Acta Mineralogica-Petrographica, Szeged 2004, Vol. 4511, pp. 49-53

ACTAMineralogicaPetrographica

EFFICIENCY OF RIETVELD-BASED XPD QUANTIFICA TI ON OF MINERAL ABUNDANCEIN GRANITIC ROCKS, A CASE STUDY

GYÖRGY A. LOVAS, GYÖRGY BUDA

Department of Mineralogy, Eötvös Loránd University, H-1117 Budapest, Pázmány Péter sétány l/C, Hungarye-mail: 10vas@ludens.elte.hu

ABSTRACT

The structure-sensitive and standardless phase quantification method based on a Rietveld analysis of X-ray powder diffraction data wascompared to some traditional techniques. It was found that the Rietveld-based method is at least equal, in several cases superior to the RIRtechnique, and can be a fast and simple substitute for the opticai modal analysis. At the same time it is shown that the points of a Rietveld-based rock classification, using the QAP rock classification of Streckeisen, can be erroneously shifted on the diagram towards the higherplagioclase containing types due to the perthitic exsolution of K-feldspar and consequent inherent overestimation of the abundance ofplagioclase.Key words: Rietveld, XRD quantification, phase analysis, granitoid

INTRODUCTION

The basic goal of this study was to find an X-ray powderdiffraction (XPD) phase quantification method that gets ridof the inherent restrictions and systematic errors of theclassical internal standard based methods (e.g. RIR: Chung,1974a, 1974b, 1975). Moreover, this XPD method shouldalso offer an attractive alternative for the time consumingopticai modal analysis as weil as a reliable comparison basedon the widely used mineral norms, calculated from bulkchemical analysis, (e.g. CIPW) for rock classification.

The Rietveld analysis has given the XPD method a newperspective providing a tool to extract classical structural asweil as real-structural information from even fairly complexor multiphase powder patterns due to its characteristics like:

1) the applied profile function explicitly takes intoaccount each peak of every phase of the sample makingpossible to handIe (as weil as reliably index the contributingsingle peaks) highly overlapping peak groups;

2) the difference diagram of the observed and calculatedpowder profile immediately indicates the presence of un-accounted for intensity contributions;

3) being basicaIly a structure refinement method, theRietveld analysis handles each and every mineral phase in itsactual, individual structural state - that is this approach isstructure-sensiti ve;

4) as every measurable hkl-group of each phase areincluded in the calculation, the effect of preferred orientationtends to be minimal.

As a kind of 'by-product', the refined scale factor of eachconstituting phase in the XPD profile of a powder mixture(e.g. minerals in a rock) is directly proportional to its relativeabundance.

Several aspects of application of the Rietveld-basedquantitative phase analysis for quantification of multicomponentstandards and mineral mixtuies were addressed by Bish et al.(1993). The present work concentrates on natural granitic rocksamples, and aims a comparison of the Rietveld-based

quantification results with two traditional rock analyticaltechniques, the opticaI modal measurement and the bulkchemical analysis-based norm ca1culations.

EXPERIMENT AL

In order to evaluate the efficiency of the Rietveld analysisas mineral quantification method for granitic rocks thefollowing approaches were used:

1) quantification of the mineral composition of artificialmixtures modelling a wide range of granitic rock compositionsby the traditio nal RIR method and Rietveld analysis;

2) applying the above methods to the quantification of aninternational standard granite sample ('GM' SZT SZEV2299-80, red granite, Meissen, Germany, analyzed by 62reference labs);

3) comparison of the results of the RIR and the Rietveldmethods for quantification of the main mineral constituentsof 'GM' granite by statistical analysis;

4) determination of the mineral composition of severalgranitoide rock samples from the Mórágy Hills belonging tothe Fazekasboda-Mórágy Mts. region (Buda, 1995) byRietveld analysis, opti caI modal measurement, and CIPWmesonorms (Parslow, 1969) calculated from the wet-chemical analyses of the whole rocks.

Ali the XPD measurements were caITied out using sidepacked powder samples of 1 !J.maverage grain size, on aBragg-Brentano geometry Siemens D5000 theta-thetadiffractometer equipped with a graphite secondary beammonochromator. The data collection was performed usingCuKa (A = 0.154178 nm) radiation scanning the 5-70°

2l'}range by 0.02° 2l'} stepsize and 5 s counting time/step.The DBWS program package (Young and Sakhtivel, 1994)

was used for Rietveld analysis, while the RIR values (Hubbardand Snyder, 1988) were individually determined on the sameequipment. The opticaI modal values were determined on largethin sections by polarized light microscopy, while the CIPWnorms were ca1culated using the MINPET program.

www.sci.u-szeged.hulasvanytan/acta.htm

-

50

Semi-quantitative chemical analyseswere carried out by means of anAMRA Y 1830iff6 scanning electronmicroscope operated at 20 kVacceleration voltage using energy-dispersive X-ray spectrometry (EDAX@VP9800 ED) at the Department ofPetrology and Geochemistry, ELTE.

