Formation of Reaction Layer and Dissolution Behavior of Alkali and Alkaline-Earth Iron Phosphate Glasses in Water

Abstract

The durability of iron phosphate glasses containing alkali and alkaline-earth oxides in water was investigated using two methods: materials characterization center (MCC)-2 static, leach test at 120°C for plate samples and product consistency test (PCT) at 90°C for granulated samples. The glass samples were classified into two types depending on the macroscopic appearance of the reaction layers, giving rise to interference colors formed on the glass surface after the MCC-2 test. The type I glasses had reaction layers with macroscopic cracks for the Li₂O- or Na₂O-containing iron phosphate (IP) glasses. Their weight loss per specific area showed a linear relation with immersion time due to the degradation of the protective layer. The type II glasses, on the other hand, exhibited superior water durability as a result of formation of the homogenous reaction layers for the K₂O-, CaO-, and BaO-containing IP glasses. The main cations released into the water in the PCT method included alkali, alkaline-earth, and phosphorous elements for both type I and II glasses. The Raman spectroscopy results suggested that the macroscopic cracks formed in the reaction layers for type I glasses are attributed to the microscopic selective dissolution of Q² phosphate tetrahedra in the glass matrix.

Fig. 11 Schematic diagrams of (a) initial stage and (b) selective dissolution of PO₄ Q² species during the formation of the reaction layer for type I glasses.

1. Introduction

The accident in the Fukushima Daiichi nuclear power plant that occurred on March 11, 2011, resulted from a catastrophic earthquake and the resulting tsunami. Operations for proper disposal of the plant have been on going to date. The treatment of the radioactive water accumulated at the plant is one of the pressing challenges. Areva (recently Orano), in cooperation with Veolia (Actiflo^TM-Rad),¹⁾ has employed a co-precipitation method, using BaSO₄ as a coagulant, for removing radioactive ⁹⁰Sr from the water. There is a plan to use the glass vitrification method to immobilize the radioactive sludge that results from the precipitation process. However, vitrification of the radioactive sludge, including BaSO₄, using conventional borosilicate glasses forms water-soluble alkali sulfate.^2,3) Therefore, to stably immobilize the radioactive sludges using alkali borosilicate glasses would be difficult.

On the other hand, studies on iron phosphate (IP) glasses, as a new host candidate for nuclear waste immobilization, have been conducted by a research group at Missouri University of Science and Technology since the late 1990s. IP glasses have excellent durability in vitrification of various nuclear wastes.^3,4) Yu et al.⁴⁾ reported that Na₂O–FeO_x–P₂O₅ glasses containing 28 mol% of Fe₂O₃ exhibit one-tenth of the dissolution rate of the conventional soda lime silicate glass used as the window material. The excellent durability of IP glasses is attributed to the chemical bonding of Fe₂O₃ and P₂O₅ constituents. Fe₂O₃ and P₂O₅ form ∼25% P–O–P bonds and the residual Fe–O–P bonds in IP glasses.⁵⁾ Substituting the durable Fe–O–P bonds for hygroscopic P–O–P bonds strengthens the glass networks against water attack.

Kitamura et al.⁶⁾ investigated the durability of BaO–FeO–Fe₂O₃–P₂O₅ glasses for immobilization of BaSO₄ as a main component of the radioactive sludge. After the materials characterization center (MCC)-2 test, a 150–230-nm-thick film of the reaction layer, comprising six-line ferrihydrite and non-crystalline phases with an Fe–O richer and no Ba (<1 at%) composition, was formed on the surface of the glass sample. This reaction layer plays a protective role against further dissolution of the glass components. The nonexistence of Ba in the reaction layer suggests that the BaO component dissolved into the leachate as Ba²⁺, thereby contributing to an increase in the local pH near the thin layer. The increase in the pH was favorable for the precipitation of hydride oxides.

