Structure study of new uranium loaded borosilicate glasses

Journal of Non-Crystalline Solids 380 (2013) 71–77

Contents lists available at ScienceDirect

Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/ locate/ jnoncrysol

Review

Structure study of new uranium loaded borosilicate glasses M. Fábián a,b,⁎, E. Sváb b, M.v. Zimmermann c a b c

Centre for Energy Research, H-1525 Budapest, P.O.B. 49, Hungary Wigner Research Centre for Physics, H-1525 Budapest, P.O.B. 49, Hungary Hasylab at Desy, Notkestrasse 85, D-22603 Hamburg, Germany

a r t i c l e

i n f o

Article history: Received 6 June 2013 Received in revised form 1 September 2013 Available online 26 September 2013 Keywords: Borosilicate glasses; Radioactive waste disposal; Neutron diffraction; X-ray diffraction; Reverse Monte Carlo simulation

a b s t r a c t The structure of multi-component borosilicate SiO2–B2O3–Na2O–BaO–ZrO2 glass loaded with 30 wt.% UO3 was investigated by neutron diffraction and high-energy X-ray diffraction. Reverse Monte Carlo modelling was applied to obtain a possible 3-dimensional atomic configuration consistent with the experimental data. It was established that the glassy network consists of tetrahedral SiO4 and of mixed tetrahedral BO4 and trigonal BO3 units. With increasing boron content the relative number of BO3/BO4 increases and the first neighbour rB\O distance decreases. For the U\O correlations two distinct first neighbour distances were determined at 1.8 Å with 1.9 oxygen atoms and at 2.2 Å with 3.7 oxygen atoms. Significant second neighbour correlations have been established between uranium and the network former (Si, B), the modifier (Na) and the stabilizer (Zr) atoms. From these observations we may conclude that uranium ions take part in the network forming. This may be the reason of the observed good glassy stability and hydrolytic properties. © 2013 Elsevier B.V. All rights reserved.

1. Introduction High-level radioactive waste (HLW) produced by spent fuel reprocessing of civil nuclear reactors currently is incorporated into an inert host material. HLW contains actinides (fission and activation products), primarily U, and Pu that is the fission byproduct of the UO2 burning process. The vitrification technology for radioactive waste management commonly favours the use of borosilicate glasses. Borosilicates are generally accepted as proper HLW isolating media [1,2], as they have a unique blend of processing and product characteristics, which make them nearly ideal for this application. They are satisfying the following major requirements: the radioactive elements become immobilized as part of the host material structure, the leaching rate of radioactive elements is acceptably low, and the encapsulation cost is acceptable. Structural characterization of these glasses is essential for the understanding of glass durability. However, relatively few structural investigations have been performed on uranium-loaded glasses [3,4] because of the large number of constituent elements. Recently, we have prepared 5-component sodium borosilicate host (matrix) glasses SiO2–B2O3–Na2O–BaO–ZrO2, which proved to be stable and capable of hosting uranium [5–8]. Network formers SiO2 and B2O3 have strong covalent bonds involving tetrahedral SiO4 and mixed tetrahedral BO4 and BO3 triangles. These network formers are located in the centre of oxygen polyhedral configuration. The polyhedra are then tied together by sharing corners, thereafter various elements of HLW occupy ⁎ Corresponding author at: Centre for Energy Research, H-1525 Budapest, P.O.B. 49, Hungary. E-mail address: [email protected] (M. Fábián). 0022-3093/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jnoncrysol.2013.09.004

positions in this three dimensional network depending on the electronegativity, ionic size etc. Na2O serves as network modifier, while BaO and ZrO2 are glass and hydrolytic stabilizers. Modifiers form weaker bond with oxygen atoms than the network formers and, act to balance the negatively charged borosilicate network. These modifiers are useful in formulating glass with potential to incorporate high molecular weight compounds. In this work we present the structure determination of a new 6-component glass system, as far as we know not studied yet, [SiO2 · B2O3 · Na2O · BaO · ZrO2] + UO3 using both neutron diffraction (ND) and X-ray diffraction (XRD) experiments. Simultaneous RMC simulation of ND and XRD data was applied to generate a reliable 3-dimensional atomic configuration to get information on the short and medium-range order. The partial atomic pair-correlation functions, the characteristic distances and coordination-numbers are calculated with special interest on the uranium surrounding. The results for one concentration of this series, we have reported earlier [9]. 2. Experimental details 2.1. Samples The glassy samples were prepared by melt-quench technique. A high temperature electrical furnace was used with a platinum crucible under atmospheric conditions. The glasses of composition 70 wt.%[(65 − x) SiO2 · xB2O3 · 25Na2O · 5BaO · 5ZrO2] + 30wt.%UO3 with x = 5, 10, 15, 20 mol% (hereafter referred to as UB5, UB10, UB15, UB20) were prepared from raw materials of p.a. grade, SiO2, Na2CO3 supplied by Reactivul (Bucuresti), BaO and ZrO2 by Merck (Darmstadt), UO3

