conch: a visual basic program for interactive processing ... · computer s & geoscienc es 32 (2006)...

Computers & Geosciences 32 (2006) 1479–1498

CONCH: A Visual Basic program for interactive processing ofion-microprobe analytical data$

David R. Nelson!

School of Physical Sciences, Curtin University of Technology, GPO Box U1987, Perth, WA 6001, Australia

Received 30 September 2005; received in revised form 2 February 2006; accepted 7 February 2006

Abstract

A Visual Basic program for flexible, interactive processing of ion-microprobe data acquired for quantitative traceelement, 26Al–26Mg, 53Mn–53Cr, 60Fe–60Ni and U–Th–Pb geochronology applications is described. Default but editablerun-tables enable software identification of secondary ion species analyzed and for characterization of the standard used.Counts obtained for each species may be displayed in plots against analysis time and edited interactively. Count outlierscan be automatically identified via a set of editable count-rejection criteria and displayed for assessment. Standard analysesare distinguished from Unknowns by matching of the analysis label with a string specified in the Set-up dialog, andprocessed separately. A generalized routine writes background-corrected count rates, ratios and uncertainties, plusweighted means and uncertainties for Standards and Unknowns, to a spreadsheet that may be saved as a text-delimited file.Specialized routines process trace-element concentration, 26Al–26Mg, 53Mn–53Cr, 60Fe–60Ni, and Th–U disequilibriumanalysis types, and U–Th–Pb isotopic data obtained for zircon, titanite, perovskite, monazite, xenotime and baddeleyite.Correction to measured Pb-isotopic, Pb/U and Pb/Th ratios for the presence of common Pb may be made using measured204Pb counts, or the 207Pb or 208Pb counts following subtraction from these of the radiogenic component. Common-Pbcorrections may be made automatically, using a (user-specified) common-Pb isotopic composition appropriate for that onthe sample surface, or for that incorporated within the mineral at the time of its crystallization, depending on whether the204Pb count rate determined for the Unknown is substantially higher than the average 204Pb count rate for all sessionstandards. Pb/U inter-element fractionation corrections are determined using an interactive loge–loge plot of common-Pbcorrected 206Pb/238U ratios against any nominated fractionation-sensitive species pair (commonly 238U16O+/238U+) forsession standards. Also displayed with this plot are calculated Pb/U and Pb/Th calibration line regression slopes,y-intercepts, calibration uncertainties, standard 204Pb- and 208Pb-corrected 207Pb/206Pb dates and other parameters usefulfor assessment of the calibration-line data. Calibrated data for Unknowns may be automatically grouped according tocalculated date and displayed in color on interactive Wetherill Concordia, Tera–Wasserburg Concordia, LinearizedGaussian (‘‘Probability Paper’’) and Gaussian-summation probability density diagrams.r 2006 Elsevier Ltd. All rights reserved.

Keywords: Secondary ion mass spectrometry; SIMS; SHRIMP; Isotope analysis; Radiometric age; Geochronology; Data reduction; VisualBasic

ARTICLE IN PRESS

www.elsevier.com/locate/cageo

0098-3004/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.cageo.2006.02.009

$Code available from server at http://conch.perthshrimp.com!Tel.: +61 8 9266 3736; fax: +61 8 9266 2377.E-mail address: [email protected].

http://www.elsevier.com/locate/cageohttp://conch.perthshrimp.com

1. Introduction

Ion microprobes are now widely applied to thedetermination of trace-element concentrations, isotopicanalyses and U–Pb and Th–Pb formation ages of awide range of minerals (e.g. zircon, monazite, xenotime,titanite, perovskite, baddeleyite and galena). In prac-tice, this technique can generate vast quantities ofnumerical data that require complex processing inorder to extract useful information. The secondary ionspectrum formed during ion-microprobe analysis ofeven chemically simple minerals is commonly highlycomplex, and will usually include monomers, molecularand multiply charged species. Furthermore, the relativeefficiency of emission of secondary ions from thesample during ion-microprobe analysis commonlyvaries unpredictably during the analysis session,requiring the periodic analysis of standards of knowncomposition in order to correct for this instrumentallyinduced drift. Data processing therefore usuallyinvolves complex calibration and isobaric correctionsbefore the measurements can be converted into ameaningful result. The monitoring and sound inter-pretation of ion-microprobe data commonly requires ahigh degree of understanding of a large number ofcomplex parameters, yet a trend is emerging in whichnon-specialist users are required to acquire and procession-microprobe data under relatively little supervision.

CONCH is a computer program coded inMicrosoft Excel Visual Basic for Applications thatallows flexible, interactive processing of ion-microp-robe geochronology, isotopic and trace elementdata. The software runs on both Macintosh andIBM-compatible platforms and will operate mosteffectively using Excel ‘97 (Macintosh) or Excel2000 (Windows). A freeware (Linux OpenOffice)version is also in preparation. The algorithmsdeveloped are applicable to the processing of ion-microprobe data obtained for a wide range ofapplications.

The aims of this contribution are to:

1. outline some features of the CONCH applicationand aspects of the data reduction approach, and

2. briefly describe its application to the processingof ion-microprobe zircon, titanite, perovskite,monazite, xenotime and baddeleyite U–Th–Pb,and Th–U disequilibrium analyses.

CONCH can generate processed data reports (asspreadsheets) and output files (as tab-delimited textfiles) containing automatically or manually edited

background-corrected count rates, ratios and un-certainties and may therefore be applied to theprocessing of secondary ion mass spectrometry(SIMS) data obtained for a wide variety ofapplications. However, for the sake of brevity, thiscontribution outlines geochronological applicationsof CONCH; only limited discussion of otherapplications has been included herein.

Copies of the latest version of the program, a‘‘User’s Guide’’ that provides a comprehensivedescription of the software (including installationinstructions and Tables describing the function ofevery control) and example input files may bedownloaded from: http://conch.perthshrimp.com orobtained on request from [email protected]

2. Input and output file formats

CONCH is currently adapted to read tab-delimited text data files such as those output bythe Sensitive High-Resolution Ion MicroProbe(SHRIMP), but the code could be adapted toprocess single-collector ion-microprobe data filesin other input formats. SHRIMP output (raw data)files, as well as CONCH-processed data files inseveral different formats, may be read directly byCONCH. All processing control parameters areassigned sensible default values but these areeditable, via a Set-up dialog (see Fig. 1), prior toreading the input data file.

2.1. Input (ion-microprobe output) files

An ion-microprobe analysis commonly consistsof a matrix of integers, each corresponding to thenumber of secondary ion counts measured over aspecified integration time, acquired during one ormore passes through the mass spectrum of interest.For example, SIMS U–Pb analysis of a zircontypically involves a series (commonly 4–10 sets) ofmeasurements of the abundance of the species90Zr2

16O+, 204Pb+, background (for example, atmass station 204.1), 206Pb+, 207Pb+, 208Pb+,238U+, 232Th16O+, 238U16O+ and possibly also238U16O2

+. The count rate measured at each massstation (after subtraction of the background countrate) is proportional to the secondary ion intensityof each species, determined using an ion counter orfaraday cup.

