Structural Characterization of Natural and Processed Zircons with X-Rays and Nuclear Techniques

1Departamento de Fı́sica, IFLP-CCT-CONICET, Facultad de Ciencias Exactas, Universidad Nacional de La Plata, CC 67, 1900 La Plata, Argentina 2Facultad de Ciencias Agrarias y Forestales, Universidad Nacional de La Plata, IFLP-CCT-CONICET, 60 y 119 s/n, 1900 La Plata, Argentina 3Departamento de Fı́sica, IFLP-CCT-CONICET, Facultad de Ciencias Exactas, Universidad Nacional de La Plata, CICPBA, CC 67, 1900 La Plata, Argentina 4Department of Engineering “Enzo Ferrari”, University of Modena and Reggio Emilia, Via P. Vivarelli 10, 41125 Modena, Italy 5Department of Engineering and Architecture, University of Parma, Parco Area delle Scienze 181/A, 43124 Parma, Italy 6Department of Science and Methods of Engineering, University of Modena and Reggio Emilia, Via Amendola 2, 42100 Reggio Emilia, Italy 7Department of Chemistry “G. Ciamician”, University of Bologna, Via Selmi 2, 40126 Bologna, Italy


Introduction
Zirconium is the 18th most abundant element on Earth.In nature it mainly occurs as the free oxide ZrO 2 (baddeleyite), but more commonly as zircon (ZrSiO 4 ).Mineral sands, the common name for zircon in connection with its most typical ore, are mainly exploited in the refractory and ceramic industries.e former application originates from zircon excellent thermophysical properties such as the low coefficient of thermal expansion, low electrical and thermal conductivity, and good corrosion resistance [1].
On the other hand, the widespread use of zircon sand in the ceramic industry (49% of the total consumption) depends on its use as a glazing and opacifying constituent.Natural zircon sands whose average grain size, for example, in granite rocks, is about 100-200 m are the major source of zirconium minerals for industrial applications; in this grain size range zircon can be used either as a feedstock to fused zirconia or in zirconium chemical plants as foundry sand or directly in the manufacture of some refractory materials.Other industrial applications require lower particle size usually restricted to two basic types: zircon flour (40-50 m) and micronized zircon ( < 5 m) used as an opacifier [2,3].e use of micronized zircon and zircon flour in ceramic products has been forecast to reach the 56% in 2012 of the total worldwide industrial consumption.Australia (the main producer), together with South Africa, United States, and India, accounts for over 80% of the global zircon production [3].
Beside about 2% Hafnium (Hf), zircon sands typically hosts REEs (Rare Earth Elements) and relevant amount of naturally occurring radionuclides, namely, tetravalent uranium (U) and thorium (), replacing zirconium in the crystal lattice.Uranium and thorium concentration in zircon sands range, respectively, between 5-4000 ppm and 2-2000 ppm, depending on age and location of the deposits [4].e levels of U and in zircon sands are therefore frequently high enough to include these materials within the so-called NORMs, Naturally Occurring Radioactive Materials requiring handling and management according to appropriate radioprotection standards [3].As parent nuclides of 238 U, 235 U, and 232 radioactive families they are accompanied by their respective radioactive descendants which altogether contribute to significant radioactivity levels in zircons compared to average rocks.Cumulative radioactivity of zircon sands ranges between 5000 and 12000 Bq/kg depending on ore age, weathering, and deposit location [5,6].Due to radioactive decays of U and families, in particular in the form of alpha radiation with energy ≥4,6 MeV whose emission is accompanied by relevant recoil interactions, the structure of zircon can be heavily damaged over geological times, resulting in a partially aperiodic state, the so-called metamict state [7][8][9].e metamict state can be described as a mixture of crystalline and amorphous domains whose extent depends on the dose of radiation absorbed by zircon [8,9].For this reason exposure time, or, in other words, the age of zircons together with the enrichment in uranium and thorium, plays a relevant role in the degree of metamictization of each deposit.e amorphous domains due to radiation can be restored by recrystallization during thermal processes on geological times [10].e crystalline and amorphous domains can be efficiently characterized by several instrumental techniques such as the widely employed electron microscopy (SEM [11], TEM [12]), X-Ray Diffraction (XRD) [8,13], Raman spectroscopy [7], and, less frequently, Neutron Diffraction (ND) [14], while the structure of the domains damaged by irradiation is not well known due to the aperiodicity of the system [10,15].e aperiodic or amorphous domains remain poorly studied since short range techniques of nanoscopic resolution are needed.e EXAFS experiments by Farges and Calas in metamict zircons from Madagascar and Japan show the coexistence of Zr(VII) and Zr(VIII) coordination, Zr(VIII) corresponding to the fundamental structure of the crystalline zircon and Zr(VII) produced by metamictization [15,16]. is observation allowed Farges to suggest that the metamict zircon structure could be supported by the existence of three different types of oxygen: (a) oxygen included in Zr-rich domains; (b) oxygen included in Si-rich domains; (c) oxygen as bridges among domains, O bridges being bound to two Zr, and one Si as in the crystalline structure of zircon.Furthermore, Trachenko et al. [17] have studied the effects of radiation damage on structural changes in zircon by computer simulations, proposing damaged regions with nonuniform density, increasing it from depleted core to a densified boundary.
Further information on the surrounding of Zr and Hf both in natural zircons from Canada and in synthetic zircons and hafnon was collected by the determination of the quadrupole hyperfine interactions [18,19].e remarkable sensitivity of the techniques employed in these studies allowed obtaining valuable in depth information, but the variability of situations and states presented by natural zircons due to their highly differentiated geological histories requires the contribution of extensive and demanding experimental research to fulfill statistical representativeness of zircon complexity.
In this work structural analysis of several zircon samples has been carried out using short range nuclear techniques, that is, Perturbed Angular Correlation (PAC) spectroscopy and Positron Annihilation Lifetime Spectroscopy (PALS) whose results were compared and integrated with those from the long-range X-Ray Diffraction.In particular XRD and PAC were used to both determine the amorphous fraction derived from cumulative alpha-decay events and analyze the atomic nanoconfigurations and defects of zircon sands [19].e study covers the comparative analysis of sands of different origins (United States, Australia, and South Africa) with the aim of characterizing the type and the extension of the regions constituting the metamict state in untreated materials from geographically distinct areas as well as the modifications observed as a consequence of the industrial milling process and the thermal treatment (annealing) in the production line.