72

M. Fábián et al. / Journal of Non-Crystalline Solids 380 (2013) 71–77

(supplied by Reactivul, Bucuresti), B2O3 by Sigma-Aldrich Co. (Hungary). B2O3 was isotopically enriched in 11B in order to reduce the influence of the high neutron absorption of 10B present in natural boron. The 11B isotope enrichment was 99.6% as determined by Inductively Coupled Plasma Mass Spectroscopy (ICP-MS) technique [10]. The glasses were synthesized by melting the previously homogenized powder mixtures at 1350–1450 °C; the melting temperature decreases with increasing boron content. The melted mixture has been kept at the melting temperature for 2 h, meanwhile the melt was periodically homogenized by mechanical stirring. Thereafter the melt was cooled by 50 °C, kept there for 30 min, and was quenched by pouring it on a stainless steel plate. Powder samples were prepared by milling the quenched glasses in an agate mill. The density of the samples was measured by chemical method, using a hydrostatic balance based on Archimedes' principle. The measured density of the glasses was 3.75 ± 0.04, 3.68 ± 0.04, 3.62 ± 0.03 and 3.53 ± 0.03 g·cm−3 for UB5, UB10, UB15 and UB20, respectively. The number density, which is an important input parameter for the RMC modelling (see later in Section 3) was calculated from these experimental data, and resulted in an average number density of ρ0 = 0.08 atoms · Å−3, being the same value for all compositions within limit of error. The elemental composition was verified by Prompt Gamma Activation Analysis (PGAA) [11,12], the measured values were used in data evaluation procedure, as they are tabulated in Table 1. The samples proved to be fully amorphous, no crystalline phase was detected. For borosilicate glasses it is an often problem, that the glass is hydrolytic, and with time it becomes humid. In our previous studies we have prepared and studied several compositions based on silicate and borosilicate networks of 4-components (Si, B, Na, Ba – oxide glasses), and it was established that the amorphous samples partly crystallize and absorb hydrogen [5]. By adding Zr and U these properties could be improved [7–9]. The amorphous and the hydrolytic states of these new samples were periodically checked by ND measurements at ambient conditions. Anyhow, we have handled the samples with the greatest care by keeping them in a desiccator. 2.2. Neutron and X-ray diffraction experiments Neutron diffraction measurements have been performed at the 10 MW Budapest research reactor using the ‘PSD’ neutron powder diffractometer [13]. Monochromatic wavelength of λ0 = 1.068 Å was used. The diffraction spectrum was measured in the momentum transfer range of Q = 0.95–9.8 Å−1. The powder specimens of about 3–4 g were filled in cylindrical vanadium sample holder of 8 mm diameter, 50 mm height and 0.07 mm wall thickness. The specimens had to be handled with a special care due to their radioactivity. Correction and normalization procedures utilized to obtain the total structure factor S(Q) from the measured pattern was described in our previous work [8]. Fig. 1/a shows the ND experimental S(Q) data for the investigated samples together with the results of RMC simulation (details of the RMC modelling will be discussed in the next section). The high energy X-ray diffraction measurements were carried out at the BW5 experimental station [14] at HASYLAB, DESY. The powdered samples were filled into quartz capillary of 2 mm in diameter (wall Table 1 Elemental composition (at.%) of uranium loaded multi-component waste glasses measured by PGAA method. The error is ± 0.1 at.%. Samples

