A Review on Photocatalytic Glass Ceramics: Fundamentals, Preparation, Performance Enhancement and Future Development

Photocatalytic technology is considered as one of the most attractive and promising technologies to directly harvest, convert and store renewable solar energy for generating sustainable and green energy and a broad range of environmental applications. However, the use of a photocatalyst in powder or coating forms restricts its applications due to its disadvantages, such as difficulty in recovery of nano-powder, secondary pollution, low adhesion between photocatalytic coating and substrate material, short service life of photocatalytic film and so on. The investigation and application of photocatalytic glass-ceramics (PGCs) in water purification, bacterial disinfection, self-cleaning and hydrogen evolution have received extensive attention due to their inherent advantages of low cost, easy fabrication, transparency, chemical and mechanical stability. Real-time solutions to energy shortage and environmental pollution faced by the development of human society can be provided by rationally designing the chemical composition and preparation methods of glass ceramics (GCs). This review introduces the concept and crystallization mechanism of PGCs and expounds on the basic mechanism of photocatalysis. Then, the key point difficulties of GCs’ design are discussed, mainly including the methods of obtaining transparency and controlling crystallization technologies. Different modification strategies to achieve better photocatalytic activity are highlighted. Finally, we look forward to further in-depth exploration and research on more efficient PGCs suitable for various applications.


Introduction
Energy shortage and environmental pollution are the two major problems that have faced the development of human society since the 21st century. Therefore, the search for clean and renewable energy and the development of efficient energy storage and conversion technologies are imperative. Photocatalytic technology can directly capture, convert and store renewable solar energy, resulting in sustainable green energy and a wide range of environmental applications. This opens up more tangible possibilities for the development of sustainable energy and technologies [1][2][3].
Since the discovery of TiO 2 as a photocatalyst by Fujishima and Honda [4] in 1972, a lot of exploration into its photocatalytic properties has been carried out [5][6][7][8][9]. In addition to TiO 2 , the photocatalytic activity of bismuth-and zinc-based compounds also has been preliminarily investigated [10][11][12][13]. Most of the research have been conducted by taking raw materials in powder form. Furthermore, TiO 2 coated photoactive self-cleaning Pilkington Activ™ glass has been used for photocatalytic degradation of common pharmaceutical

Photocatalytic Glass-Ceramics (PGCs)
Since the first proposal of GCs by Stookey [26] in the 1950s, numerous researchers have investigated the crystallization of glasses, and many functional GCs have been developed commercially so far [27,28]. GCs are composite multiphase materials with crystallites dispersed in super-cooled liquid and glassy matrices [29]. The parent glasses can be fabricated by melt-quenching techniques above the liquidus temperature. Then, the crystals are nucleated and grow to form GCs, typically with heat treatment. The crystallites formed in the glass not only improve the original properties of the parent glass, but also may provide completely new properties to the parent glass [30]. The composition,

Photocatalytic Glass-Ceramics (PGCs)
Since the first proposal of GCs by Stookey [26] in the 1950s, numerous researchers have investigated the crystallization of glasses, and many functional GCs have been developed commercially so far [27,28]. GCs are composite multiphase materials with crystallites dispersed in super-cooled liquid and glassy matrices [29]. The parent glasses can be fabricated by melt-quenching techniques above the liquidus temperature. Then, the crystals are nucleated and grow to form GCs, typically with heat treatment. The crystallites formed in the glass not only improve the original properties of the parent glass, but also may provide completely new properties to the parent glass [30]. The composition, microstructure and phase assemblage determine the properties of GCs [31]. Different microstructures can be obtained by adjusting the composition and heat treatment process. Furthermore, another significant advantage of GCs is their transparency. To obtain a TGC, it is necessary to minimize the light attenuation caused by scattering and absorption. In fact, in order to ensure the transparency of the GCs, the nanocrystals are made sufficiently small (<30 nm) by controlling the crystallization process. On the other hand, when large crystals, up to micrometer scale, are present in TGCs, the refractive index difference between the crystalline and the glassy phase must be very small (<0.01) to avoid loss of transparency. The properties of GCs combine the important features of glass and polycrystalline ceramics and have many potential applications, such as optical, dental, machining, thermal, construction, bioactive and electrical applications, etc.

