Introduction

Calcium carbonate and calcium-magnesiumcarbonate in the form of limestone, dolomite, marl,chalk, and Oyster shell are one of the most widelyutilized non-metallic materials in the industrial world. The largest use of limestone or calcium carbonate is in the cement industry where it is used as a source ofCaO and also in the concrete industry where it is usedas the primary coarse aggregate. Following thecement industry, the second largest user would be thelime industry.

Geological materials present numerous problemsas a result of the preponderance of low atomicnumbered elements in a variable mineralogical andelemental matrix. X-ray fluorescence can provide theanalyst with an accurate and precise method providing the analytical techniques are properly addressed andare consistent from sample to sample. The most ser-ious problems to solve are absorption and enhance-ment effects, mineralogical differences amongsamples, and particle size effects which often in-

fluence the intensities of the analytical lines.Consequently, relative intensities of a standard and an unknown sample are often only approximate mea-surements and not directly proportional to theconcentration since the matrix, in addition to theconcentration of the assayed element as related to themeasured characteristic radiation must be correctedfor enhancement, absorption, and possible peakoverlaps. These matrix effects are generallyconsidered as absorption, enhancement, peak over-laps, mineralogical differences, and inhomogenity ofthe sample particles. Consideration of these problems, thereby providing a useful and workable procedure by X-ray fluorescence, has been approached by the useof internal standards [1], comparison to standardsapproximate in composition to the unknowns [2],fusion and dilution with transparent materials such alithium-tetra-borate [3-5], reduction of particle sizedby fine grinding [6-9], and mathematical corrections[10-14]. The method utilized by the author employsthe powder method, fine grinding and pelletizing, andempirical calculations for corrections due to ab-sorption, enhancement, and peak overlaps.

16 The Rigaku Journal

The Rigaku Journal

Vol. 16/ number 1/ 1999

CONTRIBUTED PAPERS

ANALYSIS OF LIMESTONES AND DOLOMITES BY X-RAY FLUORESCENCE*

BRADNER D. WHEELER, PHD.

Rigaku/USA, 199 Rosewood Drive, Danvers, MA 01923, USA

Sources of calcium are generally widespread and quite extensive. These sources are of limestone, dolomite, marl,chalk, and oyster shell. Cement plants account for nearly half the demand while more than two hundred lime plants in theUnited States and Puerto Rico consume about twenty five percent. Since the chemical composition of the limestone orother sources of calcium is critical in the cement and the lime industry, particularly for the deleterious compounds such asNa2O, MgO, P2O5, and K2O, accurate determinations are critical. Due to the tonnage per hour, these determinations mustbe made rapidly and accurately. X-ray fluorescence can thereby satisfy this need for accuracy and precision.

Production of lime is performed by calcining limestone and the industry is generally located and concentrated in theStates of Michigan, Pennsylvania, and Missouri. The resulting product is quicklime-CaO, or hydrated lime-Ca(OH)2.Substantial amounts of quicklime is further processed into calcium carbide in order to produce acetylene gas. In this case,the determination of P2O5 is critical since minor amounts of phosphorous in the acetylene gas can cause prematureexplosions. Other uses for lime are well known in the treatment of water, the paper and pulp industry, and the steelindustry for the production of slag to remove impurities. Dolomitic lime is heavily utilized in the manufacture ofmagnesite refractories by reacting the dolomitic lime with brines from the Michigan basin to produce MgO and CaCl2.

Sample preparation for these materials has been performed by grinding and pelletizing or fusion with Li2B4O7. Inaddition to the chemistry, matrix and interelement effects will be discussed as related to the chemical analysis by X-rayfluorescence.

* Presented at the ASTM Committee on Lime, C-7, December 8,Nashville, Tennessee.

Analysis of any material by X-ray fluorescence isbest applied to materials where the compositionalrange is reasonably small. Calcium/calcium-mag-nesium carbonate rocks fall into this category eventhough the calcium/ magnesium ratios plus theargillaceous fractions are quite variable. In order tosuccessfully apply an X-ray fluorescence technique,the characteristics affect the reproducibility andaccuracy must be identified and corrected. Thesevariables which can cause errors in the analysis aredeviations in the particle size, mineralogy, andinterelement effects due to varying chemicalcomposition among samples.

