a values and have separate oxygen pool lineages [15]. In contrast, isotopically "related" CO2 must either be measured independently for 517O and 51SO (vs. VSMOW) or measured absolutely for 17R and 18R [22];

3)This calibration metho bees dha nstandara d laboratory practice thesn I . e proceedings, Francey and Allison have recommended that working standards be calibrated for 45R and 46R only (Annex 1). This method of calibration avoids unnecessary propagation of uncertainty in a by relating measurements against the reference standard directly.

113 TABLEI I RESULTS OF SIMULATIONS Figure Symbo (%o)U ± l C B 5 S18O ± U (%o) vs workind gst workins v d gst

1 D -30.971 ±0.104 -29.967 ± 0.004

+ -30.98 0.109± 6 -29.96 0.006± 4

X -31.04 0.117± 2 -29.964 ± 0.004 + -31.015 -29.967 2 a -30.97 0.105± 4 -29.966 ± 0.056

+ -30.993 ±0.106 -29.966 ± 0.056

X -31.05 0.112± 2 -29.96 0.054± 8

+ -31.015 -29.967 3 u -30.974 ± 0.030 -29.96 0.056± 6

+ -30.99 0.032± 0 -29.965 ± 0.056

X -31.05 0.030± 2 -29.963 ± 0.056

• -31.015 -29.967 4 a -30.972 ± 0.106 -29.968 ± 0.062

+ -30.98 0.109± 6 -29.96 0.068± 2

X -31.048 ±0.112 -29.965 ± 0.062

4 -31.015 -29.967

5 D 0.00 0.10± 1 4 -0.002 ± 0.062

+ 0.00 0.10± 1 6 -0.002 ± 0.062

X 0.001 ± 0.106 -0.002 ± 0.062 + 0.000 0.000

The 513C and 518O distributions are expressed by mean value ± expanded uncertainty (U) where U = 2 • u. Here, u is the standard deviation of the population of simulated delta values.

"interrelated 2 samples"CO , where asam(l a)= &am (2), must originate fro mcommoa n oxygen pool and be linked through mass-dependent isotopic fractionation exclusively. Figur showe1 distributione sth possiblf so e S13C 5d 18(ploOan ) (ploa t value) b t s from (hypothetically) infinitely precise measurements. The 518O distributions exhibit only 0.004 %o expanded uncertainty (U = 2 • u), demonstrating that 518O is not significantly influenced by variation . a However n i s , §13C value e dispersear s %o)1 d % 0. 3 .oove 0. = ranga r U ( e

Text cont. on p. 120.

114 (a) -se.a

-ae. a

-31.8

U O

-3i„a

-31.3 -

e.45 e.se e.es a.se Fractionation Coefficien * USMOUu S (U t)

(b) -29.8

-S3.9 ac •>4 JE La 3

Ö•HD -39.e

l _L i e.45 e.se e.56 e.ee

Fractianatian Coefficient

Figure 1: Scatterplot of results from Monte Carlo simulations for infinitely precise measurements (see Tabl inpur fo eI t parameter Tablstatistics)d r fo san I eI ; 813C Sd 18(ploan O ) (ploa t distribution) b t plottee sar d versus possible oxygen isotope fractionation values between working standar VSMOWd dan . Three distributions (300 points each) are presented in each plot, generated through different assumptions : Eq.7° : : Eq.8;+ : Eq.9;x : ;IAEA-198^ 1 reference value sectioe Se . n 1.3.

115 (a) -38.8 -

-38.9 - n c

-31. 6 - JC L.

-31.1 -

-31.2 -

e.45 e.so e.65 e.ee Fractionation Caafriciunt «US v« USHOU)

(b) -29.8

B L.

•O C

L 0 3

2 -38.

-38.1 - t

e.45 e.68 e.ss e.ee Fractianation Co«rrici«nt

Figure 2: Scatterplot of results from Monte Carlo simulations for measurements with typical repeatabilities (see Table I for input parameters and Table II for statistics); 513C (plot a) and 518O (plot b) distributions are plotted versus possible oxygen isotope fractionation values between working standard and VSMOW. Three distributions (300 points each) are presented in each plot, generated through different assumptions Eq.7: : Eq.8a : Eq.9 + : x ;+ ; ; IAEA-1981 reference value sectioe Se . n 1.3.

116 (a) -36.8

-38.9 •D 1.

a -31.8 "ic JC L Q 3 •3 U -31.1 o

• TJ -31.2

-31.3 -

e.45 e.ee e.ss e.ea

Fractianation Coefficient (Sampl* ÜB US)

T (b) -29.8 -

•ai. TJ -as. 9

•H J£ L O 3

O -i -38.8

-38.1 J_ l l

8.4S 6,se e.ss e.ee Fract.ianmt.Lan Comfficimnt. (Sampl) US t euf Figur : e3 Scatterplo f resulto t s from Monte Carlo simulations (see Tabl inpur fo eI t parameters and Table II for statistics); 513C (plot a) and 5I8O (plot b) distributions are plotted versus possible oxygen isotope fractionation values between sampl workind ean g standard. Three distributions (300 points each) are presented in each plot, generated through different assumptions: °: Eq.7; Eq.9: Eq.8: : IAEA-198+ X + ; ; 1 reference value. These uncertainly distributions of delta values are expected among intralaboratory measurements using one working standard, or interlaboratory measurements using working standards calibrated as recommended in these proceedings. sectioe Se n 1.3.

117 (a) -38.8

-38.9 73 c

a c -31. jr L O

O -31. 1 n

-31.2

-31.3 I I

e.45 B.BÔ 8.65 9.68

FrcctJ.anat.ian Co«rr±ci.«n USMOU)m u S (U t

1 1 I 1 (b) -29.8 -

-29.9 a c

L O 3

O ID -38.6 -

-38.1 - i i i

e.4B e.68 a.55 e.ea

FractjLOnation Co«fTiCi«nt (US u» USflOU)

Figure 4. Scatterplot of results from Monte Carlo simulations (see Table I for input parameters and Table II for statistics), 513C (plot a) and 5I8O (plot b) distribution e plottesar d versus possible oxygen isotope fractionation values between working standar VSMOd dan W Three distributions (300 points each e presentear ) eacdm h plot, generated through different assumptions ° Eq 7, + Eq 8, x Eq 9, ^ IAEA-1981 reference value These uncertainly distributions ot delta values are expected among interlaboratory measurements using unrelated working standards calibrated for 13R, !7R and 1SR See section 1 3

118 (a) a.3

a.a

c0)

L. 0 3 • D 0 -e.i

-e.a

-0.3

e.45 a.50 B.es e.60 Fractionation Coefficient (US >j» USMOU) (b) a.2 -

tl L e.i n c

a c

aL 3 e,e

o 0>

-e.i

8.4S 0.50 8.SB e.68

Fractionatian Coeffici«n m USMOUv S (U )t Figur : e5 Scatterplo f resulto t s from Monte Carlo simulations (see Tabl r inpufo eI t parameters and Table II for statistics); 5 C (plot a) and 5 O (plot b) distributions are plotted versus possible oxyge13 n isotope fractionatio18 n values between working standar VSMOWd dan . Three distributions (3ÜO points each e presente)ar eacn di h plot, generated through different assumptions: °: Eq.7; +: Eq.8; x: Eq.9; +: IAEA-1981 reference value. These uncertainty distributions of delta values arc expected among intedaboratory measurements using unrelated working standards calibrated 13r an Rdfo R , 17 18R. See section 1.3.

119

Compariso f thesno e distributions with those generated when measurements possess typical precision values (Fig.2) demonstrates that measurement repeatability defines O518 uncertainty but adds insignificant uncertainty to 513C. The means of the three S13C distributions arising three th f eo 17 e RysMous fro e mth w value mucoffses e a sar y 0.08s hb a t %o. Distributionn si Fig.2 represent the interlaboratory (global) uncertainties expected when unrelated samples and unrelated working standards are differentially measured by conventional methods. The IAEA- 1981 value is well within all ranges expressed. In contrast, Fig.3 illustrates the relatively good inlralaboratory distributions of possible 513C values for unrelated samples, given that only one working standard is used. These

distributions also represent expected interlaboratory error when working standards are calibrate recommendes da thesn di e proceedings. Reproducibilit unrelater Cfo Sf y 13 o d samples 0.030%c= U ( onls )i y slightly worse tha interrelater nfo d sample 0.020%o= U s( , measurement

repeatability). This indicates that variatio fractionation ni n coefficient between sampled an s

commoa n working standar minoa s di r sourc uncertaintf eo significanA C 5n .13 y i t effect arises

from using differenRtsMo 17 w values to define the composition of the working standard; the

V maximum offset is again 0.08%o. Tne IAEA-1981 value for SCI3 is within the expanded uncertainty rauge of one distribution (+), and plots between the other two distributions. Figure illustrat5 & s4 reproducibilitie e eth s expected during intercomparisons, when splis measured i sampla an t CO f eo d against many unrelated working standards. Surprisingly,

e rang2th f Seo 13 C value s aboui s 0.1%ot — 0.3% mucU — )( ch greater than intralaboratory reproducibility (Fig.3), and (coincidentally?) equal to the dispersion of 613C measurements

during the last IAEA intercomparison [2]. This dispersion is independent of the magnitude of

46 13 17

measured COS d CO045 an (cf . Figs.4-5) effece C5Th n . o tfro m using differenMoRt w 2 VS

value s dependeni s e magnitudth n 2 o t f inpuo e t deltas e smalleth ; e isotopith r c differences between sample and standard, the smaller the 17RysMOW effect on S13C. Therefore, unless the working standar s calibratei d s outlineda thesn di e proceedings, variation fractionationn i coefficient between working standards commona and reference majora material be sourcecan of uncertainty in consequential values of 813C.

4 Covariatio1. n Between 5Sd 1813OCan

Because conventional reduction algorithms must assum a efixe d oxygen isotope

fractionation coefficient between sampl d workinan e g standard a functiona, l relationship

18 between calculate resuldn value Oca Sd t an Cwhe5f 13 s o tru e nth e valu differa f eo s froe mth value assumed have W .e determine magnitude dth f thieo s effec simulatiny b t g compositions 13 18 hypothetican ote f 2 samplesCO l , keepin constanR g varyind an t througR g e naturahth l range. Three wervalueR 17 r e sfo calculate eacr dfo h simulated gas, usin 0.5= ,ga r Eq.5fo , 0.54d value0.5R an .46 2 d sThen an wer R e,45 constructed through Eqs.10-11 workine th ; g

45 45 4 13 17 R = CO/ CO2 = R + 2- R (10) 18 13 17 17 2 = 2 -2-+ R R- (R+ R) (11)

45 46 standard was defined as NBS-19-CO2 with a = 0.52 vs VSMOW. Simulated 5 CO2 and 5 CO2 values (from Eq.l) were reduced to 513C and S18O (using the IAEA-1981 algorithm [9]); results 13 13 13

are displayed in Fig.6. The magnitudes of the biases in S C values (S Ctrue - S ^,,,) were independent of the value oRf 13 within the natural range. Inspection of Fig.6 reveals that resultin functionalle gt fixedar valuet no bu e Sr , 13ar sCfo y relate . Whea d = 5o d18t an na O s assumei e algorithms th 0.5a ,n i d , 513C value e welar s l behave d deviationan d e ar s insignificant e covariatioTh . n effect increases, however valuea s a , s depart from 0.5. Values of 513C from sample sworkine havinth s 0.5= (v 2ga standard) wil biasee b l 0.0d> 1 %o when J518O j > 10%o. Since the true value for a is not usually known for a particular measurement,

120 -J T 8.15 -

8.10 - O x O 8.05 X X O O P) a -* a X i a a +J 0.00 a a a a -4 Tl X C O X n X a X -4 O ta -0.85 X ._

o -0.18 o

-0.15

-56 -25 58

d«lt«-18 O

Figure 6: Covariation between CS andO 5 values (from the IAEA-1981 algorithm) 1S

as a function of the actua13 l oxygen isotope fractionation relationship (a) between sample and working standard. All hypothetical CO2 samples were constructed witvalue e^C^,,,r hon efo . Symbols designate tru: eD valu: a f eo : 0.520.5x ; 0.54: ;0 . Bias (S^C^-S13^,, independens )i value th f f eo o t 13 6 Ctrvic. See section 1.4. biae th s canno correctede b t totae th ; l uncertaint partiall5f s 13i yo C yfunctioa 5e 18th Of no value e uncertaintieTh . s from this effect have been covere r Montou n di e Carlo treatment described above.