In this study, the effect of other alkali and alkaline-earth oxides on water durability was investigated. Two methods of water durability evaluation––MCC-2 static, high-temperature leach test at 120°C for plate samples and product consistency test (PCT) at 90°C for granules––were adopted. The two methods were carried out on R₂O and R′O–FeO–Fe₂O₃–P₂O₅ glasses (R = Li, Na, K and R′ = Ca). The durability of the IP glasses was characterized by the weight loss, dissolved ions in solution, microstructure of the reaction layer, and structure of the phosphate networks.

2. Experimental Procedures

2.1 Glass preparation

The glass is composed of 15R₂O–85 (0.35FeO_x–0.65P₂O₅) and 20R′O–80 (0.35FeO_x–0.65P₂O₅) in molar ratio, where R = Li, Na, K; R′ = Ca, Ba; and x = 1–1.5. The compositions are denoted by xBaIP35 glasses, where x denotes the concentration of alkali or alkaline-earth oxides. A binary IP glass cullet with a batch composition of 35Fe₂O₃–65P₂O₅ (mol%) was prepared via the conventional melt quenching method. The cullet was ground to a particle size of less than 120 µm. The premelted IP glass cullet and commercial reagents of Li₂O (cationic purity 99.99%), Na₂CO₃ (99.99%), K₂CO₃ (99.9%), CaCO₃ (99.99%), and BaCO₃ (99.9%) were used as the raw materials. Twenty grams of the raw materials were weighed in various compositions and mixed in an alumina mortar. The mixed batch was calcined at 200°C for 1 h to reduce the moisture content. It was then melted in a Pt crucible at 1100–1200°C for 1 h, depending on the composition. The melt was poured onto graphite molds. The casts were further annealed at a temperature near the glass transition temperature for 1 h, and then cooled to room temperature at 1.0°C/min.

2.2 Water durability tests

The water durability of the glass samples was evaluated using two method: MCC-2 static leach test⁷⁾ and PCT.⁸⁾

2.2.1 MCC-2 static leach test method

A glass sample plate of dimensions 10 × 10 × 3 mm³ with the six surfaces polished to mirror surfaces was suspended in a PTFE-sealed vessel (TAF-SR-50, Taiatsu Techno Co., Japan) by a fluoroethylene resin thread and immersed in 50 mL of ultrapure water at 18.2 MΩ·m. The vessel, including the suspended glass plate, was held at 120°C in an oven with a high-precision temperature control (DF411, Yamato) for 24–168 h. After holding for the desired time, the vessel was removed from the chamber, and then cooled in air for 15 min. The glass sample plate was removed from the vessel and dried at 60°C after removing the fluoroethylene resin thread. The sample surface was then observed with a digital microscope (DVM6, Leica). The weights of the sample before and after immersion were recorded. The weight loss per specific surface area, ΔW/S (kg/mm²), was calculated, where ΔW is the weight change (kg) and S (mm²) is the sum of the areas of the six surfaces. The measurements were performed three times for each immersion condition, and then, the average value of ΔW/S was calculated. The solution pH was also measured before and after the immersion tests.

2.2.2 PCT test method

The ground glass samples of particle size 75–150 µm were ultrasonically cleaned with ethanol and ultrapure water. Three grams of the sample was immersed into 30 mL of ultrapure water in the PTFE sealed vessel. The vessel was held at 90°C in an oven for 168 h. Thereafter, the solution was diluted with 3% nitric acid solution and the cation concentrations of the solution were determined using ICP-OES (PerkinElmer Optima 2000 DV, Norwalk, USA), except for 15LiP35 glass, due to the difficulty in evaluating Li-ion concentration. The normalized elemental mass release NR_i (kg/m²) was calculated for two times run in each sample using the equation   

\begin{equation} \mathit{NR}_{i} = \frac{C_{i}}{f_{i} \cdot (S \cdot \text{V$^{-1}$})} \end{equation} (1)

where C_i is the concentration of element i in solution (µg/mL), f_i is the mass fraction of element i in the glass (mass%), and S·V⁻¹ is the ratio of the surface area of the sample to the volume of the leachate (m⁻¹). The specific surface area of the ground glass samples was measured using the BET method with the BELSORP-mini II, MicrotracBEL, apparatus.

