Review Article
Structural Analysis Methods for Characterizing Multicomponent Melts and Glasses Processed at High Temperatures
2023 Volume 63 Issue 5 Pages 767-778
Details
2023 Volume 63 Issue 5 Pages 767-778
Recent structural-analysis research was reviewed for characterizing multicomponent melts and glasses processed at high temperatures. Multicomponent melts and glasses are often formed in slags during pyrometallurgical processing and the vitrification of highly radioactive wastes from nuclear power plants. Melt and glass physical properties were interpreted based on structural analyses using different methods. For example, the viscosities of different molten-silicate compositions were explained based on Raman and magic-angle-spinning magnetic resonance (MAS NMR) spectra. In addition, the thermal conductivities of multicomponent vitrified glasses were examined based on the structures elucidated using MAS NMR spectroscopy. X-ray absorption spectroscopy, which provides information about element chemical states, is used to analyze the local structure around specific elements in both melts and glasses. These studies show that structural analysis methods provide important information for designing and controlling melts and glasses processed at high temperatures. The structures characterized using different analysis methods were presented to elucidate the physical properties of multicomponent melts and glasses processed at high temperatures and propose prospects and directions for further research of process-oriented methods.
Although common oxide glasses exhibit a silicate network structure, the addition of alkali or alkaline-earth metals such as Na or Ca, respectively, affect glass properties. The silicate network comprises different Si–O bonds wherein O atoms are either bridging or nonbridging (BO or NBO, respectively).1,2,3,4,5,6,7) Figure 1 shows a schematic of different T–O bonds wherein T represents Si or Al atoms bridged with O ones, and M1 and M2 represent monovalently and divalently charged cations, respectively.2) To express different Si–O bonds in silicate glasses and melts, Qn species represent the number (n) of BOs in T–O bonds. In crystals, atoms are arranged in a regular sequence, and structural units are repeated in an orderly manner. In glasses, on the other hand, atoms are arranged in a disorderly manner. In these arrangements, SiO4 tetrahedra are preserved at nearly the same Si–O distance and O–Si–O bond angles as those in the first coordination and almost identical to those in corresponding crystals. However, by breaking the glass network by modifying M1+ and M22+ cations such as Na+ and Ca2+, respectively, glasses exhibit short-range but not long-range order. These structural characteristics are observed in the broad scattering profiles of glasses and sharp peaks in X-ray diffraction (XRD) patterns.8) Although melt or liquid and glass structures appear disordered, their structures are closely related to various physical properties.
For melt, glass, and crystal structures, the corresponding temperature-dependent volumetric characteristics of the melt (liquid), glass (supercooled liquid), and crystal are shown in Fig. 2. To prepare glass, a raw-material mixture is heated to a high temperature (e.g., 1400°C) and then cooled to form a melt (e.g., liquid or flux). The transition from the melt to the glass is explained based on the volumetric (e.g., shrinkage or expansion) temperature change. When a melt is cooled above the crystal melting point (Tm), the volume shrinks along the solid line and continues shrinking without any gradient change, even after reaching Tm. However, the melt is thermodynamically unstable below Tm and is called a “supercooled melt/liquid.” At lower temperatures, the liquid becomes glassy at the glass-transition temperature (Tg) and contracts exhibiting a different coefficient of thermal expansion to become a room-temperature glass, which melts at approximately Tm when reheated.
Melt, glass, and crystal volumes plotted as functions of temperature. Glass-transition and -melting temperatures are denoted by Tg and Tm, respectively.
For slags and mold fluxes used in steel mills, the oxide volume is temperature dependent, and the control of the oxide viscosity and other physical properties is important for producing iron and steel.9,10,11,12) Therefore, physical properties such as the viscosity and surface tension of the relevant oxides have previously been investigated using various methods.13,14,15,16,17,18,19,20,21,22,23) However, because of errors in experimental data collected under different conditions at different laboratories, systematic experiments must be conducted.9) To measure physical properties and elucidate reaction mechanisms, structures have previously been analyzed using methods such as Raman spectroscopy and 29Si magic-angle-spinning magnetic resonance spectroscopy (MAS NMR), which are powerful for detecting SiO2-network polymerization and depolymerization in melts and glasses. SiO2-network structural units represent a combination of different Qn species exhibiting BOs and NBOs, as shown in Fig. 3. These structural descriptions imply that in addition to cations, BOs and NBOs play important roles as network modifiers for determining melt and glass properties.
Combination of different Qn-species in silicate glasses; silicate structures of (a) Q4–Q4, (b) Q4–Q3, and (c) Q2–Q3. M+ and M2+ are singly and doubly charged cations, respectively, acting as network modifiers. (Online version in color.)