1.5 Conclusions and Recommendations

r unrelateFo d sample workind an s g standard f COo s e IAEA-1982,th 1 algorithm generates S13C and 518O values within expanded uncertainty (U) distributions of alternative algorithms thae variouus t 17d s RysMow an historican valuea a s f A o d sexplici - an l t data

121 reduction standard the IAEA-1981 algorithm should continue to be employed. However, 45 46 isotopic analyse2 shoulCO " dot s includ e reportinth e f 5go CO5d an 2CO e isotopi2th , c composition and source of the working standard, and algorithm/assumption identifiers, to 45

provide a link with any future changes. Working standards should be calibrated for S CO2 and

CO546 vs. RM8544-CO (NBS-19-CO) and linked directly to the VPDB scale before reducing 2 2

measurement 5d 2 1San O detailes ,a C 5° Francey o dt b s Allisod yan n (Anne . Reportx1) f o s measurements should include not only measurement repeatability but also uncertainty based on natural variations in a with respect to the calibration procedures utilized; these uncertainties s higa ma s ±0.he a y (U)b o e a-elfec1% Th . t would also influence isotopic analyse f gaseso s

thaO N oxygen td a rel an n yo n suc SO isotop s ha e fractionation assumption. 2

2 Further independent measurements of S17O and 518O, and absolute ratio determinations, are needed on terrestrial and atmospheric samples of interest, and on gases used as working standards, to place tighter constraints on the fraclionation coefficient among and between samples and typical working standards. Further measurements of absolute ratios of carbon and oxygen isotopes in standard reference materials are also encouraged.

1318 45 46 47 . 2 Calculatio f S no Sd COan from S CO2, 0 CO0d CO2an 2 Measurements

o naturat e Du l variation e mass-dependenth n i s t relationship between poolf o s 13 18 4:>

terrestrial oxygen isotopes, conventional methods for determining 5 C and 5 O from 5 CO2 and COS46 measurements are inherently limited in achievable reproducibility; Section 1

describe2 s limitations than translatca t e into expande r <5fo B dCo uncertaintie% 1 0. f o ) (U s

independent of intralaboratory measurement repeatability. Here, a method is introduced for

1S 45 46 47

calculation of C5 13 and OS from COS , CO5 and CO5 measurements, independent of 2 2

the relationship among oxygen2 isotopes. Measurements are reported that demonstrate the feasibility and current limitations of this technique.

1 "3-Measurement2. " Algorithm

46 47 45 46 47

From measured CO545 , COS and CO5 , values foRRr, anRd are calculated 2 2 2 using Eqs.l,10,11 thira d dan , definitio r isotopinfo c CO2:

17 18 13 17 2 47R = 47CQ2/44CQ2 _ 2.«R."R + 2- R- R + R-( R) (12)

solves i e valuR d13 Th f througeo iterativn ha e quadratic expression, Eq.13,

(13)

where:

A(i 5-= ) 13R(i 3-- ) 4SR (14)

B - 4-46R - (45R)2 (15)

C = 4-45R-46R - 8-47R - (45R)3 (16)

higheo N r order terms have been neglected e relationshipth ; e exactar s . Seedin intR g45 o °R(1) and ten iterations gives a 13R(11) value with the required precision; this value is used usinR 1S gd Eqs.10-12an R solvo t 17 r efo .

122 47 o determinT e requireth e d leve f 8o l CO2 measurement repeatabilit- "3 e th r fo y measuremcnt" (3-M) approach, Monte Carlo methods (section 1.2) were used to generate distributions of 513C and 51SO values from various levels of measurement uncertainty. These 47 simulations showe algorithM d3- thae e th s i repeatabilitlimitemtth y b d f <5o y CO2 46 measurement - sthes e must approach 0.03%o repeatabilite (equath o t l f Syo CO2 during conventional measurements) to rival the "2-measurement + a-assumption" (2-M) technique. In natural abundance CO2, the m/z 47 ion beam is about two orders of magnitude less intense than the m/z 46 ion beam, so unconventional measurement methods were required to reach repeatabilite th y requirements.

Measurement2.2 s

13 Two CO2 samples from our inventory, differing in S C compositions by about 45 %o, were measured on a Finnigan4 MAT 252. The Faraday cup detector array was physically arrange beamn io 44-4thao z ds 2 sm/ t wer7CO e collected individuall simultaneouslyd yan e Du . to software limitations only two ion beam ratios could be simultaneously acquired, so automated methods were used that repeatedly alternated acquisition configurations between m/z 44, 45, e resistor/capacitoTh . 47 , 45 , 44 rz detectoe pairth m/ n d si an r 6 arra4 y were selecte thao ds t 47 all output voltages were similar. To improve repeatability of 5 CO2 measurements, pressures of CO2 (40 kPa) about eight times larger than normal were used [23]. This required an increase in ion source pumping capacity that was achieved with differential pumping; gas pressures during acquisitions wer /uP0 e5 a (50 sourc n /xP3 0io pbar 0 d pbare a(3 e an th n i )n )i analyzee th sourcn rio operate s tubee ewa Th . d wit emission ha , electron currenmA 1 nf o t energy of 80 eV, and accelerating voltage of 8 kV. The output signal of the mass 44 peak was automatically centered and adjusted to 4 V to start each set of measurement cycles; the idle time for each half-cycle was 3 minutes. Peak 44 overlap onto the mass 45 signal was measured 45 and found to contribute less than 0.001 %o to the measured values for 0 CO2. Each gas was 45 measured repeatedly; the grand mean (%o) and standard uncertainty (u) were S CO2 = 41.560 46 47 ± 0.006, 5 CCs = 10.645 ± 0.007, and 5 CO2 = 53.756 ± 0.030.

3 Result2. Conclusiond an s s

Grand mean standard san d uncertainties were reducealgorithmsM 3- d d an usin M , g2- and uncertainly distributions were generated by Monte Carlo methods; 5I3C results are plotted as frequency histograms in Fig.7. The top three histograms arise from 2-M algorithms using three discrete values of 17R\/sMo\v; trie bottom histogram depicts results of the 3-M algorithm. e resulM uncertaint 3- Th s slightli te th f yo ysmalle distributionsM r2- thae th nl Al . distributions overlap, although some barely. A sensitivity analysis of the two algorithms indicate methoM s 3- tha e mors di th t e susceptibl isobario et c interferences owin reliancs it o gt e on the low abundance (ca. 0.0048%) of the m/z 47 ion beam. Isobaric interferences of 2.5 ppbv (parts-per-10 volumey 9b positiot thiz a ) m/ s lean biaseno ca dt Sn si 13 f 0.10%cCo e Th . same effect for the 2-measurement method requires a 1200. ppbv perturbation at the m/z 45 position. Becaus have t echaracterizew eno d these gase isobarir sfo c interferences statisticaa , l comparison of these 513C distributions is unwarranted at this time. The 3-M method is best applicabl e differentiath o t e l measuremen f well-characterizeo t d standard materialse W . continue to develop this approach through improvement in measurement repeatability and identificatio f errono r sources.

4)Certain commercial equipment, instruments r materialo , e identifiear s thin i d s papeo t r specify adequatel e experimentath y l procedure. Such identification doct implno s y recommendatio r endorsemenno Nationae th y b t l Institut f Standardeo Technologyd san r no , doe t implsi y thae materialth t r equipmenso t identifie e necessarildar e besyth t availablr efo purposee th .

123 ae

60 u c (a) or o

43. a 44 44.2 44.4 44.6

89 -

U C 0> (b) e3r 49 o> £ 29

9 - 43.8 44 44. S 44.4 44.6

89 ~

68 - >. O (c) 3 4Ö or o> ÙI 29

43.8 44 44.2 44.4 44.6

109 ~

U C. 69 0) (d) " 40

43.8 44 44.2 44.4 44.6 Delta-1 . Workinvs 3C g Standard (%o)

Figur : e7 Frequency histogram f Montso e Carlo 513C results (plotthiougM 2- sà e hth

C) and 3-M (plot d) algorithms using highly-repeatabie 8^CO2, 5 CO2 and 46 6 CO2 measurements differene Plotus c sa- t assumptions ) Eq.7(a :) (b , sectioc Se . 3 . n2 9 q E ) (c , 8 q E 47

124

3. Requirements, Production and Measurements of Isotopic Gas Standards To detect small but significant temporal and spacial variations of C6 13 in carbonaceous trace gases, inlerlaboratory reproducibility of 0.01 %o (u) for CO2 is a goal of the global

atmospheric monitorin modelind gan g communities measuremene th ; t reproducibility goalr sfo othen i 135 C r carbonaceous gase somewhae sar t more relaxed [24]. These goals require eth e reductioon f generao ne us algorithl e globath m d (Annel an accessibilit ) 1 x f o y intercomparison standards prepared with isotopic homogeneity among replicates belter than 0.01%o(u). The isotopic measurement quality requirements for our group program are similar. We e integratinar g 513C measurements witr establisheou h measuremenC 14 d t capabilito t y characterize atmospheric aerosols and carbonaceous gas species [25]. Multi-isotopic signature f thesso e species conservativs a , e tracers, improve discrimination powe r sourcfo r e apportionment, chemical transport/transformation, and receptor modeling [26]. Because chemical separation and purification of gases are accompanied by some degree of isotopic fractionation, we are producing isotopic gas standard mixtures to explore these effects and to achieve maximum measurement reproducibility. We are developing an approach to

characterize puref thegased C 6r an n sfo blen d them with each othe witd an rh nitroger no M 13

Scrubbers 20 Liter and Traps Variable (4) Volume

(2)

(1)

Figure 8: Gas isotope standard production system. (1) Set of seven parallel glass lines, prepinched into several open breakseals; valves are driven by pneumatic solenoids; (2) 20 liter stainless steel variable volume; (3) Capacitance manometer ) Scrubber(4 ; purifications trapd ga r san sfo ) Meta(5 ; l bellows pump ) Vacuu(6 ; m system molecula- r dra gs inle residuapumpd Ga an t ) (7 ; l gas analyzer. See section 3.1.

125 synthetic thiair.n I s5 sectio describe nw firse eth t stage productio:a n metho purr dfo e isotopic gas standards. Measurements on a prototype standard demonstrate that gas isotope standards and reference materials can be prepared with the reproducibility necessary for atmospheric programs.

3.1 Productio Isotopif no Standards cGa s

Using information gained from prior studies [2-3, 27] e hav,w e designed, builtd an , teste dparallel-procesa manifols sga d (Fig.8 alloo )t precise wth e preparatio replicatf no e units of pure gas and gas mixtures. Our system consists of a 20 L stainless steel variable volume, metal bellows pump, and circulation loop with a set of seven parallel glass tubes prepinched into several open breakseals. A quadrupole mass spectrometer is used to test for leaks and to identify impurities a high-throughpu d an , t Russian doll cryotrap [28 s availabli ] o removet e condensable impurities such as water vapor. To keep the internal pressure constant (about 70 kPa) during the breakseal filling procedure, the variable volume can be precisely adjusted. After each fill, a torch is used to quickly seal the prepinched areas of the tubes. Over one thousand 400/zmole replicate isotopin a f so standars cgeneratee ga b n dca e d th fro file f o ml on variable volume; the method is applicable to most gases. e methoTh testes dwa d with pur2 fro mecylindeCO a r storer inventoryou n di . Seven prepinched borosilicate tubes (9 mm outer diameter) were annealed overnight at 600 °C; one d tubs positionean e systewa ea th s expande kP d m wa 0 an d7 2 evacuated o t dCO e Th . recirculated througice/methanoy dr e hth l cryotrap overnight. Then cryotrae th , isolates pwa d s allowega ane o recirculatth dt minutes0 3 r fo e . Afterwards e breaksealth , s were torch- sealed, another prepinched tube installed, evacuated, and the gas expanded and recirculated for another 30 minutes through the new tube. The sequence was repeated several times and resulted in 28 replicates of the prototype standard. Ordinarily the seven tubes would be installed and filled together, but this method simulated the actual procedure without requiring an excessive number of tubes. The replicates produced were measured against a working standard prepared from the same gas to determine preparation reproducibility and possible isotopic differences among tubes.

3.2 Measurements

All delta values reported were measured on a Finnigan MAT 252. Automated methods were used that alternated acquisitio. 47 , 45 , n 44 configurationz m/ d an 6 4 , s 45 betwee , 44 z nm/ The ion source was operated with an emission current of 1 mA, electron energy of 80 eV, and accelerating voltage of 10 kV. Normal gas pressures (5 kPa) were used in the inlets. The output signal of the m/z 44 peak was automatically centered and adjusted to 5 V to start each measuremenf o t se t cycles idle ;th e tim eacr efo h half-cycl minutes5 s ewa . Five replicates have 45 been measured repeatedly; grand means (%o) and standard uncertainties (u) are: S CO2: 0.047 46 47 0.005± ; S CO2: 0.23 0.0060± Sd an CO; 2: 0.2 40.08± significano N . t differences,'within r betweeo n tubes, were observed amon replicatese gth degree Th . f reproducibiliteo quits yi e sufficien e needatmospherie th th f r so fo t c monitoring community continue W . measuro et e remaining replicates to monitor long-term stability of the isotopic compositions.

5) 14 fM is the C content expressed as the fraction of modern carbon, as related to the oxalic acid standard SRM 4990B.

