Lt

AF

a

ARRAA

KL(AUTN

1

irrootafAomborAei

s

(

0d

Talanta 78 (2009) 800–804

Contents lists available at ScienceDirect

Talanta

journa l homepage: www.e lsev ier .com/ locate / ta lanta

aser-induced breakdown spectroscopy for determination of uranium inhorium–uranium mixed oxide fuel materials

rnab Sarkar, Devanathan Alamelu, Suresh K. Aggarwal ∗

uel Chemistry Division, Bhabha Atomic Research Centre, Trombay, Mumbai 400 085, India

r t i c l e i n f o

rticle history:eceived 12 November 2008eceived in revised form 18 December 2008ccepted 19 December 2008vailable online 30 December 2008

eywords:

a b s t r a c t

Laser-induced breakdown spectroscopy (LIBS) has been developed for determining the percentage ofuranium in thorium–uranium mixed oxide fuel samples required as a part of the chemical quality assur-ance of fuel materials. The experimental parameters were optimized using mixed oxide pellets preparedfrom 1:1 (w/w) mixture of thorium–uranium mixed oxide standards and using boric acid as a binder.Calibration curves were established using U(II) 263.553 nm, U(II) 367.007 nm, U(II) 447.233 nm and U(II)454.363 nm emission lines. The uranium amount determined in two synthetic mixed oxide samples using

aser-induced breakdown spectroscopyLIBS)dvanced heavy water reactors (AHWR)raniumhoriumuclear fuel

calibration curves agreed well with that of the expected values. Except for U(II) 263.553 nm, all the otheremission lines exhibited a saturation effect due to self-absorption when U amount exceeded 20 wt.% inthe Th–U mixture. The present method will be useful for fast and routine determination of uranium inmixed oxide samples of Th and U, without the need for dissolution, which is difficult and time consum-ing due to the refractory nature of ThO2. The methodology developed is encouraging since a very good

obta

analytical agreement wasthe work.

. Introduction

The Indian nuclear power program has been conceived bearingn mind the optimum utilization of domestic uranium and thoriumeserves with the objective of providing long-term energy secu-ity to the country. Keeping in mind that India has to fall backn its vast thorium resources, which account for about one-thirdf the world’s thorium reserves (∼300,000 t) [1,2], third stage ofhe Indian nuclear power program is based on Th–233U fuel cyclend developing advanced heavy water reactors (AHWRs). Two dif-erent compositions of Th–U mixed oxide have been proposed forHWR fuel containing 3 and 3.75 wt.% of 233U [3]. The Th–U mixedxide pellets are generally prepared by the conventional powderetallurgy route, which has been tested and used for U- and Pu-

ased fuels. The qualitative as well as quantitative characterizationf the fuel materials for major as well as minor constituents isequired as a part of chemical quality assurance of nuclear fuels.t present, chemical and mass-spectrometric methods are being

mployed to determine the composition of major/minor elementsn Th–U mixed oxide with the desired accuracy and precision.

Laser-induced breakdown spectroscopy (LIBS) is an emissionpectrometric technique with several advantages over the conven-

∗ Corresponding author. Tel.: +91 22 25593740; fax: +91 22 25505151.E-mail addresses: skaggr@barc.gov.in, skaggr2002@rediffmail.com

S.K. Aggarwal).

039-9140/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.talanta.2008.12.046

ined considering the limited resolution of the spectrometer employed in

© 2008 Elsevier B.V. All rights reserved.

tional emission spectrometric techniques, especially for nuclearapplications [4]. The basic theory of the technique has beendescribed in several reviews [5–7]. However, LIBS has not beenextensively used for characterization of nuclear materials. Amongthe few notable work in the field of nuclear industry by LIBS includethe work carried by Fichet et al. which showed the applicabilityof LIBS for impurity quantification in both UO2 and in PuO2 [8].Shen and Lu have showed the applicability of uranium detection upto 462 ppm in glass samples using laser-induced breakdown spec-troscopy in combination with laser-induced fluorescence [9], whereas Cremer and co-workers showed a detection limit of 0.1 g L−1 foruranium in liquid by direct liquid analysis method [10]. With thehelp of very high-resolution spectrometer, the isotopic ratio of Uand Pu had been also determined by LIBS [11–13]. In our labora-tory, we have developed LIBS methodology to determine U and Thin aqueous solutions individually as well as in the presence of eachother [4]. The applicability of LIBS for the determination of traceconstituents in a thoria matrix has also been reported recently byus [14].