Figure 4 shows a schematic of the main bands corresponding to different silicates in the Raman spectra of alkali and alkaline-earth glasses and melts, which are usually classified as high-, mid-, and low-frequency regions in ranges 1200–800, 800–700, and 700–400 cm−1, respectively.2) Although Raman bands in high- and mid-frequency ranges are explained by different silicate vibration modes, spectral bands are not always clearly separated, and bands in the low-frequency range appear broad because of mixed stretching and bending modes.1)
Schematic of major band groups maximized as function of silica content in alkali and alkaline-earth glass and melt series. Low-frequency band group is shown as single continuously varying band in range 700–400 cm−1.2)
To store spent nuclear fuel long term, glasses must be vitrified at high temperatures. Therefore, a method must be developed for fabricating stable glasses.24,25,26,27,28,29,30,31) SiO2 is the main component in silicate glasses used to store spent nuclear fuels, and borate (B2O3) and/or phosphate (P2O5) is/are added to modify the glass properties. Because of their high strengths and neutron-absorption capacities, borosilicate glasses are attracting considerable attention for storing spent nuclear fuels. Because the viscosities and thermal conductivities of these glasses must be controlled at high temperatures, glass structures have also been studied using 11B MAS NMR.29,30,31)
Additionally, fluorides are used for electrolytically extracting rare metals and aluminum from ores and scrap metals melted at high temperatures.32,33) Accordingly, the production of molten alkali/alkaline-earth halide salts is very important. LiF–NaF–KF (FLiNaK) is a typical alkaline fluoride-based flux used at high temperatures. The FLiNaK melting point decreases at the ternary eutectic temperature.34) To elucidate the electrolysis mechanism of rare metals in molten salts and silicates, X-ray absorption spectroscopy (XAS) has previously been used to analyze the local structures of rare-metal-containing melts and glasses.35,36,37) For example, the in-situ structural analysis of molten FLiNaK provides much information for controlling melt properties in metallurgical processes.37)
The macroscopic chemical composition of multicomponent glasses and melts analyzed by chemical and physical methods is useful for comparison with phase diagrams. Microscopic information on specific elements is also important because they influence physical properties through their local structure. The local structure of such specific elements can be studied by analytical methods such as Raman spectroscopy, MAS NMR, and XAS, and these structural analysis methods are also useful in discussing the overall properties of glasses and melts. To control various physical properties of multicomponent melts and glasses, multiple aspects of melt and glass structures must be studied. Therefore, melt and glass structures and corresponding physical properties analyzed using several methods are reviewed herein. Although some glass compositions may contain traces of crystalline phases, which hinder glass preparation using conventional quenching methods, prospects and potential applications of structural analysis are discussed by considering the advantages and disadvantages of different analytical methods.
Many efforts have been made for systematically measuring the viscosity of melts exhibiting different compositions and establishing measurement procedures using reference materials under constant conditions. Because CaO–SiO2–Al2O3 ternary melts are fundamental in slag formation, their viscosity measurements are essential for elucidating the individual-component roles to determine melt properties. To date, although the viscosities of multicomponent oxides containing Ca, Mg, Fe, Mn, Na, and K have been satisfactorily plotted as functions of nonbridging oxygen ions per tetrahedrally coordinated cation (NBO/T) ratios,6) the viscosities of CaO–SiO2–20 mass% Al2O3 (CaO/SiO2 = 0.67, 1.00, or 1.22) ternary melts and alkali oxides were measured at 1873 K for melts exhibiting different alkali oxide (Li2O, Na2O, or K2O) compositions to obtain systematic viscosities under different conditions at high temperatures. Figure 5 shows the effect of the alkaline oxide content on the viscosities of CaO–SiO2–Al2O3 ternary melts. The results indicate that viscosity decreased with increasing basicity (CaO/SiO2). The viscosities of these quaternary melts decreased with increasing Li2O or Na2O content and increased with increasing K2O content.16) Furthermore, CaO–SiO2–20 mass% Al2O3 (CaO/SiO2 = 0.67, 1.00, or 1.22) ternary melts containing alkaline-earth oxides such as MgO or BaO exhibited similar viscosity trends.16) The influences of basicity and alkali and alkaline-earth oxide contents on the melt viscosity are explained based on structural data obtained from 27Al and 29Si MAS NMR spectra.
Effect of alkali oxide (R2O) content on viscosity of CaO–SiO2–Al2O3 melts at 1873 K.16)
In fact, it has been reported that alumina dissolved in silicate melts and glasses significantly changes not only viscosity, but also chemical durability, hardness, thermal expansion, compressibility, and other properties.7) This is thought to be due to the fact that the addition of alumina improves resistance to devitrification and largely reduces the compositional range of liquid immiscibility.
In a similar manner, the viscosities of different Na–Si-oxide melt compositions have previously been measured to investigate the effects of both N and F contents on the viscosities of Na–Si–O–N–F melts. The Na2O·2SiO2(NS2)–Na2O·2SiN4/3–Na2F2·2SiO2 pseudo-ternary diagram in Fig. 6 shows the sample compositions.17) At a Na2O·2SiN4/3/(Na2F2·2SiO2 + Na2O·2SiN4/3) molar ratio of 0.2, the fluoroxynitride-containing sample compositions are denoted as NS2–3N–3F, NS2–6N–6F, and NS2–12N–12F.