126 3 Isotopi3. Standardss cGa : Statu t MISsa T

Facilities now exist at NIST to prepare a variety of isotopic gas standards. In conjunction with the IAEA, we are collecting feedback from the measurement and modeling communitie e mosth tn o susefu l chemica isotopid an l c composition packagind an s g options. Currently r inventorou , y contains several candidate natura biogenid an l c gases (Table IIId )an we are interested in identifying and locating other possibly important standard gases and mixtures. At this time, however, only the IAEA-NGS gases [2] are available for informal distributio6 n from our group.6

TABLI EII ISOTOPIC COMPOSITIONS7 AND SOURCES OF CANDIDATE STANDARD GASES 14 Gas 13 18 f ( c) Source 5 CVPDB (%') 8 0VPDB (%°) M

CO2 -10.4 0.2 1.4 Grain Fermenter

CH4 -43.6 - 1.3 Brazilian Sewer Naturas Ga l -29 - 0 IAEA: NGS-1 (81% CH4) Natural Gas -45 - 0 IAEA: NGS-2 (53% CH4) Natural Gas -73 - 1 IAEA: NGS-3 (99% CH4)

Acknowledgements e authorTh : s wis thano ht k E.L. Bakke, C.A.M. Brenninkmeijer, R.N. Clayton, T.B. Coplen, J.M. Hayes, and M.H. Thiemens for valuable discussions, and numerous reviewer r theisfo r helpfu practicad an l l comments.

References

(1993)O IS [1. ] Guidexpressioe th o et f uncertaintno measurementn yi 199O IS : 3 (E). International Organizatio r Standardizationfo n (Geneva, Switzerland). p 1 10 :

[2] HUT, G., Consultants' Group Meeting on Stable Isotope Reference Samples for Geochemical and Hydrological Investigations; IAEA, Vienna, Austria; 16-18 September 1985; published 1987: 42p.

] [3 COPLEN, T.B., KENDALL , HOPPLEC , , "Compariso,J. stablf no e isotope reference samples," Natur (19832 e30 ) 236-238 (amended).

[4] CRAIG, H., "Isotopic standards for carbon and oxygen and correction factors for mass- spectrometer analysi f carboso n dioxide," Geochim. Cosmochim. Act (19572 a1 ) 133- 149.

6)Please contact RMV: (301) 975-3933; FAX (301) 975-2128; EM [email protected].

7)A11 values are nominal only.

127 [5] DANSGAARD, W., "The isotopic composition of natural waters, with special reference to the Greenland ice cap," Medd. Gr0nl. 165 (1961) 1-120.

[6] DEINES, P., "Mass spectrometer correction factors for the determination of small isotopic composition variations of carbon and oxygen," Int. J. Mass Spectrom. Ion Phy. 4 (1970) 283-295.

[7] MOOK, W.G., GROOTES, P.M., "The measuring procedure and corrections for the high-precision mass-spectrometric analysis of isotopic abundance ratios, especially referring to carbon, oxygen and nitrogen," Int. J. Mass Spectrom. Ion Phys. 12 (1973) 273-298.

] [8 HALAS , "CorrectioS. , n factor variour fo s s collectorn typeio f so S-carbon-13n i s - 5 , oxygen-1 o-sulfur-3d 8an 4 measurement Russian)n "(i , Isotopenpraxi (19773 1 s ) 321- 324.

[9] GONFIANTINI, R., "The 5-notation and the mass-spectrometric techniques," Stable Isotope Hydrology: Deuteriu Oxygen-1d man Watee th n 8i r Cycle, edsd . J.Ran t .Ga . GonfiantiniR , IAEA-TECDOC-210, IAEA, Vienna (1981) 35-84.

[10] SANTROCK , STUDLEY,J. , S.A., HAYES, J.M., "Isotopic analyses mas e baseth sn do spectrum of carbon dioxide," Anal. Chem. 57 (1985) 1444-1448.

, W.J. B.L.LI , [11NI ,] , J1N, D.Q., CHANG, T.L., "Measuremen absolute th f to e abundance of oxygen-1 VSMOW,7n i " Kexue Tongbao (Chinese Science Bulletin (19883 )3 ) 1610- 1613.

[12] CHANG, T.L., LI, W.J., "A calibrated measurement of the atomic weight of carbon," Chinese Science Bulletin 35 (1990) 290-296.

[13] BAKKE, E.L., BEATY, D.W., HAYES, J.M., "The effect of different CO2 ion correction methodologies on 518O and 513C results," Paper presented to the Geological Societ f Americayo , October 1991 Diegon . Sa , CA ,

[ 14] BIGELEISEN, J., WOLFSBERG, M., "Theoretical and experimental aspects of isotope effect chemican i s l kinetics," Adv. Chem. Phys (19581 . ) 15-76.

[15] MATSUHISA , GOLDSMITHY. , , J.R., CLAYTON, R.N., "Mechanismf o s hydrothermal crystallizatio kbar,5 1 d " an GeochimC ° f quart0 no 25 t za . Cosmochim. Acta 42 (1978) 173-182.

[16] CLAYTON, R.N., MAYEDA, T.K., "Oxygen isotope n eucritesi s , shergottites, nakhlites chassignites,d an , " Earth Planet. Sei. Lett (19832 6 . ) 1-6.

[17] ROBERT, F., REJOU-MICHEL, A., JAVOY, M., "Oxygen isotopic homogeneity of the Earth: new evidence," Earth Planet. Sei. Lett. 108 (1992) 1-9.

[18] THIEMENS, M.H., JACKSON , MAUERSBERGERT. , , SCHUELERK. , , B. , MORTON, J., "Oxygen isotope fractionation in stratospheric CO2," Geophys. Res. Lett. 18 (1991) 669-672.

[19] THIEMENS, M.H., "Mass-independent isotopic fractionation theid an s r applications," Isotope Effect n Gas-Phasi s e Chemistry . J.Aéd , . KayeS SympAC , . Ser. 502, Washington, D.C. (1992), 138-154.

[20] HAMMERSLEY, J.M., HANDSCOMB, D.C, Monte Carlo Methods, New York, John Wile Sonsy& : (1964) 178p.

128 [21] BAERTSCHI , "AbsolutP. , e 18contenO standarf o t d mean ocean water," Earth Planet. Sei. Lett. 31 (1976) 341-344.

BIEVREE D [22LAETERE D ] , P. , , J.R., PEISER, H.S. REED, W.P., "Reference materials by isotope ratio mass spectrometry," Mass Spectrom. Rev. 12 (1993) 143-172.

[23] DOUTHITT, C, "High precision analysis of large CO samples," Finnigan MAT,

unpublished (1990). 2

[24] FRANCEY, R.J., ALLISON, C.E., "Recommendations for the measurement and reportin e resultgth f stablso e isotope analysi f carboo s oxyged nan atmospherin ni c trace gases," IAEA Research Coordination Meetin "Isotopn go e Variation Carbof so n Dioxide and Other Trace Gases in the Atmosphere," 2-5 November 1992, Heidelberg, Germany; K. Rozanski, Scientific Secretary (1992).

[25] KLINEDINST, D.B., McNICHOL, A.P., CURRIE, L.A., SCHNEIDER, R.J., KLOUDA, G.A.REDENN ,VO , K.F., VERKOUTEREN, R.M., JONES, G.A., "Comparative study of Fe-C bead and graphite target performance with the National Ocean Science (NOSAMSS sAM ) facility recombinato source,n rio " Nucl. Inst. Meth. Phys. Res. (1994) submitted.

[26] CURRIE, L.A., "Source apportionment of atmospheric particles," Environmental Particles, (BUFFLE, J., VAN LEEUWEN, H.P., Eds.), Lewis Publ., Ann Arbor (1992) 3-74.

[27] COPLEN, T.B., KENDALL, C., "Preparation and stable isotope determination of NBS- NBS-1d an 6 1 7 carbon dioxide reference samples," Anal. Chem (19824 5 . ) 2611-2612 (amended).

[28] BRENNINKMEIJER, C.A.M., "Robust, high-efficiency, high-capacity cryogenic trap," Anal. Chem (19913 6 . ) 1182.

129 HIGH PRECISION STABLE ISOTOPE MEASUREMENTS ATMOSPHERIF O C TRACE GASE] S[1

C.E. ALLISON, R.J. FRANCEY CSIRO Divisio f Atmospherino c Research, Mordialloc, Victoria, Australia

Abstract: An overview of the CSIRO-DAR stable isotope program is presented. The broad scope program ofthe applicationand results ofthe atmosphericto science illustrated.are Some problems associated with the maintenance of a stable isotope program are described and the CSIRO-DAR data reduction procedure for stable isotope measurements is described.

Overvien ParA : t1 w

1.1, Introduction

Commonwealte Th h Scientifi Industriad can l Research Organisation Divisiof no Atmospheric Research (CSIRO-DAR) has a significant interest in solving problems concerning the physics, dynamics and chemistry of the atmosphere. Four research programs address specific objectives in the areas of atmospheric pollution, atmospheric processes, climate modellin globad gan l atmospheric change. globae Th l atmospheric change program focuse chemistre th n e so th f yo troposphere including observationa modellind lan g studies. Specific objectives include: • Determination of global distributions and trends of long-lived radiatively active (greenhouse) and ozone-depleting gases, their isotopic composition and precursors from atmospheric observations coree archive,ic d san samplesr dai . • Modelling of atmospheric transport and exchange of these gases for interpretation of observations and prediction of future atmospheric concentrations. • Quantification of rates and processes controlling emissions of climatically active trace gases from natural, agricultural and urban/industrial systems in Australia. • Assistanc developinn ei g national inventorie gaseouf so s emissions. • Support and scientific collaboration in the Cape Grim Baseline Air Pollution Station. • Developmen enhancemend tan instrumentatiof to calibratiod nan n technique requires sa d achievo t scientifie eth c objectives.

As a key component of the study of the distributions and trends of long-lived radiatively active gases, CSIRO-DAR operates GASLAB (Global Atmospheric Sampling Laboratory) which is involved in addressing the reduction of uncertainties in the global carbon cycle, through coordination of measurements of trace gas concentrations and isotopic composition with transport modelling.

GASLAB maintains state-of-the-art instrumental facilitie analysie th r sfo f so radiatively active trace gases and their isotopic composition. Whole air samples are

131 collected from a number of land-based sites from the South Pole, Antarctica, to Alert, Ellesmere Island, Canada, through collaboration with international organisations. Figure1 indicates some of the sampling locations. Samples are also collected on Antarctic supply voyage aircrafn o d san t over-flight variout sa s location specifir sfo c projects.

One of the main interests of GASLAB is the high precision measurement of the stable isotopes of carbon dioxide, CC>2. This paper concentrates on this aspect of the CSIRO-DAR research interests.

1.2. Isotopic ratio and composition Isotopic ratio is usually expressed as the ratio of the less common heavy isotope to more th e common light isotop elementn a f eo carbor 13oxyger = ,fo C/ e.gfo 3 .12 d nr! C nan =7 1 =8 r O/r! O/d Oan O. Isotopic compositio usualls ni y expressed usine gth 17 18Î6 16 6-notation as the difference in isotopic ratio between a sample material and a reference material (subscript respectively)R d an sS .

rl3 Srl3- R r!3R

r!8- r!8 Ô180 = Rs r!8R

Isotopic composition is often expressed in units of parts-per-lhousand or "per mil", %0, obtaine multiplyiny db above gth e expression 1000y sb . Many processe relevancf so o et atmospheric studies, particularly biological processes, discriminate agains heaviee tth r isotopes, therefore, isotopic composition expressed in the 0-notation is usually negative.

1.3. Cape Grim in situ record

CSIRe Th O stabl e stud th initiate s f eyo wa wit] isotope2 2 [2 197hn dCO i 7 CO f so being extracted cryogenically from whole air at the Cape Grim Baseline Air Pollution Monitoring Station, in north-west Tasmania, and returned to GASLAB for mass spectrometer analysis.

Although commenced in 1977, the Cape Grim in situ stable isotope program did not produce consistent high quality data untidateo t and] l, [3 , 198analyse2 datf so a have concentrate perioe th n do from 1982. From 1982 until mid-199 0VG602a D mass spectrometer was used exclusively for the stable isotope analysis. In mid-1990 a new mass spectrometer, a Finnigan MAT252, was commissioned and both mass spectrometers were use 2 analyssampleo dt CO e eth s extractee th f 199o 1 Capt dd a en ee Grith t mA [4]. VG602D was de-commissioned.

The Cape Grim in situ CO2 stable isotope record is presented in Figure 2. The solid lines represent data from the VG602D and the diamonds represent data from the MAT252.

132 •Macquarie Island

FIGUR . LocationE1 globasome f so th f eo l sampling site whict sa collectes i r hai analysir dfo GASLABy sb . -7.4 (a) Ô13C -7.5

-7.6

-7.7

- - - - -7.8

-7.9 82 83 84 85 86 87 88 89 90 91 92 1.6 ) Ô18(b O

1.4

1.2

1.0 '

0.8 • - - -

0.6 83 84 85 86 87 88 89 90 91 92

13 18 FIGURE 2. The (a) S C and (b) 0 O isotopic records for Cape Grim in situ CO2 perioe foth r d 1982 through 1992 inclusive solie Th d. line represent 5-poinsa t average, correspondin approximateln a o gt wee6 y4- k average VG602e th f ,o D datadiamonde Th . s represen same th t e averag MAT25e th r efo 2 ô data l Al . measurements are in per mil, %o, against VPDB. The slow convergence in 518O has been addressed eîsewhere[4].