Quantitative chemical analyses wereperforrned using a JEOL JXA-733electron microprobe analyser (EMP A)equipped with three wavelength-dispersive spectrometers (WDS), at theGeochemical Research Laboratory ofthe Hungarian Academy of Sciences.The measurements were carried out with:

15 kV accelerating voltage, 40 nA beamcurrent, beam defocus 10 /lm diameter, 5x 4 s counting time. Analyticallines: Lafor Pb, Ka for Al, Ca, S, CI and P, K~ forAs. The following standards were used:PbSe (synthetic) for Pb, artificial glassfor Al, apatite for Ca, P and CI,chalcopyrite for S, GaAs for As. Formatrix correction conventional ZAF

method was applied.

I

J

I

RESULTSANDDISCUSSION

The 'GM' standard granite samplewas used for optimizing the refinementstrategy and parameters of the Rietveldtreatment. The best fit was achieved

using a 5 parameter orthogonalpolynomial fit for background treatment,the pseudo Voigt function for profiledescription, 5 FWHM region to handiethe tail regions, March-Dollas functionfor preferred orientation correction,sample displacement correction and nocorrection for absorption, microextinctionand surface roughness. Structural modelsfor the main constituents (quartz, K-feldspar (low-microcline), plagioclase(iow-albite), biotite (3T) and chlorite(clinochlor)) were taken from the ICSDdatabase. The test mixtures were

prepared using the following purecomponents:quartz- from the VelenceMts., (Hungary), microcline and albite -from Ytterby (Sweden), biotite - fromErdosmecske, Mecsek Mts., (Hungary).Rietveld refinement of each of these

single phases (Fig. 1 shows refinementresults for microcline ) indicated that:1. No refinement of atomic positionalparameters gives significantly better fit2. Refinement of an overall isotropicthermal parameter (Bover) is justified,3. The traditional Rietveld formula

describes properly the peak asymmetrydown to 300 2'O and a first order

Gy. A. Lovas and Gy. Buda

5000

4000

~ 3000>,

~ 2000

1000

0.0" II'"

Aw,n:QU,\lnZ

on.00

o:.0202

9.3!iC..fI6.5l'"'51

N."IHJ

8.0 Nil

II III 1181111, -''' ,,., ,,_,1111'"..""11".."_.."".""-." '.-''''.1 I ',1 I '. II II." ,. ,.,32.0 40.0 48o 560 64.0

2-lheta

1.6.0

Fig. 1. Rietveld analysis of the porpyroblastic microcline granitoid rock fromErdosmecske, Mecsek Mts., S-Hungary.

Cagliotto formula gives suitabledescription of the angular dependenceof the FWHM, 4. Cell parameters haveto be included in the refinement.

The comparative study of thequantification of the artificial mixturesprepared and analyzed using the above-derived optimum experimentalparameters, shed 'light on the followingdifficulties:

1) a heavy overlap of the strongest(that is the most desirable as analyticalline) XPD reflections of the main rockforming minerals is very frequent;

2) the traditional intensity measure-

ment is very much hindered by theuncertainty of the base line deter-mination due to the even more heavilyoverlapping weak reflections;

3) ali main mineral constituents,except quartz, are inclined to showpreferred orientation of various degrees.

Ali these facts provide explanationwhy the Rietveld analysis was found tobe superior to the RIR results in eachcase where one or more of the above

phenomena occurred (the differencebetween the weight percentage derivedby the above methods can be as high as20%) (Fig. 2).

Fig. 2. Comparison of Rietveld and RIR/Chung phase quantification of thestandard mixture #4.

www.sci.u-szeged.hu/asvanytan/acta.htm

#4 Mixture #4 Mixture100%

100% 90%90% 80%80%70% " 70%"

,g 60% ,g 60%'u; 'iijo 50% g, 50%c.E 40% E 40%o Ou 30% u 30%

20%20%

10%0% 10%

Rietveld \N% RIR 0%Type of Measurement Rietveld W/o RIR

Type of Measurement

Rietveld W% RIR.. Corundum

QUaltz 35,9 35,0 36,7o Biotite

K-Feldspar 5,5 5,0 6,1O Ragioclase Plagioclase 32,3 35,0 39,8.. K-Feldspar Biotite 15,2 15,0 Il,2!!ilQuartz Corundum 11,2 10,0 6,1

L 100,0 100,0 100,0

Efficiency of Rietveld-based XPO quantification of mineral abundance in granitic rocks

The mineral-by-mineral comparison and its statisticalanalysis in case of the 'GM' sample even more explicitlypointed out how sensitively the RIR method reacts to theunavoidable preferred orientation, while the Rietveldanalysis was considerably affected only at the extremely highconcentrations. The Rietveld analysis gave a slightly betterfit even in the case of quartz showing that it can adapt to theslightest deviances from the structural model.