Nominal compositions of samples in Na2O·2SiO2–Na2O·2SiN4/3–Na2F2·2SiO2 pseudoternary diagram. Compositions of fluoroxynitrides are on line at Na2O·2SiN4/3/(Na2F2·2SiO2 + Na2O·2SiN4/3) molar ratio of 0.2.17)
The experimental results showed that the melt viscosity increased with increasing nitrogen concentration and decreased with increasing fluorine concentration at 1673 K. Furthermore, temperature dependence of the viscosity of molten fluoroxynitrides on line at Na2O·2SiN4/3/(Na2F2·2SiO2 + Na2O·2SiN4/3) molar ratio of 0.2 has been investigated, as shown in Fig. 7. This indicates that that fluorine effectively reduced the viscosity when nitrogen and fluorine were both added to the melt. These results were attributed to dopant (e.g., N and F)-induced changes in the silicate network, and the influences of these elements on the oxide viscosities were discussed based on corresponding NBO/T ratios.17)
Temperature dependences of the viscosity of the sample melts of Na–Si–O–N–F system.17)
Although melt or glass viscosities are important fundamental parameters for vitrifying highly radioactive liquid wastes at high temperatures, multicomponent-oxide viscosities depend on many factors.25) Radioactive raw materials (e.g., liquid wastes and glasses) are fed from the top of the furnace, and melts are poured from the bottom of the furnace and injected into a canister. Therefore, the melt viscosity and convection must be considered and used to homogenize the melt, respectively.
2.2. Thermal ConductivityAlthough melt and glass thermal conductivities have also been investigated for different NBO/T ratios, the correlation is not necessarily clear.9) Several methods have been developed for measuring the precise thermal conductivity or diffusivity of melts such as molten salts at high temperatures.38,39,40,41,42,43,44,45,46,47) The thermal conductivities of silicate-containing melts are especially important for elucidating metallurgical processes.48,49,50,51,52) For example, the thermal-transfer properties of metallurgical silicate melts have previously been investigated to design and control metal solidification and refinement.47) The results also showed that the thermal conductivities of silicate melts depend on the chemical composition, particularly the NBO/T ratio—which represents the polymerization degree of the silicate-network structure and is associated with the thermal conductivity of slags in the narrow liquid region of the ternary Al2O3–CaO–SiO2 diagram.47)
In nuclear power plants, melt and glass thermal conductivities are essential parameters because fission products (FPs) and minor actinides (MAs) are generated when fuel is irradiated in nuclear reactors,53,54,55,56,57,58,59) concentrated as radioactive high-level waste (HLW) solutions, and stored in tanks equipped with stirring and cooling devices to release heat from radioactively decaying materials.26,27) In this process, vitrification furnaces use Joule heat to melt glasses by passing an electric current through the melt, which is then solidified. Therefore, in addition to viscosity, melt thermal and electrical conductivities are important parameters for operating vitrification furnaces.25,26,27,28)
To study the influences of the Na and Ca contents on the thermal conductivities of borosilicate melts, melt properties were systematically measured using a specially developed laser flash method.29,30) The thermal conductivities of B2O3–SiO2-based glasses (B1–B4) and 27–29-mass% CaO–B2O3–SiO2-based glasses (C1–C4) prepared using different B2O3/SiO2 ratios were measured using the laser flash method and a cell configured for front heating and detection in the range 1548–1573 K, and the results are summarized in Fig. 8.29) Clearly, the melt thermal conductivities decreased and increased with increasing B2O3/SiO2 ratio to and from 0.5, respectively. The thermal conductivities of the CaO–B2O3–SiO2-based glasses were higher than those of the B2O3–SiO2-based ones. Additionally, the thermal conductivities of the Na2O–B2O3–SiO2-based glasses were between those of the B2O3–SiO2- and CaO–B2O3–SiO2-based ones.29,30) The characteristic thermal conductivities of borosilicate glasses are attributed to borate-induced changes in the NBO/T ratio in silicate networks.