Each poinaveragn a s analyseti 5 f eo representd san approximateln sa wee6 y4- k average. Comparison of data from the two mass spectrometers is the subject of a separate paper [4] and was the trigger for the work described in Part 2 of this paper. While both 613C and 518O records exist, discussion will be limited to the 813C record.

Three phenomen 0 e cleae C th a ar n ri record seasonae firse th s i tTh . l signal 13 presen date th a n tsuggestini g exchange with terrestrial biota, possibl ycombinatioa f no exchange with southern hemispher mont6 e4- planta hd laggesan d signai from northern hemisphere plants secone Th genera.a s datmospherii f o l C decreaso° e th c n eCC>i 2 from 1982 until 1989, due to the release of isotopically depleted CO2 into the atrao.sphere from

134 :•

-7.5 -. oQ JQ s$ o.*? ïS Q-O P* ïi d ù -8.- - 0 •.r. * • . •** » , «, ; *• -8.5 -• '4 i Point Barrow

-7.5 -. ,' * •;

-8.0 -.

-8.5 -- Mauna Loa

-7.5--

-8.0 -

-8.5-- South Pole

84 85 86 87 89 90 91 92

FIGURE 3. The 813C records for 3 global sampling sites. The solid circles (dashed GASLAe Unesth e )ar B measurements opee Th n. square splina dato e t st ar ae fi obtained by NOAA-UC (Ref [5]). No NOAA-UC data are avauable for the South Pole.

135 fossil fuel. The third is the cessation of this decrease in 1989 with the 13C isotopic composition of CC>2 remaining relatively constant. This "flattening" is the subject of a separate paper [5]. 1.4. Global sampling sues

In 198 prograa 4 establishes mwa samplesr collecai o d t L 0 1 t, chemically dried, from a network of global sampling sites. The air samples were returned to GASLAB wher trace concentrationes th ega s were measure CC> d extractes dan 2 wa (abou) dmL 3 t cryogenicall isotops ga r yfo e ratio mass spectrometer analysis recenA . t upgradf eo GASLA allowes Bha sampline dth g progra reducmrequiremenr o t ai e eth 1 o t t froL 0 m1 L, and modifications to the CC>2 extraction facility have reduced the amount of air required perforo t mroutina e isotopi cC(>2f o analysi .abouL e i ii , 0 abou1 o tst mL 0 t3

Figure 3 shows the 513C records obtained from the whole air sampling program from three sites (solid circles) [5] .thre e Eacth ef h o site s exhibit changsa leso et s negative slop 5n ei C from about 1988. Superimposed ove GASLAe rth B recore dar 13 measurements mad separata y eb e sampling program (open squares) excellene Th . t agreement betwee independeno tw e decadne th th f o te program d 1982-199en e th t sa 2 lends support to the CSIRO record over the whole period, particularly in view of the consistency of calibration standards and procedures in the CSIRO program.

7.5. Analysis stable2 CÛ e isotopeoth f record Total CO2 in the atmosphere is a composite of CO2 from various sources or reservoirs. Likewise, the isotopic composition of the atmospheric CO2 is a composite of isotopie th c signature differene th f so t reservoirs. Neglecting fractionations whicy hma occu transferrinn ri 2 betweegCO n reservoirs construcn ca e ,w relationshipta , from Dalton's law of partial pressures, which combines the concentration and isotopic 13 compositio 2 fro reservoirsl mCO al f no . Usin isotopie th g r 5 fo Cc compositioe th f no designatind CO2an , varioue gth s reservoir wit 2 subscripte hth Œ> f so y,..., sx e ,w determin observee eth d isotopic compositio atmospheree th f no .

-i^| o t o ff~*/~\ i Ç13/"" 1*d ^/~^i r/^/~\ i i v /~^/~\ i ~ " *-^x+y+ l*-'*-'2Jx+y+ = " ^-xt^^vlx "*" " ^ylV'-'zJy ~*~ -•• * additioe Th isotopicallf no atmosphere yth distinco t 2 tCO e will affec overale th t l compositio atmosphere th f no determines ea thiy db s relationship. demonstrato T e this, conside atmospherie rth 0 c35 concentratioe b o t 2 CÜ f no ppmv and the isotopic composition of this CÛ2 to be 013C = -8%c. The addition of 1 ppmv

2 witCO h 5 C -25%o= , typica terrestriaf lo l plant carbon, will increas atmospherie eth c 13 CO2 concentration to 351 ppmv, and decrease the §13C by about -0.05%c. Likewise, the additio wit2 ppm1 h Œ> f 8nf o v13 o C -10%o= , suc fros ha m atmospheri whic2 n i c CÛ s hi equilibrium with ocean mixed-layer carbon, would increas concentratiotota2 e eth lCO o nt 351 ppmv, but would decrease the atmospheric 013C by about -0.006%c.

Using these results possibls i t ,i constraio et globae nth l carbon budgey tb considering separately the changing atmospheric carbon content and the changing atmospheric isotopic composition separatn ca e lefe W e.tth hand sid Equatiof eo intn3 o componentso tw f o m su .e th

4 d(CaSa)/d ot= a d(Ca)/d Ct+ ad(oa)/dt 136 Detailed analysi thif so s relationship [5,6,7 groupes ]ha d significant componentf so the isotopic constraint on the global carbon budget as a combination of net and gross biospheric fluxes, Nb and Gb, and net and gross ocean fluxes, Ns and Gs. This allows us consideo t release rth carbof eo n int atmosphere oth e from fossi biospherie lth fueld an , c and ocean sources and sinks for carbon. GASLAB has accumulated examples which demonstrat majoe eth r influence atmospherin so c 5 13exampleo C;tw givee sar n here.

7.5.7. Net biosphere term, Nb

dominane Th t influenc atmospherin eo c 013 timn Co e scales less than centuries si the photosynthetic kinetic fractionation by terrestrial plants. The influence over seasonal time scales is illustrated in Figure 4(a), for the Canadian station at Alert, Canada (R.J. Francey, C.E. Alliso N.B.Ad nan . Trivett, private communication) seasonae Th . l decrease in CC»2 (northern hemisphere summer) is accompanied by an increase in 013C due to the preferential uptak lightee durinth 2 f eo r CÜ g photosynthesis.

Figure 4(b) shows 013C plotted against C(>> concentration (open triangles) and

against inverse CC>2 concentration (solid diamonds). The plot of 013C versus CC>2 indicates

13 the change in C013 to be about -0.05%o per ppmv. The plot of C0 versus 1/[C(>2] concentration shows the isotopic composition of the C(>2 being used in photosynthesis to aboue b t -25.8%o.

7.5.2. Gross air-sea exchange, Gs

Figure 5 shows the 013C latitudinal gradient obtained on a ship cruise between Hobart, Tasmania, and the east coast of Antarctica in January, 1993 (RJ. Francey, C.E. Allison and H.M. Beggs, private communication). The observed 813C gradient is the revers thaf eo t expected fro southware mth d drif northerf o t n hemisphere air, whics hi depleted in 013C due to fossil fuel release. We expect, and have confirmed elsewhere, that no significant influence is expected on Ô13C as a result of net exchange with the oceans. observee Th d chang howevers ei , consistent wit change hth air-sen ei a fractionation factor 14°e th Co t change surfaca du se n ei e temperature [8]grose .Th s exchang CC>f eo 2 wiîh oceane th s transmits this signaatmosphere th o lt e absenc e eve exchanget th n ne i f eo . Rapid atmospheric mixing reduces the signal in 513C in the atmosphere due to air-sea fractionatio arouny nb orden da magnitudef ro , however sees a , Figur n signai e th , es 5 li still detected. 1.6. Other isotopic measurements

Atmospheric C(>2 isotopes can be measured for past atmospheres by sampling air archivestyper o ai hel tw f so n di . Measuremen numbea f to isotopif ro c ratios, other than 13 18 6 C and 6 O of CO2, are possible in GASLAB. The isotopes of N2 and O2 can be 0e measure13 methaneth f C o n ca s da ,

1.6.1. Air archives - archive tanks

Sinc establishmene eth Cape th f eo t Gri m baseline station bees ha nr store,ai r dfo future analysis of atmospheric constituents in high-pressure stainless steel tanks. The 013C of atmospheri 2 sinccCO eaccessee 197b n 8ca d through this resource atmospherin ca s a , c trac compositions ega isotopid san c measurement othen so r species.

137 -7.4 365 .•-—I ppmv

(a) -7.9 355

-8.4 345 92.0 92.5 93.0

0.0027 0.0028 ppmv" 0.0029 -7.4 ——I——

-8.0

= -25.8±0.3%o A813 = -0.050 ± 0.001%o / ppmv -8.6 340 350 360 ppmv 370 yeae (soliFIGURrOn 2 cycl) dCÜ (a n circlese. i E 4 : ppm vrigh- t ordinaled )an S13C (open squares : per mil - left ordinale) measured at Alert, Canada (82°N), showing the correlation between C(>2 concentration and 013C isotopic composition, ) Plo(b Sf t13o CC>. Cvs 2 concentration (open triangles) shows plant 813C signature of -0.05 00.00± 1 %ppmvr ope . Plo 8f 13to invers. C vs concentratio 2 eCÛ n (solid diamonds) indicate extinctioe sth n removee 5th 13 f Co d -25.%oCC>e 3 b 0. .o 28t ±

1.6.2. Air archives - Antarctic ice-sheets GASLAB collaborate drilline th n s i Antarctif go c ice-core extractd san trapper sai d icee th . n Dependini site drillingdecadef eth o w recen s n ancienfe s g a o a a r r s ,ai o t a s ta hundredw fe a thousand accessedyearse b n ,ca . More recen samplesr tai n te o t , w frofe ma extracteyearse b n ,ca d fro fire mnth layer accumulatine ,th g snow layer whics hi compacting to form ice. Measurements of 513C for CÛ2 in this firn air show reasonable

agreement, after correctio gravitationar nfo l separation, with modern atmospheric 813C measurements and with C5 13 measurements made on the air extracted from ice-cores. (It is

138 16.0 -7.70

0.0 -

- -7.80

RGUR surfaca . EPlo5 se f to e temperature (SST (soli) ,°C d lin elef: t axis ôd l3C)an (%o) atmospherif o c CO2 (circles : right axis) vs. latitude (°S). anticipated that even better agreement wil achievee lb d when full correctio nonr nfo - equilibrium diffusion effect made)s i s . Combinatio moderne th f no , ice-corfird nan e 13 records provide detailesa d histor 2 fro 6f ymo CO Cn pre-industriai l presentimee th o st t and over glacial time-scales. Measuremen othef to r isotopic ratios, suc 0s h815a d 18NOan 2 respectivelyO d an fo2 N r , assis understandinn ti mechanisme gth s which drivs ega inclusion in ice formation. 1.6.3. Other isotopes

13 If sufficien availabls i morer r o ai t6 e L ) th 5 eCatmospheri f ( e o b n ca 4 cCH measured using extraction techniques developed jointly with Daviw dNe Lowe th f eo Zealand National Institut Watef eo Atmospherir& c Research Ltd. (NIWA Paud )an l Quay Universite ofth Washingtonf yo ,combustes i USA 4 CH . d with atmospheri foro t 2 mcO CO2 and H2O which are trapped cryogenically. The CO2 is then separated from the H2O and analysed. Long term measurements from Cape Grim are not yet available but preliminary analysis gives good agreement with the atmospheric record constructed for Baring Head, New Zealand [9].

1.7. Problems

A number of problems have been identified in the 013C measurement of the atmospheri summarisec1 COx 2Bo . "pitfallse sth have "w e identifie whicd dan e har described below.

Box 1. Pitfalls in 613C measurement : The CSIRO experience. Precision target Individual sample 0.010 %o Annual average 0.005 %c Gas standards

CO2 flasks (o-ring taps) ~ +0.02 %c L / year Air tanks ±0.1d 0% Mass spectrometer fractionations Reservoir "bleed" effect « 0.003 %o I hour 05 % "Sample size" effect : Ar ~ 1 %o * &2-6V < °- ° Ar ~ 20 %o A2.6V « 0.4 %0 Ion corrections 17O algorithms Ar ~ 1 %o - 0.05 %o

0.22+0.0= 0 1% O 2 N Sampling site bias ±0.0= o 1% Cape Grim cuff

Boundary layer = -0.01 %0 Interannual ?