UB5 UB10 UB15 UB20

Elemental composition (at.%) Si

B

Na

O

Ba

Zr

U

15.2 14.4 15.9 11.1

3.4 5.9 9.1 13.7

12.7 12.7 11.8 8.1

61.0 59.8 55.5 61.1

1.3 1.4 1.2 0.5

2.8 2.4 2.9 2.3

3.1 3.1 3.3 3.1

thickness of ~ 0.02 mm). The energy of the radiation was 109.5 keV (λ0 = 0.113 Å). Raw data were corrected for detector dead time, background, polarization, absorption, and variations in detector solid angle. The XRD structure factors were obtained up to 25 Å−1. Fig. 2/a shows the XRD experimental S(Q) data for the investigated samples together with the results of RMC simulation (details of the RMC modelling will be discussed in the next section). 2.3. Experimental results The overall run of the ND and XRD S(Q) curves is fairly different as they are shown in Figs. 1/a and 2/a. The S(Q) curves are compared in the overlapping Q-range for UB5 composition in Fig. 3. to see the details better. The intensities or/and the positions of the oscillations posses unlike values: for the neutron S(Q) the peak positions are at 1.3, 2.0, 2.9, 5.3, 8.2 Å, while for the X-ray S(Q) at 1.3, 2.1, 3.7, 5.5, 7.4, 9.3 Å. The two first peak positions are at similar values, but their intensities are significantly different. This is due to the different weighting factors, wij, of the partial structure factors, Sij(Q) defined as: SðQ Þ ¼

k X

wij Sij ðQ Þ;

ð1Þ

i; j

wij

ND

ci c j bi b j ¼2 32 ; k X 4 ci b j 5

ð2Þ

i; j

wij

XRD

ci c j f i ðQ Þf j ðQ Þ ðQ Þ ¼ 2 32 ; k X 4 ci f i ðQ Þ5

ð3Þ

i; j

where ci, cj are the molar fractions of the components, bi, bj are the coherent neutron- and fi(Q), fj(Q) are the X-ray scattering amplitudes, and k is the number of elements in the sample. Note, that the neutron scattering amplitude is constant [15], while the X-ray scattering amplitude is Q-dependent [16] and for each atom in a somewhat different way. In order to see the weight of the different atom pairs in the total XRD S(Q) functions for the two radiations, Table 2 presents wND ij and wij (Q = 0.8 Å−1) for several atom pairs. It can be seen that the Si\O atom pairs have a significant contribution for both radiations, however, for ND experiment it is more dominant than for XRD, while the B\O plays a significant contribution only in the neutron experiment. The O\O contribution has a dominant weight in the neutron experiment, in contrast to the X-ray case, where it is much weaker. On the other hand the uranium surrounding, as a heavy element, appears with a significant weight in the XRD experiment. The experimental data were Fourier transformed to characterize the average atomic distributions. The atomic total pair-correlation function, g(r) was calculated (for more details see our recent work [8]): Z

Q max

g ðr Þ ¼ 1 þ

1 2πρ0 r

Q ½SðQ Þ−1 sin Qr dQ

ð4Þ

o

where ρ0 is the average number density, and Qmax is the integration limit determined by the actual upper limit of the experiment. The value of Qmax determines the r-space resolution Δr ¼ Q2π . In the ND max

experiment Qmax = 9.8 Å−1, which provides a rather low resolution ~0.6 Å, while in the XRD experiment Qmax = 25 Å−1 leading to a much higher resolution ~0.25 Å. The results are shown in Figs. 1/b and 2/b for the ND and XRD experiments, respectively. Some qualitative findings for the local structure can be made from the total g(r) functions. In the neutron case a broad first peak appears at 1.6 Å peak position for UB5, and it slightly shifts to lower values,

M. Fábián et al. / Journal of Non-Crystalline Solids 380 (2013) 71–77

73

Fig. 1. Neutron diffraction total structure factors (a) and atomic pair-correlation function (b) of uranium loaded borosilicate waste glasses. Experimental data (symbols) and RMC simulation (solid line). The difference between the experimental and calculated S(Q) data is shown at the bottom of the figure.

1.5 Å for UB20 composition (see Fig. 1/b). Due to the low r-space resolution this peak is the convolution of B\O and Si\O atomic pairs, for other atomic pair contributions we cannot make any conclusions. The second broad peak is centred at 2.6 Å, this can be attributed to the O\O contributions. The X-ray g(r) is more informative due to the relatively high r-space resolution. The first coordination sphere splits into well-resolved peaks centred at 1.6, 1.8 and 2.2 Å, and an asymmetry appears at low r-side which becomes even a small resolved peak at 1.4 Å for UB20 sample. The peaks may be attributed to B\O (1.4 Å), Si\O (1.6 Å), U\O1 (1.8 Å) and U\O2 (2.2 Å); the vertical lines in Fig. 2/b are guide for the eye. The small peaks in the 2.5–3.3 Å interval cannot be identified; they contain the Na\O, Zr\O, Ba\O and O\O distributions. It is noteworthy, that at higher r-values a relatively high intensity broad peak appears at around 3.7 Å. As far as in the neutron g(r) the intensity of this peak is considerably weaker, we may suppose that the uranium second neighbour correlations play a significant role in its formation. 3. Results from Reverse Monte Carlo modelling The investigated uranium loaded multi-component borosilicate glasses contain k = 7 different atoms, resulting in 28 gij(r) functions, according to the formula k(k + 1)/2. Several atomic pair-distributions overlap with each other, thus from the total g(r) functions it is not a realistic expectation to obtain the characteristic structural parameters. In such cases the reverse Monte Carlo simulation method [17] proved to be a very effective and widely used computer modelling procedure, however, because of the very high number of simulated parameters the obtained results of an actual RMC run have to be handled very carefully. The RMC minimizes the squared difference between the experimental S(Q) and the calculated one from a 3-dimensional atomic configuration. The RMC algorithm calculates the one-dimensional