Experimental Procedure
e samples studied in the present work are natural zircons from the United States, Australia, and South Africa (provided by Endeka Ceramics) in two different stages of their industrial elaboration: sands (raw material) were named according to their origin as USA, AUS, and SUD, respectively (see Table 1 for chemical composition and granulometry), and sands micronized at a size of 50 = 4 m were named as USA4, AUS4, and SUD4.e milling process was conducted with an air-jet milling system (INCO) that is able to produce, without the help of moving and classifying parts, micronized particles typically 100% below 5 m (laser particle size analyzer, Mastersizer 2000, Malvern) without causing damage to the products.e micronized sands were subsequently annealed at 1000 ∘ C for 2 h in air atmosphere using an electric oven and the processed samples obtained will be indicated as aUSA4, aAUS4, and aSUD4, respectively.

Radioactivity
Measurements.e radioactive content of the raw sand samples investigated and of the corresponding micronized derivatives was determined by high resolution -ray spectrometry using a HPGe p-type detector (Ortec, now Ametek).Aliquots of about 50 g of each sample were accurately weighted and made up to a standard geometry in a polystyrene jar.All the samples were counted with the same procedure.e gamma spectra were acquired for 24 hours each in order to optimize radioactivity measurements.
Qualitative and quantitative analysis of spectra were carried out relatively to a mixed radionuclide standard solution using the Ametek-Gamma Vision soware; analytical results were checked by certified reference materials, namely, UTS-3 and DH1-a both by CANMET.Analysis of NORM -emitters was performed using a library of -emissions chosen to this scope based upon the highest photonic yields % among those available for the members of all the three natural radioactive families [20].

