Sulfur behavior in silicate glasses and melts: Implications for sulfate incorporation in nuclear waste glasses as a function of alkali cation and V2O5 content

https://doi.org/10.1016/j.jnoncrysol.2006.09.041Get rights and content

Abstract

The presence of sulfur in radioactive waste to be incorporated in borosilicate glasses entails difficulties mainly due to the relatively low solubility of sulfates in the vitreous phase. In this work a study is presented on the effects of the ratio R = [Na2O]/[B2O3], the type of sulfate added and the addition of V2O5 on the incorporation of sulfates in borosilicate glasses. Glass samples were prepared at the laboratory scale (up to 50–100 g) by melting oxide and sulfate powders under air in Pt/Au crucibles. XRF and ICP/AES chemical analysis, SEM/EDS, microprobe WDS and Raman spectroscopy were employed to characterize the fabricated samples. The main experimental results confirm that the incorporation of sulfates in borosilicate glasses is favored by the network depolymerization, which evolves with the ratio R. The addition of V2O5 seems to accelerate the kinetics of sulfur incorporation in the glass and, probably, increase the sulfate solubility by modifying the borate network and fostering the formation of voids of shape and size compatible with the sulfur coordination polyhedron in the glassy network. The kinetics of X2SO4 incorporation in the glass seems to be slower when X = Cs.

Introduction

Silicate and borosilicate glasses are systems of great importance for both glass industry and earth sciences, in particular in the field of domestic/nuclear glasses, volcanology and the study of the mass transfer. Sulfur solubility in silicate melts is a particularly relevant aspect in the domain of nuclear glasses [1], [2] and in basaltic melts produced in subduction zones [3].

Sulfur is present in several kinds of nuclear waste destined for confinement in a glass matrix. In particular, the most common high-sulfate content waste species are generally residuals of incineration or of ion-exchanging resins or effluents of radioactive solutions. Such species are difficult to incorporate in borosilicates glasses, the most widely used compositions for nuclear waste vitrification, due to the poor miscibility of molten sulfates in a borosilicate melt [2], [4], [5], [6], [7], [8]. This feature may constitute an important technological issue in the development of a vitrification process as it can dictate the radioactive waste-load-limiting factor [6]. The formation of an immiscible sulfate layer floating on the surface of the borosilicate melt during the waste vitrification process is highly undesirable for at least three reasons:

  • the immiscible layer may include radioactive fission products [2] and therefore hinder their incorporation into the glass matrix;

  • it is normally a good conductor of electricity and heat [7], reducing the efficiency of the melting process both in the induction heated melter with cold crucible and in the liquid-fed ceramic melter with electrodes;

  • it can be highly corrosive for crucible and furnace [8].

Currently the maximum amount of S6+ (generally reported as weight percentage of SO3) that can be included in nuclear borosilicate glasses without formation of an immiscible phase is limited to about 0.6 mol%, i.e. not higher than the sulfate solubility in silicate melts estimated about 40 years ago [5]. Within a given range of glass composition, this value can vary as a function of the temperature, the reactivity and red-ox conditions during the elaboration process, and of the partial pressure of sulfur-containing species over the glass melt [9], [10]. Moreover, most of the numerous studies appeared in the last two decades on the vitrification of sulfur-bearing species (e.g. [11]) led to the conclusion that the glass frit composition and structure considerably influence sulfur solubility in the final glass.

