Effect of Y₂O₃ Concentration on the Surface and Bulk Crystallization of Multicomponent Silicate Glasses

Abstract

Multicomponent silicate glasses are crystallized by Y₂O₃ addition. Depending on the Y₂O₃ concentration, different crystalline phases evolve. In the absence of Y₂O₃, a multicomponent glass crystallizes as ZnSnO_3, while with the addition of just 3% of this oxide, ZnSnO₃ no longer crystallizes and ZrSiO₄ appears instead. Different yttrium silicate crystals are formed in all glasses containing Y₂O₃, but, while α-Y₂Si₂O₇ and β-Y₂Si₂O₇ are favored at low Y₂O₃ concentrations, the γ-Y₂Si₂O₇ and y-Y₂Si₂O₇ phases are favored at the maximum Y₂O₃ content. At a 12% Y₂O₃ concentration, barium and calcium silicate crystalline phases also evolve. Interestingly, the crystalline phases appearing on the surface of the material present different microstructures compared to crystals developed in the bulk. While the crystallized surface presents a tabular-shape type, crystallization in the bulk is of a prismatic type at low Y₂O₃ concentrations and of a globular (spherical) type at higher concentrations. The main crystal size ranges between 0.85 and 0.75 micrometers, but most of the crystals coalesce to form larger superstructures depending on the Y₂O₃ concentrations.

1. Introduction

Global warming is leading to an increase in the use of air-conditioning systems in cities with a high population, which also leads to an increase in the environmental temperature due to the abuse of these systems. This is a continuous process that leads to ceaselessly incremental temperature values in cities, an effect that is known as the urban heat island (UHI) effect. To mitigate the UHI effect, several strategies have been proposed and tested in the best cases [1,2,3,4]. Two of these strategies consist of the use of white glass enamels and/or glass beads [5]. White glass enamels can display high reflectivity to solar radiation due to the presence of micrometric crystals of high refractive indexes within their structure [6,7], whereas in the case of glass beads, the high retro-reflectivity effect is accomplished due to their spherical shape, leading to the reflection of most solar radiation and, thus, avoiding its transmission to the inner side of buildings [8,9]. In this case, glass beads are small transparent glass spheres deposited and fixed over a white substrate, with these being mainly constituted of zirconium silicate (ZrSiO₄) because of its high refractive index (n_d = 1.94–1.96). Nevertheless, the use of bead compositions such as barium titanate (n_d = 2.43) has also been reported, although in this case, the colorimetric appearance of the substrate is changed [9].

White substrates such as glass enamels, in which different forms of ZrSiO₄ crystallize, have been widely used for a long time in the ceramic tile industrial sector [10], but the rising price of zirconium oxide (ZrO₂) has led to the development of new materials in which other crystalline phases, such as gahnite, also appear (ZnAl₂O₄; n_d = 1.79–1.80) [11]. At the same time, glasses with high refractive indexes (n_d = 1.59–1.62) might also be used [12], although for those with very high refractive indexes (n_d = 2.26–2.33), it is also necessary to employ some other expensive oxides such as La₂O₃ and Nb₂O₅ [13]. There are other oxides that are less expensive (SnO₂, ZnO, BaO, CaO, MgO, etc.), which can be used for preparing high-refractive-index glasses and are also used as raw materials for the manufacturing of glass enamels [14,15,16]. In the case of Y₂O₃, it can serve as a material for the preparation of high-refractive-index glasses as a network former or as a network modifier, depending on its concentration [17]. With these premises, in the literature, there are reports on several glass compositions containing Y₂O₃ in combination with some other oxides, such as Li₂O [18], Al₂O₃ [19], MgO [20], ZnO [21], and ZrO₂ [22]. The incorporation of different oxides (Y₂O₃, SnO₂, ZnO, BaO, etc.) into the base of a silicoborate glass generally leads to the obtainment of high-refractive-index glasses that can be used in optical applications [18,23,24,25]. However, when these glasses are heat-treated at high temperatures, they are prone to crystallizing. The effect of ZrO₂ and ZrSiO₄ has been widely studied for their incorporation in multicomponent glasses and glass enamels in the ceramic tile sector. However, this effect has not been analyzed for Y₂O₃, even though this oxide has interesting properties related to its high refractive index and the possibility to produce glass enamels of high whiteness. The objective of this work is to study the effect of Y₂O₃ concentration on the crystallization behavior of a multicomponent oxide silicoaluminate glass containing Li₂O₃, Na₂O, K₂O, MgO, CaO, BaO, ZnO, ZrO₂, SnO₂, and B₂O₃. These crystallized materials can be used as glass enamels for high-reflecting surfaces in ceramic tiles for roofs and façades, home ovens, fillers for white paints, etc.