Particle Size

Reproducible and accurate results by the powdermethod in the quantitative analysis of mineralogicalsamples requires proper sample preparation in order to minimize intensity fluctuations as a function ofvariations in the particle size and distribution. Burn-stein [16] has illustrated that the fluorescent intensityfrom a pure material will increase as the particle size is decreased. In limestone and dolomites the intensitiesfrom several elements may all increase, decrease, orone may decrease while others increase. Campbelland Thatcher [17], measuring calcium in Wolframitewhere the calcium may be present as a carbonate,tungstate, or phosphate supported Burnstein's work.Differences in intensities were observed for equalconcentrations of calcium in three chemical stateswhen the particle size is large as compared to theeffective depth of penetration of incident X-ray.Extensive grinding illustrated the intensities from thevarious mineralogical forms approach a commonvalue by reducing the absorption within the individualparticles to a small value (-325 mesh). Figure 1illustrates the relation of intensity with grinding timeor a reduction in particle for Ca-Mg carbonate rocks.Reduction in particle size causes a reduction in theintensities of iron, sulfur, and potassium while theintensities of calcium and silica are increased. As thesize of the individual particles is reduced, theintensities stabilize. Further reduction in particle sizethrough continued grinding does not promote anyadditional improvement in the intensities, Referring to the example on Figure 1, the minimum grinding timewould be five minutes. A lesser amount of time couldcause significant intensity/concentration deviationsamong the standards and samples. Now that a grinding time has been established, determining the properpelletizing pressure must be determined. A similarstudy was performed as illustrated on Figure 2.

Examination of Figure 2 reveals that in order toreproduce a consistent pellet, a pelletizing pressure offifteen tons per square inch should be used.

lnterelement Effects

Quantitative analysis by X-ray fluorescence ofany material requires that the measured intensity of aparticular element is proportional to the percent com-position. A matrix such as a limestone or a dolomite

Vol. 16 No. 1 1999 17

Fig. 1 Grinding time versus intensity.

Fig. 2 Pelletizing pressure versus intensity.

may reveal that the intensity of an element may not bedirectly proportional to the concentration due to result of an additional element within the sample. This non-linearity is frequently referred to as the interelementeffect and may also be referred to as enhancement orabsorption. When the characteristic radiation of oneelement excites another element, enhancement oc-curs. Absorption is observed when one element in thesample matrix has an absorption edge on the lowenergy side of the element of interest or has a massabsorption coefficient larger than the element ofinterest at the energy level of that element. Examplesof these effects are illustrated on Figure 3 (MassAbsorption Coefficient vs Energy in KEV). Since theSi Kα line occurs just on the high energy side of theAl K-edge, secondary fluorescence will take placeand conversely, silica is strongly absorbed byaluminum. A similar case is observed in the po-tassium-calcium System where calcium is stronglyabsorbed by potassium since the Ca Kα line lies juston the high energy side of the K K-edge as illustratedon Figure 3. An additional complication is the fact that iron has a high mass absorption coefficient at theenergy levels of the lower Z elements, thereby actingas a strong absorber.

Although absorption and enhancement effectscan be severe, mathematical corrections can easily beapplied. Numerous methods have been proposed [10-14]. The author has proposed a method described byLaChance [11] where a relationship is established that the relative intensity of a characteristic line in a binary system is directly proportional to the weight fractionof a given element (A) plus (B) however must sum tounity. The expression would be as follows;

( ) ( )R C C ABA a b= +1 α (1)

( ) ( )R C C BAB b a= +1 α (2)

Defining αAB and αBA

( ) ( )

( )αµ θ µ θ

µ θ µ θAB

B

A=

+ ⋅ −

+ ⋅1 1 2 2

1 1 2 2

1csc csc

csc csc(3)

( ) ( )

( )αµ θ µ θ

µ θ µ θBA

A

B=

+ ⋅ −

+ ⋅1 1 2 2

1 1 2 2

1csc csc

csc csc(4)

where

(µ1)A and (µ1)B= the mass absorption coefficients of elements Aand B at the effective wavelength for theexcitation of the A radiation

(µ2)A and (µ2)B=the mass absorption coefficients of elements Aand B at the effective wavelength for theexcitation of the B radiation

θ1 and θ2 = the angle of incidence of the primary X-ray beam andthe take off angle of the secondary radiation

Ca and Cb =the weight fractions of elements A and B

Ra and Rb = the relative intensities of elements A and B expressedas ratios of net intensities of the elements A andB in the sample to the net intensities for the pure elements A and B

Calibration of the standards involves an iterativeprocess according to Equations 3 and 4 which estab-lishes the a coefficients which are then assigned toEquations 1 and 2. The unknown sample data is thenprocessed through multiple regression analysisutilizing Equations 1 and 2.