Mechanism of Photocatalysis
In 1967, photocatalysis was discovered by Fujishima Akira and Kenichi Honda, also known as the "Honda-Fujishima Effect", which is a pioneering study of photocatalysis. In 1972, Fujishima and Honda [4] first reported a photochemical system consisting of a TiO 2 photoelectrode and platinum electrode, capable of splitting water into hydrogen and oxygen, thus opening up the new field of semiconductor photocatalysis. The photocatalysis process has potential applications in the evolution of hydrogen, reduction of carbon dioxide (CO 2 ), wastewater treatment, air-cleaning, surface self-cleaning, detoxification and bacterial disinfection [32,33].
Thus far, the fundamental mechanism of photocatalysis has been well proposed, as shown in Figure 2. Basically speaking, the heterogenous photocatalysis process is initiated with the absorption of light (or radiation) having energy equal to or greater than the band gap energy (E g ) of targeted semiconductor material [34]. E g is defined as the difference between the filled valence band (VB) and the empty conduction band (CB). When photons with energy equal to or higher than E g reach the surface of the photocatalyst, they will cause molecular excitation. Some electrons are excited from VB to CB of the semiconductor and create holes in the valence band. After the initiation of these photogenerated electron (e − )-hole (h + ) pairs, these can either recombine and dissipate the absorbed energy in the form of heat or can migrate to the surface for catalytic redox reactions [35]. During catalytic redox reactions, these holes (h + ), being good oxidizing agents, oxidize adsorbed water (H 2 O) to form hydroxyl radicals (OH − ). This OH − will then lead to the production of strong oxidizing HO • radicals. Meanwhile, the negative electrons being strong reducing agents reduce adsorbed oxygen (O 2 ) to form a superoxide anion (O 2 •− ). This superoxide anion also produces HO • radicals via the formation of HO 2 • radicals and H 2 O 2 [35]. These reactive oxidizing species (ROS) react with adsorbed pollutants such as organic compounds and decompose them, thereby degrading them into CO 2 , water and, sometimes, other less harmful species [36].

Classification of PGCs
Photocatalytic GCs can provide corresponding photocatalytic activity due to the crystals precipitated during the heat treatment. There are two distinct crystallization mechanisms involved in the preparation of GCs, i.e., surface and volume (bulk) crystallization. Research has shown that the devitrification of glass with Trg < 0.58 (Trg = Tg/Tm, where Tm is melting temperature) is dominated by volume crystallization, and surface crystallization may also occur [39,40]. Homogeneous nucleation is more favorable in this case because the maximum nucleation temperature is lower than Tg. If Trg > 0.6, the nucleation rate near Tg is low and the heterogeneous nucleation becomes dominant, which is often observed in the surface crystallized GCs. It has been found that GCs with a small density difference (<10%) between the crystalline and glassy phases typically undergo homogeneous nucleation and, thus, appear to be dominated by volume nucleation [41]. However, if the density difference is large, the high elastic deformation energy (ΔGE) at the interface may inhibit bulk crystallization, and GC is more inclined to surface crystallization [42].
Either surface crystallization, bulk crystallization or a combination of the two can be used to design GCs with desired properties. Different from other photocatalytic materials, photocatalytic active GCs have the advantages of being easy to produce and process into different shapes (sheets, plates, fibers and rods) and sizes and having excellent chemical stability, leading to broader applications in photocatalysis. As shown in Figure 3, photo- The electron-hole recombination step is undesirable as it will result in low photocatalytic efficiency and waste the energy supplied by the photon. Therefore, reducing the charge recombination rate in the body and on the surface of the semiconductor is one of the most effective methods to improve the photocatalytic activity. For the surface charge utilization, the large onset overpotential and sluggish kinetics are two key factors limiting the surface photocatalytic efficiency of reduction and oxidation reactions. Usually, shortening the diffusion length of photo-generated charge carriers or constructing interfacial electric fields could efficiently reduce the recombination rates, thus substantially enhancing the photocatalytic activity [37]. Moreover, to design highly efficient photocatalysts suitable for various photocatalytic applications, light harvesting efficiency (η abs ), charge separation efficiency (η cs ), charge migration and transport efficiency (η cmt ) and charge utilization efficiency (η cu ) should be comprehensively considered and optimized [38]. Despite the significant advances in heterogeneous photocatalysis, there are still many challenges related to the further enhancements of light harvesting (especially for the visible light region) and reduction in the recombination rates of photo-induced electrons and holes, separation and utilization.