XRD
Analysis.e X-Ray Diffraction spectra were recorded at room temperature using Cu-K radiation in a Philips PW1700 diffractometer.Quantitative analysis of the samples was performed by the combined Rietveld RIR (Reference Intensity Ratio) method.10 wt% of corundum (NIST SRM 674a annealed at 1500 ∘ C for 1 day to increase the crystallinity to 100 wt%) was added to all samples as an internal standard.e mixtures, ground in an agate mortar, were side-loaded in an aluminum flat holder in order to minimize the preferred orientation problems.Data were recorded in the 5 ∘ -140 ∘ 2 range (step size 0.02 ∘ and 6 s counting time for each step).e phase fractions extracted by the Rietveld RIR refinements, using GSAS soware and EXPGUI as graphical interface, were rescaled on the basis of the absolute weight of corundum originally added to the mixtures as an internal standard and therefore internally normalized [22].e background was successfully fitted by a Chebyshev function with a variable number of coefficients depending on its complexity.e peak profiles were modeled using a pseudo-Voigt function with one Gaussian (GW) and two Lorentzian (, ) coefficients.Lattice constants, phase fraction, and coefficients corresponding to sample displacement and asymmetry were also refined.e cut-off value for the calculation of the peak profiles in all refinements was 0.05%.

PALS Analysis.
Positron Annihilation Lifetime Spectroscopy (PALS) is based on the capability of positron to sense the electron density in the material; in consequence it is highly sensitive to open volumes in solids.In matter, a positron and an electron mainly annihilate by the emission of two gamma rays of 511 keV.Under given circumstances, both particle and antiparticle can form a bound state, called positronium (Ps) [23].e ortho-positronium (o-Ps, the spins of the two particles are parallel oriented) is commonly annihilated with an electron of the medium to give two gamma radiations.e measured positron and o-Ps lifetimes are related to the open volume defect size allowing characterizing the defect structure of the material under study.PALS measurements were done in a conventional fast-fast coincidence system with two scintillator detectors (one BaF 2 and one plastic BURLE) provided a time resolution (FWHM) of 260 ps.A 22 NaCl (10 Ci) radioactive source deposited onto a kapton foil (1.42 g/cm 3 ) and sandwiched between two identical samples was used.e analyzed sands were uniaxial pressed into 8 mm diameter pellets of around 1 mm thick.e source contribution and the response function were evaluated from an Hf metal reference sample using the RESOLUTION code [24].Positron lifetime spectra of 3 × 10 6 counts each were recorded at room temperature and analyzed with the POSITRONFIT program [24].Aer background and source contribution correction, the lifetime spectra were decomposed into different exponential decays, each positron state being characterized by a positron lifetime, , with certain intensity, (normalized).e 22 (), the perturbation function, for a static quadrupole interaction has the following expression: e frequencies, , are related to the quadrupole frequency = /20ℎ, by = () .e () coefficients are known functions of the asymmetry parameter, , defined by = ( − )/ , where are the principal components of the EFG tensor.e exponential function takes into account a Lorentzian frequency distribution of width around the values due to disorder and/or impurities or defects present in the lattice.
On account of the −3 dependence of the quadrupole hyperfine interaction, the technique is extremely localized and different surroundings can be determined.us, a relative fraction or intensity (! ) and the corresponding EFG parameters ( , , ) can be determined for each neighborhood.Finally, the experimental ratio () is fitted by () = 22 ∑ !22 (), where ! are the relative fraction of nuclei experiencing a given perturbation 22 () [25,26].PAC measurements were performed at room temperature using a standard setup with four conical BaF 2 scintillation detectors with a time resolution of 0.6 ns (FWHM).Details on the experimental setup adopted in this work can be found elsewhere [27].