Average thermal conductivities of B2O3–SiO2 (B1–B4) and CaO–B2O3–SiO2 (C1–C4) melts plotted as functions of B2O3/(B2O3 + SiO2) ratio in ranges 1248–1423 and 1548–1573 K, respectively.29)
The typical HLW-solidification-glass compositions developed in Japan are 46.6, 14.2, 5.0, 3.0, 3.0, 3.0, 10.0, 10.1, and 5.1 mass% SiO2, B2O3, Al2O3, Li2O, CaO, ZnO, Na2O, liquid-waste fission products, and other oxides, respectively.25) In liquid wastes, the percentages of oxides and other materials may currently be in the range 20–25%. During furnace operation, raw materials are mixed on the molten-glass surface. During vitrification, platinum-group elements such as ruthenium, which is a fission product,54,55,56,57,58,59) are almost completely insoluble in glass and dispersed as particles in the glass melt. These particles influence both the melt viscosity and electrical conductivity in a complex manner.55,56) During vitrification, the melt contains platinum-group (mainly RuO2) particles, which increase the glass viscosity. The viscosity increased when 14% platinum-group elements (precipitated as particles) were added to borosilicate glass.57) However, because the RuO2 density is approximately 7 and much higher than the glass-matrix one (approximately 2.5), the particles settled. Hence, in hot glass melts, particles are distributed more downward on average, and the melt viscosity near the flow nozzle varies with both position and time. Therefore, the furnace operation requires precise temperature control, and the melt flow is notably non-Newtonian.
2.3. Electrical ConductivityAlkali and alkaline-earth cations both play an important role for determining melt and glass electrical conductivities. Additionally, not only Na+ cations but also F− anions reportedly both contribute to silicate-glass electrical conductivities and unlike Na2O or CaO, NaF is negligibly involved in breaking Si–O–Si bonds.60) F− anions contributed approximately 1/8 the contribution of Na+ cations to the electrical conductivity. When NaF is added to silicates melted in air, it negligibly interacts with them and dissociates into Na+ and F− ions in the melt. Moreover, ionic conductivities contribute to molten-salt electrical conductivities.60,61) These characteristics contrast with those of metallic alloys, wherein the electrical conductivity is controlled by dissolved-alloying-element contents and interpreted according to the Wiedmann–Franz law.62)
To elucidate the electrical conductivities of multicomponent blast-furnace slags, the influences of Na2O and K2O contents on the electrical conductivities of CaO–MgO–Al2O3–SiO2 melts have previously been measured using the four-electrode method63,64) by adding either Na2O or K2O or both Na2O and K2O. The results showed that the electrical conductivity gradually monotonously increased and decreased with increasing Na2O and K2O contents, respectively, indicating that the large K+ ionic radius may have decreased the electrical conductivity.
During HLW-glass vitrification, electrical conductivity data sensitive to both chemical compositions and operation conditions must also be accumulated because the operation of liquid-waste solidifiers requires the adjustment of the electric current to control the Joule heating.55,56,57,58) Owing to the molten-glass viscosity, RuO2 particles settle downwards in the furnace. Therefore, the RuO2-particle concentration will likely be higher near the bottom of the furnace.56,57) Thus, when the RuO2-particle concentration increases and conduction-percolation threshold is exceeded at the bottom of the furnace, the conductivity increases, thereby suppressing Joule-heat generation at the center of the furnace and preventing melting.
To elucidate the mechanisms underlying the characteristic physical properties of melts and glasses, the structures of melts and glasses such as silicates, aluminosilicates, and borosilicates have previously been analyzed using Raman, NMR, and X-ray absorption spectroscopies. Although these methods are different, they provide useful information about melt and glass structures, which can be described as monomers, dimers, chains, and three-dimensional network structures in oxide systems. For instance, when the nonbridging-oxygen-per-silicon ratio (NBO/T: T = Si) is approximately 2 or less, Si–O monomers, dimers, and chains are mixed depending on the other elements. To analyze the structures of Na- and K-containing aluminosilicate melts and glasses, Raman spectroscopy was applied, and theoretical and measured viscosities were compared.65) The viscosity results indicated that Na and K did not randomly mix in the aluminosilicate melts, and Raman spectra suggested that the glasses exhibited different TO2 environments. Although Raman spectroscopy has previously been used in numerous studies to elucidate the physical properties of oxide melts and glasses at high temperatures,66,67,68,69,70,71,72,73,74) Raman spectra alone only provide relatively qualitative information and do not provide any information about Raman absorption cross-sections, which depend on molecular polarizabilities.75) However, because NMR experiments are usually more time consuming, Raman spectra adequately measured in high-resolution mode can provide quantitative information.