2.3 Microstructure of the reaction layer

The microstructures of the reaction layers of the glass sample surfaces after MCC-2 tests were characterized using FE-SEM (S-5500, HITACHI, Japan). The reaction layer on the 15KIP35 glass immersed for 168 h, as a representative sample, was observed using a 200 kV TEM (JEM-2100, JEOL Ltd., Japan). The obtained result was compared with the microstructure of the 20BaIP35 glass evaluated in the previous study.⁶⁾

2.4 Glass structure

The phosphate network structures of the glass samples were analyzed by Raman spectroscopy using a laser Raman spectrometer (inVia, Renishaw, United Kingdom) in the range of 400–1400 cm⁻¹, with a 532 nm YAG laser as an excitation light source. The Raman spectra were collected using the laser beam focused by a 50× objective lens with an exposure time of 10 s and an accumulation number of 20.

3. Results

3.1 MCC-2 static, high-temperature leach test method

Microscopic interference colors were observed on all surfaces of the glass samples with various initial compositions and immersion times. Figure 1 shows the surfaces of all glass samples after 168 h of immersion. The interference colors on the glass surfaces imply the formation of thin reaction layers of thickness less than the visible wavelength. As for 15LiIP35 and 15NaIP35 glass samples, several macroscopic cracks and spalling were observed, which are attributed to stress corrosion (type I). However, cracks are seldom observed on the surfaces of 15KIP35 and 20CaIP35 glasses with the formation of a relatively homogeneous reaction layer, as in 20BaIP⁶⁾ (type II). Figures 2 and 3 show the variations in the weight loss per specific area, ΔW/S, and pH value with the immersion time, respectively. Figure 2 shows that ΔW/S increases linearly with the immersion time for type I 15LiIP35 and 15NaIP35 glasses. As for the type II 15KIP35 and 20CaIP35 glasses, it increases nonlinearly with the immersion time, attaining a maximum value where it remains approximately constant. This saturation in ΔW/S is attributed to the formation of homogeneous reaction layers, which is in good agreement with our previous observation regarding the immersed 20BaIP35 glass.⁶⁾

Fig. 1

Appearances of glass samples immersed at 120°C for 168 h by MCC-2 static leaching method: (a) 15LiIP35, (b) 15NaIP35, (c) 15KIP35, (d) 20CaIP35.

Fig. 2

Variations in weight loss per specific area with immersion time at 120°C in ultrapure water.

Fig. 3

Variations in solution pH with immersion time at 120°C in ultrapure water.

3.2 PCT test results

Figure 4 shows the normalized mass release data of all cationic elements for the 15NaIP35, 15KIP35, 20CaIP35, and 20BaIP35 glasses. All NR_i values are over two orders of magnitude lower than the United States Department of Energy (DOE) requirements for both high- and low-level activity wastes,^9,10) as shown in Table 1. All glasses released cations preferentially in the order R or R′ > P ≫ Fe, where R = Na, K and R′ = Ca, Ba.

Fig. 4

Normalized elemental mass release NR_i in glass powdered samples immersed at 90°C for 168 h by using the PCT method.

Table 1 United States DOE requirements for normalized elemental mass release NR_i by using the PCT method.

3.3 Microstructure of the reaction layer

Figures 5(a)–(d) show the FE-SEM micrographs of the cross-section of the reaction layers on the glass samples immersed at 120°C for 168 h in the MCC-2 static leach tests. The FE-SEM micrographs revealed that the reaction layers having a thickness of ≤300 nm or less were formed in all IP glasses. The existence of nanosized voids at the interface between the reaction layer and glass matrix suggests a selective dissolution of a weak part of the glass components. Further studies are required for a detailed understanding of the formation mechanisms of the heterogeneous microstructures.

Fig. 5

FE-SEM photographs of the cross-section of the reaction layers on the glass samples immersed at 120°C for 168 h by MCC-2 static leaching tests: (a) 15LiIP35, (b) 15NaIP35, (c) 15KIP35, and (d) 20CaIP35.

