2.1 Glass preparation

Phosphate glasses with the composition of 2Ag₂O–(14+x)ZnO–14MgO–(70-x)P₂O₅ (x = 0-18 mol%) were synthesized using high-purity raw materials, including Ag₂O (DAEJUNG, 99.0%), ZnO (Thermo Scientific, 99.9%), MgO (Kojundo Chemical Laboratory Co., Ltd., Japan, 99.0%) and NH₄H₂PO₄ (DAEJUNG, 99.0%) powders. The chemical compositions of the prepared glasses are listed in Table 1. For each composition, 40 g of the batch materials were homogeneously mixed for 1 h using a 3D mixer. The blended powders were placed in Pt–Rh crucibles, melted in an electric furnace by heating to 1400 °C at a rate of 10 °C/min, and held for 1 h to ensure complete fusion. The melts were then transferred to an annealing furnace and annealed at 575 °C (near T_g) for 1 h. Photographs of the synthesized glass samples are shown in Figure 1.

Fig. 1. Photograph of synthesized glass samples

2.2 Measurement and calculation methods

The amorphous states of the fabricated glasses were confirmed by X-ray diffraction (XRD, D8 Advance, Bruker), and their densities were measured using a gas pycnometer (Anton Paar). The optical transmittances of the glasses in the wavelength range of 200–1000 nm were evaluated using a UV–visible (vis) spectrophotometer (V-770, JASCO Corporation). The chemical states of the elements within the glass matrix were analyzed via X-ray photoelectron spectroscopy (XPS) using binding energy spectra. Core-level peaks (e.g., O 1s) were fitted using a combination of Gaussian–Lorentzian functions. The binding energy scale was calibrated with reference to the C 1s peak at 284.8 eV. The thermal properties were evaluated by thermogravimetry-differential scanning calorimetry (TG–DSC, HITACHI STA 200RV); the samples were heated up to 900 °C at 10 °C/min rate. The glass structure was analyzed using Fourier transform infrared (FT-IR). FT-IR spectroscopy (Frontier, PerkinElmer) was performed in the wavenumber range of 400–1600 cm⁻¹.

The atomic packing density, C_g was calculated using Equation (2), where X_i is the molar fraction of oxide i and V_i is the ionic volume of oxide i.

The ionic volume V_i was determined using Equation (3), where N_A is Avogadro’s number; m and n are the number of cations and oxygen atoms in the oxide A_mO_n, respectively; r_A is the ionic radius of the cation; and r_O is the ionic radius of the oxygen anion. The ionic radii were obtained from Shannon and Prewitt^{[7, 16]}.

The free volume was calculated using Equation (4) ^[17].

In addition, we quantitatively evaluated the reducibility of silver in the glass by calculating its optical basicity, oxygen-ion polarizability, and ionicity/covalency. The theoretical optical basicity of multicomponent glass can be calculated based on the following equation proposed by Duffy and Ingram^{[9, 18]}:

where X₁, X₂, …, X_n are equivalent fractions based on the amount of oxygen each oxide contributes to the overall glass stoichiometry, and Λ₁, Λ₂, …, Λ_n are basicities assigned to the individual oxides^[19].

The oxygen-ion polarizability, α₀ ²⁻ was also calculated based on the optical basicity values, using Equation (6):

According to this relationship, the increment in the value of α₀ ²⁻ enhances the value of Λ_th.

The overall ionicity and covalency of the glass were calculated by first determining the ionicity of each constituent oxide, as follows (7) ^{[20, 21]}:

The fractional ionic character, f_AB of two elements A and B was calculated according to Pauling’s algorithms described in Ref.^[22], where ($\chi_A - \chi_B$) is a measure of the electronegativity difference between elements A and B^[23].

Subsequently, the overall ionicity of the glass (I_glass) was determined using the following equation, where X_i represents the mole fraction of each oxide, and f_i denotes the ionicity of each oxide, calculated based on the electronegativity differences between the constituent elements.

The covalency of the glass (C_glass) was determined using Equation (9).

For each glass composition, five specimens were prepared for density measurements. All values are reported as averages.

3.1 Physical and thermal properties

XRD analysis (Figure 2) verified that every glass specimen was fully amorphous, displaying a broad diffuse halo and no resolvable Bragg peaks. A faint yellow tint was also noted, consistent with partial reduction of Ag⁺ to metallic Ag⁰; however, any nanoparticles formed are inferred to be too small for XRD detection. In line with this interpretation, Saad et al.^[24] observed by TEM nearly spherical Ag nanoparticles of ~10–30 nm while the corresponding XRD patterns still indicated an amorphous phase. Likewise, Rahman^[25] reported that Ag-NP–containing glasses retained amorphous profiles in XRD. Prior work has further analyzed the detectability of Ag⁰ by XRD as a function of nanoparticle size^[26]. Taken together, the lack of silver reflections here is best explained by sub-critical particle dimensions, underscoring the strong size dependence of Ag diffraction features.

Table 2 summarizes the physical and thermal properties of the synthesized glasses. As P₂O₅ was progressively replaced by ZnO, both density and atomic packing density decreased approximately linearly, while the free volume increased. This trend indicates that the molar mass of the substituting oxide plays a decisive role in governing the bulk density: P₂O₅ (141.94 g/mol) is considerably heavier than ZnO (81.38 g/mol). The concurrent decrease in packing density and increase in free volume further suggest network depolymerization, consistent with replacing the network former (P₂O₅) with the network modifier (ZnO). Structurally, this substitution raises the fraction of non-bridging oxygens (NBOs), disrupts bridging linkages, and diminishes overall polymerization of the phosphate network. Thermal properties were evaluated using the glass transition temperature (T_g) and the dilatometric softening point (T_d). T_g reflects the combined influence of oxide melting points, cation radii, field strength, and network connectivity; in the present system, the oxide melting points appear to dominate. Specifically, the much higher melting point of ZnO relative to P₂O₅ rationalizes the observed increases in both T_g and T_d with increasing ZnO content^[27].