Classification of PGCs
Photocatalytic GCs can provide corresponding photocatalytic activity due to the crystals precipitated during the heat treatment. There are two distinct crystallization mechanisms involved in the preparation of GCs, i.e., surface and volume (bulk) crystallization. Research has shown that the devitrification of glass with Trg < 0.58 (Trg = Tg/Tm, where Tm is melting temperature) is dominated by volume crystallization, and surface crystallization may also occur [39,40]. Homogeneous nucleation is more favorable in this case because the maximum nucleation temperature is lower than Tg. If Trg > 0.6, the nucleation rate near Tg is low and the heterogeneous nucleation becomes dominant, which is often observed in the surface crystallized GCs. It has been found that GCs with a small density difference (<10%) between the crystalline and glassy phases typically undergo homogeneous nucleation and, thus, appear to be dominated by volume nucleation [41]. However, if the density difference is large, the high elastic deformation energy (∆G E ) at the interface may inhibit bulk crystallization, and GC is more inclined to surface crystallization [42].
Either surface crystallization, bulk crystallization or a combination of the two can be used to design GCs with desired properties. Different from other photocatalytic materials, photocatalytic active GCs have the advantages of being easy to produce and process into different shapes (sheets, plates, fibers and rods) and sizes and having excellent chemical stability, leading to broader applications in photocatalysis. As shown in Figure 3, photocatalytic active GCs have various applications in the environmental fields, such as degradation of water pollutants, air purification, surface self-cleaning, antibacterial, photocatalytic hydrogen evolution and photocatalytic reduction in CO 2 , etc.

Volume Crystallization
In volume crystallized PGCs, the photocatalytic active crystallites are dispersed on the surface as well as in the bulk of GCs. Indeed, volume crystallization is often required to obtain a wide range of novel TGC materials with good optical homogeneity. In this case, nuclei in glass continue to grow linearly with volume until they are hindered by adjacent crystals. This is followed by secondary growth, which causes smaller crystals to fuse in favor of larger crystals (see Figure 4a).  In volume crystallized PGCs, the photocatalytic active crystallites are dispersed on the surface as well as in the bulk of GCs. Indeed, volume crystallization is often required to obtain a wide range of novel TGC materials with good optical homogeneity. In this case, nuclei in glass continue to grow linearly with volume until they are hindered by adjacent crystals. This is followed by secondary growth, which causes smaller crystals to fuse in favor of larger crystals (see Figure 4a). containing a high content of a ZnO component (up to 40%) for the glass composition SiO2, ZnO, B2O3, Na2O, K2O, P2O5, Li2O and BaO. This GC showed proper photocatalytic activity for the degradation of Red SPD Cotton dye as a model compound. Maximum removal (76%) was achieved in the case of the sample containing 30 wt% of ZnO and heat treated at 450 °C for 10 h. Singh et al. [48] also fabricated transparent ZnO crystallized GCs from the 9.35K2O-9.35CaO-42.05ZnO-23.36B2O3-9.35SiO2-6.54Al2O3 (in mol%) glass system. The results showed that the degradation ability of the photocatalytic Rhodamine B (RB) dye is enhanced with increasing crystallization time from 30 to 90 min at 700 °C due to the reduced band gap (3.75 to 3.14 eV) of GC samples heat-treated for 30 to 90 min. As shown in Figure 6, they also evaluated the antibacterial properties of the samples by Escherichia coli (Gram-negative bacteria). It is evident that ZnO crystallized glass powders (ZnO-1, ZnO-2 and ZnO-3) have significant anti-E. coli activity, and the antibacterial properties increase with the increase in the crystallization in the glass matrix. Interestingly, it was found that the ZnO-3 glass could completely eradicate bacteria under UV irradiation for 60 min, which indicated its promising antibacterial properties.  It is well known that TiO 2 has excellent photocatalytic activity. Therefore, an attempt can be made to form TiO 2 crystals in glass to obtain PGCs. Yazawa et al. [43] have successfully prepared rutile-type TiO 2 GCs from the SiO 2 -Al 2 O 3 -B 2 O 3 -CaO-TiO 2 glass system. The photocatalytic activity of this GC was about four times larger than that of a TiO 2 coated photocatalyst fabricated by the sol-gel process. In this direction, Singh et al. [44] have fabricated the stable rutile phase of TiO 2 nanocrystals in the 5CaO-65B 2 O 3 -10BaO-Al 2 O 3 -20TiO 2 -10ZnO (mol%) glass by a heat treatment process and investigated the effect of different heat treatments on the transparency and photocatalytic activity of GCs. The parent glass sample was referred to as the CBBATZ sample. The parent glass was heat treated with three different crystallization times (3, 5 and 10 h at 630 • C) and the resulting GCs were designated CBBATZ-1, CBBATZ-2 and CBBATZ-3, respectively. All heat-treated samples were transparent in the wavelength range of 800-405 nm. Hence, we can easily see that, even with the formation of TiO 2 crystals in glass samples, the samples were still transparent in the visible region. It means that the most important property of glass (transparency) is still preserved in glass nanocomposite. Figure 5 shows the photocatalytic activity of the photocatalytic ink test and MB dye degradation test. With the increase in heat treatment time, the photocatalytic activity increased. Compared to CBBATZ, CBBATZ-1 and CBBATZ-2 samples, GC CBBATZ-3 showed the best degradation performance of Resazurin ink and MB dye. From CBBATZ-1 to CBBATZ-3, the bandgap of the samples decreased and the TiO 2 crystallinity increased, so the photocatalytic activity of CBBATZ-3 was more impressive. GCs containing anatase-type TiO 2 crystals or (TiO) 2 P 2 O 7 crystals were obtained by Jie Fu [45] through thermal treatment of binary TiO 2 -P 2 O 5 glasses with 69 mol% and 76 mol% TiO 2 . Consistent with the above, the activity of photocatalytic degradation of methylene blue (MB) was enhanced with the increase in anatase TiO 2 crystalline phases. In the other report of Jie Fu [46], a GC containing Nasicon-type crystal MgTi 4 (PO 4 ) 6 was prepared by heat-treating MgO-TiO 2 -P 2 O 5 -SiO 2 glasses, which exhibited a strong ability to decompose methylene blue (MB) and acetaldehyde and super-hydrophilicity under UV irradiation.  Like TiO 2 , many other photocatalytic crystallites can also be used to form photocatalytic activity GCs. Margha et al. [47] have prepared a transparent photoactive GC material containing a high content of a ZnO component (up to 40%) for the glass composition SiO 2 , ZnO, B 2 O 3 , Na 2 O, K 2 O, P 2 O 5 , Li 2 O and BaO. This GC showed proper photocatalytic activity for the degradation of Red SPD Cotton dye as a model compound. Maximum removal (76%) was achieved in the case of the sample containing 30 wt% of ZnO and heat treated at 450 • C for 10 h. Singh et al. [48] also fabricated transparent ZnO crystallized GCs from the 9.35K 2 O-9.35CaO-42.05ZnO-23.36B 2 O 3 -9.35SiO 2 -6.54Al 2 O 3 (in mol%) glass system. The results showed that the degradation ability of the photocatalytic Rhodamine B (RB) dye is enhanced with increasing crystallization time from 30 to 90 min at 700 • C due to the reduced band gap (3.75 to 3.14 eV) of GC samples heat-treated for 30 to 90 min. As shown in Figure 6, they also evaluated the antibacterial properties of the samples by Escherichia coli (Gram-negative bacteria). It is evident that ZnO crystallized glass powders (ZnO-1, ZnO-2 and ZnO-3) have significant anti-E. coli activity, and the antibacterial properties increase with the increase in the crystallization in the glass matrix. Interestingly, it was found that the ZnO-3 glass could completely eradicate bacteria under UV irradiation for 60 min, which indicated its promising antibacterial properties.