2. Experimental Section

2.1. Glass Preparation and Crystallization

All glasses prepared in this work were obtained from the following raw materials: silica quartz, H₃BO₃, Al₂O₃, Na₂CO₃, K₂CO₃, MgCO₃, CaCO₃, BaCO₃, ZrO₂, ZnO, SnO₂, and Y₂O₃, with all of them being of 99.9% purity. The appropriate amounts of these raw materials were thoroughly mixed for 2 h and then heat-treated in a Pt crucible. The thermal cycle to which these raw materials were subjected was a first pre-treatment at 900 °C for 1 h to fully oxide carbonates and, subsequently, the temperature was raised to 1500 °C and maintained for another 2 h to allow melting and homogenization. The melted glasses were then poured on metallic plates and left to cool to room temperature. Finally, the obtained glasses were annealed at 550 °C to remove internal stresses. A fraction of the glass was cut into small pieces sized 2 × 2 × 0.5 cm, and the remaining materials were agate-milled to obtain particles below 50 µm.

The crystallization schedule was in accordance with the differential thermal analysis (DTA) results, as will be described in the Results section. All DTA curves presented exothermic peaks that were assigned to the formation of crystals inside the glass matrix. The crystallization procedure consisted of heating at a rate of 5 °C/min, followed by holding times of 30 min and then cooling at a rate of 5 °C/min. After this process, all glasses were stored for further characterization. Figure 1 shows an image of the as-prepared and crystallized glasses, and the white color obtained after crystallization can be observed due to the formation of different crystals, as will be described below.

2.2. Characterization

The chemical composition of the obtained glasses (

Table 1. Chemical compositions (wt %) of the prepared glasses.

Component	AK-0Y	AK-3Y	AK-5Y	AK-12Y
SiO2	50.40	48.60	44.90	38.60
Al2O3	14.10	14.20	14.45	14.80
Na2O	0.86	0.86	0.86	1.50
K2O	1.04	0.88	0.97	1.10
MgO	6.50	6.40	5.06	6.75
CaO	9.25	9.69	9.86	9.90
BaO	5.10	5.00	5.06	3.90
ZnO	4.26	3.82	3.53	3.03
SnO	3.81	3.65	3.86	4.16
ZrO2	4.00	3.72	3.78	3.50
Y2O3	0	2.85	5.06	11.90
Refractive index	1.651	1.653	1.654	1.655

) was analyzed by X-ray fluorescence (Philips, MagiX, Eindhoven, The Netherlands). The materials were characterized by X-Ray diffraction (XRD) using a Bruker D8 Advance fitted with a Cu K_α1 (1.540598 Å) anode and working at 40 kV. All the measurements were carried out in the 2θ range of 5–70°. Raman spectra were obtained with a spectrometer (Renishaw InVia, Gloucestershire, UK) using a 514 nm laser in the region from 100 to 1500 cm⁻¹ and for each spectrum 10 scans were collected (background subtraction was performed in all the cases by assuming a 6-order polygonal spline line). UV–Vis–NIR reflectance was obtained with a spectrophotometer (PerkinElmer, Lambda 950, Waltham, MA, USA) equipped with an integrating sphere in the 280–2500 nm spectral range. The microstructures of the crystalline phases developed in the glass matrix were observed by using a field emission scanning electron microscope (FE-SEM, Hitachi S-4700, Chiyoda, Tokyo, Japan) on the fractured surface attacked by hydrofluoric acid (HF, 5%) for 5 s, rinsed and dried at 100 °C for 1 h, and finally coated with Au. Energy dispersive analysis (EDS) on fracture surfaces was carried out by using a Noran X-ray instrument coupled to the FE-SEM (Thermo Fisher Scientific, Waltham, MA, USA). Each analysis was carried out for 10 min at 20 kV.

