**1. Introduction**

The use of glass-making processing is favorable for the fabrication of Na+ conducting electrolyte tubes, which has been the key to the technological development of 1 MW Na/S secondary battery plants. However, the processing technique cannot be applied to wellknown β- and β″-aluminas (e. g., NaAl11O17 and NaAl5O8) and Nasicons (Na1+*x*Zr2P3-*x*Si*x*O12) because their high inclusion of Al2O3 or ZrO2 brings about the inhomogeneous melting or crystallization from glasses. Alternatively, Nasicon-like glass-ceramics were synthesized using the composition with lower content of ZrO2 (*m*Na2O·*x*ZrO2·*y*P2O5·(100-*m*-*x*-*y*)SiO2 [*m*=20, 30 mol%]), however, the conductivities (σ) attained were, at most, as high as σ300=2×10-2 S/cm at 300°C with the activation energies (*E*a) of ca. 30 kJ/mol. These low conductivities were attributed to the crystallization of the poorly conductive rhombohedral phase in these Nasicon-like materials. Na5YSi4O12 (N5), which comprises 12-(SiO4)4- tetrahedra membered skeleton structure (Fig. 1), is another Na+-superionic conductor with σ300=1×10-1 S/cm and *E*a=25 kJ/mol. A pioneering work on N5-type glass-ceramics has been performed by Banks *et al*. on the family of N5-type materials by substituting Y with Er, Gd or Sm. However, their results were not completely satisfactory because of the relatively lower conductivities of σ300<2×10-2 S/cm than the reported values of N5. This discrepancy may possibly have arisen from the occurrence of a less conductive metastable phase during crystallization, as is discussed below.

Contrary to the results of Banks *et al*., the present authors have produced glass-ceramics with σ300=1×10-1 S/cm and *E*a=20 kJ/mol, which were based on the phosphorus-containing N5-type materials discovered in the Na2O-Y2O3-P2O5-SiO2 system. These N5-type materials have been obtained, as well as Na3YSi3O9 (N3)-type materials, with the composition formula originally derived for N3-type solid solutions and expressed as follows,

$$\text{Na}\_{3\times3\times y} \text{Y}\_{1\times} \text{P}\_{y} \text{Si}\_{y} \text{O}\_{9} \text{ (x\lessapprox 0.6, y\lessapprox 0.5)}\tag{1}$$

With the aim of searching for more conductive glass-ceramic N5-type materials, the verification of the validity of the generalized composition formula

$$\text{Na}\_{3+3x\cdot g} \text{R}\_{1\cdot a} \text{P}\_{y} \text{Si}\_{3\cdot g} \text{O}\_{9} \tag{2}$$

for the synthesis of other kinds of rare earth N5-type glass-ceramics was studied first. Formula 2 is rewritten with formula 3 according to the formula N5.

Preparation of Na<sup>+</sup>

**2. Materials** 

and

Then,

**2.1 Preparation of glasses and glass-ceramics** 

**2.2 Measurements and characterization** 

atomic ratios of [Na]/[P+Si] and [Na]/[Y] were also prepared.

Superionic Conductors by Crystallization of Glass 85

Precursor glasses were prepared from reagent-grade oxides of anhydrous Na2CO3, R2O3 (R=Y, Sc, In, Er, Gd, Sm, Eu, Nd, La), NH4H2PO4 and SiO2; the mechanically mixed powders according to formula 2 or appropriate compositions shown below were melted at 1350°C for 1 h after calcinations at 900°C for 1 h. The melts were quickly poured into a cylindrical graphite, then annealed at 500°C for 3 h, giving NaRPSi glasses. The composition parameters studied were in the range of 0.2<x<0.6 and 0<y<0.5 of formula 2. As shown below, grain boundary conduction properties are discussed in relation to the properties of glasses. For the evaluation of the composition dependence of conductivity in Na+ conducting glasses, various sodium yttrium silico-phosphate glass specimens with different