CONCH distinguishes between standard analysesand analyses of unknowns from different samples inthe input file by the matching of part of the analysis

ARTICLE IN PRESSD.R. Nelson / Computers & Geosciences 32 (2006) 1479–14981480

http://conch.perthshrimp.commailto:[email protected]

label string with user-specified standard and samplelabel strings specified in the Set-up and Standardsdialogs. The user may specify a label string toidentify standard analyses in the Set-up dialog;separate labels for concentration and Pb/U stan-dards may be defined using the Run-table dialog.

2.2. CONCH processed data output files

Following editing of the counts, CONCH directsprocessed data to its Report Sheet. However, toallow ion-microprobe data obtained on the samesample during different analysis sessions to becombined, grouped and plotted, processed dataoutput by CONCH may also be written to files in aform that can also be read, combined with new dataand further reprocessed by CONCH. Exceptions tothis are the processed data output files generated bythe Counts/Ratios, Al–Mg, Mn–Cr, Fe–Ni andREE/trace element analysis types, which cannot beinput into CONCH for reprocessing. CONCH-processed data files in four different output formatsmay be input directly into CONCH. ‘‘Processed

Data’’ output files are tab-delimited text filescontaining the processed data for each unknownin each 94-entry line (or, in the case of ‘‘StandardsCalibration’’ output files, 102-entry line), termi-nated by a carriage return, in the order that theywere read from the ion-microprobe (raw data) inputfile. The ‘‘Standards Calibration’’ output file formatenables a comprehensive long-term record ofcalibration parameters obtained during the analysisof mineral standards during different sessions to beaccumulated within a single ‘‘Calibration’’ file thatmay be directly read by CONCH for grouping,plotting and reporting. CONCH can also generate,and read, files containing 204Pb-, 207Pb- and 208Pb-corrected data in a ‘‘table’’ text format ready forpublication. Counts/Ratios, Al–Mg, Mn–Cr, Fe–Niand REE/trace element analysis types generateoutput files with a much-simplified format. Anexample of a processed data output file generated bythe Al–Mg analysis type is shown in Fig. 2.

CONCH will, as a (user-editable) default action,concatenate ‘‘.data’’ to the input (ion-microproberaw data) file name to name the processed data

ARTICLE IN PRESS

Fig. 1. Set-up dialog. Default checkboxes are checked.

D.R. Nelson / Computers & Geosciences 32 (2006) 1479–1498 1481

output file. In addition to its use in determiningcalibration curves and sensitivity factors, eachstandard analysis is also processed as an unknown,using calibration values and uncertainties deter-mined from the pooling of all session standardsfollowing editing by the user.

3. Analysis run-tables

A generalized report giving ratios and theiruncertainties can be generated using CONCH’sCounts/Ratios analysis type, but for a moresophisticated treatment (correction of each speciesfor background counts, for example), CONCHrequires knowledge of the identity of the massspecies in the analysis. A shortcoming of acquisitiondata files generated by the SHRIMP acquisitionsoftware is that they do not specify which specieshave been measured. The CONCH user musttherefore provide the species measured and mea-surement order in a ‘‘run-table’’ (see Fig. 3).CONCH includes a number of commonly useddefault run-tables for each mineral (and will make acontext-informed guess, informing the user of thiswhen, for example, the number of mass stationsspecified in the CONCH run-table selected by theuser does not match that in the input file), but itmay occasionally be necessary to modify a defaultrun-table or prepare a new one. For most applica-tions, CONCH presently allows up to 16 massstations to be specified; for the Trace/Rare-earthanalysis type, up to 32 mass stations are allowable.The mass-station order may be edited using theRun-table dialog (see Fig. 3), accessed via the Set-updialog. In order for CONCH to use each speciesmeasured appropriately (i.e. as a calibration nu-merator, or to subtract the background counts fromthe counts obtained for all other species), CONCHidentifies a set of strings (e.g. ‘‘Zr2O 196’’, ‘‘Pb206’’,‘‘Backgrnd’’, ‘‘UO2 270’’, ‘‘Cup-in’’, etc.) as corre-sponding to mass species having a particularfunction. Unrecognized strings entered into the‘‘mass station labels’’ editboxes and strings usedinappropriately (use of the wrong species as acalibration numerator or denominator, or anattempt to generate a diagram requiring a speciesnot included in the run-table, for example) aretrapped and brought to the attention of the user.

All other analysis details (analysis label, numberof sets, scans, stations, counts, integration times,analysis times and secondary beam monitor counts)are assigned for each analysis via the input

(SHRIMP raw data output) file, with error-check-ing to notify the user of any values outside theexpected range.

4. Editing of raw counts data

During measurement using an ion-microprobe,the secondary ion intensity may fluctuate withtransient fluctuations in the primary ion intensityand there may be errors in peak centering of thesecondary ion species of interest on the detector.The first step in the processing of ion-microprobedata is detection of peak-centering problems andaberrant counts on each species.

CONCH enables input data for all species of eachanalysis to be displayed for manual editing on a setof plots of counts (as checkboxes) for each massstation versus analysis time, on the Counts Sheet(see Fig. 4). Counts determined using the SecondaryBeam Monitor, measured simultaneously with thoseof each species, are also plotted. Counting statistics-based ‘‘error bars’’ (7On, where n is the number ofcounts for each measurement) may be added byselecting this option in the Plot Options dialog. Aregression line is fitted through the counts obtainedfor each species. The user may delete any ‘‘aber-rant’’ counts by simply clicking on (or ‘‘un-checking’’) their checkboxes. A dropdown controlor the ‘‘Next Analysis’’ or ‘‘Previous Analysis’’menu items in the ‘‘CONCH’’ worksheet menu barenables the user to display count plots for anyanalysis.

In addition to the manual editing of counts foreach analysis, CONCH allows automatic identifica-tion and display only of scans that include aberrantcounts. For automatic identification of outliers,mass-station counts may be ratioed to counts takenon the Secondary Beam Monitor via a checkbox inthe Set-up dialog. Count rates for each mass speciesduring an analysis will not necessarily be randomlydistributed about a mean value but may changevariably as the secondary ion intensity changes. As aconsequence, standard statistical techniques basedon an assumed distribution function cannot reliablyidentify count outliers. Software identification ofanomalous count values is therefore undertakenusing a ‘‘neural network’’ approach, with thresholdvalues for parameters optimized for the detection ofthe analytical problems commonly encounteredand intended to be identified (e.g. peak-centeringerrors, baseline and peak spikes, etc.). Peak-center-ing errors occur more often during the first


ARTICLE IN PRESS

Fig. 2. Processed data output file. Example shown is of 26Al–26Mg data.


measurement of a secondary ion species. Oneparameter is optimized to identify this sort ofoutlier by comparing the difference in the slopeand scatter of all counts from each mass station,with the slope and scatter for all counts less eachone of the count measurements. A second para-meter (the ‘‘drop-out’’ parameter in Fig. 1) may beoptimized to identify other peak-centering errors(anomalously low counts) occurring after the firstmeasurement. A Background counts upper thresh-old value may also be specified in Set-up to identifyspikes at the background (e.g. 204.1) mass stations.Data from stations for which anomalous counts areautomatically detected will be displayed in diagramsof counts (plotted as checkboxes) versus analysistime, for assessment by the user. Extensive testinghas confirmed that, when the correct thresholdsettings are used, this automated data editing canreliably identify all known instrument-relatedmeasurement aberrations, including secondarybeam spikes and centering failures (drop-outs).