3. Results

3.1. Glass Characterization

The obtained glasses were all transparent and with very little brown coloration. Figure 2 shows the XRD diffractograms of the prepared glasses where there is a noticeable absence of any crystallization peak. Additionally, a broad halo appears in the range 2θ = 17–40° attributed to the silicoaluminate glass structure that indicates the absence of any long-range translational order [26]. As the amount of Y₂O₃ increases in the glass composition, a slight displacement to higher 2θ is observed. This result indicates significant modifications that this oxide is producing in the glass structure. Y₂O₃ is known to act as a network former for concentrations lower than 5% and as a network modifier for higher concentrations [20]. Therefore, it can be inferred that the XRD patterns of the AK-5Y and AK-12Y in Figure 2 show the highest displacement to a higher 2θ range. Additionally, a shift of the main band of the FTIR spectra is also observed (Figure 3) when the Y₂O₃ content is increased. In the FTIR of all the glasses, the three characteristic bands observed at approximately 1000 cm⁻¹, 725 cm⁻¹, and 460 cm⁻¹ can be assigned to Si-O-X bonds where X equals Al, Y, Zr, Sn [22,27,28]. It should be noted that in these materials, there is also the possibility of bonding to Zn [16], although due to the low concentration of this element, it is likely acting as network modifier [29].

Regarding the main band in the FTIR spectra (asymmetric stretching of the Si-O-X bond), it is centered at 1009 cm⁻¹ in the glass with no Y₂O₃. As the Y₂O₃ concentration increases from 0 to 3, 5, and 12%, this band shifts towards 1000, 988, and 966 cm⁻¹, respectively. This shift suggests that the Y₂O₃ is effectively contributing to the change likely through the contribution of Si-O-Y bonds where the Y₂O₃ acts as a network former [30]. The symmetric stretching of the Si-O-X bond is observed in the FTIR at about 725 cm⁻¹ and this band is also affected by the same displacement to lower frequencies. However, it is noticed that the O-Si-O bending appearing at 460 cm⁻¹ does not change in position, indicating that only the stretching modes are influenced by the Y₂O₃.

The Raman spectra in Figure 4 show two high-intensity and broad bands around 990 cm⁻¹ and 510 cm⁻¹, which are attributed to the Si-O-X bonds. Depending on the composition, there are additional medium-intensity bands centered around 1120 cm⁻¹, 750 cm⁻¹, and 360 cm⁻¹, as well as three small bands or shoulders at 1180 cm−1, 680 cm⁻¹, and 610 cm⁻¹. The wide bands located at 980 and 1120 cm⁻¹ are assigned to the Si-O-X bonds with X = Al, Y, Zr, Sn [31,32,33,34] and to the presence of Qⁿ units (n = 0, 1, 2, 3, 4) where Q represents the SiO₄ tetrahedra and n the number of bridging oxygens (BOs). These Qⁿ units are formed after the BOs are broken by network modifier elements (Na, K, Ca, etc.) to form non-bridging oxygens (NBOs) [32,33].

The medium-intensity band centered at 1120 cm⁻¹ must be assigned to the Q⁴ units, representing a three-dimensional SiO₂ network. On the other hand, the band at 990 cm⁻¹ should be attributed to either the Q³ (Si₂O₅²⁻), Q² (SiO₃²⁻) or Q¹ (Si₂O₇⁶) units. This indicates the appearance of NBOs in the glass composition due to the presence of network modifier elements [35]. In the Raman spectra, it is also noticed that the band at 1120 cm⁻¹ is only present in the AK-0Y and AK-3Y glasses, disappearing when the Y₂O₃ composition is 5% and 12% (glasses AK-5Y and AK-12Y). This observation is attributed to the disappearance of the SiO₄ units to form new NBOs that, together with the shift of the band centered at 1000 cm⁻¹ towards lower frequencies and it increases in intensity as the Y₂O₃ concentration increases, corroborates the effect of Y₂O₃ as a network modifier at the highest concentrations and the formation of large amounts of NBO [34].

Figure 5 displays the DTA curves of the prepared glasses, indicating several temperature phenomena such as the glass transition temperature (T_g) and the exothermic peak of crystallization (T_p) along with its corresponding temperature onset (T_x) and the melting temperature (T_m). It can be observed that for the AK-12Y sample, two endothermic T_m peaks appear, indicating at least two phases that melt at very close temperatures. As shown below, this glass has several crystalline phases that may have different melting temperatures. The actual values of these characteristic temperatures are provided in

Table 2. Characteristic temperatures (T_g, T_x, T_m) and glass stability parameters.