* Ar = 645 = rl3(sample) - r!3(reference)

140 1.7.1. Precision

Measuremen usee 0f b constraito 13o dn t C ca sourcee nth sinkd r san sfo atmospheric CÜ2- The deconvolution of atmospheric 0 C data, for example, puts very 13 severe constraints on the measurement for both spatial and temporal differences of 8 C 13 [7,10]. Entin conclud. al t ge e that temporal precision, i.e trende .th 8f more 13o , th C es i important constraint precisior Ou . n individuatargen a r fo t l 5 C isotopic analysis si 13 0.01%e. For an annual average, the precision target drops to 0.005%o based on averaging to reduce the precision of individual measurements. This assumes that most errors are random.

standards1.7.2s Ga . maintaie W nsuita standardpurf 2 eo eCO s (Tabl whic) eI usee har d botr hfo measuring sample monitorind san long-tere gth m stabilit standardsr ou f yo . Occasional measuremen carbonatf o t e standards, suc NBS19s ha , also serv checo et precisioe kth f no standardsr ou . Measurements ove year5 1 r s have shown that differences between sub- sample standardpur2 e th eCÛ f so lese sar s than 0.02%o. Further have w , e evidence that our standards are either all maintaining close to initial values, or that all standard« are

13 TABLE I. The pure CO2 reference gases used in GASLAB. All 5 CypDB 618OvpD r mil. B pe value ,%o n i e sar

Standard Name Flask Acquired 513yPDCB

STOlf 15L glass 1977 -6.384 -13.230 ST02f 15L glass 1978 -6.383 -13.121 ST03f glasL 50 s 1978 -6.396 -13.176 ST04f 15L glass 1990 -6.405 -13.178 ST05f 15L glass 1990 -6.400 -13.158 CG06# 0.3L s/steel 1992 -37.2 -6.7 GS-19t 0.25L s/steel 1992 -7.502 -0.193 GS-20t 0.25L s/steel 1992 -8.622 -0.991 OZTECH-3* IL steel 1990 -3.759 -21.511 OZTECH-30* steeL 1 l 1990 -30.281 1.315 OZTECH-40* IL steel 1990 -40.651 -29.744

f CAR cylinde2 S ACO r HC453 # Air Liquide, Melbourne, Australia. Approximate values. $ Universit Groningef yo n * Oztech Tradin. gCo

141 -7.8, IS, :3 -80 n *%

\ Ai 00. V f 'ticf " - -82 «*k *4». (a) S44L-001 (d) S44L-OOr* ** . -8 4 , * , „ , . . r .„,,..» 3 9 Aug9n Ja 33 3 9 9 y Ocl9v 3 Ma 9 No Mau 3 la 3 r9 2 9 Oct9c De 2 3 3 9 9 Jun9v y No Ma 3 3 3 9 9 Aug9t r Oc Ma 3 3 9 n Ja 2 9 c De Oc 2 t9

1.5, ' ifj** S**?!*.1* »M"11 • *• • no. i "l '•if î 1 *i*Ly| .^^ [L (b) S22L-005 (e) S22L-005 %%**! ;

3 3 9 9 v y No Ma Jun93Ma 3 3 r9 9 t Oc Allg93 9 3v No 3 9 t Oc 3 9 g Au 3 9 o Ju 3 9 y Ma M I3 IT9

78. i 1-5 1

-80 4ffAr^4|iM '' ^*'fcrf^%d ^ lik^^c4 B ^

, 00.

-82 (c) ALVZ861 (f) ALVZ861

_O 4 Jnn91 Sep91 Dec 91 Apr 92 Jul92 Ocl92 Jan 93 May 93 Ang93 Nov 93 Jm 91 Sq)91 Dec91 Apr92 Jul92 Ocl92 Jan93 May93 Aag93 Nov93

FIGUR standardr Ai . E6 s use checo dt performance kth MAT25e th f eo 2 mass spectrometer. Note that the time scales vary. The 813C measurements are shown in panels (a), (b) and (c). The S18O measurements are shown in panels (d), (e) and (f)- All measurements are in per mil, %o, against VPDB discontinuite Th . y observe cylinder dfo r S44L-00 refillino t e du g1s i cylindee th Julyn ri , 1993 discontinuite Th . y observe cylinder dfo r ALVZ86n a 1s i instrumental effect.

driftin identicat ga l rates. External check isotopie th n so c compositio some f no th f eo standards indicates the former. Further, we believe that observed shifts in CÛ2 are due to polymee th r o-ring seals use containerssome n do th f eo .

alse oW maintai calibratio standardr s nsuita a ai e f eo us r sfo n r gasesai e Th . standards have not been calibrated to the same precision as the CÜ2 standards but have been monitored using the pure CC>2 standards. Significant variation in the behaviour of the standardr ai bees sha n observe indicated dan s thastorage tth e vessel importann a e sar t variable. Figur presente6 resulte sth long-terf so m monitorinr ai r thref go ou f eo standards, ALVZ861, a 60L high pressure (45 bar) cylinder, S22L-005, a 22L low pressure

142 bar2 ( ) cylinder S44L-001pressurd w ,an lo L bar2 e( 44 ,a ) cylinder. ALVZ861 appears stabl botr efo h 8 13Sd 18Can O , S22L-005 maintained good 0 138t 18CdrasticaUs bu O wa y affected, and S44L-001 showed significant drift in both 813C and 018O. We attribute these drift isotopin si c compositio smalo nt l container leake th n si . 7.7.3. Mass spectrometer fractionations numbeA masf ro s spectrometer effects have been observed whic introducn hca e errors into the measurement of isotopic composition. We identify two of these effects as "reservoie th r bleed "sample effectth d "an e size effect".

"reservoie Th r bleed effect observes "i d during long period analysif so s where eth reference gas fractionates. We estimate this effect is greatest when the reference gas has been allowed to "bleed" for long periods of time to the ion source or to waste. Measurements have characterised the maximum value of this effect to be about 0.0028%o hourr pe hou,4 base2 a r tesn do t usin gaseo gsimilatw f so r isotopic composition (Figure minimise W . 7) e this effec refilliny b t reference gth reservois ega r ever hourw yfe s during analysis. The "sample size effect" results in different isotopic compositions being measured for the same sample against reference gas when measured at smaller analysis voltages, e.g. Vsa(m/e 44) < 2 volts, than normally used, e.g. > 3 volts. The magnitude of the effect is dependen difference th n to isotopin ei c compositio sample th referencf d no ean e gase. s(C Flehoc . LeuenbergerA . ,M France. J . ,R C.E d . yan Allison, private communication). Where difference th isotopin ei c compositio smalns i l (545 %c differenconlw e )fe th ya e observed in measuring small samples is less than 0.05%c. When reference and sample gas are very different (e.g. abou observee t 20%th 0) d differenc largs a 0.4%cs s ei e a effec e Th . t most likely originates in mixing of the sample and reference gas in the ion source region. For normal samples analyse GASLAy db B thi problea s t effecno s mi t smal e becausth ) l(i f eo differenc 8n ei 13 C standardsbetwee£ CO r nou , base HC45n do 3valuesr CC>2ai d d an ,,an (ii) the consistency of our sample sizes.

1.7.4. correctionsIon Ion correction importann a e sar t consideratio constructioe th n i n a f no atmospheric trace gas isotopic record. A number of problems in the ion correction procedures have been identified and are discussed below, in Part 2 of this paper.

7.5./. Sampling site bias

Other effects have been observed which reflect a particular bias introduced into the isotopic composition during sample collection. For instance, vegetation near the sampling

introducy sitema e 813C increasvariation a o t decrease r eo n du photosynthetin ei c activity. ObservedC 8 13 may thus represent local phenomena rather than large scale features, such as ocean/atmosphere exchange deconvolutios A . n studies [5,6,7] require high precision i 813C, effects suc this ha s identifieneee b do t described dan individuan a n do l basis.

Part 2: Data Reduction Data reductio importann a ns i t aspec stablf to e isotope measurements. Because relative rather than absolute measurement made sar e error propagatn sca e through

143 Aô(%o)

0.18 n

D 0.16 - D • 045 0.14 - ———5 4 t —fî 0.12 - o 546 Aö46 =+0.0063 0.10 - ---- fit 46 0.08 - 0.06 - 0.04 - 0.02 -\

0.00 8 6 1 12 20 24 Time / hours

FIGURE 7. The reference gas bleed effect measured over a 24 hour period for sample and reference gas of similar isotopic composition soli e dashed Th d.an d lines represent linea (soli5 r bes§4 de tcircles th fit o st d )an 046 (open squares) respectively. The AÔ values are in units of per mil per hour (%c I hour). calculations and comparisons can be made which do not have common points of reference. incrementae th o t e Du l natur developmene th f eo datr ou a f reductioto n proceduren ,a historical approach is used to describe the procedure. Some material has been presented earlier in Part 1. 2.1. Introduction The CSIRO Division of Atmospheric Research has a number of research projects which use the facilities of the Cape Grim Baseline Air Pollution Monitoring station (jointly administered by the Bureau of Meteorology and CSIRO-DAR) in north-west Tasmania.

-7.50 VG-DAR method %0 (a) -7.60 -- B MAT-MAT method

-7.70 -.

-7.80 -.

Q B

-7.90 --

-8.00 00.50 90.75 91.00 91.25 91.50 91.75 92.00

-7.50 (b) VG-DAR method -7.60 -. Q MAT-DAR method

-7.70 --

-7.8- - 0

-7.90 --

-8.00 -4- 90.50 90.75 91.00 91.25 91.50 91.75 92.00 FIGURE 8. The Cape Grim in situ 013C isotopic record for the period of the mass spectrometer comparison VG602e Th . D data (represente soliy db de linesth d )an MAT252 data (squares) represen wee6 t4- k averages breae Th bot.n k i h records earln i rejectioyo t 199e du datf 1s no i a from that tim e8e period13CTh ) (a , records constructed using the CSIRO-DAR data reduction procedure for the VG602D data and the Finnigan MAT252 recommended ion correction procedure MAT25foe rth 5e 132 CTh data ) record(b , s constructed usin CSIRO-DAe gth R data reduction procedure for both data sets.

145 suce On h progra Cape th ms ei Gri msitn i u CC>2 stable isotope program whic bees hha n operating in its present form since 1982. About 3 mL of CC>2 is extracted cryogenically from air, stored in a 100 mL glass flask and transported back to DAR for analysis.

From 1982, the Cape Grim CC>2 samples were analysed on a VG602D stable isotope mass spectrometer. In August, 1990, a Finnigan MAT252 stable isotope mass spectromete installes rwa replaco dt VG602e eth CC>d Dan 2 samples were analyse botn do h instrument perioa r sfo ovef dyeare o ron .

When the data for this comparison period was analysed, there was clear disagreemen same th r e sampletfo s concerning both 5 013d 18C 0an e O 18.Th O record appears to converge during the period of the mass spectrometer comparison and is assigned to an instrumental effect (Figure 2(b)). The 513C record shows a significant constant offset, illustrate Figurn di e 8(a).

A number of potentially important factors, including sample storage time were considered, however, with normal turn-around time and procedures, these were too small concernf o te ob . The main cause for the disagreement was identified to be the different data reduction procedures recommended by the respective instrument manufacturers. Investigation of the data reduction procedures showed that systematic differences could be expected observeds a , thamagnitudd e ,th an t differencee th f eo dependens wa s e th n to differenc isotopin ei c compositio sample th f no e CC>2 fro reference mth e gas modifieA . d data reduction procedur establishes ewa applied dan boto dt h set dataf effece so Th f . o t procedurthiw sne shows ei Figuren i 8(b).

After describin differeno tw e gth t data reduction procedures date ,th a reduction procedure now in use is described.

2.2. General background date goae th a f Th reductioo l n procedur enablo t s ei e r!3e C/,th C ratiod an , r 18, the 18O/16O ratio, to be measured 13 12 . The ratios r!3 and r!8 in a sample gas are usually expressed by reference to the same ratios in a standard or reference gas using the 0-notation.

r!3c 013C = - 1 r!3R

r!8c - 1 r!8R The subscripts S and R to refer to ratios in the sample and reference gas respectively. We also use the lower case letters "s" and "r" to designate specific sample and reference gases as respectively, e.g. s!3, r!3, etc .... When used by itself, an "r" term without a complementar term" y"s , represent sample2 alse CO y o W rati.sa an reserv n oi uppee eth r

146 refe o rati a t calibrate cas a " o rt n oi e "R d reference materiale ,th sucr R13fo s h a 5 ,R4 absolute ratios in a calibration material such as NBS19.

1318 presence Duth o et isobarif eo c specie COn si 2, measure t S Sno d Ce Oan ar d directly. Instead, combinations of 12C, 13C, 160,17 O and 18O form three ion currents measured at mass-to-charge ratios (m/e) 44, 45 and 46, referred to as 144, 145 and 146 respectively.

12 16 144: C O2

13 16 12 16 17 145: C O2 ; C O O

12 16 18 12 17 13 16 17 146: C O O; C O2; C O O As 144 contains only the most abundant species, C and O, the ratio of the ion currents lese th s o abundant e du t species 146d an , wit5 i.e.14 h respec mose th to t abundan n tio 12 16 current, r46144d reportee an . ,ar 5 r4 s da

5 r4 145/145= 4

r46 = 146/144 6

2 moleculesymmetre Duth CO o e t th configurationf o ,yo ther tw e e2 ar CO r sfo containing one 17O or 18O atom. Combining the symmetry considerations, the combination 2 moleculeCO f so s contributin current relationshin e io th e d th 5 san go r4 t r pfo . 18 r combinations d a 6 an allow6 r4 7 1 expres r4 d o r ratiot e , d an s th s u 13 an f 5 ssr o s r4

2rl3rl7 + (rl7)2 8

Specifically, the quantities we measure for a sample CO2 gas are the difference in s45 and reference th f o 6 eexpresses r4 C(>ga d 28-notatione an th 5 n di r4 fro6 e s4 mth .

s45 845 = - 1 Lr45 J

0 1 1 -

2.3. The VG602D procedure

The VG602D data reduction procedure used at CSIRO DAR has been described in detail elsewhere briefl[11]s i t ,bu y summarised here.