3.1. 27Al, 29Si, and 31P MAS NMR SpectroscopiesOwing to the application of a high magnetic field and improved spectral resolutions under optimized operation conditions, MAS NMR spectroscopy is a powerful method for analyzing aluminate structures at high temperatures.76,77,78,79,80) Information about Al and Si has previously been obtained by applying multinuclear NMR measurements,7,77,80) and the results provide comprehensive information about aluminosilicate structures.7) The structures of alkaline- and alkaline-earth-based silicate glasses have also previously been characterized using single 29Si MAS NMR.81,82,83) Equilibrium constants were estimated for different SiO2-structural-unit Qn values, and Li, Na, and K cationic powers were investigated based on a thermodynamic model for glasses. XRD has also been used to study silicate structures and elucidate structures for alkaline-earth (e.g., MgO and CaO–SiO2) and iron-oxide (FeO–SiO2) silicate melts for which interference and pair-distribution functions were used to elucidate melt ionic distributions.81,82,83)
Alkali- and alkaline-earth-oxide-containing CaO–SiO2–Al2O3 glass structures have previously been systematically investigated using multinuclear 27Al and 29Si MAS NMR spectroscopies and CaO–SiO2–Al2O3 (CaO/SiO2 = 0.67; Al2O3 = 20 mass%)–R2O (R = Li, Na, or K) quaternary glass samples similar to those used for melt viscosity measurements.16) To determine the influence of the R2O content on glass structures, 27Al MAS NMR spectra were measured for (32CaO–48SiO2–20Al2O3)–R2O glasses prepared using a constant R2O content, as shown in Fig. 9.84) Different 27Al NMR spectral components were attributed to pentahedrally [Al(5)] and tetrahedrally [Al(4)] coordinated Al84) and suggest that the pentahedrally [Al(5)]-coordinated Al fraction in CaO–SiO2–Al2O3 was reduced by adding alkali oxides. In addition, the 29Si NMR spectra generated for CaO–SiO2–Al2O3 indicated that the silicate-anion polymerization degree decreased with increasing Li2O or Na2O content because the relative area of the Q3 component was relatively small for Li2O- and Na2O-containing glasses.
Effect of alkali-oxide content (10.8 mol%) on 27Al MAS NMR spectra of 32 mass% CaO–48 mass% SiO2–20 mass% Al2O3 [CAS(1)] glass.84)
29Si MAS NMR spectra provide important information about the polymerization of silicate anions in Na–Si–O–N–F glasses.17) Although the influences of both the N and F contents on the glass structure were investigated, NaF was detected as a precipitate in solidified glasses. Figure 10 shows the 29Si MAS NMR spectra measured at 11.7 T for Na–Si–O–N–F glasses and corresponding Gaussian-fitted simulation curves.17) The solid-black, broken-blue, and red lines represent the experimentally measured spectra, fitted Qn-component curves, and fitted-curve summation, respectively. The spectrum of the NS2 glass revealed a main peak, indicating that the Q3 species was the major silicon-atom component in the NS2 glass. Two shoulder peaks were assigned to Q4 and Q2 species. The Q4-component signal intensified with increasing nitrogen and/or fluorine content(s) in the NS2 glass, suggesting that the Q4 species content increased with increasing nitrogen and/or fluorine content(s). By adding nitrogen and fluorine to the silicate glasses, the structural changes around the silicon atoms corresponded to silicate-melt viscosity changes at high temperatures.17)
29Si MAS NMR spectra generated at 11.7 T for glasses. Solid-black, dashed-blue, and dashed-red lines represent experimental data, fitted peaks for Qn components, and sum of fitting peaks, respectively.17) (Online version in color.)
The Qn distributions estimated based on the Qn-species area fractions in the NMR spectra were essentially like the Qn distributions reported in a previous study.78) Although the NBO/T ratio does not monotonically change depending on the species or alkali-oxide content, the ratio indicates the silicate-anion depolymerization degree. Regarding silicate-anion polymerization in glasses and melts, XRD patterns revealed that the SiO4 unit is reportedly distorted by adding alkaline-earth and iron oxides to melts.78,79)
31P MAS NMR spectroscopy is often used to analyze the structures of phosphate-anion-containing glasses.88,89,90,91,92,93,94) For example, solid-state MAS NMR spectroscopy is a powerful method for obtaining information about short-range orders and networks in phosphate glasses, which are important for fabricating electrochemical devices in electronic and related industries.92) Phosphate glasses have also attracted considerable attention as materials suitable for storing radioactive wastes, and 31P MAS NMR spectroscopy is a promising method for analyzing these materials.94) Additionally, the structures of phosphate-containing glasses should be analyzed using 31P MAS NMR spectroscopy because phosphate glasses exhibit functional characteristics such as storage-material latent heats.95)
3.2. 11B, 17O, and 19F MAS NMR SpectroscopiesBorosilicates exhibit high neutron-absorption capacities, and the stable 11B isotope is used in NMR spectroscopy. 11B MAS NMR spectroscopy has previously been used to analyze borosilicate glass structures, which exhibit high Na+ cationic mobilities.96) 11B MAS NMR spectra have also been utilized to analyze CaO–B2O3–SiO2 structures for immobilizing HLWs and elucidating the thermal conductivities of B2O3–SiO2, Na2O–B2O3–SiO2, and CaO–B2O3–SiO2 melts.29,30) The molten-salt thermal conductivities were calculated by combining the measured thermal diffusivities, specific heats, and densities. Because B2O3 and SiO2 exhibited tricoordinate and tetrahedral structures exhibiting three and four oxygen species, respectively, the B2O3–SiO2 melt was believed to exhibit a sparse network structure.