	AK-0Y	AK-3Y	AK-5Y	AK-12Y
Tg (±6 °C)	721	725	735	759
Tx (±8 °C)	959	963	969	972
Tm1 (±1 °C)	1121	1145	1147	1150
KH	1.47	1.31	1.31	1.13
β	26.35	21.08	22.48	20.87
TC (°C)	238	238	234	213

. The glass transition temperature corresponds to the temperature at which the atoms suffer a structural relaxation, and the structural units rearrange due to the high viscosity of the solid glass [26]. Based on these temperatures, several crystallization parameters can be calculated such as the glass stability (K_H, Equation (1)), the glass-forming ability (β, Equation (2)), and tendency to crystallize (T_C, Equation (3)) [36]:

$$K_H=\raisebox{1ex}{$\left({\mathrm{T}}_{\mathrm{x}}-{\mathrm{T}}_{\mathrm{g}}\right)$}\!\left/ \!\raisebox{-1ex}{$\left({\mathrm{T}}_{\mathrm{m}1}-{\mathrm{T}}_{\mathrm{x}}\right)$}\right.$$

(1)

$$\beta =\raisebox{1ex}{${\mathrm{T}}_{\mathrm{x}}{\mathrm{T}}_{\mathrm{g}}$}\!\left/ \!\raisebox{-1ex}{${\left({\mathrm{T}}_{\mathrm{m}1}-{\mathrm{T}}_{\mathrm{x}}\right)}^2$}\right.$$

(2)

$$T_c={\mathrm{T}}_{\mathrm{x}}-{\mathrm{T}}_{\mathrm{g}}$$

(3)

Among these parameters, K_H and T_C represent the resistance against crystallization while the parameter β represents the relative ease of obtaining a glass after cooling the melted composition. Table 2 provides the corresponding values for the characteristic temperatures and the calculated stability parameters. It is observed that T_g, T_x, and T_p increase with the Y₂O₃ concentration, indicating that this oxide is being incorporated into the glass network as its concentration is raised. However, the calculated parameters K_H, β, and T_C show a continuous decrease with the concentration of Y₂O₃, a result that is interpreted as a decrease in the stability against crystallization as the Y₂O₃ amount is raised. According to the literature, it has been demonstrated that the incorporation of low concentrations (below 3%) of Y₂O₃ into different glass compositions produces an increase in the stability of the obtained glasses because the Y³⁺ ions become tetrahedrally coordinated and act as a network former [20].

3.2. Crystallization Process

The DTA results presented in Figure 5 and the data of Table 2 were used for the subsequent treatments applied in all glasses to develop the crystal phases. According to the position of the exothermic peaks, the glasses were heat-treated at 1040 °C for 10 and 30 minutes. In all cases the transparent glasses were transformed into white glass-ceramic materials due to the crystallization process and the formation of the different crystalline phases. No significant differences were observed by varying the dwell time at the maximum temperature of crystallization. Therefore, in this work, we only present the results corresponding to 30-minute dwell time since crystallization was enhanced. Figure 6 shows the high-intensity XRD reflections corresponding to the crystallized phases. An interesting change occurs when just 3% Y₂O₃ is added to the glass composition since the crystallization peaks of the AK-0Y sample completely disappear and are substituted by several other phases that are maintained at the other different concentrations. This indicates the powerful effect of the Y₂O₃ on the crystallization process.