partial atomic pair correlation functions gij(r), and they are inverse Fourier transformed to calculate the partial structure factors, Sij(Q):

Sij ðQ Þ ¼ 1 þ

4πρ0 Q

rZ max

h i r g ij ðr Þ−1 sin Qr dr;

ð5Þ

0

where rmax is the half edge-length of the RMC simulation box, and ρ0 is the number density. The structure of the computer configuration is modified by moving the atoms randomly until the calculated S(Q) (see Eqs. (1)–(5)) agrees with the experimental one within the experimental error. Moves are only accepted if they are in accordance with certain constraints (see below the ones applied in this work). In the present work we have utilized our previous experiences obtained on a systematic study of glassy materials starting from the simple B2O3 [18] and SiO2 [19], through the two component B2O3–Na2O [18] and SiO2–Na2O [19], followed by the three component SiO2– B2O3–Na2O and the five component borosilicate SiO2–B2O3–Na2O– BaO–ZrO2 host glass systems [6–8]. For all samples we have applied RMC modelling to obtain atomic structure parameters, and the results have indicated that the RMC simulation is a successful method to obtain stable and reproducible structural data for the investigated amorphous systems. In this study we have used the RMC software package developed by the group of Pusztai [20]. For the RMC starting model a disordered atomic configuration was built up with a simulation box containing 5000 atoms with ρ0 = 0.08 atoms · Å−3 and rmax = 19.84 Å for all investigated specimens. In the RMC simulation procedure two types of constraints were used; the minimum interatomic distances between atom pairs (cut-off distances) and connectivity constraints. We have used connectivity constraints for the two basic network formers, for the Si\O and B\O atom

Fig. 2. X-ray diffraction total structure factors (a) and atomic pair-correlation function (b) of uranium loaded borosilicate waste glasses. Experimental data (symbols) and RMC simulation (solid line). The difference between the experimental and calculated S(Q) data is shown at the bottom of the figure.

74

M. Fábián et al. / Journal of Non-Crystalline Solids 380 (2013) 71–77

Fig. 3. Comparison of neutron (open square) and X-ray (solid square) experimental data for UB5 sample.

pairs, and also for the U\O atom pairs. The latter will be discussed in the next section. It is widely accepted in the literature [21,22] and our previous results on similar compositions [7–9 and references therein] have concluded that silicon has tetrahedral 4-fold oxygen coordination (SiO4), while boron is surrounded both by 3-fold (BO3) and 4-fold (BO4) coordinated oxygen atoms. Thus, in the RMC calculations each Si atom was forced to be surrounded by 4 oxygen neighbours between 1.5–2.0 Å, and B atoms were forced to have either 3-fold or 4-fold oxygen neighbours between 0.9–1.7 Å. In the RMC simulation procedure we have used as starting cut-off distances, the characteristic values obtained in our previous work for UB10 composition [9]. Several RMC runs have been performed by modifying the cut-off distances in the way, that the results of each run have been carefully checked to obtain reasonable data for each gij(r) and coordination number distributions, Nij(CN). The RMC calculation converged well and the final S(Q) matched the experimental one, as it is displayed in Fig. 1/a for the ND and in Fig. 2/a for the XRD case. As far as the experimental and the calculated data practically fully overlap, we have calculated and illustrated the deviation curves in the same figures. The actual set of average cut-off distances used in the final RMC runs are the following: Si\O 1.53 Å, B\O 1.1 Å, O\O 2.47 Å, Na\O 2.05 Å, Zr\O 1.98 Å, Ba\O 2.55 Å, U\O 1.75 Å, U\B 2.8 Å, U\Si 3.3 Å, U\Na 3.3 Å, U–Zr 3.4 Å and U\Ba 3.4 Å. It should be noted, that the cut-off values were calculated independently for each composition by analyzing the best fit obtained for the experimental and calculated S(Q) curves. The actual cut-off values were very similar for each composition, therefore here the average values are given. The most informative atomic pair correlation functions and coordination number distributions obtained for the investigated samples are shown in Figs. 4 and 5. The nearest neighbour distances for several atom pairs are tabulated in Table 3, while the average first neighbour coordination numbers are summarized in Table 4. In order to characterize the network structure of the RMC generated atomic configuration, the atomic partial pair-correlation functions and the coordination number distributions have been analyzed.