In the case of the natural granitoid samples not only theresults of the two main XPD methods were compared, but alsotheir output was corroborated with the results of their opticaImodal analysis as weil as with the caJculated CIPW norms ofthe same samples. There are slight differences in the comparisonresults between localities (Erdosmecske, Kismórágy, Mórágyquarry), some main trends, however, can still be recognized:

1) the Rietveld analysis results approximate the opticaimodal values significantly better than the RIR results do;

2) the greatest discrepancies were found in the case ofbiotite. It is obviously due to its extremely severe preferred

c:.2~ 60%oa.E8 40%

Erdösmecske

100%

80%

20%

0%Rietveld Mesonorm

Type of Measurement

c:o:;::; 60%'Uioc-E 40%o

(,)

Erdösmecske100%

80%

20%

IIIChlorite

O Biotite

O Ragioclase

III K-Feldspar

IIIQuartz

0%Modal Mesonorm

Type of Measurement,

Fig. 3A. Comparison of phase quantification of the granitoid tTomErdosmecske, Mecsek Mts., S-Hungary, by different methods.

Rietveld RIR

51

orientation (that is to the fact that one method can handie thisphenomenon by modeling, while the others can not);

3) in ali cases the XPD quantification results fit better themeso norm values than the opticaI modals.

The accuracy of the opticaI modal analysis dependheavily on the grain size (as well as on the number ofmeasurements), consequently large megacrysts are fairlydifficult to handle. Accordingly the resulting modal valuesare more representative to the small-grained rock fraction('base material'). The mesonorm caJculations are based onbulk wet-chemical analysis; consequently it is independent ofthe grain-size of the rock. Since the XPD measurements arealso can'ied out on powdered rock samples, their better fit withthe corresponding mesonorms is expected (Fig. 3A, B, C, D).

A comparison of rock classification based on the QAProck classification diagram of Streckeisen (1974) wasperformed using phase abundance values obtained by threedifferent methods (Rietveid, opticai modal, and CIPWmesonorms) (Fig. 4). Comparing the Rietveld based

100%90%80%

c: 70%o:E 60%II)8..50%

,g 40%u 30%

20%10%0%

Kismórágy

Rietveld MesonormType of Measurement

,..:8 60%

.~50%c-E 40%ou 30%

20%10%

Kismórágy100%

90%80%

70% IIIChlorite

0%Rietveld RIRMadai Mesonorm

Type of Measurement, ,

Fig. 3B. Comparison of phase quantification of the granitoidtTom Kismórágy, Mecsek Mts., S-Hungary, by different methods.

www.sci.u-szeged.hu/asvanytan/acta.htm

Rietveld Madai Mes anam} RlR

Quat1z 23,4 25,1 33,5 21,0K-Felds par 24,0 22,1 18,9 39,0Plagiaclase 35,3 37,0 38,6 36,0

Biatite 16,0 14,8 9,0 4,0Camndul11 1,3 0,9

L 100,0 100,0 100,0 100,0

Rietveld MadaI Mesollonn R1R

Quartz 24,8 19,3 33,5 21,0K-Feldspar 16,4 21,4 18,8 39,0Plagioclase 37,3 40,8 38,6 36,0

Biotite 21,1 17,6 9,0 4,0Chlorite 1,4 0,9

" 100,0 100,0 100,0 100,0"--

52 Gy. A. Lovas and Gy. Buda

Mórágy, quarry

100%

90%

80%

c: 70%

~ 60%

.~ 50%a-E 40%oU 30%

20%

10%

0%

II O1lorite

o Biotite

O Ragioclase

II K-Felds par

O Quartz

Rietveld Mesonorm

Typeof Measurement

Mórágy, quarry100%

90%

II K-Felds par

O Quartz

80%

70% II O1lorite

o Biotite

O Ragioclase

c:.2 60%....