The XRD peaks of the AK-0Y sample correspond to the reflections of ZnSnO₃ crystals (JCPDS-028-1486), a phase that disappeared with the sole incorporation of the Y₂O₃, as mentioned above. In the AK-3Y, AK-5Y, and AK-12Y samples, zirconium silicate (ZrSiO₄, JCPDS-003-0655) is encountered as the main crystalline phase growing together with other minor crystalline phases such as barium silicate (BaSiO₃, JCPDS-026-0179), calcium silicate (Ca₂SiO₄, JCPDS-029-0369), wollastonite (CaSiO₃, JCPDS 01-084-0655), and several yttrium silicates (α-Y₂Si₂O₇ JCPDS-021-1457, β-Y₂Si₂O₇ PCPDS-021-1454, γ-Y₂Si₂O₇ JCPDS-042-0167, y-Y₂Si₂O₇ JCPDS-048-1623). The crystallization complexity increases as the Y₂O₃ concentration rises. The sample with 3% Y₂O₃ mainly exhibits the formation of the ZrSiO₄ phase and most of the other minority crystals are present in very small amounts. Increasing the concentration of Y₂O₃ to 5% leads to the increased appearance of these other minority phases. A close analysis of the XRD patterns reveals that both the α-Y₂Si₂O₇ and γ-Y₂Si₂O₇ crystalline phases are still present in the samples containing 3% and 5% Y₂O₃ while these phases decrease in the AK-12Y sample, and the y-Y₂Si₂O₇ crystallizes instead, acquiring a higher intensity than that of both α- and γ-yttrium silicate phases. Additionally, in all the XRD patterns there is a broad halo attributed to the glassy (or non-crystallized) phase which decreases in intensity with the Y₂O₃ concentration. This effect is due to the fact that this oxide leads to an increase in the tendency to form crystals inside the glass. Vomacka and Babushkin [22] already demonstrated that the presence of ZrO₂ and Y₂O₃ in a glass material increases the rate of nucleation and crystal formation, exerting a positive influence on glass crystallization.

The glassy phase mentioned above can be better analyzed through Raman spectroscopy carried out on the crystallized samples (Figure 7). The spectra of the samples AK-0Y and AK-3Y are similar to the ones presented in Figure 4, typical of a glassy material (no crystals were detected). In the spectra of the samples containing up to 5% Y₂O₃, the main band is broad and located at about 1000 cm⁻¹, shifting to 1019 and to 1026 cm⁻¹ for the AK-3Y and AK-5Y samples, respectively. This relative displacement of the band indicates that some of the Y³⁺ cations may cease to act as network modifiers and become network formers. In the spectrum of the sample containing 5% Y₂O₃, small bands appear at about 510, 880, 960, and 1010 cm⁻¹, which become well-defined peaks in the spectrum of the sample AK-12Y. At the highest Y₂O₃ concentration, several other bands are also noticed at 330 cm⁻¹, 390 cm⁻¹, 450 cm⁻¹, and 660 cm⁻¹.

ZrSiO₄ has characteristic Raman bands at 1010, 450, and 330 cm⁻¹ [37] whereas Ca₂SiO₄ and CaSiO₃ show their fingerprint bands at 1010, 960, 860, 660, and 330 cm⁻¹ [38]. On the other side, different yttrium silicates provide Raman spectra composed of several bands, with the most important ones located at about 400 cm⁻¹, 880 cm⁻¹, and 905 cm⁻¹ [39,40]. Since most of these bands are very close to each other, the Raman spectra of the polycrystalline materials result in an overlapping of the spectra of the different constituents, leading to wide bands that are difficult to uniquely assign. Nevertheless, it should be emphasized that all the Raman spectra show the presence of a glass phase intricated with the crystallized ones, indicating that the developed crystals are widespread all over the glass matrix as described below.

The microstructures of the crystallized glasses as observed by FE-SEM are shown in Figure 8 and Figure 9. The micrographs show that the samples present two zones of crystallization, one starting on its surface (Figure 8) and the other appearing in the bulk (Figure 9). On the outer surface of the samples, a rough crystallized fracture is observed while the microstructure of the bulk fracture surface appears smooth with widespread crystals. The length of this rough crystallized fracture in the outer surface varies from 42 ± 4 for 0% Y₂O₃ to 53 ± 5, 560 ± 12, and 592 ± 12 for increasing amounts of this oxide. This observation suggests that the surface crystallization evolves inwards towards the bulk of the sample. Similar results have been reported by several other authors in different materials, such as Zheng et al., who studied yttrium aluminosilicate glasses [34], Wang et al. who described the same observations in zinc borosilicate glasses [41], and Wisniewski et al. for MgO/Y₂O₃/SiO₂/Al₂O₃/ZrO₂ glasses [42]. According to these observations, it can be deduced that for Y₂O₃ concentrations below 6%, the Y₂O₃ does not act as a crystallization promoter and the crystallization is mainly driven by the glass constituents. However, for concentrations above 6%, this oxide promotes the crystallization of the different phases.