The Si\O and Si\Si pair-correlation functions are fairly stable and do not change in dependence of concentration. The Si\O first neighbour distance is at 1.58 ± 0.03 Å which agrees within limit of error with the value 1.60 Å characteristic for the multi-component borosilicate host glass without uranium [8]. However, both distances are somewhat shorter than that for simple SiO2 glass, where rSi\O = 1.608 Å and 1.62 Å from ND and XRD, respectively [23]. Similarly, for Si\Si the 3.0 Å distance is at the same value as for the multi-component glasses and somewhat shorter than that for SiO2, where rSi\Si = 3.077 Å [23]. The Si\O coordination numbers show a slight decrease with increasing boron content, as it is tabulated in Table 4, namely from the value of 3.95 atoms decreases to 3.8 atoms, however, the changes are within limit of error. The B\O bond length slightly decreases from 1.55 Å to 1.35 Å (see Table 3) and the average coordination number also decreases from 3.4 to 3.2 atoms (see Table 4) from the UB5 to UB20 glasses with increasing B2O3 content. In our previous high resolution neutron diffraction study of the multi-component matrix glasses [8] we have established two possible local network configurations with a larger (1.60 Å) and with a shorter (1.40 Å) B\O bond distances corresponding to the 4-fold and to the 3-fold oxygen coordinated boron atoms. Taking into consideration the relatively low resolution of the present neutron diffraction experiment, the agreement is reasonable, especially for the UB5 and UB10 compositions. For all samples studied in this work 3- and 4- fold oxygen coordinated boron atoms were obtained and with increasing boron content the number of 3-fold coordinated boron atoms is increasing (see Fig. 5/b), which leads to the decrease of B\O nearest neighbour distance to 1.35 Å, in accordance with our earlier results reported for the multi-component borosilicate glasses [8]. In our earlier study on B2O3– Na2O glass we have discussed [18 and references therein], that Na+ ions convert the trigonal BO3 units into tetrahedral BO4 units, and the linkage of BO3 and BO4 form superstructural units as proposed originally by Krogh-Moe [24] and later by NMR studies [22,25] and also by model calculations based on empirical potentials [26]. In the present work the ratio of the B2O3/Na2O concentration increases from UB5 to UB20 samples, thus the sodium charge compensation effect for BO4 decreases, which leads to the increase of the number of BO3 with respect to BO4 units. The U\O pair distribution functions were obtained with a very good reproducibility from the RMC simulations, due to their relatively high ~17% weighting factor in the XRD experiment (see Table 2). Fig. 4/h displays gU\O(r), where two well-resolved peaks appear at relatively short distances, centred at U\O1 = 1.8 ± 0.05 Å and U\O2 = 2.2 ± 0.05 Å for all investigated samples. The first neighbour peak positions could be determined with a relatively high accuracy, because they do not overlap with other atomic pair-distributions. 4. Discussion Here we focus our interest on the analysis of the atomic surrounding of uranium ions and on the role of uranium in the network forming. The experimental g(r) data (see Fig. 2/b) and RMC modelling, without any

Table 2 Several weighting factors, wij (%) of uranium loaded waste glasses. The wXRD (Q) values are given at Q = 0.8 Å−1, while wND ij ij doesn't depend on Q. Samples