The thermal conductivity of the Na2O–B2O3–SiO2 melt was higher than that of the B2O3–SiO2 one, suggesting that Na+ cations were bonded to BO4− anions. Consequently, B2O3–SiO2 exhibited a denser network structure and longer phonon mean free paths. Ca2+ cations, on the other hand, bonded with two BO4− anions to form a denser network structure exhibiting longer phonon mean free paths. Therefore, the thermal conductivity of the CaO–B2O3–SiO2 melt was higher than that of the B2O3–SiO2 one, suggesting that the bonds between the Ca2+ cations and NBO/T were broken in the network. Thus, 11B MAS NMR spectroscopy was used to analyze the structures of the CaO-free and -containing B2O3–SiO2 melts. Figure 11 shows the 11B MAS NMR spectra of the B2O3–SiO2 (B1–B4) and CaO–B2O3–SiO2 (C1–C4) glasses corresponding to the samples shown in Fig. 7.30) The spectra generated for different B2O3–SiO2 compositions revealed that B exhibited a fundamentally tricoordinate structure with three oxygen species, whereas B exhibited a partially tetracoordinated structure with four oxygen species in CaO–B2O3–SiO2. Therefore, the content of the tetracoordinated B structure in the CaO–B2O3–SiO2 (C1–C4) samples increased with decreasing thermal conductivity for the CaO–B2O3–SiO2 melts.
11B MAS NMR spectra measured at 18.8 T for (a) B2O3–SiO2 (B1–B4) and (b) CaO–B2O3–SiO2 (C1–C4).29)
11B and 29Si MAS NMR spectroscopies have also been utilized to analyze the structure of molten Na2O–B2O3–SiO2 for which the thermal conductivity was measured.97,98) When the Na2O content was below 30 mol%, changes in the relative content of tetracoordinated boron influenced the thermal conductivity, indicating that B predominantly comprised tetraborate units.
Because 17O NMR spectroscopy is directly related to O coordination, it is also a powerful method for elucidating oxide structures from an oxygen-centered perspective.99,100) In previous studies on the viscosities of CaO–SiO2–Al2O3–K2O melts, 17O MAS NMR spectra were measured to analyze the structures of melts exhibiting different compositions. The results revealed that K+ cations compensated for the AlO4− anion negative charge, while Ca2+ cations primarily generated NBO/T in the melts.
X-ray photoelectron spectroscopy (XPS) O 1s spectra have previously been analyzed to elucidate the O chemical states and determine the O coordinations in Na-containing silicate and aluminosilicate glasses.101,102) XPS spectra measured in high-resolution mode revealed that the spectral component attributed to NBO/T was detected in silicate glasses, suggesting that although the XPS signals were sensitive to the sample surface, NBO/T could be discriminated from BO atoms.
In addition, because F− is an important anion in steelmaking slags, 19F NMR spectroscopy has previously been applied to analyze slag structures and F, Si, and Al chemical contents in slags.103) The results revealed that slag structures were not affected by the F, Si, and Al contents. Additionally, chemical structures were proposed for slags containing different Al contents. Because rapid MAS reduced the magnitude of the 19F–19F dipole interactions, high-resolution NMR spectra were generated for the slags. The peak chemical shifts were used to identify the partner element directly bound to F and revealed that F–Ca was the main component. Additionally, the Al in the slags may have subdivided the fluorine chemical structure.103) Furthermore, 19F NMR spectroscopy has previously been used to characterize the structures of Li-containing oxides.104)
3.3. X-ray Absorption SpectroscopyX-ray absorption spectroscopy (XAS) has previously been used to analyze the chemical states of elements in CaO–SiO2–MgO slags containing metals such as Fe and Cr, which are produced during steel making. The Fe and Cr chemical states in oxide slags exhibiting different glass compositions under different conditions were investigated using XAS to measure the X-ray absorption near-edge structures (XANESs). Figure 12 shows the normalized Fe K XANES spectra generated for different glass-slag compositions melted under different atmospheric conditions.105) All the spectra revealed equal absorbance at the energy positions indicated by the arrows, indicating that the slags contained a mixture of ferrous Fe(II) and ferric Fe(III) ions. The spectra exhibiting the lowest and highest absorption-edge energies (indicated by thick solid and broken lines) were measured using Fe(II)- and Fe(III)-containing reference materials, respectively. The other spectrum was represented by a weighted sum of both spectra. The sample Fe(II)-to-Fe(III) ratios were calculated based on the respective weighted ratios of all three spectra and revealed that the valence ratio changed depending on the oxygen partial pressure during melting, basicity (CaO/SiO2), and Fe and Cr mixture. This method may be applied to analyze the Fe(II) and Fe(III) chemical states in silicate-based slags.83) The results revealed that Fe was both divalent and trivalent, while Cr was divalent, trivalent, and hexavalent. Although the chemical state of the Fe–Cr mixture was confounded by the individual Fe and Cr chemical states, XAS was useful for quantitatively analyzing the chemical states of transition metals, which (unlike alkali and alkaline-earth metals) can be multivalent.