.~ 50%a-E 40%oU 30%

20%

10%

0%

Madai Mesonorm

Type of Measurement

Fig. 3e. Comparison of phase quantification of the granitoid fromMórágy quary, Mecsek Mts., S-Hungary, by different methods.

classification with those, derived from CIPW catanormcalculations (Buda, 1995) it was found that the Rietveld-based compositions tend to indicate higher quartz andplagioclase containing rock types (granodiorite)

Rietveld RIR

CONCLUSIONS

The Rietveld"based XPD phase quantification methodproved to be an efficient tool to determine the mineralabundance in granitic rocks. Its most desirable feature is thestructure sensitivity. It means that it takes into account theactual structural state of the constituting minerals ascompared to the external standardization inevitable in theRIR method. The comparative measurements on artificialmineral mixtures (modeling a wide range of graniticcompositions), on natural samples (granites from theMecsek Mts.), as weil as on the international granite rockstandard 'GM', proved the superiority of the title methodover the RIR technique. It is mainly due to its capability ofhandling heavily overlapping peak groups as weil as its

"GM" standard garnitic rock

100%

90%80%

c: 70%o:i:160%1/)

8.50%E 40%oU 30%

20%

10%0%

Rietveld MesonormType of Measurement

[.dq

"GM"standard granitic rock

vrcd100%

90% e80%

70%gf

l:j

fI1

c:.2 60%....

.~ 50%a-E 40%oU 30%

20%

10%

c

0%

Rietveld Standard Mesonorm RIR

Type of Measurement

Fig. 3D. Comparison of phase quantification of the 'GM'international standard granitic rock sample by different methods.

efficiency in the treatment of the effect of preferredorientation.

Comparing the Rietveld results to the opticai modalmeasurements and calculated CIPW norms of the same

natural samples it was found that the Rietveld based XPDmethod is a suitable replacement of the time consumingopticai modal analysis. The comparison with the CIPWmesonorrns, however, pointed out the need of cautioustreatrnent of porphyric constituents in grain size dependentmethods like the later.

The comparison of Rietveld-based rock classification withthose of modal-based ones pointed out an inherent feature ofthe Rietveld quantification; no distinction can be madebetween a self-contained and a perthitic albite phase.

Accordingly the Rietveld-derived plagioclase content ishigher than the amount of the modal plagioclase.Consequently the Rietveld-based rock classification iserroneously shifted towards the plagioclase-richer rocktypes. The conclusions of this study are in good accordance

www.sci.u-szeged.hu/asvanytan/acta.htm

Rietveld Modal Mesonorm RlR

Quartz 26,3 28,3 29,6 23,0K-Felds par 23,1 29,7 18,5 38,0Plagioclase 36,1 38,9 34,0.).),.)

Biotite 14,5 7,2 12,9 4,0Chlorite 1,5 1,0

2: 100,0 100,0 100,0 100,0

Rietveld Standard Mesononn RIR

Quartz 31,6 32,6 33,2 26,0K-Feldspar 28,5 25,5 28,1 31,0Plagioclase 26,6 35,7 36,7 40,0

Biotite 4,4 6,1 2,0 3,02: 100,0 100,0 100,0 100,0

Efficiency of Rietveld-based XPD quantification of mine ral abundance in granitic rocks

. Erdosm"'skc. Rlw'Cld

O ErdöSl11ccskc. Mo(!.1l

~ Enlosl11ccskc.Mcsollonll

~ Kismór.íg)'. Rlct\'cld

1> Kl>mór.íg!. Modul

~ Kismór.így.M,'SOlIonn. Mor.'Ig)'.Ricl\'cld

O MO";Ig,\', whxJ,,[

liiiIMór.i!<!'J\'1csollonn

A'1''''''''''1<

,U''''''."""",J,

i

\' \'

"",/;" ; ,, \! .{ \

\

I ; ~I ~ 00.10

I ",.~4J~'( ",,,,<>ló.';".tr- \\\"'"~''''',...~",

""'",'!'''A P

Fig.4. Rock classification by Streckeisen's QAP discriminativediagram based on phase abundances obtained by differentquantification methods.

with the results published by HiIl et al. (1993) for a widerrange of rock types, Their finding concerning the superiorityof using capillary packed sample, that is a transmission typedata collection, over the reflection-type one reassure theexpectation of the authors that changing the diffractiongeometry from Bragg-Brentano to Debye-Scherrer wouldfurther improve the results by eliminating the preferredorientation phenomenon.

ACKNOWLEDGEMENTS

A special thank is due to Sándor Háden for helping infieldwork and sample preparation, as weil as to LíviaRudnyánszky for the precious grain size, magnetic, andheavy liquid separations. This work was financiallysupported by OTKA under contract no. T 014969.