UB5 UB10 UB15 UB20

Weighting factors, wij

ND XRD ND XRD ND XRD ND XRD

Si\O

B\O

Na\O

Zr\O

Ba\O

O\O

U\O

U\Si

U\B

U\Na

U\Zr

Si\Si

15.4 7.5 14.0 7.5 14.2 7.1 10.1 7.0

5.5 0.5 9.3 0.5 13.0 1.1 20.0 2.4

11.3 4.7 10.8 4.7 9.2 3.9 6.5 3.9

4.9 5.6 4.1 5.6 4.5 5.0 3.6 5.6

1.6 4.0 1.6 4.0 1.3 3.3 0.6 1.9

43.1 7.8 40.7 7.8 34.6 6.4 38.9 10.0

6.3 17.6 6.2 17.6 6.0 16.9 5.6 21.9

1.1 8.5 1.1 8.5 1.2 9.4 0.7 7.7

0.4 0.5 0.7 0.5 1.1 1.5 1.4 2.6

0.8 5.3 0.8 5.3 0.8 5.2 0.4 4.2

0.3 6.5 0.3 6.5 0.3 6.9 0.3 6.5

1.4 1.8 1.2 1.8 1.4 2.0 0.6 1.2

M. Fábián et al. / Journal of Non-Crystalline Solids 380 (2013) 71–77

75

Fig. 4. Partial atomic correlations for the uranium loaded borosilicate glasses for UB5 (square), UB10 (open circle), UB15 (star), UB20 (triangle): a) Si\O, b) Si\Si, c) B\O, d) O\O, e) Na\O, f) Ba\O, g) Zr\O and h) U\O atom pairs.

connectivity constraint for the U\O neighbours, have shown that the first neighbour U\O distribution shows two well resolved distances at 1.8 and 2.2 Å. The average U\O coordination number is 5.5 atoms calculated for the 1.6–2.4 Å interval, which covers both U\O1 and U\O2 distributions. As far as, RMC modelling leads to a “most random” configuration without application of constraints, a relatively broad and symmetric coordination number distributions have been obtained, as we have reported in our previous work on UB10 sample [9]. In this work we have applied connectivity constraints for both U\O1 and U\O2 distributions in the RMC modelling, taking into consideration the overall good agreement of our findings for the two first neighbour distances at 1.8 and 2.2 Å and the average CNU\O = 5.5 atoms with the corresponding data obtained from the literature [27–31] on uranium borosilicate crystalline phases and glasses. Based on the above works, UO2+ uranyl groups can be characterized with two axial oxygen atoms 2 between 1.77–1.85 Å and with four to five equatorial oxygen atoms in interval 2.21–2.25 Å. These bond length values show that the uranyl ions have a well-defined local structure that includes two oxygen atoms in axial position and four additional oxygen atoms in equatorial plane and, forms UO6 cluster, with mixed U(V) and U(VI) coordination numbers.

Based on the similarity of the two characteristic first neighbour U\O distances determined in this work and in the above listed references and in the similar value of the average first neighbour U\O coordination number for the entire interval (1.6–2.4 Å), it is reasonable to suppose, that in our system the local uranium coordination is similar to the above systems. Thus, we applied connectivity constraints for the shorter and for the longer first neighbour U\O distances. In the RMC calculations each uranium atom of the 3-dimensional atomic configuration was forced to be surrounded by 2 oxygen neighbours at ~1.8 Å and by 4 oxygen neighbours at ~2.2 Å. The actual intervals of the connectivity constraints are indicated in Table 4 in brackets. The corresponding U\O1 and U\O2 coordination number distributions generated by the RMC algorithm are shown in Fig. 5/c,d. The connectivity constraints are fulfilled excellently, and the distributions do not show differences for the four specimens. The calculated average coordination numbers are CNU\O1 = 1.9 ± 0.2 atoms and CNU–O2 = 3.7 ± 0.2 atoms and the total first neighbour U\O coordination number is ~5.6 atoms. Significant correlations may be observed at higher r-values between uranium and the network former Si and B atoms, and with the modifier cations Na and Zr, as well, as it is displayed in Fig. 6. It should be noted,

76

M. Fábián et al. / Journal of Non-Crystalline Solids 380 (2013) 71–77

Fig. 5. Coordination number distributions for UB5 (black), UB10 (stripe), UB15 (cubes), UB20 (grey) glasses from RMC modelling: a) Si\O, b) B\O, c) U\O1, d) U\O2, e) Na\O and f) O\O atom pairs.

that usually the second neighbour partial correlation functions calculated from RMC simulation do not show such characteristic correlations. Although, these correlation functions are rather noisy because of the relatively low number of the contributing atoms in the RMC simulation box, clear correlations may be observed for U\Si at 3.4 ± 0.1 Å, for U\B at a somewhat shorter distance 3.0 ± 0.1 Å, and for the modifier cations at higher distances, U\Na at 3.8 ± 0.1 Å and U\Zr at 3.7 ± 0.1 Å. This indicates that uranium atoms are connected through an oxygen atom with the network former and modifier atoms, which build up the network structure. The basic glassy network consists of mixed

Table 3 Nearest neighbour atomic distances, rij (Å) in uranium loaded borosilicate waste glasses. The error is ± 0.03 Ǻ for Si\O; ± 0.05 Ǻ for Si\Si, B\O and U\O; while it is about ±0.1 Ǻ for other atom pairs, as estimated from the data obtained from different RMC runs. Atom pairs