Sample Preparation

As previously discussed, particle size anddistribution can have an effect on the intensities ofmost elements with the most severe being at the low Zend of the periodic table. An illustration of this fact,Figure 1, displays the effect of grinding time trans-lated into smaller sized particles with increasedgrinding time. Eight (8) separate samples of singlestandard of five (5) grams and 0.1 grams of Na-stearate as a grinding aid were placed in a tungstencarbide rotary swing mill and ground for one (1) toeight (8) minutes. The resulting powder was thenpelletized under 15 tons per square inch with boricacid as a backing material. Each pellet was analyzedwith the resulting intensities plotted as a function ofgrinding time (Figure 1). The grinding curve indi-cated a minimum grinding time plus one (1) minutefor a total of six (6) minutes.

18 The Rigaku Journal

Fig. 3 Elemental absorption curve.

Instrumentation

A Rigaku RIX 3100 X-ray spectrometer with a 4kW generator was utilized for this analysis and wasoperated under the instrumental operating parametersas described on Table 1.

Results and Conclusions

The samples utilized in this study were seven (7)standards supplied by the National Bureau ofStandards [18], Ash Grove Cement Company [19].Hercules Cement Company [20], and well charac-terized unknown samples from Medusa CementCompany [21]. The standards and unknowns werederived from diverse geographical and geologicalareas. In addition the mineralogical structure variedfrom calcite (rhomboidal-R-3c), vaterite (hexagonal-P63mmc), aragonite (orthorhombic-Pnm), and ara-gonite (orthorhombic-Pmca) as listed in appendix A.

The results of analysis is contained on Table 2,while the individual calibration curves are containedin appendix B. A typical spectra of limestone is con-tained in appendix C. Calcium oxide and magnesiumoxide, the main components in Ca/Mg carbonaterocks, exhibited absolute errors of approximately twoto four percent relative utilizing a simple least squaresregression analysis. Utilizing the theoretical alpharoutine as outlined by LaChance [11] with a multipleleast squared analysis, these errors were reduced to0.07 and 0.05 percent respectively. It should be noted

that the CaO in the unknowns was initiallydetermined by a KMnO4 titration with no attempt todifferentiate between CaO and SrO. In the KMnO4

titrimetric determination of CaO, both the CaO andSrO are both precipitated as a calcium-strontiumoxalate and when titrated with KMnO4, the oxalateion is being determined. As a result, the determinedvalue by this method will be CaO plus SrO. In alimestone or a dolomite, the SrO could be from 0.05to 0.3 percent. Therefore, since X-ray fluorescencedetermines CaO and SrO separately, the SrO and CaO determination by X-ray fluorescence must be com-bined in order to agree with the KMnO4 titrimetriccalculation.

Conclusions

The results of analysis illustrate that X-rayfluorescence is a viable technique for the analysis oflimestone and dolomite. It further illustrates that, with proper sample preparation, mineralogical differencesbecome insignificant. Data reduction the use oftheoretical alphas automatically solves the problemsassociated with absorption and enhancement effect.

References[1] Dwiggins, C. W. Jr., and Dunning, N. H., Quantitative

Determination of Traces of Vanadium, Iron, and Nickelin Oils by X-ray Spectrography, Anal. Chem. 32, 1137-1141, 1960.

Vol. 16 No. 1 1999 19

Table 1. Instrumental Operating Conditions for Limestones and Dolomites.

Element Na2O MgO Al2O3 SiO2 P2O5 SO3 K2O CaO TiO2 MnO Fe2O3 SrO

ANODE Rh

ANODE VOLTAGE, KV 30 50

ANODE CURRENT, MA 130 80

SLIT Coarse

CRYSTAL RX-40 PET Ge LiF

DETECTOR FP-C SC

FILTER None

PEAK 2θ ANGLE 17.05 14.15 144.65 109.04 141.10 110.80 70.00 61.95 86.18 62.96 57.49 25.13

COUNTING TIME, S 40 20

PHA LOWER 50 100

PHA-UPPER 450 300

ATMOSPHERE Vacuum

[2] Brown, 0. E., “Use of X-ray Emission Spectroscopy inChemical Analysis of Cement, Raw Materials”, And Raw Mix, ASTM Annual Meeting, 1963.