From a structural point of view, the most important parameters defining the molecular arrangement in a sodium-borosilicate glass are the relative ratios K = [SiO2]/[B2O3] and R = [Na2O]/[B2O3] (where [·] indicates mol%), as suggested in the model proposed by Yun and Bray [12] and by Dell et al. [13]. This model, based on experimental 11B NMR results, describes the structure of SiO2–B2O3–Na2O glasses with K  8, by dividing the compositional range in sub-regions as a function of the values of R and K. More recent publications reporting 11B, 17O, 23Na and 29Si MAS-NMR and Raman studies on both molten and sol–gel prepared alkali borosilicate glasses [14], [15], [16], [17], [18] have shown that in general alkali cations are more randomly mixed to the borate and silica networks than was assumed in [12], [13]. In particular, in [12], [13] the formation of tetra-coordinated boron atoms in Na2O · B2O3 · 2SiO2 danburite-like units (e.g. [14]) was overlooked, which affected a rigorous definition of the structural sub-regions described in the model of Yun, Bray and Dell. This model is nonetheless useful for an approximate description of the structure evolution as alkali cations are added to the glass melt. In such a model, each structural sub-region is characterized by a different effect of the addition of network modifier Na2O, which finally leads to a higher depolymerization degree of both silica and borate networks as R increases. A fundamental point in this description is the critical composition R (=0.5 + 1/16K). As R is smaller than R Na+ cations interact as charge – compensators: for four-coordinated boron atoms [19] in the domain R  0.5 and for 1/2(Na2O · B2O3 · 6SiO2) reedmergnerite units [12] in the range 0.5 < R  R. The value of R should be changed to 0.5 + 3/16K if the formation of danburite-like units is also taken into account, as in [14], however the glass structure evolution is likely to be even more complicated, and further experimental results are certainly needed for a sounder definition of R.

For R > R additional alkali cations cause depolymerization of the glassy network, starting to form non-bridging oxygens (NBOs) in silica tetrahedrons. NBOs are then formed also in borate units at higher sodium contents (for R  0.5 + 0.25K).

A first goal of the present investigation is the study of the sulfate solubility in the depolymerized borosilicate melt as a function of the alkali cations (Na, Cs) concentration. Several authors report that sulfate solubility is improved by an increase of network modifier content, i.e. by an increase of network depolymerization [6], [7], [8], [20], [21], [22], [23]. This effect can be described in simple terms starting from the chemical model proposed by Ooura and Hanada [21]. Using the Fincham–Richardson notation [24], the formation of a non-bridging oxygen (NBO), i.e. the glass depolymerization, can be expressed by the following reaction:O0 + O2−  2Owhere O0 indicates a bridging oxygen (BO), O2− a free oxygen (FO) and O a non-bridging oxygen (NBO). As suggested by Papadopoulos [10], in general the concentration of O2− can be neglected in a borosilicate glass containing more than 33.3 mol% of SiO2, and the number of NBO can be related to the network modifier content. Thus Ooura and Hanada assumed that all O2− in the glass came from the main equilibrium governing the sulfate incorporation in the glass [25]:SO42−  SO3 (g) + O2−This assumption is acceptable if the formation of more reduced sulfur species can be neglected, as for most practical cases at temperatures below 1800 K and oxygen potential higher than 10−6 bar [26], [27]. The model of Ooura and Hanada combines the equilibrium constants of reactions (1), (2) to obtain a quadratic dependence of sulfate solubility as a function of the number of NBO in the glass. Combining reaction (2) with the equilibriumSO3  SO2 + 0.5O2and using the more general treatment proposed by Pinet and Di Nardo [28], one obtains the following formula for the sulfate activity in the glass:aSO42-fSO2[NBO]2fO2[BO]In formula (4), fα is the fugacity of the gaseous species α. Dependence on the gaseous atmosphere (fSO2 and fO2) and on the glass basicity will be considered in a further development of this research.

Another point investigated in this work is the effect of the addition of V2O5 to control the sulfate behavior in the glass and melt. The Radon Russian research team [29] suggested that the presence of V2O5 in the glass matrix enhance the liquid state miscibility of sulfates and glass melt, leading to the incorporation, in the final glass, of a higher waste volume. According to Stefanovskii [4], [29], the capacity of a vitreous matrix to incorporate sulfates depends on the compatibility between the crystallo-chemical parameters of the sulfate and those of the cations contained in the host-glass. These parameters are: the cation-oxygen binding energy; the Dietzel field strength zc/a2 where z = charge of the cation and a = (cation radius + O2− radius); the strength of the bond zc/k where k = coordination number of the cation; the size and the polarization of the polyhedrons. According to this criterion, the cations present in the glass may be distributed in the series:SIV6+PIV5+VIV5+BIII3+SiIV4+BIV3+AlIV3+ZnIV2+where the subscript roman numbers indicate the cation coordination. In the presence of a single network modifier (e.g. Na+), the closer are the ions in the series, the better is their reciprocal compatibility for the formation of/incorporation in the glass. Thus for instance sulfate and phosphate tetrahedrons have a better compatibility than sulfate and silicate tetrahedrons. In agreement with these considerations, phosphate glasses are known to have a good affinity for sulfate incorporation [30]. This feature is also linked to the high reactivity of phosphate glasses, mainly attributed to the presence of a Pdouble bondO bond in the phosphate tetrahedron, that implies on the other hand a poorer chemical durability.