Surface Crystallization
When bulk crystallization is activated, surface crystallization will dominate. Surface crystallization generally relies on surface nucleation initiated through tribochemical activation or other methods. Then, surface crystallization proceeds vertically to the interior of the glass, and the crystal usually exhibits a strong preferential growth orientation (see Figure 4b). Surface crystallization is more meaningful for GCs to obtain efficient photocatalytic activity. However, in general, the preferred growth direction of crystal diffusion is into the glass points to the glass interior. Thus far, although many studies have prepared surface-crystallized GCs, their photocatalytic activity has not been explored. Almeida et al. [49] prepared niobate glasses and ceramics ((xNb2O5 (0.5−x)P2O5) 0.5Li2O) by the meltquenching method and detected LiNbO3 crystals in samples with high niobium concentration. Differently, Murugan et al. [50] fabricated transparent glasses in the system (100−x)Li2B4O7−xSrBi2Ta2O9 (0 < x < 20) via a splat-quenching technique. The XRD of the glass-ceramic composites for 10 < x ≤ 20 depicts only the presence of monophasic crystalline SrBi2Ta2O9 (SBT). On the other hand, the GC composites for x ≤ 10 show the presence of Li2B4O7 (LBO) crystalline peaks in addition to those of SBT. It implies that higher (>10 mol%) concentrations of SBT prevent the crystallization of LBO glass. Further, it is confirmed that the LBO crystallization is confined only to the surface of the sample, while that of SBT was found to be a bulk crystallization process. Plate-like non-porous glass can achieve photocatalysis by crystallizing only at the surface. Laser induced crystallization is one of the most common techniques for growing the desired photocatalytic active crystals on the surface of GCs [29,51]. In addition, there are unconventional methods to engage