It should be emphasized that this surface crystallization effect is not exclusive to this multicomponent glass but has also been found in several other Y₂O₃-containing materials [19]. In these micrographs, the surface-crystallized zone also changes with the Y₂O₃ concentration in the sense that, in the absence of Y₂O₃, the crystals do not maintain any specific shape whereas tabular-shaped crystals begin to appear clearly for the 3% Y₂O₃ concentration and dominate the microstructure of the AK-12Y sample. In addition, in the presence of the Y₂O₃ oxide, either inside or over the prismatic crystals, there also appear small crystals in the shape of prisms and spheres (globular shape). The amount of these small crystals increases with the Y₂O₃ content and their size decreases as well.

The micrographs of Figure 9 present the microstructures appearing in the bulk of the crystallized samples. Widespread crystallization is observed all over the bulk sample and the number of crystalline features increases with the Y₂O₃ concentration. And just as with surface crystallization, the size of the crystal decreases as the amount of Y₂O₃ is raised. Indeed, for the 12% Y₂O₃ sample, most of the bulk material is covered with very small crystals. The shape of these crystals also depends on the Y₂O₃ content, acquiring a petal-like dendritic shape for samples with no Y₂O₃ incorporated whereas, with a small amount of this oxide, the crystals adopt a pyramidal shape. When the concentration of Y₂O₃ is then raised to 6 and 12%, the crystals become spherical or globular in shape.

4. Discussion

The above results demonstrate the strong influence of the Y₂O₃ concentration on the structure and crystallization behavior of multicomponent silicate glasses. From the XRD patterns (Figure 2), FTIR (Figure 3), and Raman (Figure 4) spectroscopies, it is demonstrated that Y₂O₃ is incorporated either as network former or modifier depending on its concentration. At low Y₂O₃ amounts, this oxide acts as a network former and its role within the glass network changes to glass modifier when its concentration is high enough. For concentrations of 5% and 12%, the Raman spectra show the disappearance of Q⁴ units (SiO₂) to form new Q³ and Q² units. These results are in accordance with those found from thermal analysis (Figure 4) where the glass stability decreases as the amount of Y₂O₃ is increased and the tendency to crystallize increases in the same sense.

Since the main crystallization peaks in the glass materials appeared between 1006 and 1032 °C, the subsequent crystallization process was carried out at 1040 °C. The glasses were heat-treated at this temperature for 30 minutes. It should be noted that there were some remainders in the sample which are still in the glassy state, though. After this crystallization process, the crystalline features developed on both the surface and the bulk of the samples. The XRD patterns of crystallized glasses (Figure 6) revealed the occurrence of different crystalline phases as a function of Y₂O₃. If the original glass is produced in the absence of Y₂O₃, ZnSnO₃ is the unique crystalline feature appearing in the material despite the multicomponent character of the glass sample. Adding just 3% of Y₂O₃ to this glass, the ZnSnO₃ no longer crystallizes and ZrSiO₄ crystallizes instead. According to these pieces of evidence, it can be concluded that Y₂O₃ prevents the crystallization of ZnSnO₃ and favors the ZrSiO₄ crystallization. Increasing the Y₂O₃ concentration to 5% leads to the appearance of some other crystalline phases.

The assignment of the diffraction peaks in the XRD patterns to their corresponding reflections allows us to conclude that these minority phases were mainly yttrium silicates (α and β) and BaSiO₃. At the maximum amount of Y₂O₃, ZrSiO₄ remains the majority phase, the reflections of the α- and β-yttrium silicate phases decrease in intensity, and some other reflections attributed to other crystalline phases start to increase. These new crystalline phases that increase, in the sample containing 12% Y₂O₃, are calcium silicates (including wollastonite), γ-yttrium silicate, and y-yttrium silicate.

These XRD and Raman spectra results permit us to determine the type of crystals and their evolution during heat treatment as a function of the Y₂O₃ concentration in the glass. These crystals are responsible for the white color of the treated samples and can be used as enamels of high reflectivity to minimize the heat transfer into houses to achieve a lower temperature than if such enamels were not used.

Although the above-mentioned crystalline phases appear in these glasses, the presence of two well-defined crystallization regions remains a common feature in all the materials. It is interesting to note the crystallization region starting at the surface of the material which grows inwards independently of the presence or absence of Y₂O₃, indicating that the surface crystallization is mainly due to the different oxides of the glass composition. The second crystallization region occurs in the bulk itself. As shown in the micrographs of Figure 8, the thickness of these surface-crystallized zones increases with the Y₂O₃ concentration, indicating that this oxide also promotes crystallization at the outer surface of the sample. This phenomenon has also been observed in yttrium aluminosilicate [31, 34, 42] and zinc borosilicate glasses [41]. In turn, as observed in Figure 9, the amount of Y₂O₃ also influences the bulk crystallization of the prepared glasses and the concentration of crystals increases while their sizes decrease.