Normalized Fe K XANES spectra generated for different glass-slag compositions melted under different atmospheric conditions.105)
A previous study found that molten-oxide viscosities are influenced by not only alkaline and alkaline-earth cations but also transition elements such as Fe106) and that the alkaline-earth cationic radius decreased in the order Ba > Sr > Ca for different melt Fe2+/t-Fe ratios. In such studies, XAS-obtained XANES spectra may effectively elucidate chemical states.
Although in-situ XAS is a useful method for analyzing local structures around specified elements of interest, either experimental ingenuity or an optional unit is required for analyzing melts such as molten salts at high temperatures in situ. For example, the local Ta structure must be analyzed in molten FLiNaK at high temperatures.34) In-situ XAS spectra were generated at a synchrotron facility in which the K2TaF6 crystal structure was refined using monocrystalline XRD data.107) Extended X-ray-absorption fine-structure (EXAFS) spectra recorded at the Ta L3 absorption edge and corresponding Fourier-transform spectra generated for K2TaF7 both at room temperature and dissolved in molten FLiNaK revealed that the local structure around Ta was influenced by the seven nearest-neighboring Ta–F correlation pairs in the TaF7 unit. Compared with the coordination number of the Ta-surrounding F atoms in the K2TaF7 crystal, that of the Ta-surrounding F atoms in molten salts likely will decrease from 7 to 4.
The melt and glass structural models estimated based on experimentally measured physical properties and structures are summarized. Figure 13 shows a schematic of the network structure and channel region in sodium silicate glasses and melts.21) The network structure comprises framework species such as Si and M and SiO4 and AlO4 tetrahedra, and the channel is in the noncrystalline-structure breakage region. In the channel and network-structure regions, nonframework cations are disturbed as network modifiers and charge compensators for Si–O–M BO atoms, respectively, suggesting that melt and glass physical properties are both strongly affected by the channel density.
Schematic of planar network structure comprising Si and M framework species (e.g., SiO4 and AlO4 tetrahedra) and channel in noncrystalline-structure breakage region. Nonframework cations are disturbed in both channel and network structure region as network modifiers and charge compensators for Si–O–M BO atoms, respectively.21) (Online version in color.)
To elucidate the influences of nitrogen, fluorine, and sodium atoms on the 29Si MAS NMR spectra of the melts, a model was proposed. Although the spectra indicate that Si–N bonds are more covalent than Si–O ones, quantifying the polymerization degree of the entire nitrogen-containing system is difficult using only the 29Si MAS NMR spectrum. Figure 14 shows schematics of the possible planar structures obtained for different anionic compositions of oxide, oxynitride, oxyfluoride, and fluoroxynitride glasses.17) For simplicity, the atomic coordination numbers and silicon-atom Qn distributions were not considered, suggesting that the terminal fluorine atoms bonded directly to the silicon ones to form Si–F bonds and depolymerize silicate anions. Additionally, free fluorine atoms complexed with sodium cations, indicating that the fluoride-ion content varied with the SiO2 concentration in the fluorine-doped system.
Schematics of possible planar structural variations produced for different anionic compositions of (a) oxide, (b) oxynitride, (c) oxyfluoride, and (d) fluoroxynitride glasses. For simplicity, atomic coordination numbers and silicon-atom Qn distributions were not considered.17) (Online version in color.)
Figure 15 shows the network models proposed to explain the thermal conductivities of the B2O3–SiO2, CaO–B2O3–SiO2, and Na2O–B2O3–SiO2 melts.30) These models were proposed based on experimentally measured thermal conductivities and NMR spectra and, thus, are not fully accurate because the B2O3–SiO2 thermal conductivities were lower than those of both CaO–B2O3–SiO2 and Na2O–B2O3–SiO2. Therefore, the models should be further refined. We will soon combine the experiments proposed herein to construct accurate atomistic structural models for melts and glasses.
Network models proposed to explain thermal conductivities of B2O3–SiO2, CaO–B2O3–SiO2, and Na2O–B2O3–SiO2 melts.29) (Online version in color.)
Clearly, having a measurement- and observation-derived rule of thumb would be useful for predicting the properties of melts and glasses at high temperatures. For example, the lower the glass melting point, the higher the glass solubility (i.e., ability to melt and unite various components). However, whether the glass melting temperature and chemical durability both decrease in the order silicate > borosilicate > borate > phosphate glasses remains unclear. Although solvency theoretically increases in the order silicate < borosilicate < borate < phosphate glasses, solvency is practically much more complex.25) For instance, because phosphate glasses are highly solvent, they dissolve considerable iron oxide, which markedly and considerably increases both the chemical durability and glass melting temperature, respectively, which in turn hinder glass production. Thus, it is necessary to control the properties related to the viscosity, conductivity, and compatibility of the glass container in order to produce stable glass with high strength and durability, and to obtain design guidelines for glass that allow tuning of chemical composition and structure. Against this background, the properties of glass have been studied from a wide range of perspectives along with its structure, and much knowledge about glass has been obtained, especially in the pyrometallurgical processes.108,109,110,111,112,113,114,115,116,117) These studies are expected to accumulate a great deal of knowledge about new fields of glass and to be applied to the control of new glass properties.