Surface Crystallization
When bulk crystallization is activated, surface crystallization will dominate. Surface crystallization generally relies on surface nucleation initiated through tribochemical activation or other methods. Then, surface crystallization proceeds vertically to the interior of the glass, and the crystal usually exhibits a strong preferential growth orientation (see Figure 4b). Surface crystallization is more meaningful for GCs to obtain efficient photocatalytic activity. However, in general, the preferred growth direction of crystal diffusion is into the glass points to the glass interior. Thus far, although many studies have prepared surface-crystallized GCs, their photocatalytic activity has not been explored. Almeida et al. [49] prepared niobate glasses and ceramics ((xNb 2 O 5 (0.5−x)P 2 O 5 ) 0.5Li 2 O) by the melt-quenching method and detected LiNbO 3 crystals in samples with high niobium concentration. Differently, Murugan et al. [50] fabricated transparent glasses in the system (100−x)Li 2 B 4 O 7 −xSrBi 2 Ta 2 O 9 (0 < x < 20) via a splat-quenching technique. The XRD of the glass-ceramic composites for 10 < x ≤ 20 depicts only the presence of monophasic crystalline SrBi 2 Ta 2 O 9 (SBT). On the other hand, the GC composites for x ≤ 10 show the presence of Li 2 B 4 O 7 (LBO) crystalline peaks in addition to those of SBT. It implies that higher (>10 mol%) concentrations of SBT prevent the crystallization of LBO glass. Further, it is confirmed that the LBO crystallization is confined only to the surface of the sample, while that of SBT was found to be a bulk crystallization process. Plate-like non-porous glass can achieve photocatalysis by crystallizing only at the surface. Laser induced crystallization is one of the most common techniques for growing the desired photocatalytic active crystals on the surface of GCs [29,51]. In addition, there are unconventional methods to engage bulk photocatalytic active crystals, such as using porous/etched/leached GCs and using a powdered form of GC.
Bulk crystallization is more beneficial in obtaining GCs with high transparency, while surface crystallization is of great significance to the photocatalytic effect of PGCs. In addition, if the transparency is high enough, the photocatalytic crystals on the inner surface of GCs can be maximally activated during the photocatalytic process, resulting in efficient photocatalytic performance. A detailed introduction to the transparency and controlled crystallization of GCs is presented in Section 3.

Design and Fabrication of Photocatalytic Glass Ceramics (PGCs)
Compared with other photocatalytic materials, PGCs have various advantages. However, due to the inherent drawbacks of GC materials, rational design must be carefully considered to obtain the best photocatalytic performance. Figure 7 summarizes the design considerations for the detailed composition, structure and performance of the GC photocatalysts.
Catalysts 2022, 12, x FOR PEER REVIEW 10 of 40 bulk photocatalytic active crystals, such as using porous/etched/leached GCs and using a powdered form of GC. Bulk crystallization is more beneficial in obtaining GCs with high transparency, while surface crystallization is of great significance to the photocatalytic effect of PGCs. In addition, if the transparency is high enough, the photocatalytic crystals on the inner surface of GCs can be maximally activated during the photocatalytic process, resulting in efficient photocatalytic performance. A detailed introduction to the transparency and controlled crystallization of GCs is presented in Section 3.

Design and Fabrication of Photocatalytic Glass Ceramics (PGCs)
Compared with other photocatalytic materials, PGCs have various advantages. However, due to the inherent drawbacks of GC materials, rational design must be carefully considered to obtain the best photocatalytic performance. Figure 7 summarizes the design considerations for the detailed composition, structure and performance of the GC photocatalysts.

Transparency
Transparency is one of the most important properties of GCs. TGCs are most wanted in the photocatalytic oxidation process to reduce the turbidity in the photoreactor and to activate more photocatalyst particles. However, the formation of crystals within the glass