Interatomic distances, rij (Å) Samples

Si\O Si\Si B\O O\O Na\O Ba\O Zr\O U\O U\Si U\B U\Na U\Zr

UB5

UB10

UB15

UB20

1.6 3.0 1.55 2.5 2.05 2.7 2.0/2.25 1.8/2.24 3.4 2.95 3.7 3.7

1.55 3.0 1.5 2.5 2.10 2.7 2.0/2.25 1.84/2.24 3.4 2.95 3.8 3.7

1.55 3.0 1.35 2.5 2.10 2.7 2.0/2.25 1.75/2.20 3.4 2.9 3.8 3.8

1.6 3.0 1.35 2.5 2.05 2.7 2.0/2.3 1.8/2.25 3.6 2.85/3.15 3.9 3.8

tetrahedral SiO4, tetrahedral BO4 and trigonal BO3 units, which are partly connected by uranium atoms. This is a fingerprint that uranium ions take part as network former or partly as network former, which leads to the well-defined glassy structure with a high chemical and hydrolytic stability. 5. Conclusions We have performed ND and high-energy XRD measurements on multi-component borosilicate host glasses loaded with 30 wt.% UO3 in order to get structural information on the basic network structure and on the uranium surrounding. The S(Q) data sets were simulated by RMC modelling to obtain a possible 3-dimensional atomic configuration consistent with the experimental data.

Table 4 The average first neighbour coordination numbers, CNij (atom) for the uranium loaded borosilicate glasses. The actual intervals (Å) are indicated in brackets. The error is estimated from different RMC runs: ±0.05 atom for Si\O; ±0.1 atom for B\O; ±0.3 atom for Na\O and O\O and ±0.2 atom for U\O atom pairs. Atom pairs

Coordination numbers, CNij Samples

Si\O B\O Na\O O\O U\O1 U\O2

UB5

UB10

UB15

UB20

3.95 (1.45–1.85) 3.4 (1.3–1.7) 3.6 (2.0–2.45) 5.9 (2.4–2.95) 1.90 (1.75–1.95) 3.69 (2.1–2.4)

3.92 (1.45–2.0) 3.4 (1.3–1.75) 3.7 (2.0–2.45) 5.6 (2.4–3.0) 1.97 (1.66–1.9) 3.71 (2.1–2.35)

3.80 (1.45–2.0) 3.1 (1.1–1.95) 3.6 (1.95–2.5) 5.2 (2.4–2.9) 1.89 (1.68–1.85) 3.80 (2.15–2.32)

3.87 (1.4–2.0) 3.2 (1.3–1.75) 3.6 (2.0–2.37) 5.2 (2.35–2.95) 1.92 (1.7–1.9) 3.60 (2.05–2.4)

M. Fábián et al. / Journal of Non-Crystalline Solids 380 (2013) 71–77

77

Fig. 6. Second neighbour correlations for the uranium loaded borosilicate glasses for UB5 (square), UB10 (open circle), UB15 (star), UB20 (triangle): a) U\Si, b) U\B, c) U\Na and d) U\Zr.

It was established that the basic network structure consists of tetrahedral SiO4 units and of mixed tetrahedral BO4 and trigonal BO3 units, being rather similar to the network structure of the corresponding host glass as reported in our previous study [8]. With increasing boron content the relative number of BO3/BO4 increases and the first neighbour rB\O distance decreases. For the U\O correlations two distinct first neighbour distances were determined at 1.8 ± 0.05 Å with 1.9 ± 0.2 oxygen atoms and at 2.2 ± 0.05 Å with 3.7 ± 0.2 oxygen atoms. Furthermore, significant second neighbour atomic pair correlations have been established between uranium and the network former (Si, B), the modifier (Na) and the stabilizer (Zr) atoms. From these observations we may conclude that uranium ions take part in the network forming. This may be the reason of the observed good glassy stability and hydrolytic properties. Acknowledgement The research was supported by the European Community's Seventh Framework Programme under grant agreement No.226716 and No.283883-NMI3, and by the Hungarian research fund OTKA-PD109384. References [1] K.S. Chun, S.S. Kim, C.H. Kang, J. Nucl. Mater. 298 (2001) 150. [2] L. Dewan, L.W. Hobbs, J.M. Delaye, J. Non-Cryst. Solids 358 (2012) 3427. [3] M. Karabulut, G.K. Marasinghe, C.S. Ray, G. Waddill, D.E. Day, Y.S. Badyal, M.L. Saboungi, S. Shastri, D. Haeffner, J. Appl. Phys. 87 (2000) 2185. [4] Y. Badyal, M. Karabulut, K. Marasinghe, M.L. Saboungi, D. Haeffner, S. Shastri, Mater. Res. Soc. Symp. Proc. 556 (1999) 297. [5] M. Fábián, E. Sváb, Gy. Mészáros, L. Kőszegi, L. Temleitner, E. Veress, Z. Krist. (Suppl. 23) (1996) 461. [6] M. Fábián, E. Sváb, Gy. Mészáros, Zs. Révay, E. Veress, J. Non-Cryst. Solids 353 (2007) 1941.