Glass structures used for solidifying and storing HLWs can be characterized using several structural analysis methods. This paper describes glass and melt structures analyzed using Raman, NMR, and XPS spectroscopies, XRD, etc. In addition, although high-energy (synchrotron radiation) X-ray diffraction, neutron diffraction118,119,120,121) and anomalous X-ray scattering122) are useful for characterizing atomic structures, these methods must only be used as needed. The long-term storage of spent nuclear fuels will become increasingly important and require the development of HLW vitrification techniques.123,124,125) Several studies on glass reactions have previously used Raman spectroscopy to analyze glass structures and characterize alteration and corrosion, and the results provide useful information, which cannot be obtained using NMR spectroscopy.125,126,127,128,129) Therefore, it is important to use reliable data from an interdisciplinary perspective to find true information about the structure of glass, although data by each structural analysis depends on conditions such as signal sensitivity and resolution.
So far, technological developments and policies related to renewable energy have certainly progressed over time. However, it is important to note that renewable energy technologies and policies have not always been sufficient.130,131,132) There is a strong need to achieve carbon neutrality by 2050 in order to avoid global environmental changes, especially global warming, caused by use of fossil fuels. Therefore, the safe use of nuclear energy needs to be expanded, and vitrification of high-level radioactive waste is an important related issue. As a result, vitrification technology for HLW has become indispensable, and efforts to address related research issues are also required.133,134)
Furthermore, it is also important to investigate many factors influencing the properties of glasses, in order to predict properties of glasses formed in vitrification.135,136,137) Particular techniques have been developed to measure the residual stress of glass and glass-ceramics, especially as it relates to the mechanical properties of glass.138,139) The mechanical properties of glasses formed by high-temperature processing are an important issue in HIW management for long-term storage. Thus, since usage of nuclear energy should be extensively reconsidered, the vitrification of HLW has become important.
This paper discussed viscosities, thermal and electrical conductivities, analyses, and proposed structural models of glasses and corresponding melts. Because the reproducibility and accuracy of analytical results must be validated using reference materials and correct procedures, different analytical results must be integrated to obtain accurate and reliable information, which can be compiled into an extensively used materials database.140,141)
In order to avoid global warming, the steel industry must minimize greenhouse gas (GHG; e.g., CO2) emissions and become carbon neutra.142) Toward this goal, reduction processes using fossil fuels to reduce CO2 emissions, hydrogen, and other carbon substitutes are currently being considered. In parallel, nuclear energy must be used safely, and the treatment and storage of spent nuclear fuel is also important. This makes it necessary to urgently develop a vitrification technology for HLW based on the previous knowledge of high-temperature processes.
In molten glasses, Na+ cations dissociated from halogen anions, and Na+ and other cations decreased the viscosities of halogen-anion-containing molten glasses. The thermal conductivity of molten Na2O–B2O3–SiO2 was higher than that of molten B2O3–SiO2, indicating that thermal conductivity depends on the molten-salt composition. Na+ and Ca2+ cations, on the other hand, bonded with one or two BO4− anions, thereby densifying the network and lengthening the phonon mean free paths compared to those of B2O3–SiO2. These results suggest that electrical conductivity was mainly affected by both cations and halogen anions including Na+ and F−, respectively, indicating that Na+ cations dissociated from halogen anions in melts. Therefore, glass physical properties should be controlled by tuning both the glass chemical composition and production process parameters. In addition to borosilicate glasses, phosphate ceramics have attracted considerable attention for immobilizing actinides and nuclear fission products. State-of-the-art mineralogical approaches and iron and steel technologies should be utilized for characterizing and synthesizing these glasses.
This review was written under the Cooperative Research Program of the “Network Joint Research Center for Materials and Devices” and was supported by Grants-in-Aid for Scientific Research (16K14443, 16H04543, and 21H01854) from the Japan Society for the Promotion of Science, High-Efficiency Rare-Element Extraction Technology Area of the Tohoku Innovative Materials Technology Initiatives for Reconstruction, and ISIJ Research Promotion Grant. Synchrotron radiation experiments were performed at the Japan Synchrotron Radiation Research Institute (SPring-8) and Saga Light Source. The authors would like to express their sincere gratitude to Prof. Y. Waseda, Prof. T. Nakamura, Prof. N. Sato, Prof. K. Nakashima, Prof. N. Saito, Prof. H. Ohta, Dr. K. Kanehashi, and Dr. K. Saito for their valuable discussions.