Transparency
Transparency is one of the most important properties of GCs. TGCs are most wanted in the photocatalytic oxidation process to reduce the turbidity in the photoreactor and to activate more photocatalyst particles. However, the formation of crystals within the glass matrix reduces transparency and becomes translucent, sometimes even opaque. The reduction in transparency is mainly due to high optical scattering, and more ionic/atomic absorption occur [28], as described by: where I T and I 0 are the intensity of the transmitted and the incident light, respectively, α and σ are the absorption coefficient and the turbidity of the sample, respectively, and l is the sample thickness. Therefore, in order to maintain the high transparency of the glass, the scattering loss (i.e., turbidity) must be minimized. For GCs containing crystals with sizes much smaller than the wavelength of light, the crystals are widely separated (for diluted systems). The Rayleigh-Gans scattering model gives the turbidity as [52]: where N is the number density of the crystallites, R and V are the radius and volume of the crystal phase, respectively, n is the refraction index of the crystal and ∆n is the refractive index difference between the glassy and the crystal phase, and k = 2π/ λ, where λ is the wavelength of light. From the equation above, it is clear that the turbidity is inversely proportional to λ4, which is known as Rayleigh scattering. In addition, the turbidity is extremely sensitive to the size of the particles (∝R3) as well as the difference in the refractive index of the (∝∆n2). Practically, highly transparent TGCs (in the visible range) without a noticeable scattering effect can be obtained for crystal radius R < 15 nm and refractive index difference ∆n < 0.1. This condition does not have to be stringently obeyed for a sufficiently small ∆n (~0.01).
It is known that the Rayleigh-Gans model overestimates the light attenuation due to scattering for concentrated particle dispersions. A quasi-continuum model has been developed by Andreev [53] and Hopper [54] for the description of interfering individual particles that produce a large compensation effect. In this model, in addition to requiring the particle size to be much smaller than the wavelength of light, the distance between particles should not be less than the particle radius, but, rather, at least six times the particle radius. The turbidity under such condition is given by: where θ = a + W/2 is the mean phase width (W: the inter-particle distance). According to this model, high transparency can be still achieved for particle size < 30 nm and refractive index difference (∆n) up to 0.3. In summary, it is clear that two possibilities exist for producing a transparent crystalglass system. In one case, nucleation and crystal growth rates should be controlled to precipitate nanocrystals (NCs) and avoid overgrowth and coarsening. The dispersed crystalline phases are sufficiently small in size that they produce no effective scattering, even at the shorter wavelengths of the visible spectrum. In the other case, the refractive index difference between all crystalline and amorphous phases must be minimized, and the birefringence within any single crystalline phase must be small.
The above statement is limited to the case where the crystallite size is much smaller than the wavelength of incident light. For scattering of particles comparable in size to the wavelength of light, only ∆n~0 in the GC can maintain high transparency.
Actually, the magnitude of transparency of GCs affected by crystallite size and refractive index difference is mainly determined by chemical composition and the heat treatment process. Prasad et al. [55] reported the transparent glasses in the composition (100 − x)Li 2 B 4 O 7 -x(SrO-Bi 2 O 3 -Nb 2 O 5 ) (where x = 10, 20, 30, 40, 50 and 60, in molar ratio) and obtained the compositional dependence of the refractive indices at 570 nm of the GCs. The result indicated that the refractive index increased notably with the increase in the SrBi 2 Nb 2 O 9 composition x. In addition, Molla et al. [56] fabricated Bi 4 Ti 3 O 12 (BiT) crystals in the K 2 O-SiO 2 -Bi 2 O 3 -TiO 2 (KSBT) system by conducting controlled heat treatments at different temperatures and durations. From the spectra in Figure 8a, it is seen that with increasing BiT content in the glass at 1150 • C, transmission of the glass in the wavelength range of 300-1100 nm decreases. In addition, the effect of melting temperature on the transmission properties of BiT45 glass was investigated, and the transmission properties in the visible wavelength range decreased significantly with increasing melting temperature ( Figure 8b). in the K2O-SiO2-Bi2O3-TiO2 (KSBT) system by conducting controlled heat treatments at different temperatures and durations. From the spectra in Figure 8a, it is seen that with increasing BiT content in the glass at 1150 °C , transmission of the glass in the wavelength range of 300-1100 nm decreases. In addition, the effect of melting temperature on the transmission properties of BiT45 glass was investigated, and the transmission properties in the visible wavelength range decreased significantly with increasing melting temperature ( Figure 8b). In addition to chemical composition, the crystal size after heat treatment also has a distinct influence on transparency. Lin et al. [57] presented a simple and universal method for fabricating various kinds of metal or semiconductor nanoparticle (MSNP) glass composites by using metallic Al as a reducing agent in the raw materials of the glass batches. Bi crystals were precipitated in 10Na2O-48B2O3-2Al-40SiO2-4Bi2O3 glass. The heat treatment temperature and holding time can control the particle size, and the heat treatment can also change the color of the glasses. In general, the larger grain size will bring about the increase in turbidity, and the Mie scattering effect will become prominent for the large refractive index difference between the fabricated nanocrystals and glass matrix [58]. Indeed, the crystal's size must remain smaller than the incident wavelength to avoid too much light scattering in order to maintain the transparency of GC. Masai et al. [59] prepared a single TiO2 phase GC in the 5CaO-65B2O3-10Bi2O3-Al2O3-20TiO2 glass composition. Although the refractive index difference between the rutile TiO2 phase and the glass matrix is relatively large (n > 0.8), the obtained GC still has high transparency. This is mainly due to the size of the crystals being much smaller than the wavelength of the incident light.
In the case of polycrystalline precipitation, control of transparency becomes more complicated. The GCs in the Bi2O3-K2O-TiO2-P2O5 system were elaborated by Haily et al. [60]. Different crystalline phases are obtained in the matrix glass as the heat treatment proceeds. The color of the glass samples containing Bi2O3 is brown as well as the glass free of Bi2O3 presents blue. The XRD pattern of xBi2O3-(20−x)K2O-30TiO2-50P2O5 (x=5, 10, 15) GCs implied the appearance of KTi2(PO4)3 and BiPO4 phase. With a further increase of Bi2O3 content, there is an increase in the content of the BiPO4 phase. The complexity of crystal types determines the opacity of GCs. In addition to chemical composition, the crystal size after heat treatment also has a distinct influence on transparency. Lin et al. [57] presented a simple and universal method for fabricating various kinds of metal or semiconductor nanoparticle (MSNP) glass composites by using metallic Al as a reducing agent in the raw materials of the glass batches. Bi crystals were precipitated in 10Na 2 O-48B 2 O 3 -2Al-40SiO 2 -4Bi 2 O 3 glass. The heat treatment temperature and holding time can control the particle size, and the heat treatment can also change the color of the glasses. In general, the larger grain size will bring about the increase in turbidity, and the Mie scattering effect will become prominent for the large refractive index difference between the fabricated nanocrystals and glass matrix [58]. Indeed, the crystal's size must remain smaller than the incident wavelength to avoid too much light scattering in order to maintain the transparency of GC. Masai et al. [59] prepared a single TiO 2 phase GC in the 5CaO-65B 2 O 3 -10Bi 2 O 3 -Al 2 O 3 -20TiO 2 glass composition. Although the refractive index difference between the rutile TiO 2 phase and the glass matrix is relatively large (n > 0.8), the obtained GC still has high transparency. This is mainly due to the size of the crystals being much smaller than the wavelength of the incident light.
In the case of polycrystalline precipitation, control of transparency becomes more complicated. The GCs in the Bi 2 O 3 -K 2 O-TiO 2 -P 2 O 5 system were elaborated by Haily et al. [60]. Different crystalline phases are obtained in the matrix glass as the heat treatment proceeds. The color of the glass samples containing Bi 2 O 3 is brown as well as the glass free of For GCs with almost the same composition of the glassy and crystalline phases, the refractive indices can be easily matched. This type of GC can minimize scattering loss and provide high transparency. During the crystallization of such GCs, local structural changes are minimal and ion diffusion paths are the shortest. Examples of such TGCs can be found in chapter 2 in Ref. [61]. Sharma et al. [62] exhibited the formation of Bi 4 B 2 O 9 crystals embedded in 2Bi 2 O 3 -B 2 O 3 (BBO), the host glass matrix. In BBO GCs, the crystals are uniformly dispersed in the glass matrix in the nano range. The crystal and the matrix glass composition are almost identical so that the transparency of the GC is appreciable. Unfortunately, as the heat treatment time increases, the GC becomes darker. Likewise, Singh et al. [63] precipitated a TGC with SrBi 2 B 2 O 7 crystals from SrO- From the UV-Vis absorption spectra, transmission spectra of the GC showed almost 60-80% transparency in the 500-800 nm spectral range. Despite the presence of SrBi 2 B 2 O 7 crystals on the surface, GC was still quite transparent, probably due to the close matching of the refractive indices of the crystal and glass phases. However, the transparency of the SBBO GC is gradually lost, accompanied by asymmetric, tightly packed variable grains on the surface, increasing in size from 10 µm to 50 µm.
It should be noted that the same chemical composition of both the glass matrix and crystal does not guarantee the transparency in the GC, and the matching degree of their refractive indices and interface microstructure between the glassy matrix and precipitated crystallites must also be considered. Bertrand et al. [64] achieved a new GC in the Bi 2 O 3 -Nb 2 O 5 -TeO 2 (Bi 0.5 Nb0.5Te 3 O 8 ) system. Although the glassy matrix and the crystals have an identical chemical composition, GCs tend to become opaque with crystal nucleation, and growth occurs gradually. As the crystallization process progresses, the optical transmittance of GCs decreases due to the mismatch of refractive indices as well as some micro cracks that appear at the glassy matrix/crystalline droplet interface at some point.