The VG602D mass spectrometer has a two Faraday cup collector system which measures directly the ratio of 145 to 144 to provide 845 but measures the ratio of 146 to the sum (144 +145) to provide 846. A "slit-correction" is applied to convert this measured "846" int truoa e 846.

147 The relationship between 17O and 18O in the sample and reference gases is the equilibrium expression given by Craig [12].

s!7 r!7 samplr Fo wits ega h isotopic composition referencclose th o et e gas wits a , h atmospheric samples, thi approximates si followss da . 2 1 =518/7 §1 2 Also approximat e useth s di e relationship give Mooy n b Groote d kan s [13r fo ] convertin int6 g omeasure04 05d d 1813OCan an . 5 d04

R45545 R17S46 = ———— - ———— ij 3 R1 R13

4 1 - ———— l 6 =54 — SyuJ l R45R17] L R18 J MAT252The 2.4. procedure The MAT252 has discrete collectors for each ion beam (144, 145 and 146) and no "slit-correction necessarys "i date aTh . reduction procedure use methobasee s di th n do d of Santrock et al. [14] which differs considerably from the VG602D method in implementation. The relationship between r!7 and r!8 is expressed as

r!7 = &(rl8)a, 15

where a = 0.516 and k = 0.0099235. This compares with the expression used for the VG602D (Equatio wher) 1 [r!7/(rl8)n1 = ek 0.5value = use k a .Th f ed o a]an (0.0099235) represents the hypothetical VPDB material. Equatio combines i 5 n1 d with Equationgiv o followint e e8 th d an s7 g expression -3£2(sl8)2ct + 2fcs45(sl8)a + 2sl8 - s46 = 0, 16

whic solves hi d numericall relationshie Th giv. o yt 18 es Equatiof po uses i provid5 o dnt 1 e s!7, and s!3 is calculated from Equation 7. These are then expressed as 513C and 518O using Equation. 2 d an s1

2.5. Common conversion

Two common correction applie e 5 5d 18se 13ar OCth an o dobtainet d usine gth 13 18 VG602D and MAT252 methods. The first converts the ion corrected 5 CS and 5 OS to the international VPDB material via 513CR and 518C>R measured for the reference gas.

13 13 13 13 13 0 C(VPDB) = 5 CS + 5 CR + 0.001(5 CS0 CR) 17

Î8 18 18 18 18 Ô O(VPDB) = 5 OS + 5 OR + 0.001 (6 OS5 OR) 18

148 A second correction is applied to correct for the presence of N2O which is removed cryogenically wit COe hth 2.

13 13 S C(N20) = S C(VPDB) + 0.233(N20/C02) 19

18 I8 0 0(N2O) = 0 0(VPDB) + 0.338(N20/C02) 20

N2O is the concentration of N2O in parte per billion (ppbv) and CO2 is the concentration partn i millior 2 spe ofCO n (ppmv) constante Th . s 0.23 0.33d 3dependene an 8ar t upoe nth

relative ionisatio havd an e n bee efficiencieCO d n determinean O N f so d usine gth 2 2 VG602D [3]. The corrections to the VPDB scale and for N2O are applied identically to both VG602 MAT25d Dan 2 data. 2.6. Comparison of VG6Q2D MAT252and data To compare the two data reduction procedures a set of input data was generated and "corrected procedures"o usintw e g th inpu e resulte Th .t th datpresentee d sar a an dn i Table II.

similare ar 6 difference 04 ,th Wherd an valuee 5 eth betwee 64 f so n data obtained using the two methods is small. However, when the 045 and S46 are dissimilar, either of opposite sig witr no h appreciable differenc magnituden ei difference ,th e becomes quite large.

Using values for 845 and 646 of -l%o and 13%o, typical for Cape Grim CO2

analysed agains major tou r reference gas observee ,w ddifferenca calculaten ei d S Cf o

13 -0.05%o and OS18 of +0.05%o. TheC 8 difference is almost exactly that observed as the

TABL . ComparisoEII resulte th f no s obtaine correctiodn froio e mth n procedures user dfo data fro VG602e mth MAT25d Dan 2 mass spectrometers milr . pe valuel ,%c n i Al . e sar equale Wher ar differenc e 6 ,th magnitud e 64 eth d an inpue sig5 d th betwee f et64 nan o e nth two method smalls si largr Fo .e difference magnitudn si e and/or sig differencee nth e sar appreciable. Values given in the final row are typical for atmospheric air measured against our working reference gas ST03.

Input VG602D MAT252 difference 813p S18O 6 64 5 64 ° ^VPDB ° ^VPDB " CVPDB o ÜVPDB u ^VPDB u WVPDB

-20.000 -20.000 -20.676 -19.956 -20.701 -19.976 -0.025 -0.020 -10.000 -10.000 -10.338 -9.978 -10.351 -9.988 -0.013 -0.010 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 10.000 10.000 10.338 9.978 10.353 9.988 0.015 0.010 20.000 20.000 20.676 19.956 20.708 19.975 0.032 0.019

-20.000 20.000 -22.028 19.956 -22.204 20.071 -0.175 0.115 20.000 -20.000 22.028 -19.956 22.211 -20.070 0.183 -0.114 -1.000 13.000 -1.507 12.972 -1.560 13.018 -0.053 0.046

149 046 = -50,. (a) S13Cvs. 045

046 = 0

(b) Ô180vs.ô46

546

-0.3-L FIGUR . ECalculate9 d differences betwee originae nth l VG602 MAT25d Dan 2 methods plotted as (a) S13C vs. 045 for various values of 846, and (b) 518O vs. 546 for various value solie 045f Th dso . squar eacn eo h plot represent typicasa l sampl atmospherif eo c CÜ2 measured against our working reference gas ST03. The expected differences are about -0.05 +0.0%d oan 5 5 d 18%6r O13an o Cfo respectively.

discrepancy in the Cape Grim in situ record (see Figure 8). The Cape Grim in situ 018O record is not sufficiently precise over the comparison period to identify the expected difference. Other calculations showe difference dth e betweemethodo tw systemati e e b nth o s t c as demonstrated in Figure 9.

2.7. Origin of discrepancy The discrepancy arises from the different values for the isotopic ratios of the reference material in each data reduction procedure.

150 TABLE III. Compariso 5f n13o C, 5 isotopi18d Oan c hypothetica e ratioth r sfo l VPDB CO2, CSIRO-DAe th R reference CO2, HC453 NBS1e th d 9,an VPDCOr NBS1Fo 2d -B an e 9th ratio calculatee sar CC>r dfo 2 generated fro carbonate mth ephosphori % usin 0 g10 c acit da 25°C. All isotopic ratios are in per mil, %c.

V-PDB HC453 NBS19

13 5 CvpDB 0.000 -6.396 1.95 18 o OvpDB 0.000 -13.176 -2.20 r!3 0.01123720 0.01116533 0.01125911 7 r! 0.0003808033 0.0003782863 0.0003803842 r!8 0.002088349 0.002060833 0.002083755

For the ion correction we need to consider two reference gases, VPDB CO2, the hypothetical international reference gas, and HC453, the CSIRO DAR reference CO2. The isotopic compositions use thesr dfo e gase eacn si h method presentee sar Tabln di e III. Significant differences are apparent which arise as a consequence of three factors: probleme Th 1. s associated with assignin isotopie gth c compositio VPDf no B CO2, assigned by back calculation from NBS19. assumee Th . d2 relationship betwee18d O.an O n17 3. Which of the two data reduction procedures, specifically the a of Equation 15, is most appropriate. removeo T least a r to , minimise influence ,th thesf eo e factor have sw e developeda approacw ne reportine th o ht isotopif go c measurements.

2.8. Reporting of isotopic measurements Our method for reporting isotopic measurements is to relate all sample

measurements back to the hypothetical VPDB CO2 as 845 and 046 measurements. This means thacalculatioe th t assumptioe 8f th 5n 13d o 18 d COan an , n abou relationshie tth p between 17O and 18O, is required only once, and then only to report final results using the VPDB scale.

Measurements of the isotopic ratios in a sample are made against a reference material calibrated against NBS19 acceptee ,th d international reference material which defines the VPDB scale. The sample analysis and reference calibration measurements are all measured as 045 and 646 so it becomes a simple procedure to construct the 545 and sample th f o e 6 wit54 h respec NBS19o t exampler Fo . sampla f ,i e (SA analyses )i d agains referencta (WSs ega givo )t e 545(SA-WS 546(SA-WS)d bees an ) ha nS W d an , measured directly against NBS1 givo 9t e 545(WS-NBS19 S46(WS-NBS19)d an ) e ,w evaluate 045(SA-NBS19 046(SA-NBS19d )an followss a ) .

545(SA-NBS19) = (545(SA-WS) + 1)(045(WS-NBS19) + 1) 21

546(SA-NBS19 (546(SA-WS)= 2 2 1)(046(WS-NBS19)+ ) 1 )+

151 defin6 04 ed Thesdirectlan 5 eisotopi64 e yth c compositio sample th f no e against VPDB as the NBS19 reference material defines the VPDB scale. Note the 8s are not in units of per mil. To report isotopic composition as 613C and 018O it is necessary to perform ion corrections, which requires that we: . 1 Defin absolute eth e isotopic abundanceVPDBn i O 18 .d 13f an so C O , 17 . 2 Assign isotopic compositio NBS1o nt 9 relativ VPDBo et . . 3 Defin relationshie e th sampl e th n i e O poolp 18 betwee d . an O n17 This procedur describes ei Annedetai e dReporn i e th th n li xo Workinf t to g Group 2, in this volume [15]. As indicated in the Annex, the data reduction procedure has been tested by a number of laboratories and identical results have been obtained in all cases.

Par Summar: t3 acknowledgmentd yan s

An overview of the CSIRO DAR stable isotope program has been presented. The use of high precision stable isotope data in establishing the various contributions of the biosphere and oceans as sources and/or sinks for atmospheric CC>2 has been illustrated. problemSome th f eo s associate maintaininn di ghigh-precisioa n stable isotope measurement program have been identified. Strong emphasis has been placed on the data reduction procedure used to present results.

GASLAB was established and is operated by CSIRO with significant financial support from the Australian government departments with responsibility for the environment, science, energy/industry and foreign affairs, and with additional support from private industry. Sample collection is almost entirely dependent upon collaborative agreements with many agencies and universities (especially the Bureau of Meteorology and the Antarctic Divisio Australian i similad )an r agencie USAe th n s,i Canadaw IndiNe , d ,UK a an Zealand. International suppor somr tfo e aspect GASLAB'f so s operatio coms nha e from TIGEK U e RIAEe th programth d Aan .

References

[Î] This paper is based on two talks presented to the International Atomic Energy Agency Consultants' Meeting on Stable Isotope Standards and Intercalibration, Vienna, Austria, 1 -3 December, 1993.

[2] GOODMAN, H.S., "The 13C/12C ratio of atmospheric carbon dioxide at the Australian baseline station, Cape Grim "Carbon "i n Dioxid Climatd ean eAustralia: n Research" Edited by G.I. Pearman, Australian Academy of Science, Canberra, (1980) 111-114.

[3] FRANCEY, R.J., GOODMAN, H.S., "Systematic error in, and selection of, in situ 513C", Baseline 1983-84, (Eds. RJ. Francey and B.W. Forgan), Department of Scienc eBurea- Meteorologf uo CSIRd yan O Diviso Atmospherif no c Research, Australia (1986) 27-36.

152 ] [4 ALLISON, C.E., FRANCEY, R.J., LANGENFELDS, R.L., WELCH, E.D., "High precision compariso Capf no e Grim CC>2 stable isotope measurements usino gtw mass spectrometers", Baseline 1991 (Eds. C. Dick and J.L. Gras), Department of Administrative Services and CSIRO (1994) in press.

[5] FRANCEY, R.J., TANS, P.P., ALLISON, C.E., ENTING, LG., WHITE, J.W.C., TROLIER , "Change,M. oceanie th terrestrian sd i can l carbon uptake since 1982", Nature (submitted).

] [6 FRANCEY, R.J., ALLISON, C.E., ENTING, LG., BEGGS, H.M., TILBROOK, ,B. "The observed distribution of 013CO2 in the atmosphere". Presented at the 4th Internationa conference2 lCO , Carquieranne, France, September n 1993i d an , preparation for publication.

] [7 ENTING, LG., FRANCEY, R.J., TRUDINGER, C.M., GRANEK synthesiA " , H. , s inversio concentratioe th f no 5 f atmospherin13d o Can c CO2". Tellus (1994 pressn )i .