Results and Discussion
e U and contents determined in the investigated samples (see Table 1) are comparable to those published by Bruzzi et al. [5] and to those regularly checked by the authors from Bologna University in the long term collaboration with important ceramic companies of the Modena Ceramic district.Consequently the NORM levels detected suggest that the zircons analyzed in this work are partially metamicted.e XRD patterns obtained in natural zircon sands only reveal the principal lines corresponding to ZrSiO4.Rietveld RIR refinements were performed on natural and micronized samples.e obtained cell parameters and quantitative analysis are shown in Table 2 while Figure 1 shows the refinement results for the micronized samples.A small decrease in lattice parameters is observed with respect to the standard ZrSiO4, probably consistent with the presence of interstitial defects typical of metamictic state.In agreement with U and content (Table 1) USA sand is the most crystalline material as compared with the other two samples investigated.In addition, the milling process reduced grain size and amorphous content (Tables 1 and 2), without changing cell parameters.e lifetime components resolved in the PALS spectra differ in the natural and processed zircon (Figure 2).e evolution of the spectra appears to be consistent with the structure of natural zircon where the longest lifetimes ( = 1800 ps) originate from positronium annihilating in cavities.e lowest lifetimes present both in natural and in processed zircon seem rather to correspond to positron annihilating in zircon crystalline bulk ( = 300 ps) or in vacancy-type defects of smaller size ( = 500 ps) than the cavities where positronium annihilates.
If the formation of o-positronium in the cavities of the natural sands is assumed, using the semiempirical relation between the o-positronium half-life (ns) and the average cavity dimension ( Å) where it is annihilated it is possible to estimate that the natural sands cavities are approximately of 80 Å3 ( ≅ 1800 ps).
As a consequence of the micronization process the cavities disappear and only crystalline materials with regions displaying intrinsic defects coexist.Further annealing heals the defects, leading to a crystalline state (Figure 2(c)).
e results of the quantitative Rietveld RIR analysis are in good agreement with PALS observations.Table 2 shows that the amorphous content decreases around 30% from zircon natural sands to the micronized ones favoring the crystalline ZrSiO 4 phase.
e PAC spectra (Figure 3(a)) and their associated Fourier transform (Figure 3(b)) for natural sands are typical for polycrystalline samples.All of them were analyzed using four quadrupole components according to the perturbation factor (3) (see Table 3).
Fitting the PAC spin rotation curves both for micronized sands and for those thermally treated revealed different combinations of the four interactions determined in the natural sands.e whole evolution of the relative fractions obtained for all the samples is shown in Figure 4. Relative to the micronization of the sands, PAC shows no changes in the amounts of the interactions.e thermal annealing of the micronized sands instead indicates a remarkable increase of the fraction ! for interaction 1 ; in addition, a systematic decrease of 2 fraction is observed.e decrease is higher for the SUD than for the USA and AUS zircon sands.e comparison of the !values in Table 3 and Figure 4 shows that the most relevant hyperfine interaction ( 1 , 1 = 101 Mrad/s and 40 < ! 1 % < 49) corresponds to Zr lattice site in crystalline zircons in agreement with quadrupole frequency values reported previously [19,27].With significant relative abundance (20 < ! 2 % < 35) very asymmetric and slightly distributed interaction 2 of quadrupole frequency ( 2 = 104-109 Mrad/s) similar to that of the crystalline zircon is also observed in the analyzed samples. is interaction has been observed in fully metamict zircons [19,29].e above-mentioned interactions are accompanied by the interactions of similar or lower abundance 3 (11 < ! 3 % < 22) and 4 (8 < ! 4 % < 9).PAC studies on synthetic zircon and mullite-zirconia system [27] determined the presence of a similar 3 interaction associated with aperiodic regions. is interaction which exists in synthetic zircons cannot be attributed to the metamict state.Interaction 4 has not been reported in natural zircons so far.us, PAC results show that the metamict state of zircon may consist in an atomic arrangement with two different Zr surroundings described by 2 and 4 interactions.Furthermore, PAC allows validating the model of the metamict state of three regions (a Si-rich region, a Zr-rich region, and a bridge region) suggested by Farges [15] since the model admits two nonequivalent sites for the Zr(VII).e fact that the 2 interaction has a quadrupole frequency similar to that of crystalline zircon ( 1 ) but more asymmetric and slightly distributed gives an experimental support to assigned it to Zr probes involved in the bridge zone since they resemble the zirconium environment in the crystalline zircon.On the other hand, quadrupole parameters of the 4 interaction are close to those reported for the Zr(VII) in the tetragonal metastable form of ZrO 2 [30]; therefore this interaction can be directly associated with the Zr-rich regions proposed in the model.Simple calculations of the quadrupole interaction based on the zircon structure, using the point charge model [31] for the first Zr coordination, provide additional support since they suggest the existence of interactions of high asymmetry parameter for configurations of one oxygen vacancy in the ZrO 8 group.
Summing up, in terms of the model proposed by Farges [15] for the metamict state of the zircon, it is possible to assign the following interpretation to the quadrupole interactions observed in the described experiment: (i) 1 and 3 interactions are the corresponding ones to probes immersed in crystalline zircon ( 1 ) that coexist with aperiodic zircon ( 3 ).(ii) 2 and 4 correspond to two nonequivalent zirconium sites of the metamict zircon in Zr-rich regions ( 4 ) and in the bridge region that contains Zr and Si ( 2 ).
In this circumstance, the addition of the relative fractions of 2 and 4 represents the extent or degree of metamictization  3 for the fitting parameters.
of the zircons: 29, 44, and 42% for USA, AUS, and SUD, respectively (see Table 3).e values show a reasonable agreement with those expected from the content of radioactive elements determined in the sands (see Table 1) and consequently with radiation dose [9].According to Farges and modeling the Zr-rich region as the core of a sphere surrounded by the bridge region as shown in Figure 5, it is possible to estimate an order of a magnitude for the size of the metamict regions based on the relative abundance of the 2 and 4 interactions.
In fact, the ratio of the relative fractions ! 2 /(! 2 + ! 4 ) is proportional to the ratio between the volume containing PAC probes experiencing 2 interaction and the volume containing all PAC probes of the metamict zone. is last volume ratio is plotted in Figure 6 versus the bridge region thickness , for different spheres sizes .If we assume that the metamictic region has a radius of about 40 Å, similar to other author's proposals (Weber et al. 1999 and [17]), our experimental results of the ratio ! 2 /(! 2 + ! 4 ), shown in Table 4, allow estimating the width bridge zone in about 15 Å (dashed sectors of Figure 6).e three applied techniques, XRD, PALS, and PAC, report similar degree of metamictization for all natural sands in agreement with the determined U and content.
e proposed model based on PAC results to describe the metamictic region is consistent with the presence of the small cavities determined by PALS within the Zr-rich region which is the less dense one.is feature is also in agreement with the computer simulations reported by Trachenko et al. [17].
A gradual diminution of the metamictic state as processing advances (micronized and annealed samples) is observed by all the techniques.However, while XRD and PALS show an apparent complete recovery of the crystalline state, PAC senses residual Zr environments inside the metamictic state.