Controlled Crystallization
The fabrication of GCs can be viewed as the synthesis of crystals through solid-state confined reactions in supercooled glass. The crystals formed in the glassy matrix are the product of the reaction between the glass components, yet in the solid state. This solid-state reaction can be triggered by increasing the temperature (i.e., thermal treatment), which is the most popular technique for GC fabrication. Glass can be made by fusion quenching techniques above liquid temperatures. A schematic diagram of glass' transition to GC is shown in Figure 9. In most instances, the cooling rates should be greater than the critical rate to obtain glass (q > qc), where q c = T L −T N δ t , with TN being nose temperature and t being time. TN represents the temperature at which the crystal fraction of 10-6 is achieved in the shortest time (δ t ). GCs are usually obtained from parent glass by heat-treating for crystal nucleation and growth. Some GCs are formed directly during cooling from melt using cooling rate q ≤ qc. 2Bi 2 O 3 -B 2 O 3 (BBO), TGC is obtained by melt-quenching and heat treatment [62]. The XRD of an as-quenched sample illustrates the presence of nanocrystals, which indicates that a small amount of crystals is precipitated during the cooling of the melt. Moreover, with heat treatment of the as-quenched glass at 380 • C for 1, 2 and 3 h to control the crystallization process, the samples all precipitated Bi 4 B 2 O 9 crystals. Catalysts 2022, 12, x FOR PEER REVIEW 14 of 40

