3.1 Influence of time and temperature
The glass-forming region and properties of nitrogen-enriched glasses in the La-Si-O-N were reported by Hakeem et al. [1, 2]. These glasses were prepared by using La metal as a precursor and containing up to 68 e/o of N and 62 e/o of La, showing a larger glass-forming region than previously reported [27–31]. The effect of time and temperature are, to a certain extent, interrelated. By increasing the temperature, the melting process speeds up, but the melt decomposition with consequent weight loss occurs. The amounts of silicide did not show apparent variation with glass composition or preparation temperature but were found to decrease with increasing synthesis-holding time, as shown in Fig. 1(a,b) of a sample (La11.03Si10O20.42N10.75 having La 45.24 e/o and N 43.38 e/o) held at 1800°C for 2 and 4 hours, respectively. The optical micrograph in Fig. 1b shows a much lower volume fraction precipitates than Fig. 1a.
3.2 Influence of raw materials and nitrogen pressure
All of the glass samples were prepared from a similar source of powders of La, SiO2, and Si3N4 and using identical nitrogen pressure, i.e., 1 atmospheric pressure. So, it is difficult to ascertain the effects of raw materials and nitrogen pressure in the present study. However, Korgul et al. [12] concluded that the transparency of the glasses was undoubtedly improved by using high-purity quartz powder instead of precipitated silica. Furthermore, it was observed that the substitution of Si3N4 by AlN (in Sialon system) and increasing nitrogen pressure show no distinction improvement in the transparency. Sharafat et al. [4] reported glasses in the Sr-Si-O-N system turn to more transparent by using SrD2 than Sr metal as a modifier.
Furthermore, the glasses made by using SrD2 were in accordance found to contain much smaller amounts of Sr silicides. Pasto et al. [32] reported that the origin of the impurities was in the raw materials; additionally, impurities were pick-up during powder processing and melting. Iron is probably the most common impurity found in the batch material for glass (impurity from the precursors) and can readily oxidise silicon nitride at quite low temperatures to form FeSi2. Furthermore, decomposition was not noticeable diminish by the use of high purity Si3N4 as a batch component. As mentioned early that no iron impurities were present in the used raw materials for the preparation of nitrogen-enriched glasses in the La-Si-O-N system, both SEM and TEM investigation endorse no existence of iron silicide in the present glasses.
3.3 Observation of interface between glass and niobium crucible
All the glasses were prepared by standard procedures, as described in the experimental section. It was observed that there is no reaction between the La-Si-O-N melt and niobium crucible used in the study, although the melt did adhere to the surface of the crucible. Similarly, no reaction between the melt and Nb crucible was observed in other AE-Si-O-N [3, 4, 22, 33] systems prepared by a similar synthesis route as described for La-Si-O-N glasses. However, it was observed that the La-Si-O-N glass in which AlN was added, do react with the Nb crucible and some of AlN remained undissolved in the liquid because of its strong bonding and stability of the AlN. Therefore, aluminium had to be added in oxide (and perhaps in metallic) form, i.e., as Al2O3 instead of AlN, as an admixture to the composition. Experimental observation indicates that Al2O3 reacts and dissolves into the liquid much faster than AlN. Figure 2 reveals some undissolved AlN near the crucible wall. It is thus easier to dissolve Al2O3 than AlN in the La-Si-Al-O-N system. Recently, Natalia et al. [34] have reported the dissolution of Nb metal from Nb crucible in the phosphorus based oxynitride glasses. The addition of Al into La-Si-O-N system has shown that the glasses can be prepared at lower temperature and in a shorter time. Furthermore, the addition of Al to the La-Si-O-N system enhances transparency and minimizes losses of components during preparation.
3.4 Optical microscopy observations
Light microscopy was used for the examination of polished samples. Most of the glasses were found to be opaque and generally had a deep brown colour and reflecting white spots, as shown in Fig. 3a. Additionally, further experiments confirm that glasses contain small amounts of spherical La silicide (LaSi2) particles in the matrix. The amount was estimated in the range of around 2–4 vol.% and varied upon the starting composition. There was no elemental silicon observed by the light microscope and SEM analysis. Figure 3b shows the character of the distribution of opaque grains. The glass samples contain a very high frequency of equidimensional rounded opaque grains reaching sizes of a few microns, although most of them are much smaller. The opaque grains are unevenly distributed, with larger, more separated grains in the more transparent parts of the glass, while dusty impregnations of microscopic opaque particles occur in the dark-coloured parts of the glass.
3.5 Scanning electron microscopy
All obtained glasses were examined by SEM to observe the surface morphology of the X-ray amorphous glass samples. SEM investigations on glasses showed homogeneous microstructures, with no evident substructural features, as shown in Fig. 4. However, the glasses do contain small amounts of spherical particles of La silicide, as noticed by the optical microscopic analysis as well. La-Si interaction is complex, it might occur since during synthesis, part of the initial Si3N4 and SiO2 is reduced at an early stage to form a silicide with La metal. At evaluated temperatures, the silicides are molten, which accounts for their spherical particle shape. The reaction mechanism might involve a direct reaction between La metal and Si3N4 to produce LaxSiy + N2 or LaN + Si, with a subsequent reaction of La and Si. The La silicides gradually dissolve into the glass with increasing temperature and longer holding time. According to the Lanthanum-silicon phase diagram [35], LaSi2 melts at 1730°C and La3Si2 melt, particularly at 1470°C. Additionally, few compositions showing much lesser silicides precipitation, which indicates that the formation and the quantities of silicides largely depend on the synthesis route and heat treatment procedure and not much on nitrogen concentration present in the stoichiometry.