Automated editing thus enables convenient, con-sistent and efficient identification of aberrantcounts. The threshold values of these parametersmay be edited in the Set-up dialog.

Following editing, CONCH determines a countrate for each peak at the mid-time of the analysis bylinear regression of the (edited) peak counts.Provided a background mass station has beenidentified in the run-table, the background countrate (calculated as the arithmetic average rate ifmore than one Background mass station has beenspecified) will then be subtracted from each of themid-time count rates. Ratios of interest and theiruncertainties are then calculated using mid-timecount rates. Uncertainties based on countingstatistics cannot take into account non-linearchange (for example, exponential growth ordecay) in the secondary ion intensity during theanalysis. However, this systematic ‘‘within-analysis’’effect should not be confused with uncertaintiesapplicable to an entire analysis; augmentation or

ARTICLE IN PRESS

Fig. 3. Run-table dialog.

D.R. Nelson / Computers & Geosciences 32 (2006) 1479–14981484

reduction of uncertainties, due (for example) to thesuccess of exponential or spline curve fitting of thechange in secondary ion intensity during the courseof the analysis, may introduce systematic (i.e. non-random) biases into the distribution of ratiosobtained for pooled analyses about their weightedmean values. Secondary ion intensity ‘‘noise’’(arising from instability in the primary ion intensityfor example) may increase the scatter of analysesaround the weighted mean value beyond that basedon counting statistics but because this additionaluncertainty is random, it will not significantly biasthe mean value of pooled analyses. Additionaluncertainty that may be attributable to a high levelof random ‘‘noise’’ in the secondary ion intensity isoccasionally evident in the slightly elevated w2

parameter obtained for pooled analyses (see below).

5. Correction for the presence of common Pb

In order to calculate 206Pb*/238U (throughout thiscontribution, an asterisk indicates radiogenic-Pbisotopic abundance), 207Pb*/235U, 208Pb*/232Th and/

or 207Pb*/206Pb* dates for Zircon, Monazite, Xeno-time, Titanite/Perovskite, Baddeleyite/Common-Pband Th–U Disequilibrium analysis types, measuredPb/U, Pb/Th and Pb isotopic ratios must be correctedfor the presence of non-radiogenic (or common) Pb.The contribution from the common-Pb correction touncertainties in Pb*/U, Pb*/Th and 207Pb*/206Pb*dates varies as a function of age and can besubstantial, particularly for dates younger than ca.1Ga (see Fig. 5). Abundance of the species 204Pb isnot affected by radiogenic decay, so the amount ofcommon-Pb present is proportional to the 204Pbabundance detected during the analysis. The defaultmethod for correcting for the presence of commonPb uses the 204Pb abundance, but CONCHalso offers the option of making no corrections,or corrections to Pb*/U and Pb*/Th ratios forthe presence of common Pb using the 207Pb- and208Pb-methods of Compston et al. (1984); theseresults are written to, and can be read directly andplotted by CONCH from the ‘‘Processed Data’’output files. Assumptions required by these meth-ods for the determination of meaningful Pb*/U and

ARTICLE IN PRESS

mad

LaP

O2

1468514948

15442

1598216161

1637916205

CeP

O2

1342313691

14148

14480

1484115042 15117

Pb2

04

2321

12

19 1922

18

Bac

kgnd

0

1 1

0

1

0 0

Pb2

0625114 2502124759

23788 23696 23641

23161

Pb2

07

3026

3102

29252976

2832

28992846

Pb2

08249480 249385245946

240329237422

233776230964

Th2

32

150675

152516151814

152299151853

149109

148277

UO

254

2882529037

2855328268

27980 27967

27280

ThO

2 26

4

724772

728754725757

722767

718829

712639710755

UO

2 27

0

3662636479 36447

35828

36141

3569135467

SB

M42922

42285

Fig. 4. Counts Sheet.


Pb*/Th ratios are outlined in Compston et al.(1984).

Common Pb detected during ion-microprobeanalysis of high-quality standards may have beenadded during conductive Au-coating of the surfaceof the epoxy mount. The proportion of ‘‘inherent’’common Pb derived from within the mineralstructure during analyses of unknowns may beestimated by subtraction of the session average204Pb counts (i.e. that derived from the mountsurface) obtained for analyses on standards (whichusually have insignificant inherent common-Pb)from the measured 204Pb counts. This ‘‘inherent’’common Pb may have been incorporated into themineral during or possibly sometime after crystal-lization. Therefore, a common-Pb isotopic compo-sition appropriate for the age of the mineral, such asthat of the ore Pb growth curve of Cumming andRichards (1975), is appropriate. For analyses ofunknowns with 204Pb counts substantially higherthan that measured on standards during the analysissession, CONCH incorporates an option for speci-fying common-Pb compositions calculated using themethod of Cumming and Richards (1975). When204Pb counts (for monazite analyses, corrected for

excess 204Pb counts) on unknowns exceed a thresh-old value specified in the Set-up dialog multiplied bythe average 204Pb counts measured on the sessionstandards (also corrected for excess 204Pb countswhere appropriate), common-Pb corrections aremade using Cumming and Richards (1975) modelcompositions. The default but user-editable com-mon Pb composition is specified in the Run-tabledialog.

6. Editing of the session standards and determinationof Pb/U calibration parameters

6.1. Calculation of Pb*/U dates

Due to ionization efficiency differences duringsputtering, the representation of ion species in thesecondary ion mass spectrum differs from theirabundance in the target mineral. Provided that theanalysis conditions are constant, the magnitude ofthis ‘‘inter-element fractionation’’ effect for anyspecific mineral analyzed during an analysis sessionis usually approximately constant. CONCH correctsfor ‘‘inter-element fractionation’’ between Pb and Uduring any single analysis session, and determines

ARTICLE IN PRESS

0.50.00

0.05

0.10

0.15

1.0

206Pb

207Pb

204Pb

1.5 2.0 2.5Age (Ga)

Unc

erta

inty

con

trib

utio

n (G

a)

3.0 3.5 4.0 4.5

Fig. 5. Comparison of contribution to uncertainties in 207Pb/206Pb dates from counting statistics uncertainties associated withmeasurement of 204Pb, 206Pb and 207Pb abundances. At younger dates, uncertainty in measurement of 204Pb dominates age uncertainty,whereas for dates older than 2.5Ga, uncertainty in 206Pb and 207Pb predominate. Curves assume 750 ppm U. For lower U concentrations,204Pb uncertainty dominates to older dates (i.e. assuming 100 ppm U, 204Pb uncertainty4both 206Pb and 207Pb uncertainties for datesp3.0Ga).