Figure 6 shows the bright-field image and high-resolution TEM micrograph for a piece of the reaction layer formed on the surface of the 15KIP35 glass immersed in ultrapure water at 120°C for 168 h. The fringe pattern observed in the micrograph indicates the presence of a crystalline phase. Table 2 lists the results of TEM-EDS analysis of the cationic elements for point A of the bright-field image shown in Fig. 6. Oxygen element was qualitatively detected in the entire area of the TEM sample; in contrast, no K element (<1 at%) was detected on the reaction layer. The reaction layer had a richer Fe–O composition than the glass. Figure 7 shows the selected-area electron diffraction pattern of point A in the reaction layer, which indicates a broad ring corresponding to the noncrystalline phases and Debye rings corresponding to the crystalline phases of six-line ferrihydrite and FePO₄.^11,12)

Fig. 6

Bright-field image and high-resolution TEM photograph of a piece of the reaction layer on a 15KIP35 glass sample immersed at 120°C for 168 h.

Table 2 Cation fraction of point A (Fig. 6) in the reaction layer of the 15KIP35 glass sample. The fractions were analyzed by TEM-EDS.

Fig. 7

Electron diffraction pattern of the reaction layers formed in the 15KIP35 glass and Miller indices of Debye rings of six-line ferrihydrite and FePO₄ phases.

3.4 Glass structure

Figure 8 shows the Raman spectra of the as-prepared IP glass samples before the immersion tests. The Raman scattering bands are attributed to the symmetric vibration modes of P–O–P bonds for Q¹ and Q² PO₄ tetrahedra in the region of 700–800 cm⁻¹ and an overlap of symmetric and asymmetric vibration modes of non-bridging P–O bonds for Q⁰, Q¹, and Q² PO₄ tetrahedra in the region of 900–1400 cm⁻¹ (Qⁿ denotes the number of bridging oxygen atoms per PO₄ tetrahedron). The Raman spectra in the 800–1400 cm⁻¹ region were decomposed into six Gaussian peaks.^13,14) Table 3 and Fig. 9 show the assignments of Raman scattering bands based on previous studies and the decompositions of the Raman spectra, respectively. The peak positions and full width at half maximum were fitted to determine the relative intensity of each band, which is discussed in the next section.

Fig. 8

Raman spectra of the alkali and alkaline-earth iron phosphate glasses studied here.

Table 3 Assignments of Raman scattering bands of PO₄ Qⁿ structures.

Fig. 9

Deconvolution of Raman spectra: (a) 15LiIP35, (b) 15NaIP35, (c) 15KIP35, and (d) 20CaIP35.

4. Discussion

After the MCC-2 leach tests, interference colors, resulting from the formation of reaction layers, were observed in all alkali and alkaline-earth IP glasses, similar to the case of barium IP glasses.⁶⁾ The reaction layers suppressed the weight losses in water, as compared to the binary IP glasses.⁶⁾ There was no reaction layer in the binary IP glasses after the leach tests. The reaction layers formed on the surfaces of the alkali and alkaline-earth IP glasses played a protective role against water attack. The formation of reaction layers with/without macroscopic cracks and the dissolution of the glass samples depended on the composition of the IP glasses. The linear relation between the weight loss and immersion time for the type I glass samples with macroscopic cracks resulted in a relatively high weight loss per specific area. The glass samples were classified into two types according to the dissolution in water, as mentioned in section 3.1 (Fig. 1). Type I had macroscopic cracks in the reaction layer for 15LiIP35 and 15NaIP35 glasses, whereas type II exhibited a homogeneous reaction layer devoid of macroscopic cracks for the 15KIP35, 20CaIP35, and BaIP35 glasses. As shown in Fig. 4, the cations released in the water after the PCT method revealed that similar elements were dissolved into the leachate in both type I and II glasses. The macroscopic cracks present in the reaction layers decreased the protective effect against water attack for type I glasses. Hence, the formation of homogeneous reaction layers devoid of macroscopic cracks, as observed in type II glasses, is a better condition for achieving excellent water durability. The TEM observation for a piece of 15KIP glass revealed that the reaction layer comprises six-line ferrihydrite and FePO₄ nanocrystals with no K element (<1 at%). This result, which is similar to the microstructure of the reaction layer on the BaO–FeO–Fe₂O₃–P₂O₅ glass surface,⁶⁾ supports the claim that the constituents of alkali and alkaline-earth oxides affect the local pH around the reaction layer and promote the formation of six-line ferrihydrite microscopic phases. Consequently, the only difference between type I and II glasses is the existence of macroscopic cracks in type I glasses.