The formulation of borosilico-vanadate glasses for nuclear waste immobilization was based on the analogies between the ions V5+ and P5+. Crystalline V2O5 consist of zigzag ribbons of square-base VO5 pyramids [31], [32], [33]. However, vitreous [34] and liquid [35] vanadium pentoxide can contain, in addition to VO5 pyramids [36], V5+O4 tetrahedra with a Vdouble bondO apex and three bridging oxygens, very similar to the PO4 units in crystalline and amorphous phosphate. The goal in [29] was to empirically create a glass matrix presenting the advantages of the phosphate glasses – i.e. the ability to incorporate a high content of sulfates – and having at the same time a better chemical durability than the phosphate glasses themselves. Stefanovskii [29] recommends that the best conditions to synthesize a borosilico-vanadate glass are at 1370–1425 K, with the following composition range: 20–30 wt% SiO2, 20–40 wt% B2O3, 10–25 wt% Na2O, 5–20 wt% V2O5 and 1–10 wt% SO3. Although these borosilico-vanadate glasses revealed to be a good compromise between high sulfate solubility and acceptable chemical durability, the very high sulfate solubility obtained by the Radon group in borosilico-vanadate glasses with low B2O3 content (6 wt%) was never reproduced by any other research team (as mentioned also in [37]). Therefore it is doubtful whether those early results should be considered as reliable or not. It is anyway certain that the presence of V2O5 has a positive effect on the sulfate incorporation in a borosilicate glass matrix, although Electron Paramagnetic Resonance (EPR) spectra reported in [29] do not show any strong coupling between V5+ ions and borosilicate network. Finally it was postulated that V5+ ions could be distributed in the voids of a boron–silicon–oxygen network, loosening it and therefore favoring the inclusion of sulfate ions into it. However, a detailed description of the experiments carried out in [29] is unfortunately not available, and a complete explanation of the effect of vanadium pentoxide on sulfate solubility in borosilicate glasses is still missing. McKeown et al. [37] performed X-ray absorption (EXAFS and XANES) studies on the valence of vanadium in borosilicate waste glasses. Such studies did not show any experimental evidence of the existence of vanadium–sulfur bonds nor of vanadium bonding to sulfate tetrahedra. Vanadium resulted to be present in the glass as V5+O4 tetrahedra and as penta-coordinated V4+O5 partially disordered sites. The population of these latter would increase in a reducing environment, obtained in [36] by adding sugar to the glass melt. Under the experimental conditions in [29], as well as in this work, vanadium can be considered to be present as V5+ only, at least in a first approximation (see also [38]).

In the current work a survey of the experimental results recently obtained at the CEA Marcoule on the sulfate behavior in borosilicate glasses is presented. Original ‘dynamic’ experiments, in which the behavior of sulfate in the vitreous matrix was studied as a function of the glass elaboration time, revealed the role of glass composition (including V2O5 and network modifier content) and reactivity conditions in sulfate inclusion dynamics. The results of such experiments were combined with a systematic Raman micro-spectroscopy analysis in order to assess the influence of the glass structure on sulfur incorporation.