206Pb*/238U ratios (and Pb/U dates) for unknowns,using the simple relationship:

aunkastd

¼AunkAstd

, (1)

where aunk is the observed206Pb*/238U ratio for the

unknown, astd is the observed206Pb*/238U ratio for

the standard, Aunk is the fractionation-corrected206Pb*/238U ratio calculated for the unknown andAstd is the ‘‘true’’

206Pb*/238U for the standard(for the CZ3 zircon standard, Astd ¼ 0:0914,corresponding to a 206Pb*/238U date of 564Ma;Nelson, 1997). The 206Pb* abundance used is theradiogenic 206Pb abundance; for standard analyses,this correction uses the 204Pb counts determinedduring each standard analysis and a common-Pb204Pb/206Pb ratio equivalent to that specified in theRun-table dialog (usually that of Broken Hillcommon Pb ore).

The observed 206Pb*/238U ratio for a standard(astd) is not constant within a session but variesunpredictability with instrument set-up, secondaryion tuning, and operator and sample para-meters. Variation in Pb*/U ratios is correlatedwith variation in the ratio 238U16O/238U (as well aswith 238U16O2/

238U and 238U16O2/238U16O varia-

tion; see below). The relationship between206Pb*/238U and 238U16O/238U variation determinedfor zircons analyzed using SHRIMP was earlierconsidered by Compston et al. (1984) to follow aquadratic law. However, Claoué-Long et al. (1995)suggested that the relationship between theseparameters may be simplified to a power law ofthe form:

206Pbþ

238Uþ¼ a#

238U16Oþ

238Uþ

! "k. (2)

The value of a in Eq. (2) must be determined fromsets of analyses obtained on the standard for eachsession, or for part of each session if the relationshipbetween these parameters is unstable. On the basisof an investigation of standard analyses obtainedduring many sessions, Claoué-Long et al. (1995)determined empirically that, for the mineral zircon,k$2.0 (see Eq. (2)) and may be assumed to beconstant between analysis sessions (although it iscommonly difficult to determine whether the valueof k is constant within a single analysis session). Asthis ‘‘power law’’ relationship results in a linearcorrelation with slope ¼ k on a loge%loge diagram,

it is usually more convenient to plot loge[Pb*/238U]

against loge[238U16O/238U], rather than use the

ratios Pb*/238U and 238U16O/238U. CONCH calcu-lates the slope and y-intercept of the loge–logecalibration correlation, using the (user-specified)calibration species pairs in the Run-table dialog,and displays these in the Standards dialog (Fig. 6).For the processing of unknowns, the user mayaccept the default slope (i.e. k ¼ 2:0 for zircon) inthe ‘‘Slope Used’’ textbox of the Standards dialog,or may insert the calculated regression slope (givenin the Regression Slope textbox) if this is preferred.For Zircon, Xenotime and Titanite/Perovskiteanalysis types, the default calibration ratios areloge[Pb*/

238U] and loge[238U16O/238U], whereas for

the Monazite analysis type, the default ratios areloge[Pb*/

238U16O] and loge[238U16O2/

238U16O].CONCH fixes the position of the correlation lineusing weighted mean loge values of the calibrationratios specified in the Run-table dialog obtained forthe (user-edited) session standards. Using theobserved 238U16O/238U (or 238U16O2/

238U16O or238U16O2/

238U, depending on that specified in theRun-table dialog) value for unknowns, the corre-sponding 206Pb*/238U ratios along the correlationline are determined.

For the determination of Th/U ratios in zircon,CONCH uses the empirically determined relationshipbetween 232Th+/238U+ and 238U16O+/238U+ forzircon described by Williams et al. (1996)

232Th238U

¼232Th16O238U16O

# 0:03446#238U16O

238U

! "þ 0:8680

# $.

(3)

208Pb*/232Th dates for zircon are calculated using:

208Pb&

232Th¼

206Pb&

238U#

208Pb&

206Pb&

%232Th238U

. (4)

CONCH determines 208Pb*/232Th dates for mon-azite applying the simple relationship given inEq. (1), where aunk is the observed

208Pb*/232Thratio for the unknown, astd is the ‘‘corrected’’208Pb*/232Th ratio for the monazite standard, Aunkis the fractionation-corrected 208Pb*/232Th ratiocalculated for the unknown and Astd is the ‘‘true’’208Pb*/232Th ratio for the monazite standard. The‘‘corrected’’ 208Pb*/232Th ratio for each unknown iscalculated using the observed 232Th16O2/

232Th ratioand assuming a linear relationship between

ARTICLE IN PRESSD.R. Nelson / Computers & Geosciences 32 (2006) 1479–1498 1487

208Pb*/232Th and 232Th16O2/232Th values:

208Pb&

232Th¼

208Pb&

232Th

! "

meas

#208Pb&

232Th

! "

std

C slope#232Th16O2

232Th

! "

meas

þ y intercept& '

,

ð5Þ

where slope and y-intercept is the slope andy-intercept of the 208Pb*/232Th and 232Th16O2/

232Thcalibration line determined from the (user-edited)session standard analyses. The calculated regressionslope and y-intercept of the calibration line aredisplayed in textboxes in the Standards dialog.Uncertainties in 208Pb*/232Th ratios for individualmonazite analyses do not include uncertaintyassociated with the determination of the Pb*/Thcalibration line, but pooled 208Pb*/232Th datesinclude an uncertainty estimate based on thereproducibility of the standard 208Pb*/232Th mea-surements during the analysis session. Uncertaintiescited for monazite 208Pb*/232Th dates may be basedon the ‘‘observed’’ scatter about the weighted mean

208Pb*/232Th ratio of pooled analyses. A minimumuncertainty for Pb*/U may be specified in the Set-updialog; a minimum uncertainty of 1% (71s) isrecommended.

6.2. Editing and assessment of data obtained forstandards

For Zircon, Monazite, Xenotime, Titanite/Per-ovskite and Th–U Disequilibrium analysistypes, loge[

206Pb*/238U] ratios obtained on thestandard are plotted in the Standards dialog plotwindow (see Fig. 6) as checkboxes againstloge[

238U16O+/238U+] (or loge[238U16O2

+/238U16O+];the ratios plotted are the default ratios, orthose specified in the Run-table dialog if this dialog’s‘‘OK’’ button was used). Standard analyses canbe deleted from the calibration line by clickingon their checkboxes within the Standards dialogplot diagram. Also shown for comparison in theStandards dialog are an uncertainty-weighted regres-sion line of these parameters for the standards and aline with the default slope (i.e. usually 2.0 for zirconusing the calibration pair loge[

206Pb+/238U+]–loge[

238U16O+]). The number of standards, calibration

ARTICLE IN PRESS

Fig. 6. Standards dialog. To alleviate congestion on calibration plot, checkbox labels have been shortened; only last 4 characters areshown.


values and slopes, intercepts, uncertainties andMSWDs for uncertainty-weighted loge[

206Pb/238U]–loge[

238UO/238U] (default ratios for zircon; loge[calibration ratios] plotted are those specified inthe Run-table dialog if the ‘‘OK’’ button on thisdialog was used) and (for monazite) 232Th/238U–238U16O/238U calibration regression valuesare also displayed in textboxes in the Standardsdialog.