2. 4 .
PAC Analysis.e Perturbed Angular Correlation (PAC) method is based on the hyperfine interactions of nuclear moments with extra nuclear fields.is technique provides a nanoscopic (i.e., short range) description of the Zr-ions nearest neighborhoods.e nuclei of 181 Hf, obtained by irradiation with thermal neutrons (RA3-reactor of the Comisión Nacional de Energía Atómica (CNEA)), of the 180 Hf natural impurities located at Zr sites constitute the radioactive probe nuclei in the zircon powders.ecascade at (133-482) keV, populated by the -decay of 181 Hf, was used to measure the quadrupole interaction of the 482 keV (+5/2) state of 181 Ta. e time differential anisotropy was calculated from the coincidence spectra (, ), where is the angle between detectors and is the time delay between the two gamma events.Aer subtraction of chance coincidence background, time spectra corresponding to angles of 90 ∘ and 180 ∘ between detectors were combined to obtain the ratio () = 2 (180 ∘ , ) − (90 ∘ , ) (180 ∘ , ) + 2 (90 ∘ , ) = 22 22 () .(2)

Figure 1 :
Figure 1: Rietveld refinements results for micronized zircon sands.Points are the experimental data while full line is the results from the fits.Column bars below represent peak position for zircon and corundum phases.e full line below represents the difference from data to refinement.

Figure 2 :
Figure 2: Lifetime components and their intensities determined by PALS for the natural (a), the micronized (b), and the annealed micronized sands (c).

Figure 3 :
Figure 3: PAC spectra for zircon sands.e experimental spin rotation curves 2 2 () are shown on (a); the Fourier transform amplitude spectra are plotted on (b).Solid lines represent the least-squares-fitted PAC spectra and their respective Fourier transforms obtained with the values given in Table3for the fitting parameters.

Table 1 :
Chemical composition (provided by Endeka Ceramics) for natural zircon sands.U and content and granulometry determined in natural and micronized zircon sands.

Table 3 :
Hyperfine quadrupole parameters fitted to the experimental PAC results.Errors are indicated as subindexes.