] [8 MOOK, W.G., BOMMERSON, J.C., STAVERMAN, W.H., "Carbon isotope fractionation between dissolved bicarbonate and gaseous carbon dioxide". Earth Planet. Sei. Lett. 22 (1974) 169-176.

[9] LASSEY, K.R., LOWE, D.C., BRAILSFORD, G.W., GOMEZ, A.J., BRENNINKMEIJER, C.A.M., MANNING, M.R., "Atmospheric methane in the southern hemispher recene th e: t declin sourcn ei e strengths inferred from concentratio isotopd nan Wastd ean data"r e ManageAi . ,J . Assoc. (1994 pressn )i .

[10] ENTING, LG., TRUDINGER, C.M., FRANCEY, R.J., GRANEK, H., "Synthesis inversio atmospherif no usin2 GISe cgCO th S tracer transport model", CSIRO Divisio Atmospherif no c Research Technica l(1993 9 Pape2 . pages5 r)No 4 . [11] FRANCEY, R.J., GOODMAN, H.S., "The DAR stable isotope reference scale for CO2", Baseline 1986 (Eds. B.W. Forgan and PJ. Fraser), Department of Science and CSIRO, Australia (1988) 40-46.

[12] CRAIG , "IsotopiH. , c standard carbor sfo oxyged nan correctio d nan n factorr sfo mass spectrometric analysis of carbon dioxide", Geochim. Cosmochim. Acta 12 (1957) 133-149.

[13] MOOK, W.G., GROOTES, P.M., "The measuring procedur correctiond ean e th r sfo high precision analysi isotopif so c abundance ratios, especially referrin carbono gt , oxygen and nitrogen", Int. J. Mass Spectrom. Ion Phys. 12 (1973) 273-298.

[14] SANTROCK, L, STUDLEY, S.S., HA YES, J.M., "Isotopic analyses based on the mass spectrum of carbon dioxide", Anal. Chem. 57 (1985) 1444-1448.

[15] ALLISON, C.E., FRANCEY, RJ., MEIJER, H.A.M., "Recommendatione th r sfo reportin stablf go e isotope measurement carbof so oxygegas"d 2 nan CO , Annen i x Recommendatione th o t Workinf so g Grou Dat: p2 a Reduction Procedures, presented at the International Atomic Energy Agency Consultants' Meeting on Stable Isotope Standard Intercalibrationd san , Vienna, Austria December3 - 1 , , 1993, this volume.

153 RECOMMENDATIONS FOR THE REPORTING OF STABLE ISOTOPE MEASUREMENTS OF CARBON AND OXYGEN IN CO2 GAS [1]

C.E. ALLISON, R.J. FRANCEY CSIRO Divisio f Atmospherino c Research, Mordialloc, Victoria, Australia H.A.J. MEIJER Centr Isotopr efo e Research, Groningen University, Groningen, Netherlands

Abstract : Recommendations are made for the reporting of stable isotope measurements of carbon and oxygen in CO2 gas. The isotopic composition of the hypothetical Vienna PDB calcite and CO 2 gas are determined from the primary standard calcite NBS19. An ion correction procedure describedis which eliminates many errors ofthe which arisecan during inter-calibration exercises. Data presentedare which allow implementations ofthe data reduction procedure to be tested.

1. Introduction

Measurement stable th usef e eso ar globaisotope n di 2 CÛ l f carboso n budget analyse constraio st n source sinkd atmospherisr an sfo [2,3]2 cCÛ . Wit requiremenha r tfo high precision in both spatial and temporal data, there is a need to report atmospheric 0 CO2 in a way that permits comparison of different collection sites (spatial) to a precision 13 of around 0.01%o betterr o , comparisod ,an lone th gf ntero m behaviour (temporala o )t precision of 0.005%o or better, with only slightly relaxed requirements for 618O. However, difficulties arise when comparin isotopi2 gCO c measurements fro face mth t thae tth mostt quantitiebu 6 laboratorie04 d s measurean 5 84 s e e repordth ar r 8d fo 18t O0an 13C CO2. The reported quantities are not directly comparable between laboratories due to differen correction io t n algorithms and/or different assumptions abou isotopie tth c composition of the primary standards. Uncertainties in ion correction algorithms, in reference gases and the potential uncertainties due to inadequate calibration or correction of measurements currently prevent comparisons between different laboratorie thio st s precision.

In the absence of universally accepted protocols, the following data reduction procedure offers a means of minimizing the uncertainties.

2. General background

The goal of the data reduction procedure is to enable r!3, the 13C/12C ratio, and r!8 18e O/ ,th usuallsampl e a 16 ar On i s ratiomeasured8 e yga b r! o d t , ratioe an 3 Th .s r! expresse referency db same th eo et ratio standara n si referencr do usins ega e gth 5-notation.

155 Ô13C = - 1 r!3R

8180= - 1 r!8R

The subscripts S and R refer to ratios in the sample and reference gas respectively. We lowee th als designate o rt o casus " "r e d letterean specifi" s"s c sampl referencd ean e gases respectively, e.g. s!3, r!3 ...c ,.et When use ter itself" y d"r mb n withoua , ta complementar term" y"s , represent CC>y ratisa an 2 on samplei alse W o. reserv uppee eth r case "R" to refer to a ratio in a calibrated reference material, such as R13, R45 for the absolute ratios in a calibration material such as NBS19. We use the upper case "S" to refer to the hypothetical VPDB material.

Due to the presence of isobaric species in CO2, 513C and 618O are not measured directly. Instead, combinations of 12C,13 C, 160, 17O and 18O form three ion currents measure mass-to-chargt da 6 14 referre, ed ratio 46 an 144 s d a 5 o san d,14 t (m/e5 4 , )44 respectively.

12 16 144: C O2

13 16 12 16 17 145: C O2; C O O

12 16 18 12 17 13 16 17 146: C O O; C O2; C O O As 144 contains only the most abundant species, C and O, the ratio of the ion currents due to the less abundant species, i.e. 145 and 146 12, with respec 16 t to the most abundant ion current, 144r46d reportee an ,.ar 5 r4 s da

3 r4 145/145= 4

4 r4 146/146= 4 symmetre th CC>o e t th e 2f y o moleculeDu configurationo , thertw e ear C(>r sfo 2 containing one 17O or 18O atom. Combining the symmetry considerations, the combinations of CÛ2 molecules contributing to the ion currents and the relationship for r45 and r46 allows us to express r45 and r46 as combinations of the ratios r!3, r!7 and r!8.

r45 = r!3 + 2rl7 5

r46 = 2rl8 + 2rl3rl7 + (r!7)2 6 Specifically, the quantities we measure for a sample CC>2 gas are the difference in s45 and s46 from the r45 and r46 of the reference gas expressed in the o-notation.

156 045 =

8 r46

Isotopic ratios are often expressed in units of "per mil", %0, obtained by multiplying the 6-values above by 1000.

3. Primary standard

The assignment of isotopic ratios to the primary CC>2 standard is crucial. As almost l measuremental CX>f so reporte e 2ar originae scale th B dmateria B d usinPD ,an lPD e gth l longeo n s i r non-existen e availableth e us e ,w t Vienna-PD primarr ou Bs a y standarde W . use NBS19, with IAEA recommended values of 6 C = +1.95%o and S O = -2.20%o 13 18 relativ hypotheticae th o et l Vienna-PDB(VPDB reference ) [4]th s ,a e material through which measurements are related to VPDB.

NBS19r fo 6 neee R4 ,w firscalculat o o do t t d T t8 an assig5 S1 d eR4 nan S137 S1 , VPDB Craig' e calciteus e s W measurement. originae th materia B f sassigo o t lPD ] n[5 l r!3, r!7 and r!8 for CC>2 from the original PDB material.

r!3 = 0.0112372 0.0003799= 7 r! 5 rl 8 = 0.002079

To assign S13 to VPDB, we assume that the ratio of C to C in VPDB calcite is identica thao lt t assigne originae th materia B o dt lPD Craigy lb 0.0112372= , 3 i.eS1 . . 13 12 ratiO VSMO16 on i o t O 18 W e (Vienn th VPDBassigo e o t T us 8 e naS1 ,w Standard Mean Ocean Water 0.0020052f IAE)o e th Ad recommendean ] 0[6 d valu 8r efo f O o 18 VPDB with respect to VSMOW of+30.9%o [4]. This gives S18 = 0.002067160680.

To assign S17 to VPDB, we assume the ratio between S17/S18 in VPDB is related to the ratio rl7/r!8 in PDB by the equilibrium approximation of Craig [5]. S17/rl7 = [S18/rl8]°-5 9 thie sus approximatione W e calculatee th ,th n i 8 r! dd valu ratioe an th S18f 7 eo d s r! ,an original PDB CO2 to calculate S17 = 0.000378866601. To assign R13, R17 and R18 to NBS19 calcite we use the IAEA recommended 13 18 values for 5 C and S O (+1.95%c and -2.20%0) and assume the rl7/r!8 equilibrium approximation describes the relationship between the 17O/18O ratios in VPDB and NBS19.

157 We summarise the stable isotope ratios for the calcites as:

VPDBcalcite 0.0112372000= 3 81 : 0 S1 0.0003788666017= 0 818 = 0.002067160680, NBS19calcit =0.01125911253 R1 e: 4 R17= 0.0003784496180 R18 = 0.002062612927.

There is no change in r 13 in converting the calcite to CO2, however we need to appl fractionatioe yth n facto f 1.01025ro fractionalioO r ,fo n durin evolutiogCO t na 18 2 25°C with 100% phosphoric acid [7], and the equilibrium approximation to calculate the rl7andr!8.

0.0112372000= 3 81 VPD : 2 B0CO 81 0.0003808033427= 0 0.00208834907= 8 81 7

NBS19CO 0.0112591125= 3 R1 2: 4 R17 = 0.0003803842280 R1 0.002083754708= 9

For the data reduction procedure we need to calculate 845 and 846 for VPDB, R45 and value6 §4 NBS1r d sfo an NBS19r 5 9fo wit 84 6 d hR4 ,an respec VPDBo t .

0.0119988066= 5 84 VPD: 2 B9CO 846 = 0.004185401492

0.0120198810= 5 R4 NBS1 : 2 90CO R4 0.004176219686= 8

§45(w.r.t. VPDB +1.756367272%)= 0 §46(w.r.t. VPDB) = -2.193768974%0 note W e thaprecisioe th t n quoted justifiet herno s e i measurement y db importans i t i t ,bu t to assign a high precision to "accepted values" for the primary standard to avoid propagatio errorsf no have W significan.e0 1 chose e us no t t figures.

4. Sample reporting

The ratios r45 and r46 in a CO2 sample can be determined accurately, without assumptions about ion correction algorithms, from measured §45 and §46 values [1]. The expression accuratelr sfo samplya 845,84 e relatinn i th 6 o e t r45 r4 e VPDg,6n th i a Bvi §45 and §46 measured for the sample gas against the primary working standard (indicated by the subscripts i) are given below.

158 845NBS19 0 1 . (»«UfÏ .. 1 A + - + - 5 S4 r45H 1 , -= s ) I looV ; Vloo) o looo o , ,... , = 6 S4 + + + ; r46v looso ; loov V J looo o

expressione Th s accommodat numbeea possiblf ro e intermediate pur2 referenceCO e gases (indicated by subscripts /, i-1,...) without compromise of the isotopic composition. The terms containing 845NBS19 and Ô46NBS19 refer to the calibration of a reference gas directly evolve s NBS1o t ga 2 25°t da 9 CO C using 100% phosphoric acid. Until such tim consistens ea t protocol widele sar y accepted reportine ,th f go isotopic composition for precise inter-laboratory comparisons should provide the 845 and §46 measured for the sample(s) along with similar information relating to the reference gas(es) including calibratio primare th o nt y standard NBS19. Wher correction eio s ni performe thesd dan e result presentede sar recommene ,w correction io e dth n algorithms used be described clearly. Further, for ion correction, we recommend the use of the procedure described in the following section.

correction Io . n5

In atmospheric CO2 measurements, the ion corrections normally required are

(1) correction for the co-extraction of N2O with CO2, (2) conversion of the measured 845 to 813C by correction for a 12C17O16O contribution to m/e 45, and 18 13 17 1216 17 ) conversio(3 measuree th 8 o f t n o correctioO y 6 b d84 r Cnfo Od COan O2 contributions to m/e 46.

The correction for N2O affects mainly the m/e 44 ion current and has been described by Moo . [8]addrese al t .W k e s here) onl(3 yd . iteman ) s(2

The main difficulty in performing ion corrections lies in correcting for the presence of 17O. The method described here is an exact calculation of r!3, r!7 and r!8 from the 545 and 846 measured for a sample. The only approximation which is required is the relationship betwee rl7/r!e nth 8 sampl e ratith referencd on i ean e gases relationshie Th . p equilibriue th s i e us me w expressio Craif no g (Equatio. n9)

5 rl7S/r!7 - R[rl8 S/rl8R]°- Santrocsugges] [9 . al t k e usin differenga t relationship base usinn do exponenn ga t 0.51f o ) 6 (a instea (squar5 0. f do e root relationship)5 0. recommen e = W .a f o e dus because of its simplicity and theoretical basis but note that at the precision levelc required for some studies very real variations in the O/ O ratio are likely [10, 11]. We therefore recommend that where specific deviations 17fro above mth 18 e relationshi chosee par n that they documentee b individuan a n do l basis.