Figure 10 and Figure 11 show the microstructures of the surface and bulk, respectively, as observed at high magnification. As observed in Figure 10, the surface crystallization of the materials containing Y₂O₃ can be described with two types of crystallites, with the largest crystals having a long prismatic shape that grows adjacent to the surface. On the other hand, the smallest crystals adopt a pyramidal shape for the 3% Y₂O₃ sample and a globular shape in samples AK-5Y and AK-12Y. The size of the small crystals distributed all over the crystallization surface is determined from the SEM micrographs and the averaged data are collected in

Table 3. Crystal sizes (μm) as observed by FE-SEM analysis (*) or XRD data (‡) and crystallization percentage (CP) of the heat-treated glasses determined from FE-SEM images and XRD patterns.

	AK-0Y	AK-3Y		AK-5Y	AK-12Y
Surface	-	0.40 *		0.29 *	0.19 *
Bulk	1.47 *	0.53 *	0.45 *	0.44 *	0.25 *
Shape	petal-like	polygon	cuboid	globular	globular
ZnSnO3	0.854 ± 0.135 ‡	-		-	-
ZrSiO4		0.828 ± 0.054 ‡		0.804 ± 0.101 ‡	0.745 ± 0.130 ‡
CP (%)	76.70	69.77		89.72	94.17

. As displayed in Figure 11, the crystals in the microstructure of the bulk material appeared with different shapes depending on the Y₂O₃ concentration. In the absence of Y₂O₃, the ZnSnO₃ crystals are petal-like or dendritic. With incorporation of up to 3% Y₂O₃, these petal-like crystals disappear (ZnSnO₃ no longer crystallizes) and, instead, star-shaped or pyramidal/cubic-shaped ZrSiO₄ crystalline features form. If the Y₂O₃ concentration is raised beyond 6%, these small crystals further evolve into a globular/rounded shape. These observations are summarized in Table 3.

As a result, it can be concluded that the crystal sizes observed by FE-SEM decrease with the increase in Y₂O₃ concentration. This decrease also depends on the shape of the crystals and their location (surface or bulk). The petal-like ZnSnO₃ crystals (0% Y₂O₃) are slightly larger than the ZrSiO₄ ones, and for the case of ZrSiO4 crystals their sizes are larger when they present pyramid/cuboid shapes than for globular crystals (6% and 12% Y₂O₃).

XRD data also permit the calculation of the crystallite size by means of the Scherrer equation [43]:

$$L=\raisebox{1ex}{$K\mathrm{\lambda }$}\!\left/ \!\raisebox{-1ex}{$D_{2\mathrm{\theta }}cos\mathrm{\theta }$}\right.$$

(4)

where L is the crystallite size, λ is the X-ray wavelength, D_2θ is the peak width at the half maximum at a particular value of the diffraction angle (2θ), and K is the shape factor, a constant which is a function of the crystallite shape (K = varies from 0.9 to 1 and it is 1 for spherical particles). This equation has been applied to the main peaks of the XRD patterns (Figure 6) and the obtained values of the crystallite sizes are collected in Table 3. In the AK-0Y sample, where only ZnSnO₃ is present, the crystallite size is about 0.854 µm. For the other samples, only the ZrSiO₄ peaks were considered, although it should be noted that other phases crystallize simultaneously (yttrium silicates, BaSiO₃, and wollastonite). Overall, there is a trend of decreasing crystal size as the amount of Y₂O₃ is raised.

The XRD patterns in Figure 6 show that the broad glass halo observed in the non-crystallized materials (Figure 2) is still present in the AK-0Y and AK-3Y crystallized samples. However, its intensity decreases with increasing Y₂O₃ concentration, indicating a higher degree of crystallization in the samples with more Y₂O_3. According to Equation (5), the crystallization percentage (CP) can be estimated by:

$$CP =\raisebox{1ex}{$100I_c$}\!\left/ \!\raisebox{-1ex}{$\left(I_c+I_g\right)$}\right.$$

(5)