[7] M. Fábián, E. Sváb, Gy. Mészáros, Zs. Révay, Th. Proffen, E. Veress, J. Non-Cryst. Solids 353 (2007) 2084. [8] M. Fabian, E. Svab, T. Proffen, E. Veress, J. Non-Cryst. Solids 354 (2008) 3299. [9] M. Fabian, Th. Proffen, U. Ruett, E. Veress, E. Svab, J. Phys. Condens. Matter 22 (2010) 404206. [10] Zs. Varga, G. Surányi, N. Vajda, Zs. Stefánka, Microchem. J. 85 (2007) 39. [11] G.L. Molnár, Handbook of Prompt Gamma Activation Analysis with Neutron Beams, Kluwer Academic Publishers, Dordrecht/Boston/New York, 2004. [12] Zs. Révay, T. Belgya, G.L. Molnár, J. Radioanal. Nucl. Chem. 265 (2005) 261. [13] E. Sváb, Gy. Mészáros, F. Deák, Mater. Sci. Forum 228 (1996) 247. [14] H. Poulsen, J. Neuefeind, H.B. Neumann, J.R. Schneider, M.D. Zeidler, J. Non-Cryst. Solids 188 (1995) 63. [15] A.C. Hannon, ISIS Disordered Materials Database 2006, http://www.isis.rl.ac.uk/ disordered. [16] D. Waasmaier, A. Kirfel, Acta Crystallogr. A 51 (1995) 416. [17] R.L. McGreevy, L. Pusztai, Mol. Simul. 1 (1988) 359. [18] M. Fabian, E. Svab, Th. Proffen, E. Veress, J. Non-Cryst. Solids 356 (2010) 441. [19] M. Fabian, P. Jovari, E. Svab, Gy. Meszaros, Th. Proffen, E. Veress, J. Phys. Condens. Matter 19 (2007) 335209. [20] http://www.szfki.hu/~nphys/rmc++/opening.html. [21] J.M. Roderick, D. Holland, A.P. Howes, C.R. Scales, J. Non-Cryst. Solids 293 (2001) 746. [22] L.S. Du, J.F. Stebbins, J. Non-Cryst. Solids 315 (2003) 239. [23] A.C. Wright, J. Non-Cryst. Solids 179 (1994) 84. [24] J. Krogh-Moe, Acta Crystallogr. 15 (1962) 190; J. Krogh-Moe, Phys. Chem. Glasses 3 (1962) 101; J. Krogh-Moe, Acta Crystallogr. 18 (1965) 77; J. Krogh-Moe, Phys. Chem. Glasses 6 (1965) 46. [25] R.K. Mishra, V. Sudarsan, C.P. Kaushik, K. Raj, S.K. Kulshreshtha, A.K. Tyagi, J. Nucl. Mater. 359 (2006) 132. [26] L.H. Kieu, J.M. Delaye, L. Cormier, C. Stolz, J. Non-Cryst. Solids 357 (2011) 3313. [27] F. Farges, C.W. Ponader, G. Calas, G.E. Brown Jr., Geochim. Cosmochim. Acta 56 (1992) 4205. [28] G.K. Liu, H.Z. Zhuang, J.V. Beitz, C.W. Williams, V.S. Vikhnin, Phys. Solid State 44 (2002) 1433. [29] S.V. Stefanovsky, A.A. Shiryaev, J.V. Zubavitchus, A.A. Veligjanin, J.C. Marra, Glass Phys. Chem. 35 (2009) 141. [30] J. Ptiau, G. Calas, D. Petitmaire, A. Bianconi, M. Benfatto, A. Marcelli, Phys. Rev. B 34 (1986) 7350. [31] S.V. Krivovicchev, C.L. Cahill, P.C. Burns, Inorg. Chem. 41 (2002) 34.