The presence of Pb-hydride (PbH+) in thesecondary ion mass spectrum will result insurplus counts at the 207Pb and 208Pb mass stationsdue to 206Pb1H+ and 207Pb1H+, respectively, andthus, measurement of elevated 207Pb/206Pb and208Pb/206Pb ratios. Pb isobars may be monitoredby comparison of the 208Pb-corrected 207Pb*/206Pb*ratios (corrected for common Pb using the composi-tion specified in the Run-table dialog) obtained foranalyses on the standard with the expected value.Isobaric interferences at 204Pb+ may also bemonitored by comparison of the 208Pb-corrected204Pb/206Pb* ratios obtained for analyses on thestandard with the expected value for this ratio.CONCH enables corrections to be made for excess204Pb counts (e.g. for isobaric interference at 204Pb+

due, for example, to deposition on the mountsurface of 186W18O+ derived from the tungstenheating filament commonly used in the conductiveAu-coating process) detected during the analysisof standards; the average excess 204Pb countsdetermined for standard analyses during the analy-sis session may be subtracted from the measured204Pb counts obtained for unknown analysesdetermined during that session, using the ‘‘Excess-204 correction’’ editbox in the Standards dialog(see Fig. 6).

Excess 204Pb (calculated using the 208Pb-corrected206Pb count rate and the common-Pb 204Pb/206Pbratio from the Run-table dialog), ‘‘hydride’’values (based on the 208Pb-corrected 207Pb*/206Pb*ratio), and 204Pb- and 208Pb-corrected 207Pb*/206Pb*dates, with uncertainties for these parameters foreach standard analysis, plus the calculated sessionmeans, are displayed in a scrollable list-box next tothe calibration plot. Also provided within thescrollable list-box is a brief ‘‘diagnosis’’ statementindicating the significance of these and otherrelevant parameters (i.e. number of standardanalyses for which the background count rates werehigher than the 204Pb count rates) obtained forthe analysis session, as an aid to their correctinterpretation.

6.3. Determination and treatment of Pb*/Ucalibration uncertainties

Uncertainties determined by CONCH for indivi-dual analyses include those arising from countingstatistics and in the corrections for the presence ofcommon Pb (see Section 5 above). For individualanalyses, CONCH also provides uncertainties forPb*/U ratios that includes those arising fromcounting statistics and in corrections for commonPb, but augmented by a contribution associatedwith the determination of the Pb/U–UO/U calibra-tion curve (i.e. in the constant a in Eq. (2) givenabove) based on the reproducibility of the standardPb/U measurements.

Uncertainties in Pb*/U for age groups derivedfrom pooled analyses may also include uncertainty inthe Pb*/U calibration, E, based on the reproduci-bility of the Pb*/U ratios of standard measurements,

E ¼C # mnffiffiffi

np , (6)

where C is the calibration uncertainty (s) based onreproducibility of standard determinations, mn is theweighted mean Pb*/U ratio determined for pooledanalyses of unknowns and n is the number ofstandard determinations. This uncertainty may besummed in quadrature with the other sources ofuncertainty.

The validity of the addition (in quadrature) of acalibration uncertainty based on reproducibility ofPb*/U ratio measurements of standards to Pb*/Uratios obtained for unknowns has recently been thesubject of contention (Compston, 1999, 2001; Blackand Jagodzinski, 2003). Compston (1999, 2001)attributed scatter in excess of that attributable tocounting statistics and corrections for common Pbto heterogeneity in Pb/U within the standards thatwould not be expected to effect analyses ofunknowns. In contrast, Black and Jagodzinski(2003) attributed the excess scatter to instrumentalcauses that will effect analyses of both standardsand unknowns. Black and Jagodzinski (2003) andStern and Amelin (2003) presented evidence con-firming that the scatter in Pb*/U ratios determinedusing SHRIMP and the Pb/U–UO/U calibrationapproach is in excess of that attributable to samplePb/U heterogeneity. If the excess scatter in Pb*/U(and also in Pb*/UO and Pb*/UO2) is due to aninherent inadequacy arising from use of the UO/Uratio (and also UO2/U and UO2/U) to correct forthe large inter-element fractionation between Pb


and U during SIMS analysis, as proposed by Sternand Amelin (2003), then such uncertainty should bepropagated to corrected Pb*/U ratios determinedby this method. However, reproducibility in Pb*/Uis also a function of U and Pb count rates, which aredependent on U and Pb abundance in the targetmineral, and on common-Pb corrections. As aconsequence, reproducibility of the Pb*/U ratiofor standards can only approximate that appro-priate for unknowns.

For pooled analyses, CONCH calculates an‘‘observed’’ uncertainty, based on the observeddistribution of analyses about the weighted meanvalue, and an ‘‘expected’’ uncertainty determined bythe addition in quadrature of uncertainties forindividual analyses determined from counting sta-tistics and the common-Pb correction. In addition,CONCH provides the user with a choice of twoother uncertainty levels for fractionation-correctedPb*/U ratios; those based on ‘‘observed’’ and‘‘expected’’ uncertainties but with an additionaluncertainty, based on reproducibility of the sessionstandards, added to each analysis in quadrature, andthose to which an uncertainty, based on reproduci-bility of the session standards, has been added inquadrature to the pooled ‘‘observed’’ and ‘‘ex-pected’’ uncertainties. Provided that the reproduci-bility estimated from the Pb*/U ratio for standardsis appropriate for unknowns, the pooled ‘‘observed’’and ‘‘expected’’ uncertainties should be identical.The w2 parameter (see below) provides a convenientmeans of assessing whether this is the case.

Chi-square (w2) tests whether the observed fre-quencies in a distribution differ significantly fromthe frequencies that might be expected according tosome assumed hypothesis (Moroney, 1984, p. 249).The w2 parameter compares the uncertainty deter-mined from the observed distribution of individualanalyses about the weighted mean value, with theuncertainty determined by addition in quadrature ofthe uncertainty for individual analyses based oncounting statistics, the inter-element (Pb*/U andPb*/Th) calibration (where appropriate) and thecommon-Pb correction;

w2 ¼1

ðn% 1Þ#

Xn

i¼1

ðxi % mnÞ2

s2n þ s2i, (7)

where n is the number of pooled xi values (207Pb/206Pb,

207Pb*/206Pb*, 206Pb*/238U, 207Pb*/235U or 208Pb*/232Th ratios), mn is the weighted mean of all pooledxi values, sn is the uncertainty in mn calculated by

addition of all weighted individual uncertaintiesin quadrature, and si is the uncertainty in theratio xi. w2 values for grouped analyses of lessthan or equal to unity indicate that scatter aboutthe weighted mean value determined for thegrouped analyses can be accounted for by analyticalsources of uncertainty alone. A w2 value significantlygreater than unity indicates that analyses arenot normally distributed about the weightedmean value and that other (geological) sources ofuncertainty are present within the grouped popula-tion. In these cases, the 95% confidence uncertaintycited for dates should be based on the observed,rather than the expected, scatter about the weightedmean 207Pb/206Pb (or Pb*/U) ratio of pooledanalyses.