Instrumental techniques such as LIBS have huge potential forthe determination of uranium in mixed oxides of Th and U. Thoriabeing a refractory material, the dissolution of sintered thoria pel-

lets or a U–Th mixed oxide pellets is difficult and time consumingwhich can be avoided in LIBS. However, this is a pre-requisite forchemical as well as mass-spectrometric methods. In the presentwork, we report the application of LIBS for the rapid determinationof U in Th–U mixed oxide samples in the range up to 20% of ura-

anta 78 (2009) 800–804 801

npbmosi

2

2

Pedpawsdtte4waw(

2

Umobm0twwffg3Tpswtspt

TC

S

TTTTTTTT

Table 2Characteristics of spectral lines employed for determination of U by LIBS.

Element �ij (nm) Aij (×108 s−1) Ej (cm−1) Ei (cm−1)

U(II) 263.553 – – –U(II) 367.007 0.26 914.765 28154.450U(II) 447.233 – – –

A. Sarkar et al. / Tal

ium. The emission spectra of both Th and U being very complex,roper choice of emission lines of interest is very important. Cali-ration curves were obtained for four emission lines of U in Th–Uixed oxide. This paper presents details of the work carried out to

ptimize the experimental parameters like laser fluence and acqui-ition delay time and gives the results obtained for U determinationn two synthetic samples of Th–U mixed oxide.

. Experimental

.1. Instrumentation

A Spectrolaser 1000 M, from M/s. Laser Analysis Technologiesvt. Ltd. (now known as XRF scientific), Victoria, Australia wasmployed in this work. The details of the instrument have beenescribed elsewhere [4]. The instrument is equipped with a high-ower Q-switched Nd:YAG laser that yields 200 mJ of pulse energyt the fundamental IR wavelength (1064 nm) with a 7 ns pulseidth, at a repetition rate of 10 Hz. The laser is focused onto the

ample by a plano–convex lens with a focal length of 5 cm. Theiameter of the focal spot was 500 �m. The sample is placed inhe sample chamber in ambient atmosphere and is mounted on aranslation stage so that every laser shot hits on a fresh surface. Themission from plasma is then collected in front of the plasma at a5◦ angle with respect to the laser beam direction. Plasma emissionas collected by a 15-mm diameter imaging lens at a distance of

pproximately 50 mm, and focused onto silica optical cable fibershich deliver the plasma light to the entrance slit of spectrographs

Czerny–Turner configuration) equipped with CCDs as detectors.

.2. Sample preparation and analysis

Since certified reference materials (CRMs) for mixed oxide ofand Th are not available commercially, a series of synthetic

ixed oxide calibration standards were prepared. Known amountf highly pure ThO2 and U3O8 powders were weighed and thenlended and ground thoroughly for about 30 min. Initially, six Th–Uixed oxide calibration standards with U amount varying fromto 32% by weight were prepared. Subsequently two more syn-

hetic samples with U amounts of 4 and 20% were prepared andere treated as unknown to validate the calibration. The powdersere mixed with high purity (99.5%) boric acid powder, procured

rom S.D. Fine-Chem. Ltd., Mumbai, India, in equal-amount ratioor 15 min by blending and grinding thoroughly to obtain homo-eneous mixture. Mixed powder samples were then pelletized tocm diameter pellets by applying a pressure of 2 × 109 Pa for 5 min.able 1 gives the compositional data of the calibration standardsrepared in this work. The emission lines for U determination wereelected from the regions where minimal Th spectral interferences

ere observed and their intensities were normalized with respect

o the B(I) 249.774 nm. Boron was chosen as an internal standardince boric acid was used as the binder and its concentration in theellet was maintained constant irrespective of the composition ofhe mixed oxide.

able 1omposition of the calibration standards and two synthetically prepared samples.

tandards U concentration (%) Th concentration (%)

U-1-CAL 1.1 98.9U-2-CAL 2.1 97.9U-3-CAL 3.0 97.0U-4-CAL 8.1 91.9U-5-CAL 16.0 84.0U-6-CAL 32.0 68.0U-7-unknown 4.0 96.0U-8-unknown 20.3 79.7

U(II) 454.363 – 914.765 22917.451

�ij is the transition wavelength, Aij the transition probability, Ei and Ej are the energiesof the upper and lower level, respectively.