Fig. 2. XRD patterns of the synthesized glass samples.

3.2 Ag redox behavior and optical properties

The UV–vis transmittance spectra of the glass samples are presented in Figure 3. As the ZnO content in the glass network increases, a pronounced decrease in transmission is observed near 415 nm, which is attributable to the surface-plasmon resonance (SPR) absorption of metallic Ag nanoparticles formed via extensive reduction of Ag^{+ [28, 29]}. Notably, the Ag-reduction–related absorption feature disappears only in compositions containing ≤ 30 mol% ZnO. These results indicate that introducing ZnO at low levels is effective in suppressing the reduction of Ag⁺ within the glass network. On this basis, the present work identifies 30 mol% as an appropriate upper limit for the ZnO fraction that both enhances the antibacterial performance of zinc-phosphate glasses and mitigates undue reducibility, thereby defining an optimal compositional window for such materials.

The Ag 3d spectra exhibit a systematic shift toward the metallic Ag⁰ component with increasing ZnO content as shown in Figure 4. The pronounced trends in these optical properties—and their agreement with our complementary analyses—suggest that excessive ZnO loading increases the reducibility of Ag⁺, leading to reduced antibacterial efficacy and lowered optical transmittance; collectively, these effects may compromise the multifaceted applicability of these materials as antibacterial/optical glasses^[30].

Fig. 3. UV–vis transmittance spectra of glass samples

Fig. 4. Ag 3d XPS profiles of glass samples

3.3 Structural properties

As the concentration of the network-modifier ZnO increased, the relative area of the NBO-related O 1s component grew, consistent with a larger population of NBOs (Figure 5)^{[27, 31]}. Peak deconvolution corroborated this trend: the fraction of NBOs rose sharply with ZnO content, and at ~26 mol% ZnO the NBO/BO ratio exceeded unity. The deconvolution results, showing the ratio of NBO to BO, are presented in Table 3. Overall, the XPS results establish a positive correlation between NBO content and the formation of metallic Ag, indicating that NBO-rich, more ionic networks facilitate the reduction of Ag⁺ to Ag⁰. This supports the conclusion that the abundance of NBOs exerts a decisive influence on Ag redox behavior in phosphate-glass matrices.

To probe the structural characteristics of the glasses, FT-IR spectra were collected. As shown in Figure 6, distinct absorption bands were observed near 700, 900, and 1300 cm^-1, each corresponding to specific vibrational units. The prominent peak around 700 cm^-1 is assigned to the symmetric stretching of P–O–P linkages; the band near 900 cm^-1 relates to their asymmetric stretching; and the broad feature centered at ~1300 cm^-1 reflects the presence of P=O double-bond units within the glass network^[32]. Notably, all bands shifted to lower wavenumbers with increasing ZnO content. These spectral changes indicate progressive depolymerization of the phosphate network, attributable to the generation of NBOs that disrupt P–O–P and P=O connectivity. This trend signifies a transition from a highly polymerized phosphate framework to a more depolymerized structure driven by the network-modifying action of the Zn²⁺ ions^[33]. Consistent with this interpretation, glasses with higher ZnO exhibit a larger fraction of NBOs, whereas those with lower ZnO retain a higher fraction of BOs, reflecting the characteristic network-modifying role of Zn²⁺.

Optical basicity and oxide-ion polarizability were calculated using the oxide basicity values reported by Bordes-Richard et al.^[34] (Table 4). Consistent with the UV–vis results, both metrics increased with ZnO content, indicating an enhanced tendency for Ag⁺→Ag⁰ reduction. This behavior accords with Duffy’s framework, in which lower basicity limits the availability of electrons for redox reactions, and it aligns with structural models that emphasize the chemistry-driven network reconfiguration induced by modifier ions in phosphate glasses. To clarify the structure–redox correlation further, we evaluated ionicity and covalency (Table 4). Glasses with lower ZnO contents exhibited lower ionic and higher covalent character than their higher-ZnO counterparts, implying the smallest NBO fraction and, consequently, the weakest driving force for Ag reduction^{[31, 35]}. Taken together, these results rationalize the observed trend in Ag reduction through the combined effects of optical basicity, oxide-ion polarizability, and NBO content.

Our study indicates that the increasing ZnO content generates a larger fraction of non-bridging oxygens, which are more basic and electron-rich than bridging oxygens and can therefore donate electron density to Ag⁺ or stabilize excess negative charge around reduced Ag⁰ species, thereby lowering the energetic barrier for the Ag⁺ → Ag⁰ redox process. This interpretation is consistent with previous reports identifying NBO-rich environments as effective promoters of Ag⁺ reduction and nanoparticle formation in oxide glasses. As a result, the monotonic increase in NBO fraction (from FT-IR/XPS) and in optical basicity/ionicity with composition is not only a structural descriptor but also a direct indicator of the growing reducing power of the glass network^[36-^38]. However, the SPR minimum at ~415 nm does not necessarily follow this nearly linear NBO/optical-basicity trend, because the SPR response is governed not only by the total amount of reduced Ag⁰ but also by the size distribution and aggregation state of the Ag⁰ nanoparticles. Once silver reduction and clustering exceed a certain level, rapid growth and coalescence of Ag⁰ nanoparticles lead to a strong broadening and deepening of the SPR band, resulting in a disproportionate (non-linear) loss of transmittance near 415 nm^[36-^38].

Fig. 5. Deconvoluted O 1s XPS profiles of glass samples

Fig. 6. FT-IR spectra of glass samples