Backscattered electron (BSE) images provide evidence of impurities, and EDX gives the compositional homogeneity of the glass matrix. Figure 5 shows SEM micrographs of the different morphologies and crystalline phases in the partially crystalline glass matrix. Backscattered electron (BSE) images provide evidence of impurities, and EDX gives the compositional homogeneity of the glass matrix. Glass sample gives weak X-ray reflections according to the XRPD pattern, but the SEM image Fig. 5a shows the presence of two glass phases by BSE compositional analysis and black coloured undissolved Si3N4. For amorphous or poorly crystallised samples, as in the La-Si-O-N system, the determination of weak X-ray reflections is complicated due to the presence of a high-intensity background (the result of inelastic scattering), leading to an underestimation of the size of crystallites. SEM image Fig. 5b shows two glass phases and silicide particles (white spherical). Micrograph Fig. 5c shows the growth of dendrites and Fig. 5d further growth of the needle-like structure of a crystalline phase. Nitrogen enriches La containing SiON glasses can be described as a mixture of amorphous SiO2 and crystalline LaSi2, thus implying that phase separation occurred according to the microstructure viewpoint. Moreover, it was observed that the degree of crystallisation and the crystal sizes decrease with increasing nitrogen content. Generally, nitrogen rich glasses containing a high amount of silicides have Poisson’s ratio ranging from 25 to 36, resemble those of metallic glasses having Poisson’s ratio of ca. 40.
3.6 Raman spectroscopy
Raman spectra show that the glasses in the La-Si-O-N system contain elemental Si and LaSi impurities. It is thus believed that the absence of any signal in the XRD spectrum of these impurities due to their size and quality effect. However, according to Raman studies, an elemental Si gives a sharp peak at 522 cm− 1. The intensity of the Si peaks decreases with the increasing N content, and no Si peaks are observed for glasses with N contents above ca. 40 e/o. The spectra show a high background of fluorescence, which increases with increasing N and La contents, implying that the recorded spectra have, in general, a poorer quality for high N content glasses. Rouxel et al. [11] have reported that the strong luminescence observed in the oxynitride glasses could be due to silicon clusters of less than 7 nm in diameter. Figure 6 shows the Raman spectra of the La silicide particle.
3.7 Transmission electron microscopy (TEM) and electron energy-loss spectroscopy (EELS) analysis
TEM was applied to investigate the homogeneity of the glasses and to detect the presence of impurities in glass compositions. The HRTEM images in Fig. 7 show homogenous microstructure glasses with no porosity. There is no evidence of substructural features, and selected area electron diffraction (SAED) patterns confirmed the amorphous nature of the sample. Figure 8a, showing the microstructure of silicide particles of La silicide and Si-enriched particles. The majority of the La silicide are spherical and have uniform compositions. However, some of the La silicides have prominent internal structure, having entrapped particles of Si-rich phase (s), which might be Si or Si3N4, and La enriches in the surrounding glass (Fig. 8b and 8c). The size of the particles ranged from, 5 µm down to a typical size of 1 µm and less as also observed by optical microscopy investigation. These observations indicate that the silicides gradually dissolve during the synthesis. The amounts showed no apparent variation with glass composition. However, some of the samples, as shown in Fig. 9, phase separation in the length-scale of 10 to 30 nm.
The Raman spectra, as shown in Fig. 6, show a sharp and intense absorption band characteristic from elemental silicon at 520 cm− 1. The fact that no crystallinity was detected by X-ray diffraction might be due to the deficient amounts of elemental Si present in the glass matrix. Even a small amount of free silicon probably partly hinders the glass transparency, and other possible cause being the clustering or nano-phase separation evidenced by TEM, as shown in Fig. 10.
In Fig. 10a-c, EELS spectra from a sample of La-Si-O-N glass show the presence of silicon on the edges of the precipitate (ppt.). The interface between Si and glass shows no fringes indicating any ordered structure, but at the interface between Si and precipitate, ordered fringes in silicon can be seen. Figure 11 shows the EELS spectra of the amorphous region and no evidence of compositional fluctuation along with the 100 nm line scan. The size of free Si is less than 100 nm in diameter. These particles were not visible under a light microscope nor by SEM. These observations lead to the conclusion that the more nitrogen is incorporated in the glass network and the more fragile the glass becomes. Additionally, it is suggested that nitrogen favours the formation of a heterogeneous network at the nanometer scale and leads to a network structure with weak channels, resembling the ones proposed by Greaves et al. [36] and acting as a lubricant between the clusters.
3.8 Comparison with La-Si-O-N thin films
As mentioned under Sect. 1, nitrogen-rich Mg/Ca-Si-O-N thin films are transparent in the visible range. For comparison purposes, thin films in the La-Si-O-N system were grown on commercially available soda-lime float glass and sapphire substrates by magnetron sputtering. Reactive sputter deposition from silicon (purity 99.99%), lanthanum (purity 99.95%) was performed in an ultra-high vacuum (UHV) deposition system with a base pressure < 1·10− 5 Pa.>. For reactive sputtering, a mixture of nitrogen (N2), oxygen (O2), and argon (Ar) was used and the deposition time was 2 hours. The compositional analysis by EDX and X-ray photoelectron spectroscopy (XPS), which confirmed that obtained films contain a high amount of La and N. As shown in Fig. 12, thin films in the La-Si-O-N system are optically transparent and free from metallic impurities as compared to bulk La-Si-O-N glass. A fair comparison of La containing thin films and bulk glasses is difficult due to differing synthesis techniques and compositions. In summary, it is possible to obtained optical transparent glasses with high nitrogen content by optimising the process parameters and the selection of the precursors.