Identical conditions of laser fluence, repetition frequency andacquisition delay were used for all analyses. In the present exper-iment, 1 Hz repetition rate of laser pulse was used. For the LIBSanalysis, spectra of 100 shots were averaged. On each pellet includ-ing the calibration standards as well as the unknowns, triplicateanalyses, i.e., three 100 shot measurements were performed. Thestepping motor was fixed with a velocity of 0.2 mm s−1 so that eachlaser pulse focused on a fresh surface. The emission line intensitycan vary due to a variety of reasons that include fluctuations in laserpower due to dust in the beam path, time jitter, as well as due tothe changes in the focusing and collection optics. In order to reducethese effects and remove them from affecting the results obtainedby LIBS, the software is programmed to discard the raw data fromanalysis where the raw net line intensity is more than ±10% of theaverage intensity and then again re-averaging is done to get LIBSspectra.

3. Results and discussion

3.1. Selection of emission line

The four spectrographs allow simultaneous acquisition of a widerange of spectra, 180–850 nm for each laser shot, which containsalmost all the intense emission lines of the two actinide elementsunder study. Emission spectra of actinides are generally very com-plex and the situation gets aggravated when two actinides arepresent in major amount, as in the present case. Presence of alarge number of emission lines makes the spectra of mixed oxideas a band of emission lines at the present instrumental resolution(0.6 nm at 300 nm). To select an appropriate emission line for U,spectra of pure U, pure Th and a mixed oxide sample were recordedand compared critically. The pure U pellets were prepared in sucha way that the amount of U in the pure U pellets and in the mixedoxide pellets was same. The pure Th pellet was also prepared inthe same manner. Fig. 1 shows a comparison of the three spec-tra under identical conditions of analysis (laser energy of 100 mJand acquisition delay of 3.5 �s). By comparison of the emissionspectra, four spectral regions where spectral interferences fromTh spectrum were minimal in the U emission lines region wereidentified and the suitable emission lines were selected for U. Theemission lines used and their spectroscopic data are listed in Table 2[15]. Among these four emission lines, U(II) 263.553 nm and U(II)367.007 nm are among the reported prominent lines in ICP-AES[16].

3.2. Effect of laser fluence

To optimize the laser fluence, the calibration standard TU-4-CAL

was analysed at different laser fluence. Fig. 2 shows a comparisonof the spectra observed at different laser fluence. It is seen thatwith increasing laser fluence, the U(II) 367.007 nm emission linebecomes more and more prominent. At fluence above 56 J cm−2,the pellet was not intact, and crippled during analysis and hencelaser fluence was not increased further.

802 A. Sarkar et al. / Talanta 78 (2009) 800–804

Fs

3

bmacfUc

F

ig. 1. Comparison of the spectra of the individual actinides with the mixed oxidetandard under identical conditions of analysis.

.3. Temporal resolution for acquisition

The importance of optimization of the temporal resolution haseen discussed in literature [17,18]. A series of measurements wereade to determine the optimal time delay between the laser pulse

nd the start time of the LIBS spectra acquisition. Fig. 3 shows aomparison of emission spectra recorded using TU-4-CAL at dif-erent acquisition delays at a laser fluence of 56 J cm−2. For the(II) 367.007 nm emission line, the best spectral purity with suffi-ient signal intensity was obtained at an acquisition delay of 3.5 �s

ig. 2. Effect of laser fluence on the U(II) 367.007 nm emission line (TU-4-CAL).

Fig. 3. Comparison of the effect of acquisition delay on the U(II) 367.007 nm emissionline (TU-4-CAL).

which was selected as optimum acquisition delay for subsequentanalyses.

Though for the complete optimization of laser fluence and acqui-sition delay time, it would be ideal to study the effect on signal tonoise ratio (SNR) of each of these parameters keeping the other asconstant. In the present work, a rough optimization was carried outin the manner presented above, which provided reasonable goodoptimization of the experimental parameters with less experimen-tal effort.