7. Assignment of analyses to age groups

Analyses may be automatically assigned to207Pb/206Pb or common-Pb-corrected 207Pb*/206Pb*, 206Pb*/238U, 207Pb*/235U or 208Pb*/232Thage-groups (depending on that specified in Set-up)and displayed on Wetherill Concordia (Wetherill,1956), Tera–Wasserburg Concordia (Tera andWasserburg, 1974) or Linearized Gaussian dia-grams in a group fill color, or on a Gaussian-summation probability density diagram. For theassignment of analyses to age groups, CONCH usesa statistically rigorous algorithm based on thefollowing procedure. Common-Pb uncorrected orradiogenic (207Pb*/206Pb*, 206Pb*/238U, 207Pb*/235Uor 208Pb*/232Th) ratios are weighted according tothe inverse square of the individual analyticaluncertainty to determine a weighted mean ratiofor all pooled analyses obtained for the sample.Analyses are then rejected from the group using twocriteria. A w2 value (see above) is calculated for thegrouped analyses. If the w2 value is greater than auser-editable threshold value specified in Set-up(typically, a w2 threshold value of 1.75 is used),geological sources of uncertainty are assumed to bepresent within the group and the analysis whoseratio with assigned uncertainty is most differentfrom the weighted mean value will be excluded fromthe group. In addition, a measure of the difference(D) between each analysis and the populationweighted mean is calculated as follows:

D ¼xi % mnffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffis2pop þ s2i

q , (8)


where xi is the analysis ratio, si is the analysisstandard deviation, mn the population weightedmean and spop is the population standard deviation.Any ratio whose calculated D value is greater than auser-editable threshold value (typically, a thresholdof 72.5 is used) from the group weighted mean willalso be deleted from the group. The weighted meanvalue of the remaining analyses is then recalculated.This process is repeated until all remaining analysesare within both parameter thresholds. Analyses thatbelong with a valid group are then excluded fromthe pool and the process repeated until all remaininganalyses have been grouped. This grouping methodis statistically conservative, in that only the mini-mum number of clearly resolvable dates based onthe uncertainty limits assigned to each individualanalysis will be identified. Analyses with assigneduncertainties overlapping more than one age groupwill be assigned to the larger group (i.e. groupcontaining more analyses), as larger groups are

usually identified earlier during the groupingprocedure. Extensive testing has confirmed that thisautomated grouping method is robust and generallysuperior to manual (and potentially subjective)analysis grouping methods. If required, analysesmay be manually reassigned between groups on theReport Sheet (see below) and replotted. Aspects ofthe plots such as plot size, position and font sizes,may be edited via the Plot Options dialog (Fig. 7)which is summoned by the Plot Options button onthe Counts Sheet, whereas the Preferences dialog(Fig. 8), summoned from the CONCH pull-downmenu, sets default operation parameters and pro-cessing preferences.

8. Plotting of unknowns

U–Pb analyses may be plotted on either WetherillConcordia (Wetherill, 1956; Fig. 9), Tera–Wasser-burg Concordia (Tera and Wasserburg, 1974;

ARTICLE IN PRESS

Fig. 7. Plot Options dialog.


Fig. 10), Linearized Gaussian (Fig. 11) or Gaussian-summation probability density (Fig. 12) diagrams.Where 206Pb*/238U and/or 207Pb*/235U ratios areunavailable (e.g. common Pb and baddeleyiteanalyses), 207Pb/206Pb or 204Pb-corrected 207Pb*/206Pb* ratios may be plotted on Linearized Gaus-sian or Gaussian-summation probability densitydiagrams. CONCH initially calculates sensiblex- and y-axis ranges and axis labels and tickintervals based on the ranges of the data to beplotted. The plot position, size and aspect, axisranges, labeling, tick intervals and other aspects ofthe plots are user-editable and may be personalizedvia a ‘‘Plot Options’’ dialog (Fig. 7) and via the‘‘Edit’’ dialogs (Figs. 13 and 14) summoned bythe ‘‘Edit’’ button on each of the plot windows. Onthe Wetherill Concordia diagram, CONCH will (bydefault) plot the polygons in order of decreasingsize, so that smaller polygons (i.e. more preciseanalyses) are plotted on top of, and are thus notobscured by, larger polygons.

The Gaussian-summation probability densitydiagram displays two probability density curves,the first of which includes all (including discordant)analyses, and the second of which includes onlyconcordant analyses (i.e. 206Pb*/238U date withinuncertainty of 207Pb*/206Pb* date at 72s uncer-

tainty level, or within a user-specified percentage ofconcordance). To generate these curves, CONCHuses the Gaussian distribution probability densityfunction:

GðiÞ ¼1

sffiffiffiffiffiffi2p

p e%ðxi%mÞ2=2s2 , (9)

where xi is the analysis ratio, m is the mean and s theanalysis standard deviation. The probability densitydiagram is mainly used qualitatively with anarbitrary or no y-axis scale, to compare relativeprobabilities. In order to minimize rounding errorsthat result in steps in the curve, the curves have beenscaled so that the maximum peak height of thesecond (concordant analyses only) curve is a fixedpercentage (user-defined in the Plot Options dialog)of the y-axis length. The step size used to generatethe curves can be reduced via the ‘‘Probability PlotEdit’’ dialog (Fig. 14) to generate smoother prob-ability density curves once other aspects of thediagram have been ascertained.

On a Linearized Gaussian diagram (equivalent toplotting analyses on ‘‘Normal Probability Paper’’),a cumulate probability density function will plotas a straight line, enabling the assumption of aGaussian distribution of data about a meanvalue (i.e. individual analyses about a mean

ARTICLE IN PRESS

Fig. 8. Preferences dialog.


207Pb*/206Pb*, 206Pb*/238U or 207Pb*/235U 208Pb/232Th date) to be assessed. In practice, this diagramhas proven useful for investigation of causes behind

rare examples in which a group of 207Pb*/206Pb*,206Pb*/238U, 207Pb*/235U or 208Pb*/232Th analysesdo not show a Gaussian distribution about a mean

ARTICLE IN PRESS

207 / 235UPb

206

/238

UP

b

0.63

0.65

0.67

0.69

0.71

0.73

26.5 27.0 27.5 28.0 28.5 29.0 29.5 30.0 30.5 31.0

3400

3450

3500

Group 1: 25 zircon analyses: 3466 ± 3 Ma (95% confidence)

Group 2: 2 zircon analyses:00.4.1, 00.28.1, 3487 ± 5 Ma (±1s)

Group 3: Analysis 00.21.1, 3509 ± 9 Ma (±1s)

Fig. 9. Wetherill Sheet.