The Raman spectra of type I and II glasses, as shown in Fig. 8, were analyzed to understand the relation between the existence of cracks and the heterogeneity of the glass structure. The relative intensity ratio of I_NB(Q²)/I_NB(Q¹) for the symmetric vibration modes of the non-bridging oxygen of Q² and Q¹ PO₄ tetrahedra was calculated for three alkali IP glasses (Fig. 10). The ratio I_NB(Q²)/I_NB(Q¹) roughly expresses the relative fraction of Q² PO₄ tetrahedra. The Q² phosphate chains have poor water durability among Qⁿ PO₄ structures (for n = 0, 1, 2) because the highly polarized nonbridging oxygen interacts strongly with the high-polarized H₂O molecule.⁸⁾ The I_NB(Q²)/I_NB(Q¹) for type I glasses is relatively high compared to that for type II glasses. The Q² structural species remained locally in the type I glass matrix and its selective dissolution resulted in nano- and microcracks, which trigger heterogeneous macroscopic stress corrosion (Fig. 11).

Fig. 10

Relative intensities of Raman scattering bands for PO₄ Q⁰, Q¹, and Q² structures and the relative intensity ratio of Q² and Q¹ structures in three types of alkali iron phosphate glasses: 15LiP35, 15NaIP35, and 15KIP35.

Fig. 11

Schematic diagrams of (a) initial stage and (b) selective dissolution of PO₄ Q² species during the formation of the reaction layer for type I glasses.

Iron in IP glasses exists in two valency states: Fe²⁺ and Fe³⁺. The larger fraction of Q² species for IP glasses is attributed to the larger Fe^{2+ 15)} with poor durability. The similarity in the existence of minor Q² phosphate units, as well as their selective dissolution, was also observed in ZnO–P₂O₅¹⁶⁾ and ZrO₂–FeO–Fe₂O₃–P₂O₅⁸⁾ glass systems. Compared to Li⁺ and Na⁺ modifier ions for type I glasses, the existence of modifier cations with relatively large ionic radii,¹⁷⁾ and consequently, smaller cation-field strengths,¹⁸⁾ such as those for K⁺ and Ba²⁺, may contribute moderately to the reduction in the heterogeneity of Qⁿ PO₄ structures in the glass matrix. Therefore, the homogeneity of microscopic Qⁿ PO₄ units is key to the formation of homogeneous thin reaction layers comprising six-ferrihydrite and FePO₄ nanocrystals or P-containing noncrystalline nanogranular matrix⁶⁾ for excellent water durability in alkali and alkaline-earth IP glasses.

5. Conclusions

The durability of Li₂O-, Na₂O-, K₂O-, and CaO-containing IP glasses in water was studied and compared with that of BaO-containing IP glasses. Thin reaction layers of thickness <300 nm having interference colors were formed on all surfaces of the bulk glass samples after the MCC-2 immersion test. The glass samples were classified into two types depending on the presence or absence of macroscopic cracks on the reaction layer after MCC-2 tests. The type I glasses exhibited reduced durability in the presence of cracks on the reaction layers. They showed a linear relation between weight loss and immersion time. On the other hand, homogeneous reaction layers devoid of cracks were obtained in type II glasses, and hence, they exhibited better durability. The reaction layers in type II glasses played a better role in protecting the samples against water attack; hence, the dissolution of glass constituents was suppressed. The results of Raman spectroscopy indicate that the fraction of Q² PO₄ phosphate units remained heterogeneous in the type I glass matrix, which is attributed to the presence of macroscopic cracks in the reaction layers due to stress corrosion. The homogeneity of the microscopic Qⁿ PO₄ structure in the as-prepared glass matrix and the homogeneous thin reaction layers comprising six-ferrihydrite and FePO₄ nanocrystals after the immersion test are key to achieving excellent water durability for alkali and alkaline-earth IP glasses.