3.4. Calibration curves and precision

Calibration standards of mixed oxide were used to prepare cal-ibration curves of all the four normalized line intensities of Uemission lines with respect to B(I) 249.774 nm versus the corre-sponding concentration of U, as shown in Fig. 4. Each calibrationpoint corresponds to an average of three measurements at differ-ent locations of the corresponding pellet. The practice of discardingspectra with raw line intensity more than ±10% than the averagevalue was also followed here as discussed in the Section 2.2. Thenumber of spectra that are discarded under this criterion would beuseful to determine the statistical significance of the result obtainedin this study. Since there is no provision in the present softwareemployed for the spectral analysis to obtain this information, sothis detail could not be shown in Fig. 4. The error bars on calibrationpoints are the standard deviations (±1�), calculated by measuringthe normalized peak intensity of the emission line. Normalizationof emission line intensity was done after discarding outlier spec-tra as stated above. The calibration data were analyzed using theinstrument software and applying least squares regression analysis.

The calibration curves show a linear behavior below 8 wt.% Uin the Th–U mixed oxide. Above 8 wt.% U, saturation is observeddue to self-absorption of the lines in the plasma, a feature com-monly observed in laser-induced plasmas at atmospheric pressure[19–21]. For this region, a non-linear equation was used to fit theexperimental data, similar to the expression proposed by Aragonet.al. [22],

y = y0 + A e(−x/t) (1)

where x denotes wt.% of U and y stands for normalized emissionline intensity. In Eq. (1), the absolute value of “t” indicates theconcentration where the calibration slope decreases by a factorof 1/e from its value of (A/t) at x = 0. Hence this value (i.e., x = t)can be regarded as the critical amount above which laser-induced

A. Sarkar et al. / Talanta 78 (2009) 800–804 803

F 07 nm, (c) U(II) 447.233 nm and (d) U(II) 454.363 nm. The broken line corresponds to thel

sovdloapcalavd2ge

tgau

TPc

W

UUUU

Table 4Comparison of analytical results for U determination in Th–U mixed oxide sampleswith expected values.

Sample Emissionwavelength (nm)

Wt.% U A/B

LIBS (A) Expected (B)

TU-7-unknown 263.553 3.72 ± 0.43 4 0.93367.007 3.82 ± 0.29 0.96447.233 – –454.363 3.41 ± 0.52 0.85

ig. 4. Calibration curves obtained for the lines (a) U(II) 263.553 nm, (b) U(II) 367.0inear behavior of optically thin plasma.

elf-absorption becomes appreciable and therefore, the analysisf such elements in the sample should be restricted up to thisalue for using the particular calibration curve. Table 3 gives theata of different parameters in Eq. (1) for the four U emission

ines. Except from the U(II) 263.553 nm, the values of “t” for thether three emission lines are close to 20. This indicates that themount of U species in plasma for the Th–U mixed oxide com-osition with 20 wt.% U is sufficiently high. The value of “t” inase of U(II) 263.553 nm is 31.9. Extent of self-absorption is usu-lly high for resonant emission lines or lines having lower energyevel close to the ground level. Elower for both the U(II) 367.007 nmnd U(II) 454.363 nm emission lines is 914.765 cm−1, which isery near to the ground level, indicating the possibility of higheregree of self-absorption. The higher value of “t” in case of U(II)63.553 nm is not understood at present, but a proper energy dia-ram for the above transition might be able to explain the observedffect.

Table 4 show the results obtained for uranium in the two syn-

hetic samples treated as unknown. The wt.% of U determined is inood agreement with the expected value and there is no system-tic bias. The calibration curve of U(II) 447.223 nm was not used fornknown synthetic sample with U amount of 4 wt.%, as the cali-

able 3arameters of Eq. (1) obtained by fitting calibrations curves and the correlationoefficients.

avelength (nm) y0 A t R2

(II) 263.553 0.049 −0.04 31.92 0.979(II) 367.007 0.059 −0.05 19.08 0.996(II) 447.233 0.018 −0.02 18.73 0.994(II) 454.363 0.063 −0.06 20.04 0.995

TU-8-unknown 263.553 19.4 ± 1.13 20.3 0.96367.007 19.4 ± 0.14 0.96447.233 20.67 ± 2.01 1.02454.363 21.14 ± 2.44 1.04

bration curve obtained using this emission line was not sensitive inlower concentration range (Fig. 4).