REFERENCES

ALEXANDER,L., KLUG, H. P. (1948): X-ray diffraction analysis ofcrystalline dust. Anal. Chem., 20, 886-894.

BISH, D. L., How ARD,S. A. (1988): Quantitative analysis via theRietveld method. 1. of Appl. Cryst., 21, 86-91.

BISH, D. L., POST, J. E. (eds.) (1990): Modem powder diffractionmethods. Reviews in Mineralogy, 20, Mineral. Soc. of America

BISH,D. L., POST,1.E. (\ 993): Quantitative mineralogical analysis usingthe Rietveld full pattem fitting method. Amer. Miner., 78, 932-940.

BUDA, Gv. (1995): Variscan granitoids of the Mórágy Hills.(Geological excursion guidebook. (Manuscript) (Geológiai terepitanterületek, kirándulásvezetok, kézirat). MKM Project M.sz.3311/93.

53

CHUNG, F. H. (1974a): Quantitative interpretation of X-raydiffraction patterns. I. Matrix-tlushing method of quantitativemulticomponent analysis. 1. Appl. Cryst., 7, 519-525.

CHUNG, F. H. (1974b): Quantitative interpretation of X-raydiffraction patterns. II. Adiabatic principle of X-ray diffractionanalysis of mixtures. J. Appl. Cryst., 7. 526-531.

CHUNG, F. H. (1975): Quantitativc interpretation of X-raydiffraction patterns. lll. Simultaneous detcrmination of a set ofreference intensities. J. Appl. Cryst., 8.17-19.

DAVIS, B. L., WALAWANDER, M. J. (1982): Quantitativemineralogical analysis of granitoid rocks: a comparison of X-rayand optical techniques. Am. Mineral., 67, 1135-1143.

Dollase, W. A. (1986): Correction of intensities for preferredorientation in powder diffractometry: application of the Marchmodel. J. Appl. Cryst.,19, 267-272.

HILL, R. J., TSAMBOURAKIS,G., MADSEN, 1. C (1993): lmprovedpetrological mod al analyses from X-ray powder diffraction databy use of the Rietveld method I. Selected igneous, volcanic, andmetamorphic rocks. 1. PetL, 34, 867-900.

HUBBARD,C R., Ev ANS,E. H., SMITH,D. K. (1976): The Referencelntensity Ratio, Uic for computer simulated patterns. J.AppI.Cryst.,9,169-174.

HUBBARD,C R., SNYDER,R. L. (1988): Reference Intensity Ratio -measurement and use in quantitative XRD. Powder Diffraction,3,74-78.

KLUG, H. P., ALEXANDER, L. E. (1974): X-ray DiffractionProcedures, 2nd edition, John Wiley and Sons, New York, (a)549-553, (b) 365-368.

MANIAR,P. D., COOKE,G. A. (1987): Modal analyses of granitoidsby quantitative X-ray analysis. Am. Mineral., 72, 433-437

O'CONNOR, B. H., RAVEN,M. D. (1988): Application of the Rietveldrefinement procedure in assaying powder mixtures. PowderDiffraction, 3, 2-6.

PARSLOW,G. E. (1969): Mesonorms of granitic rock analyses.Mineral. Mag., 37, 262-269.

PAWLOSKI,G. A. (1985): Quantitative determination of mineralcontent of geological samples by X-ray diffraction. Am. Mineral.,70, 663-667.

RIETVELD,H. M. (1968): A profi le refinement method for nuclearand magnetic structures. J. Appl. Cryst., 2, 65-71.

SMITH,D. K., NICHOLS,M. C, SCHEIBLE,A., WIMS, A. M., JOHNSON,

J. L., ULLMANN,G. (1987): Quantitative X-ray powder diffractionmethod using the full diffraction pattern. Powder Diffraction, 2,73-77.

STRECKEISEN,A. (1974): Classification and nomenclature ofplutonic rocks. Geol. Rundschau, 63, 773-786.

WILES, D. B., YOUNG,R. A. (1981): A new computer program forRietveld analysis of X-ray powder diffraction patterns. J. Appl.Cryst., 14, 149-151.

YOUNG,R. A. (ed.) (1993): The Rietveld Method. IUCr Monographon Crystallography 5., Oxford University Press

YOUNG,R. A., SAKTHIVEL.A., Moss, T. S., PAIVA-SANTOS,C O.(1994): DBWS-941: Rietveld analysis of X-ray and neutronpowder diffraction patterns. School of Physics, Georgia lnst. ofTechn., USA

www.sci.u-szeged.hu/as~anytan/acta.htm

1