In Equation (5), I_c corresponds to the intensity of the main peak of the XRD patterns and I_g is the maximum height of the broad glass halo which is centered at 2θ = 20–40°. The calculated CP values are also collected in Table 3. According to these data, it is clear that the heat-treated glasses are highly crystallized especially the sample containing 12% Y₂O₃ where more than 94% of the whole material is crystallized. These results are also in accordance with the SEM images where two crystallization zones were observed, indicating that not only is the surface of the material crystallized but so is the bulk. Table 3 reveals that the glass without Y₂O₃ incorporated has a higher crystallization degree than the one containing 3% Y₂O₃, suggesting in this case that for this low concentration, Y₂O₃ acts as a network former of the silicate glass. However, when increasing the Y₂O₃ contents to 6% and 12%, the crystallization percentage increases very quickly (69.77% to 94.17%, respectively). Therefore, when Y₂O₃ is incorporated in larger amounts, it acts as a nucleating agent promoter of crystallization in these multicomponent glasses. In line with the results reported by Vomacka and Babushkin, small amounts of glassy phase remain in the structure of the glass-ceramic materials developed with SiO₂, Al₂O₃, ZrO₂, and Y₂O₃ compositions because a small amount of Y₂O₃ still acts as a network former [22].

Finally, most of the grown crystals have been analyzed by EDS and Figure 12 shows several FE-SEM images with the chemical analysis of different spots carried out in such samples.

Table 4. Chemical compositions determined by EDS analysis of the spots shown in Figure 12. Note that points 4 and 5 correspond to the AK-0Y sample.

Component	1	2	3	4	5	6
SiO2	19.07	27.27	51.69	28.35	25.23	35.20
Al2O3	6.08	6.38	13.01	8.41	9.71	12.28
Na2O	1.03	1.21	1.02	0.08	0.82	1.25
K2O	1.10	2.01	0.99	0.01	1.28	1.08
MgO	6.48	6.77	3.86	2.92	3.44	3.56
CaO	6.09	7.25	9.99	4.65	7.63	10.63
BaO	2.63	3.03	6.76	4.88	5.03	6.70
ZnO	7.28	7.93	4.21	12.83	3.50	5.77
SnO	7.55	1.35	1.70	33.41	11.21	1.07
ZrO2	8.89	9.92	6.77	4.46	11.59	3.44
Y2O3	33.80	26.98	0	0	20.56	19.02

collects the chemical compositions of the most representative crystals.

The results of Table 4 indicate that the petal-like crystals formed in the AK-0Y sample (points 3 and 4) are mainly formed by SnO and ZnO as concluded by XRD analysis (Figure 6). The other crystals (polygonal or globular) formed on the surface or in the bulk of the samples (points 1 and 5) are mainly composed of zirconium silicate and yttrium silicates, while those zones on the surface or in the bulk of the samples (points 2, 3, and 6) tend to present a chemical composition close to the original glass before crystallization.

5. Conclusions

The effect of different Y₂O₃ concentrations (0%, 3%, 5%, and 12%) on the structure and microstructure of multicomponent glasses has been illustrated. XRD, Raman, and FTIR analyses were used for studying the prepared glasses while XRD and FE-SEM were used for the crystallized materials. Raman results revealed a decrease in Q⁴ units when the Y₂O₃ concentration is higher than 3% which is an indication of its network modifier effect. The glass without Y₂O₃ crystallizes in ZnSnO₃ while those with Y₂O₃ crystallize with ZrSiO₄ as the main crystalline phase along with different calcium, barium, and yttrium silicates. Y₂O₃ prevents the ZnSnO₃ crystalline formation in the multicomponent silicate glasses whereas it favors the crystallization of ZrSiO₄. For the lowest concentrations of Y₂O₃, this oxide promotes the formation of α- and β-yttrium silicates. But for the 12% concentration, the γ- and y-phases of yttrium silicates are also present along with barium and calcium silicates. Despite the crystallization seeming to be a surface-driven process, it occurs either on the surface or in the bulk of the multicomponent glasses. The width of the surface crystallization increases with the Y₂O₃ concentration while the concentration of crystals in the bulk also increases with Y₂O₃. The microstructures of the crystallized materials are composed of crystals with different sizes and shapes (petal-like, polygonal, globular). These crystals become smaller as the amount of Y₂O₃ is raised despite the percentage of crystallized material increasing with higher Y₂O₃ content. The chemical compositions determined for the different crystals correspond to the different crystal phases determined by XRD.