207

/206

Pb

Pb

/206PbU

0.29

0.30

0.31

0.32

1.35 1.37 1.39 1.41 1.43 1.45 1.47 1.49 1.51 1.53 1.55

3450

3500

3550

238

Fig. 10. Tera–Wasserburg Sheet.


value. A common cause of skewed distributions isthe preferential loss of radiogenic-Pb from high-Uanalysis sites; on the Linearized Gaussian diagram,

this manifests as a ‘‘tilted conical’’ arrangement ofanalyses, whereby analyses from higher-U sites andwith smaller uncertainties, the uncertainty bars of

ARTICLE IN PRESS

207

/206

Pb

Pb

Median Rank

0.054

0.056

0.058

0.060

0.062

0.064

0.066

1 2 3 5 10 20 30 40 50 60 70 80 90 95 99

Fig. 11. An example of a Linearized Gaussian diagram generated by CONCH. Data shown are 207Pb/206Pb ratios obtained for analyses ofCZ3 zircon standard on SHRIMP-2A during 2003.

Rel

ativ

e pr

obab

ility

den

sity

207Pb/ 206Pb Age (Ma)

1700 1800 1900 2000

Fig. 12. Gaussian-summation probability density sheet.


which form a ‘‘cone’’, are concentrated at the lower(‘‘tilted’’) right-hand side of the plot. To generatethe Linearized Gaussian diagram, CONCH usesBenard’s Approximation to calculate a median rank(MR), as a percentage, for each point:

MR ¼xi % 0:3nþ 0:4

, (10)

where xi is the analysis ratio and n is the totalnumber of analyses.

Switching between any of the diagrams, assess-ment and editing of the group assignment andgeneration of the processed data file, is readilyachieved using buttons located in the diagramwindows and on the Report Sheet. Group labelsthat include a comma-delimited list of the analysislabels within the group, the group date and itsuncertainty, may be added to analyses on thediagrams using the Summary button. The diagrams

generated by CONCH may be imported into anygraphics program to produce publication-qualityfigures.

9. Report types

The analysis type specified in the Set-up dialogdetermines the Report type generated by CONCH.All are initially written to the Report Sheet (seeFig. 2 for an example). Saving of a Report text fileto disk may be initiated by the ‘‘Save As’’ dialog,summoned using the ‘‘Report File’’ button on theReport Sheet.

9.1. The Generalized Report type

The Generalized Report type, selected usingthe Counts/Ratios analysis type in the Set-updialog, may be generated from any readable

ARTICLE IN PRESS

Fig. 13. Plot Edit dialog.

Fig. 14. Gaussian-summation probability density diagram Edit dialog.


ARTICLE IN PRESS

DATEREPORT- 204Common-Pb Correction method Analyses grouped by 207Pb*/206Pb*

Group No. 1

Sample 206/238 +/- +/-+/-+/-diff diffdiffdiff207/235 208/232 207/206

64.02.1 0.528 0.008 -0.18 13.742 0.226 -0.2 0.0984 0.0016 0.29 0.18878 0.00055 0.0664.03.1 0.534 0.008 0.57 13.974 0.231 0.78 0.1003 0.0016 1.41 0.18965 0.00076 1.1764.04.1 0.537 0.008 0.88 13.995 0.225 0.89 0.0987 0.0016 0.47 0.18899 0.00054 0.4364.05.1 0.551 0.009 2.38 14.283 0.232 2.07 0.0986 0.0016 0.42 0.18817 0.00056 -164.06.1 0.529 0.008 -0.09 13.794 0.224 0.02 0.0982 0.0016 0.16 0.18922 0.00057 0.864.07.1 0.532 0.008 0.34 13.88 0.226 0.39 0.1004 0.0016 1.49 0.18907 0.00057 0.5464.08.1 0.522 0.008 -0.87 13.599 0.23 -0.8 0.0987 0.0016 0.48 0.18887 0.00087 0.1464.10.1 0.534 0.008 0.52 13.91 0.226 0.52 0.0985 0.0016 0.37 0.18894 0.00062 0.364.12.1 0.537 0.008 0.83 14.051 0.228 1.11 0.1001 0.0015 1.35 0.18993 0.00078 1.564.14.1 0.525 0.01 -0.46 13.744 0.308 -0.14 0.0931 0.0019 -2.52 0.18994 0.00178 0.6764.15.1 0.529 0.008 -0.08 13.786 0.223 -0.01 0.0959 0.0015 -1.32 0.18908 0.0006 0.5464.16.1 0.528 0.008 -0.19 13.685 0.216 -0.46 0.0981 0.0015 0.1 0.18801 0.00045 -1.5464.17.1 0.526 0.008 -0.46 13.728 0.224 -0.26 0.0969 0.0015 -0.63 0.1894 0.00065 0.9864.18.1 0.523 0.008 -0.79 13.56 0.22 -1 0.0974 0.0015 -0.3 0.18808 0.00061 -1.0564.19.1 0.523 0.008 -0.79 13.592 0.218 -0.87 0.0963 0.0015 -1.04 0.18852 0.00052 -0.4164.20.1 0.518 0.008 -1.47 13.382 0.215 -1.83 0.0964 0.0015 -0.97 0.18751 0.00072 -1.67

n=16 Mean Exp.s.e. Age(Ma) Exp.s.e. Obs.s.e. Obs.s.e. 206/238

unweighted 0.52965 0.00209 0.00196weighted 0.52952 0.00207 0.00195 2739.4 8.7 -8.7 8.2 -8.2calibrated 0.00337 0.0033 14.2 -14.2 13.9 -13.995%confidence 0.00718 0.007 30.2 -30.3 29.5 -29.7

chi-sq: 0.83

0.9

1.17

0.93

207/235 unweighted 0.05762 0.05489weighted 13.7886

13.79410.0569 0.05554 2735.3 3.9 -3.9 3.8 -3.8

calibrated 0.0895 0.0887 6.1 -6.2 6.1 -6.195%confidence 0.1908 0.189 13 -13.2 12.9 -13.1

chi-sq:208/232

unweighted 0.00039 0.00047weighted 0.09793

0.097880.00039 0.00044 1888.4 7.2 -7.2 8 -8

calibrated 0.001 0.001 18.7 -18.7 19 -19.195%confidence 0.0022 0.0022 39.9 -39.9 40.6 -40.7

chi-sq:207/206

unweighted 0.00019 0.00017weighted 0.18875

0.188890.00015 0.00015 2731.3 1.3 -1.4 1.3 -1.4

95%confidence 0.0003 0.0003 2.9 -2.9 2.9 -2.9chi-sq:

Student's t for 16 analyses: 2.13

GroupNo. 2

Sample 206/238 207/235 208/232 207/206 +/- diff +/- diff +/- diff +/- diff 64.09.1 0.524 0.008 0 13.508 0.221 0 0.0989 0.0016 0 0.18699 0.00061 0

n=1 Mean Exp.s.e. Obs.s.e. Age(Ma) Exp.s.e. Obs.s.e. 206/238

unweighted 0.52965 0.00209 0.00817weighted 0.00817 0.00817 2715.9 34.5 -34.6 34.5 -34.6calibrated 0.00858 0.00858 36.2 -36.4 36.2 -36.4

207/235 unweighted 0.05762 0.22055weighted 0.2205 0.22055 2715.9 15.3 -15.6 15.3 -15.6calibrated 0.2307 0.2307 16 -16.3 16 -16.3

208/232 unweighted 0.00039 0.00158weighted 0.00158 0.00158 1906.7 29 -29 29 -29calibrated 0.0018 0.0018 33.8 -33.9 33.8 -33.9

207/206 unweighted 0.00019 0.00061weighted 0.00061 0.00061 2715.8 5.4 -5.4 5.4 -5.4

File 8-206Pb/238U% 4-208Pb/232Th% standards unknowns cal.slope DesktopFolder: data file 1.418 2.71 8 18 2

0.52395

13.794113.5082

0.097880.09892

0.188890.18699

Fig. 15. Report Sheet showing an example of a Date Report.


ion-microprobe output (raw data) file. It is notnecessary to identify mass species in the raw datainput file using a run-table for this report type.However, if mass species are not identified, either byuse of a run-table or from a SHRIMP ‘‘.sq’’ file typein which the mass species are recorded, thenbackground corrections cannot be made to countrates. The Generalized Report tabulates count rates(as counts/s) and ratios of the count rates, using adenominator mass that may be specified in the‘‘Denominator-Y’’ textbox of the Run-table dialog,the ratio’s uncertainty based on counting statistics,and its D value (see Eq. (8)). If no denominator hasbeen specified, the default denominator is the firstspecies of each analysis read from the input file.Standard analyses are tabulated (in the order thatthey were read from the input file) separately, andaverages, weighted means and observed and ex-pected uncertainties for ratios obtained for stan-dards are also listed. Next, each standard andunknown analysis, with averages, weighted meansand observed and expected uncertainties for ratiosobtained for all analyses, are tabulated in analysisorder.

Al–Mg, Mn–Cr and Fe–Ni report types will alsotabulate mass fractionation-corrected and normal-ized 26Mg/24Mg, 53Cr/52Cr and 60Ni/62Ni ratios,their averages, weighted means and observed andexpected uncertainties, using normalization ratiosspecified in the Run-table dialog.

9.2. The Date Report type and User-editing of agegroups

The Date Report is the report type written to theReport Sheet for Zircon, Monazite, Xenotime,Titanite/Perovskite, Baddeleyite/Common Pb andTh–U Disequilibrium analysis types. The layout ofthe Date Report type (see Fig. 15) facilitates user-editing of age groups and printing. Because analysesbelonging to different groups are displayed indifferent fill colors on the diagrams, the DateReport on the Report Sheet and the WetherillConcordia, Tera–Wasserburg Concordia, Linear-ized Gaussian and Gaussian-summation probabilitydensity diagrams should be used together, so thatthe consequences of reassigning analyses among agegroups can be assessed using the diagrams. Depend-ing on the ‘‘Group by’’ option selected in the Set-updialog, analyses will be assigned to 207Pb/206Pb,207Pb*/206Pb*, 206Pb*/238U, 207Pb*/235U or 208Pb*/232Th age groups, based on the w2 and D threshold

parameters specified in the Set-up dialog that areused to define these groups. Analyses may bemanually redistributed among age groups byinserting an integer corresponding to the new agegroup in the cell to the right of the analysis (column14) to be moved on the Report Sheet and selectingthe ‘‘Recalc’’ button on the Report Sheet. Thevalidity of the move may be assessed by examina-tion of the analysis’s D value and the new group’sw-value on the revised Report Sheet and examinedgraphically on the diagrams. A ‘‘Processed Data’’output file can be written to disk using the ‘‘Files’’button on the Report Sheet.

Acknowledgements

The aouthor acknowledges thorough reviews byHiroshi Hidaka (Hiroshima University, Japan) andan anonymous journal reviewer, R. Adam Frew,Eric Thern and Andrea Biondo for discussions andcapable assistance with all aspects of the SHRIMPanalytical work, and Kenny Sia Tze Foo fordiligently undertaking a Monte Carlo investigationof uncertainty propagation in U–Pb dating.

References

Black, L.P., Jagodzinski, E.A., 2003. Importance of establishingsources of uncertainty for the derivation of reliableSHRIMP ages. Australian Journal of Earth Sciences 50,503–512.

Claoué-Long, J.C., Compston, W., Roberts, J., Fanning, C.M.,1995. Two Carboniferous ages: a comparison of SHRIMPzircon dating with conventional zircon ages and 40Ar/39Aranalyses. Geochronology Time Scales and GlobalStratigraphic Correlation, SEPM Special Publication 54,pp. 3–21.

Compston, W., 1999. Geological age by instrumental analysis:the 29th Hallimond Lecture. Mineralogical Magazine 63,297–311.

Compston, W., 2001. Effect of Pb loss on the ages of referencezircons QGNG and SL13, and of volcanic zircons from theEarly Devonian Merrions and Turondale Formations, NewSouth Wales. Australian Journal of Earth Sciences 48,797–803.

Compston, W., Williams, I.S., Meyer, C., 1984. U–Pb geochro-nology of zircons from lunar breccia 73217 using a sensitivehigh mass-resolution ion microprobe. Journal of GeophysicalResearch 89, B252–B534.

Cumming, G.L., Richards, J.R., 1975. Ore lead ratios in acontinuously changing Earth. Earth and Planetary ScienceLetters 28, 155–171.

Moroney, M.J., 1984. Facts from Figures, second ed. PengiunBooks Ltd., Middlesex, UK, 472pp.


Nelson, D.R., 1997. Compilation of SHRIMP U–Pb zircongeochronology data, 1996. Geological Survey of WesternAustralia, Record 1997/2, 189pp.

Stern, R.A., Amelin, Y., 2003. Assessment of errors inSIMS zircon U–Pb geochronology using a natural zirconstandard and NIST SRM 610 glass. Chemical Geology 197,111–142.

Tera, F., Wasserburg, G.J., 1974. U–Th–Pb systematics on lunarrocks and inferences about lunar evolution and the age of the

moon. In: Proceedings of the Fifth Lunar Conference(Supplement 5, Geochimica et Cosmochimica Acta), vol. 2,pp. 1571–1599.

Wetherill, G.W., 1956. Discordant uranium–lead ages. Transac-tions of the American Geophysical Union 37, 320–326.

Williams, I.S., Buick, I.S., Cartwright, I., 1996. An extendedepisode of early Mesoproterozoic metamorphic fluid flow inthe Reynolds Range, central Australia. Journal of Meta-morphic Geology 14, 29–47.


conch: a visual basic program for interactive processing ... · computer s & geoscienc es 32 (2006)...

Documents