Section snippets

Sample preparation and characterization

Borosilicate glass samples were obtained at the laboratory scale (up to 50–100 g) by melting oxide and sulfate powders in Pt/Au crucibles. Sample elaboration was carried out in a muffle alumina furnace at a temperature depending on the glass composition, in order to operate at a 100-Poise fixed viscosity level of the glass melt (Telaboration was therefore included between 1073 and 1473 K, with a heating rate of 250 K/h). After quenching in air to ambient temperature, the glass was crushed and

Behavior of the sulfate slag

Fig. 1(a) shows a micrograph of one of the investigated borosilicate glasses elaborated with the addition to the frit of high sulfate content (theoretical 5% in weight). The presence of an extended immiscible sulfate layer is clearly visible on the sample surface, and traces of solidified sulfate are visible inside the holes left by gas bubbles on the lateral surface of the sample (Fig. 1(b)). The size of the immiscible layer is considerably reduced if the elaboration time of the sample is

Sulfate behavior in contact with the glass melt

The study of the behavior of the sulfate immiscible slag and the results obtained with dynamic elaboration experiments permit to define four stages in the process of sulfate migration from salt phase to glass and gas:

  • (1)

    A fraction of the sulfate reacts with the glass and gets incorporated into the vitreous matrix already in the early stages of the melting process (in the first 15 min approximately).

  • (2)

    The molten sulfate that does not react with the glass during the fusion process quickly reaches the

Conclusions

The results obtained in this work point out some general effects of the glass melt composition and reactivity on sulfur incorporation in sodium–borosilicate glasses. These results can be summarized in the following points:

  • (1)

    It is confirmed that, within the investigated range of K = [SiO2]/[B2O3] (1.58  K  3.62), sulfur incorporation in the sodium–borosilicate matrix approximately increases like the square of R = [Na2O]/[B2O3] for R > R. This result corroborates the conclusions of previous publications.

Acknowledgement

The authors wish to thank T. Advocat for his valuable advice.

References (59)

  • M.J. Plodinec

    J. Non-Cryst. Solids

    (1986)
  • Y.H. Yun et al.

    J. Non-Cryst. Solids

    (1978)
  • W.J. Dell et al.

    J. Non-Cryst. Solids

    (1983)
  • G. El-Damrawi et al.

    J. Non-Cryst. Solids

    (1992)
  • S. Wang et al.

    J. Non-Cryst. Solids

    (1998)
  • L. van Wüllen et al.

    J. Non-Cryst. Solids

    (1994)
  • R. Martens et al.

    J. Non-Cryst. Solids

    (2000)
  • A.D. Pelton et al.

    J. Non-Cryst. Solids

    (1999)
  • S. Surnev et al.

    Prog. Surf. Sci.

    (2003)
  • D.A. McKeown et al.

    J. Non-Cryst. Solids

    (2002)
  • S.A. Sutton et al.

    Geochim. Cosmochim. Acta

    (2005)
  • D.R. Neuville et al.

    Geochim. Cosmochim. Acta

    (1996)
  • D.A. McKeown et al.

    J. Non-Cryst. Solids

    (2001)
  • T. Tsujimura et al.

    Geochim. Cosmochim. Acta

    (2004)
  • D.W. Matson et al.

    J. Non-Cryst. Solids

    (1983)
  • W.L. Konijnendijk et al.

    J. Non-Cryst. Solids

    (1975)
  • W.L. Konijnendijk et al.

    J. Non-Cryst. Solids

    (1976)
  • B.P. Dwivedi et al.

    J. Phys. Chem. Solids

    (1993)
  • D. Maniu et al.

    Mater. Lett.

    (1995)
  • B.N. Meera et al.

    J. Non-Cryst. Solids

    (1990)
  • D. Maniu et al.

    J. Mol. Struct.

    (1997)
  • O. Attos et al.

    J. Non-Cryst. Solids

    (1997)
  • D. de Waal et al.

    Mater. Res. Bull.

    (1994)
  • M.J. Plodinec

    Glass Tech.

    (2000)
  • M.R. Carroll et al.

    Rev. Miner.

    (1994)
  • S.V. Stefanovskii

    Phys. Chem. Mater. Treat.

    (1993)
  • A.R. Conroy et al.

    The Role of Sulphate in the Melting and Fining of Glass Batch

    (1966)
  • W.K. Kot et al.

    Ceram. Trans.

    (2000)
  • D.F. Bickfold et al.

    J. Am. Ceram. Soc.

    (1990)
  • Cited by (110)

    • Hanford low-activity waste vitrification: A review

      2024, Journal of Hazardous Materials
    View all citing articles on Scopus
    View full text