4. Conclusions

The LIBS method has been developed for the determination of Uin mixed oxides of thorium and uranium. The approach will be use-ful for the rapid determination of U in these samples, avoiding therigorous dissolution required for other analytical techniques. Thereproducibility of the determination is ±2% (1�). Though the addi-tion of boron in nuclear material is undesirable, the result showsthe potential of the LIBS method in nuclear industry. The effect ofalternative carbon-based binder such as starch, nitrocellulose, can

also be explored in these studies. Considering the poor resolutionpower of the instruments and also the relatively rough optimiza-tion procedure employed, the agreement between the expectedand experimental results are quite good, which indicated the enor-mous potential of LIBS and the possibility of using a high-resolution

8 anta 7

LtdTfic

A

ceF

R

[

04 A. Sarkar et al. / Tal

IBS for routine industrial work for U–Th mixed oxide charac-erization. LIBS also provides an independent approach based onifferent physico-chemical principle for the determination of U inh–U matrix. This would be useful for the development of certi-ed reference materials for these matrices in thorium-based fuelycle.

cknowledgements

The authors are thankful to Dr. V. Venugopal, Director, Radio-hemistry and Isotope Group, B.A.R.C. for his constant support andncouragement in LIBS work. The authors wish to thank Mr. S. Panja,.R.D., B.A.R.C. for providing high purity U3O8.

eferences

[1] J.B. Hedrick, U.S. Geological Survey, Mineral Commodity Summaries, January1998, p. 177.

[2] http://www.uic.com.au/nip67.htm[3] R.K. Sinha, A. Kakodkar, Nuclear Eng. Design 236 (2006) 683.[4] A. Sarkar, D. Alamelu, S.K. Aggarwal, Appl. Opt. 47 (2008) G58–G64.

[

[[

8 (2009) 800–804

[5] D. Cremers, L. Radziemski, Handbook of Laser Induced Breakdown Spec-troscopy, 1st ed., Wiley, West Sussex, 2006.

[6] V. Majidi, M.R. Joseph, Crit. Rev. Anal. Chem. 23 (1992) 143.[7] A.V. Pakhomov, W. Nichols, J. Borysow, Appl. Spectrosc. 50 (1996) 880.[8] P. Fichet, P. Mauchien, C. Moulin, Appl. Spectrosc. 53 (1999) 1111.[9] X.K. Shen, Y.F. Lu, Appl. Opt. 47 (2008) 1810.

[10] J. Wachter, D.A. Cremers, Appl. Spectrosc. 41 (1987) 1042.[11] H. Liu, Z. Zhang, A. Quentmeier, K. Niemax, Spectrosc. Spectral Analysis 24

(2004) 1244.12] B.W. Smith, A. Quentmeier, M. Bolshov, K. Niemax, Spectrochim. Acta Part B 54

(1999) 943.[13] C.A. Smith, M.A. Martinez, D.K. Veirs, D.A. Cremers, Spectrochim. Acta Part B 57

(2002) 929.[14] D. Alamelu, A. Sarkar, S.K. Aggarwal, Proceedings of National Laser Symposia—6,

Indore, India, 2006, p. 90.[15] http://physics.nist.gov/PhysRefData/Handbook/index.html.[16] R.K. Winge, V.J. Peterson, V.A. Fassel, Appl. Spectrosc. 33 (1979) 206.[17] B. Castle, K. Talabardon, B.W. Smith, J.D. Winefordner, Appl. Spectrosc. 52 (1998)

649.[18] B. Sallé, D.A. Cremers, S.M. Roger, C. Wiens, Spectrochim. Acta Part B 60 (2005)

479.[19] D. Alamelu, A. Sarkar, S.K. Aggarwal, Talanta 77 (2008) 256.20] V. Lazic, R. Barbini, F. Colao, R. Fantoni, A. Palucci, Spectrochim. Acta Part B 56

(2001) 807.21] C. Aragon, J. Bengoechea, J.A. Aguilera, Spectrochim. Acta Part B 56 (2001) 619.22] C. Aragon, J.A. Aguilera, F. Penalba, Appl. Spectrosc. 53 (1999) 1259.