[3] Moore, C., “Suggested Method for SpectrochemicalAnalysis of Portland Cement by Fusion with LithiumTetraborate Using an X-ray Spectrometer", ASTMReport No. E-2, 5MlO-26.

[4] Harvy, P. K., Taylor, D. M., Hendry, R. D., and Bancroft,F., “An Accurate Fusion Method for the Analysis ofRocks and Chemically Related Materials by X-rayFluorescence”, X-ray Spectrometry, 2, 33-34, 1972.

[5] Claisse, F., “Accurate X-ray Fluorescence AnalysisWithout Internal Standards", Norelco Rep. 4, 1957.

[6] Wheeler, B. D., “Cement Raw Mix Control through X-rayEmission Spectroscopy”, Proc. Third Forum the geology of Industrial Minerals, Spec. Dist. 34, University ofKansas, 1967.

[7] Hooper, P. R., “Rapid Analysis of Rocks by X-rayFluorescence”, Analytical Chemistry, 36, 1964.

[8] Wheeler, B. D., “Accuracy in X-ray SpectrochemicalAnalysis as Related to Sample Preparation”,Spectroscopy, vol. 3, No. 3, pp 24-33, March, 1998.

[9] Kester, B., AIEE Cement Industry Conference,Milwaukee, Wisconsin, 1960.

[10] Wheeler, B. D., and Newell, D., “Chemical Analysis ofMg-Cr Refractories by X-ray Fluorescence, Fall Meeting of the American Ceramic SOCIETY, RefractoriesDivision, Paper #13-RI-76F, October 9, 1972.

[11] LaChance, G. R., and Traill, R. J., “A Practical Solutionto the Matrix Problem in X-ray Analysis”, CanadianSpectroscopy, vol. 11, nos. 2-3, March-May, 1966.

[12] Lucus-Tooth, H. J., and Price, B. J., “A MathematicalMethod for the Investigation of Interelement Effects in X-ray Fluorescence Analysis”, Metallurgia, vol. LXIV, no.383, Sept., 1961.

[13] Rasberry, S. D., and Heinrich, K. F. J., Calibration ofInterelement Effects in X-ray Fluorescence Analysis,Analytical Chemistry, vol. 46, no. 1, Jan., 1974.

[14] Hasler, M. R., and Kemp, J. W., “Suggested Practicesfor Spectrochemical Computations”, ASTM CommitteeE-3, Report SM 2-3, American Society for Testing &Materials, Philadelphia, PA, pp 72-82, 1957.

[15] Sherman, J., “The Correlation Between Fluorescent X-ray Intensity and the Chemical Composition”, ASTMSpecial Publication, no. 157, pp 27-33, 1954.

[16] Burnstein, F., Particle Size and Mineralogical Effects inMining Applications of X-ray Analysis, Denver Research Institute, University of Denver, 1962.

[17] Campbell, W. J., and Thatcher, D., Advances in X-rayAnalysis, vol. 2, University of Denver, 1958.

[18] National Bureau of Standards, Washington, D.C.

[19] Ash Grove Cement Company, Kansas City,MO.

[20] Hercules Cement Company, Bethlehem, PA.

[21] Medusa Cement Company, Dixon, IL.

20 The Rigaku Journal

Vol. 16 No. 1 1999 21

Appendix A. Listing of Standards and Source

STD GEOLOGIC FORMATION REGION SOURCE

LS-1 Burlington Missouri Ash Grove Cement Co.

LS-2 Raytown Kansas Ash Grove Cement Co.

LS-3 Squamish British Columbia Ash Grove Cement Co.

LS-4 Kimswick Missouri Ash Grove Cement Co.

LS-5 Jacksonburg Pennsylvania NBC

LS-6 Jacksonburg Pennsylvania Hercules Cement Co.

LS-7 Farley Nebraska Ash Grove Cement Co.

Appendix B. Typical Limestone Spectra

22 The Rigaku Journal

Vol. 16 No. 1 1999 23

24 The Rigaku Journal

Vol. 16 No. 1 1999 25