Crystallization was carried out according to the previous report; bulk glasses were heated with an increasing rate of 75°C/h to a temperature above ca. 50°C of the glass transition point, which had been determined in advance by differential thermal analysis (DTA). This pretreatment was done in order to obtain homogeneous nucleation. After the annealing for 1 h, specimens were heated at temperatures of 800 to 1100°C, depending on the composition, for 0.5 to 72 h, thereafter slowly cooled in a furnace with a decreasing rate of 150°C/h to room temperature. These quenched glasses or glass-ceramic specimens were polished down

Ionic conductivities were measured by the complex impedance method on cylindrical glasses or glass-ceramics of typically 15 mm in diameter and 2 mm in thickness. Electrodes were prepared by sputtering of gold on polished surfaces. The applied ac field ranged from 5 to 10 MHz in frequency. The temperature dependence of the conductivity was measured in a similar way at several temperatures ranging from room temperature to 350°C. The complex impedance or admittance loci of glass and glass-ceramics were analyzed by an equivalent circuit (Fig. 2), which was experimentally found to comprise one and two semicircles in NaRPSi glasses and glass-ceramics, respectively. The two intercepting points on the real axis are interpreted as the resistance of crystallized grains (*R*G(c)) and the total resistance of grains and remaining glassy grain boundaries (*R*GB(g)). Assume the complex admittance diagram shown in Fig. 3, where the parameters L1 and L2 are set here as the radii

with 0.5 μm diamond paste, thereafter subjected to the conductivity measurements.

of the two arcs 1 and 2. Those parameters are related to one another as the following:

L1∝1/(*R*G(c)+*R*GB(g)) (4)

L2∝(1/*R*G(c))−1/(*R*G(c)+*R*GB(g)) (5)

 L2/L1=*R*GB(g)/*R*G(c) (6) Therefore, in an ideal grass-ceramic where residual glass would have negligible influence on

the total, arc 2 would be much smaller than arc 1, since L2/L1→0.

$$\mathbf{Na}\_{\mathbf{4(3^\ast,3x\cdot y)/3}} \mathbf{Y\_{4(1x)/3}} \mathbf{P\_{4y/3}} \mathbf{Si\_{(3\cdot y)/3}} \mathbf{O\_{12}} \tag{3}$$

In relation to previous works, formula 2 was employed in this work, and formula 3 is referred to in the results. The trivalent ions employed here for R3+ were Sc3+, In3+, Er3+, Gd3+, Sm3+, Eu3+, Nd3+ and La3+ as well as Y3+. These results are to be interpreted in terms of the effect of the rare earth ions on the crystallization of N5-type phase in glasses.

Fig. 1. Crystal Structure of Na5YSi4O12. Projection of the Na5YSi4O12 Structure on (100).

In the course of the fundamental studies on glass-ceramic Na3+3*<sup>x</sup>*-*<sup>y</sup>*R1-*x*P*y*Si3-*y*O9, we have interestingly found the crystallization of those N3- and Na9YSi6O18 (N9)-type phases as the precursors in the glasses. These are the analogues to the silicates N3 and N9 and therefore are the same members of the family of Na24-3*x*Y*x*Si12O36 as N5. Although we had also successfully synthesized those materials by the solid-state reactions of powders with the above composition of various sets of the parameters *x* and *y*, the metastability of those precursor phases had not been noticed in the synthesis. It has been observed that such precursor phases were transformed to the Na+-superionic conducting phase on specimens with appropriate sets of *x* and *y*. The present paper will deal with the thermodynamic and kinetic study on the phase transformation of metastable phases to the stable phase with Na+ superionic conductivity. The superiority of our present materials to the other silicate N5 will also be detailed based on the kinetic results.

The microstructure of a glass-ceramic, including neck growth among grains as well as grain size, is generally affected by the crystallization process. As the above mentioned devices utilize dc conduction properties of Na+-superionic conductors, another aim was to study the microstructural effects on the conduction properties of a whole glass-ceramic. Special attention was paid to the analysis of grain boundary properties using the Na2O-Y2O3-P2O5- SiO2 system. For the analysis of grain boundary properties, as will be discussed below, composition dependences of the conductivity of sodium silico-phosphate glasses containing Y2O3 were also studied in the Na2O-Y2O3-P2O5-SiO2 system. For convenience, the present materials are abbreviated as NaRPSi taken from the initials of the Na2O-R2O3-P2O5-SiO2 system.