Acknowledgements

The FE-SEM observation was carried out on a FE-SEM apparatus at Research and Education Center of Materials Engineering, Course of Materials Design Engineering, Faculty of Engineering, Ehime University. The TEM observation was supported by the Division of Material Science of the Advanced Research Support Center (ADRES), Ehime University.

REFERENCES

• 1)   T.  Prevost,  M.  Blase,  H.  Paillard and  H.  Mizuno: ATW-Int. J. Nucl. Power 57 (2012) 308–313.
• 2)  T.H. Lorier, J.M. Pareizs and C.M. Jantzen: Initial Sulfate Solubility Study for Sludge Batch 4 (SB4), Westinghouse Savannah River Company (WSRC)-TR-2005-00213, April 2005.
• 3)   C.P.  Kaushik,  R.K.  Mishra,  P.  Sengupta,  A.  Kumar,  D.  Das,  G.B.  Kale and  K.  Raj: J. Nucl. Mater. 358 (2006) 129–138.
• 4)   X.  Yu,  D.E.  Day,  G.J.  Long and  R.K.  Brow: J. Non-Cryst. Solids 215 (1997) 21–31.
• 5)   G.K.  Marasinghe,  M.  Karabulut,  C.S.  Ray,  M.G.  Shumsky,  W.B.  Yelon,  C.H.  Booth,  P.G.  Allen and  D.K.  Shuh: J. Non-Cryst. Solids 222 (1997) 144–152.
• 6)   N.  Kitamura,  T.  Sakamoto and  H.  Takebe: Mater. Trans. 59 (2018) 437–442.
• 7)  Power Reactor and Nuclear Fuel Development Corporation: Standard Leach Test Methods for Solidified Radioactive Waste Forms, 59 (1984) [in Japanese].
• 8)   N.  Kitamura,  A.  Nomura,  A.  Saitoh,  H.  Kobayashi,  I.  Amamoto and  H.  Takebe: J. Ceram. Soc. Jpn. 126 (2018) 948–951.
• 9)   R.K.  Brow,  C.W.  Kim and  S.T.  Reis: Int. J. Appl. Glass Sci. 11 (2020) 4–14.
• 10)  D.-S. Kim, M.J. Schweiger, W.C. Buchmiller, J.D. Vienna, D.E. Day, D. Zhu, C.W. Kim, T.E. Day, T. Neidt, D.K. Peeler, T.B. Edwards, I.A. Reamer, R.J. Workman: PNNL-14251. Pacific Northwest National Laboratory, Richland Washington 99352 (2003).
• 11)   E.  Jansen,  A.  Kyek,  W.  Schäfer and  U.  Schwertmann: Appl. Phys., A 74 (2002) s1004–s1006.
• 12)   C.O.  Björling and  A.  Westgren: Geol. Foren. Stockh. Forh. 60 (1938) 67–72.
• 13)   L.  Ma,  R.K.  Brow and  A.  Choudhury: J. Non-Cryst. Solids 402 (2014) 64–73.
• 14)   P.A.  Bingham,  R.J.  Hand,  O.M.  Hannant,  S.D.  Forder and  S.H.  Kilcoyne: J. Non-Cryst. Solids 355 (2009) 1526–1538.
• 15)   H.  Takebe,  N.  Kitamura,  I.  Amamoto,  H.  Kobayashi,  T.  Tsuzuki and  N.  Mitamura: Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. B 57 (2016) 213–217.
• 16)   H.  Takebe,  Y.  Baba and  M.  Kuwabara: J. Non-Cryst. Solids 352 (2006) 3088–3094.
• 17)   R.D.  Shannon: Acta Crystallogr. Sec. A 32 (1976) 751–767.
• 18)   V.A.  Diztel: Glastech. Ber. 22 (1948) 41–50.