159 whic6 54 TVd h ,havan perfor correction5 n eio §4 bee e e mth nus e converte,w do t afte6 r4 rd correctioan 5 r4 VPDo nt B CC> describes 2a d above using. Equation11 d an 0 s1 correction io detail e e Th th f sno method have been described previously [12].

We rearrang givo et Equationquadratie9 a d an 6 r!7 , n ci s5 . 2 1 (2S18/(S17) 3)(rl7)2- 0 2r45rl 2+ = 6 r4 7- This is then solved for r 17 as

-B + B - 4AC _ !3 2A wher 2S18/(S17)-r46= = e A theC t 2r4 I .= d nB 5 an , follow 3 2- s that

2rl7- d 5 r4 ,an = 3 r!

(r4= 68 2rl3rl- r! 7(r!7)- 2)/2.

Thes thee ear n converte 8 o d613t d 18COan .

813C = (r!3/S13 - l)1000%o 14

5 1 818 O(rl8/S18-l)1000%= o

6. Advantages of the data reduction procedure This data reduction procedure has a number of advantages:

quantitiee Th . 1 s measure experimente th 546d n i usee an , ar relat o d5 t isotopie 54 , eth c composition of the sample directly to the isotopic composition of the hypothetical VPDB. Direct comparison of 545 and §46 against VPDB means that measurements from different laboratories can be compared directly without the complication of differen correction io t n algorithms.

. 2 This method avoids problems whic arisy hema from using intermediate C(>2 working standards which have been produce processey db s whic involvy hma e O/ O 17 18 fractionations which are determined by different values of a.

correction io e Th n . procedur3 e recommende easile b n y exacs dca i change d tan f di modification to the isotopic ratios in the primary VPDB standard are required, or if change 6-valueo st s assigne referenco dt e gase madee sar .

160 7. Data reduction test data To aid in the implementation of this data reduction procedure a set of test data is provided. The lest data consists of the following: workinA . 1 g (WSstandar2 bees CO )s ha ndga directl y measured against NBS19 CC>2 (evolves ga 25°t da C with 100% phosphoric acid -7.0000% = hav o )t 5 e64 d can 046 = -1.0000%«.

2. A series of samples have the following S45 and 046 (both in %o) when measured agains: tWS

6 54 5 §4 Sample 1 -20.0000 -20.0000 2 -20.0000 0.0000 3 -20.0000 20.0000 4 0.0000 -20.0000 5 0.0000 0.0000 6 0.0000 20.0000 7 20.0000 -20.0000 8 20.0000 0.0000 9 20.0000 20.0000.

The data reduction procedure described above produces the following results (all in %o045r )fo , §46, § §13d 18Can O :

Sample §45 §46 §45(VPDB) §46(VPDB) §13C §18O

1 -20.0000 -20.0000 -25.1508 -23.1277 -26.0681 -23.0985 2 -20.0000 0.0000 -25.1508 -3.1916 -26.7489 -3.1402 3 -20.0000 20.0000 -25.1508 16.7446 -27.4230 16.8184 4 0.0000 -20.0000 -5.2559 -23.1277 -4.8233 -23.1415 5 0.0000 0.0000 -5.2559 -3.1916 -5.5042 -3.1836 6 0.0000 20.0000 -5.2559 16.7446 -6.1782 16.7746 7 20.0000 -20.0000 14.6390 -23.1277 16.4214 -23.1845 8 20.0000 0.0000 14.6390 -3.1916 15.7406 -3.2270 9 20.0000 20.0000 14.6390 16.7446 15.0665 16.7307.

These calculations have been teste verified d numbean a y db independenf ro t groups.

161 References

[ l ] Based on FRANCEY, R.J., ALLISON, C.E., "Recommendations for the measuremen reportind tan stablf go e isotope measurement carbof so oxyged nan n i atmospheric trace gases," Annex 4 to the 1st Research Coordination Meeting of the Coordinated Research Programme on Isotope Variations of Carbon Dioxide and Other Trace Gases in the Atmosphere, 2-5 November 1992, Heidelberg, Germany, (K. Rozanski, Scientific Secretary). ] [2 ENTING, LG., TRUDINGER, C.M., FRANCEY, R.J., GRANEK , "SynthesiH. , s inversio atmospherif no c CC»2 usin GISe gth S tracer transport model", CSIRO Division of Atmospheric Research Technical Paper No. 29 (1993) 45 pages. [3] ENTING, I.G., FRANCEY, R.J., TRUDINGER, C.M., GRANEK, H., "A synthesis inversio concentratioe th f no 0d 13 nan atmospheriCf o c CC^", Tellus (1994 pressn )i . [4] HUT, G., Consultants' Group Meeting on Stable Isotope Reference Samples for Geochemica Hydrologicad lan l Investigations; IAEA, Vienna» Austria; 16-18 September, 1985 (1987). [5] CRAIG, H., "Isotopic standards for carbon and oxygen and correction factors for mass-spectrometric analysis of carbon dioxide", Geochim. Cosmochim. Acta 12 (1957) 133-149. [6] BAERTSCHI, P., "Absolute 18O content of standard mean ocean water," Earth Planet. Sei. Lett. 31 (1976) 341-344. ] [7 FRIEDMAN O'NEIL, L , , J.R., "Compilatio stablf no e isotope fractionation factorf so geochemical interest," U.S. Geological Survey Paper 440-KK (1977).

[8] MOOK, W.G., VAN DER HOOK, S., "The N2O correction in the carbon and oxygen isotopic analysis of atmospheric CO2," Isotope Geoscience 1 (1983) 237- 242. [9] SANTROCK, J., STUDLEY, S.S., HAYES, J.M., "Isotopic analyses based on the mass spectrum of carbon dioxide," Anal. Chem. 57 (1985) 1444-1448. [10J BHATTACHARYA, S.K., THIEMENS, M.H., "New evidence for symmetry dependent isotope effects: O+CO reaction," Z. Naturforsch. 44a (1989) 435-444. [11] THIEMENS, M.H., JACKSON, T., MAUERSBERGER, K., SCHUELER, B., MORTON , "Oxyge,J. n isotope fractionatio stratospherin ni c CO2," Geophys. Res. Lett. 18(1991)669-672. [12] MOOK, W.G., GROOTES, P.M., "The measuring procedur correctiond ean e th r sfo high-precision analysi isotopif so c abundance ratios, especially referrin carbono gt , oxygen and nitrogen," Int. J. Mass Spectrom. Ion Phys. 12 (1973) 273-298.

162 LIST OF PARTICIPANTS

Allison, G.B. CSIRO, Divisio f Wateno r Resources, Waite Road, Urrbrae, SA , Mail, 2 :. PrivatNo g eBa Glenn Osmond 5064A S , , Australia de Bièvre, P. Central Bureau for Nuclear Measurements, Commission of the European Communities, JRG, B-2440 Geel, Belgium

Coplen, T.B. United States Department of the Interior, Geologcial Survey, Reston0922 2 ,A UniteV , d State f Americso a

Exley. R , VG Isotech, Aston Way, Middlewich, Cheshire CW10 OHT, United Kingdom

Friedrichsen. H , Institut für Mineralogie, Freie Universität Berlin, Boltzmannstrasse 18-20, D-14195 Berli , Germann33 y

Gonfiantini. R , Isotope Hydrology Section, Division of Physical and Chemical Sciences, International Atomic Energy Agency, Wagramerstrass , P.Oe1005 x Bo . , A-1400 Vienna, Austria

Halas. S , Institute of Physics, Maria Curie-Sklodowska University, Plac Marii Curie-Sklodowskiej 1, PL-20-031 Lublin, Poland

Klouda, G.A. Atmospheric Chemistry Group, National Institute of Standards and Technology, B364, Building 222, Gaithersbug, MD 20899, United States of America

Krouse, H.R. Department of Physics and Astronomy, The University of Calgary, Albert 1N4N aT2 , Canada

Longinelli, A. Inst. di Mineralogia e Petrografia, Piazzale Europa, I-134100 Trieste, Italy

Meijer, H.A.J. Centrum voor Isotopen Onderzuek, Nijenborg , NL-974h4 GroningenG 7R , Netherlands

Neubert, R. Institut für Umweltphysik der Universität Heidelberg, Im Neuenheimer Feld 366, D-69120 Heidelberg, Germany

163 Pichlmayer. F , Österreichisches Forschungszentrum Seibersdorf, Seibersdor , A-224f 4 Seibersdorf, Austria

Presser, S.J. Europa Scientific Limited, Europa Hause, Electra Way, Crewe, Cheshire CW l 1ZA, United Kingdom

Robinson, B.W. Institut f Geologicaeo Nuclead an l r Sciences, Gracefiel0 3 d Road, P.O 31312x Bo . , Lower HuttZealanw Ne , d

Rozanski, K. Isotope Hydrology Section, Division of Physical and Chemical Sciences, International Atomic Energy Agency, Wagramerstrass , P.O e5 100x Bo ., A-1400 Vienna, Austria

Stichler . W , Isotope Hydrology Section, (Scientific Secretary) Divisio f Physicano Chemicad an l l Sciences, International Atomic Energy Agency, Wagramerstrasse 5, P.O. Box 100, A-1400 Vienna, Austria

Trimborn, P. GSF-Institu Hydrologier fü t , Ingolstädter Landstraße l, D-85764 Oberschießheim, Germany

Verkouteren, R.M. Atmospheric Chemistry Group, National Institut f Standardeo Technologyd san , B364, Building 222, Gaithersburg 20899D M , , United State f Americso a

Vocke, Jr.. ,R Atmospheric Chemistry Group, National Institute of Standards and Technology, B364, Building 222, Gaithersburg 20899D M , , United State f Americso a

164 IAEA REPORT PUBLICATIOND SAN S ON STABLE ISOTOPE STANDARDS AND MEASUREMENT INTERCOMPARISON

E. Halev B.Rd yan . Payne, 1967- Deuteriu Oxygen-1d man naturan 8i l waters: Analyses compared. Science, 156, 669.

GonfiantiniR , 197 7 Repor- Consultantse th n o t " Meetin Stabln go e isotope standardd san intercalibratio hydrolognn i geochemistryd yan , Vienna, 8-10 September 1976. IAEA, Vienna

R Gonfiantini, 1978 - Standards for stable isotope measurements in natural compounds. Nature, 271, 534-536.C

GonfiantiniR , 198 4Repor- Advisore th n o t y Group Meetin Stabln go e isotope reference sample geochemicar sfo hydrologicad an l l investigations, Vienna, 19-21 Septemer 19837 7 . pp. IAEA, Vienna G. Hut, 1987 - Report on the Consultants' Meeting on Stable isotope reference samples for geochemica hydrologicad an l l investigations, Vienna, 16-18 Septembe r. IAEA 1985pp 2 4 ., Vienna

K. Rozanski, 199 1- ReporConsultantse th n o t ' Meetin C-1n go 4 Reference Materialr sfo Radiocarbon Laboratories, Vienna, 18-20 February 1981, 53 pp, IAEA-Vienna

K. Rozanski . StichlerW , GonfiantiniR , ScottM . BeukensE. , RP , n . Kromerva B ,. J d an , der Plicht, IAEA'e 199Th 2- 4C Intercomparison Exercise 1990. Radiocarbon, 506-51, 34 9 (includes results on 13C).

165 QUESTIONNAIRE ON IAEA-TECDOCs

It would greatly assist the International Atomic Energy Agency in its analysis of the effective- ness of its Technical Document programme if you could kindly answer the following questions return address and formthe the to shown below. Your co-operation greatlyis appreciated.

Title: Reference and intercomparison materials for stable isotopes of light elements Number: IAEA-TECDOC-825

1. How did you obtain this TECDOC?

] Fro[ IAEAe mth : requesn ow t tA ] [ ] Withou [ t request participans A ] IAE n [ a t Aa t meeting [ ] From a professional colleague [ ] From library

2. How do you rate the content of the TECDOC?

] [ Useful, includes informatio fount nno d elsewhere ] [ Usefu survea subjece s th a l f yo t area ] Usefu[ referencr fo l e [ ] Useful because of its international character [ ] Useful for training or study purposes [ ] Not very useful. If not, why not? becomu yo o d e awarw TECDOCe 3th Ho . f eo s available fro IAEAe mth ?

[ ] From references in: [ ] IAEA publications ] Othe [ r publications [ ] From IAEA meetings ] Fro[ m IAEA newsletters othey B r mean] [ s (please specify) [ ] If you find it difficult to obtain information on TECDOCs please tick this box

IAEA-TECDOCsf maku o e yo eo us D . 4 ?

[ ] Frequently ] Occasionall[ y [ ] Rarely

5. Please state the institute (or country) in which you are working:

Please return to: R.F. Kelleher Head, Publishing Section International Atomic Energy Agency P.O. Box 100 Wagramerstrasse5 A-1400 Vienna, Austria