Isothermal Treatment
It is not easy to make TGCs of specific crystals because the crystallization process may involve the simultaneous formation of different crystalline phases, especially for the parent glass with multiple components. Nucleation and crystal growth, which reach their maximum rates at different temperatures, are two separate processes. In general, the temperature of the maximum growth rate is 50-100 °C above the glass transition temperature (Tg). When the critical dimension r* is reached, the crystal will grow at the nucleation temperature. Nucleation stops when the temperature is raised beyond a certain range, and only a nucleus larger than r* formed in the nucleation temperature range will further grow into larger crystals. Nucleation and growth processes can also occur simultaneously because their temperature ranges overlap greatly. Typically, nucleation and the growth of crystals occur simultaneously when GCs are synthesized using a one-step method (glassceramic B in Figure 9). Differently, the two-step method means that nucleation is followed by the growth of crystals (glass-ceramic C in Figure 9). Margha [66] prepared and stabilized the bismuth molybdate β-Bi2Mo2O9 and monoclinic-scheelite BiVO4 phase in the Bi2O3-B2O3-(MoO3 or V2O5) system. The applied heat treatment program showed a significant effect on the size of the formed crystals, especially in samples with molybdenum content. Compared with two-step heat treatment (355 °C/1 h + 400 °C/2 h), one-step heat treatment (500 °C/2 h) has higher crystallinity. It is also noticeable that one-step heat treatment allows for more growth of the crystals than the two-step heat treatment. However, this will lead to overgrowth and coarsening of the crystals, which will cause the glass to lose its transparency. Two-stage isothermal treatment at two different temperatures is usually employed to separate nucleation and growth for better control over the growth of TGC crystals. In the presence of a sufficient number of nuclei, crystal growth is limited to obtain size-limited crystals, preferably less than 20 nm, to finally obtain TGC. Under this circumstance, crystal growth with only a few nuclei can form larger grains, resulting in a

Isothermal Treatment
It is not easy to make TGCs of specific crystals because the crystallization process may involve the simultaneous formation of different crystalline phases, especially for the parent glass with multiple components. Nucleation and crystal growth, which reach their maximum rates at different temperatures, are two separate processes. In general, the temperature of the maximum growth rate is 50-100 • C above the glass transition temperature (Tg). When the critical dimension r* is reached, the crystal will grow at the nucleation temperature. Nucleation stops when the temperature is raised beyond a certain range, and only a nucleus larger than r* formed in the nucleation temperature range will further grow into larger crystals. Nucleation and growth processes can also occur simultaneously because their temperature ranges overlap greatly. Typically, nucleation and the growth of crystals occur simultaneously when GCs are synthesized using a onestep method (glass-ceramic B in Figure 9). Differently, the two-step method means that nucleation is followed by the growth of crystals (glass-ceramic C in Figure 9). Margha [66] prepared and stabilized the bismuth molybdate β-Bi 2 Mo 2 O 9 and monoclinic-scheelite BiVO 4 phase in the Bi 2 O 3 -B 2 O 3 -(MoO 3 or V 2 O 5 ) system. The applied heat treatment program showed a significant effect on the size of the formed crystals, especially in samples with molybdenum content. Compared with two-step heat treatment (355 • C/1 h + 400 • C/2 h), one-step heat treatment (500 • C/2 h) has higher crystallinity. It is also noticeable that one-step heat treatment allows for more growth of the crystals than the two-step heat treatment. However, this will lead to overgrowth and coarsening of the crystals, which will cause the glass to lose its transparency. Two-stage isothermal treatment at two different temperatures is usually employed to separate nucleation and growth for better control over the growth of TGC crystals. In the presence of a sufficient number of nuclei, crystal growth is limited to obtain size-limited crystals, preferably less than 20 nm, to finally obtain TGC. Under this circumstance, crystal growth with only a few nuclei can form larger grains, resulting in a partial loss of transparency. Certain glasses cannot induce nucleation even at temperatures of very low viscosity. Then, we can use heterogeneous nucleation to initiate crystallization of the glass at lower temperatures by adding nucleating agents such as TiO 2 , ZrO 2 and noble metal nanoparticles.

Elemental Particle Precipitation
Precipitation of elemental particles in a glass matrix can be accomplished by thermal decomposition of unstable precursors or reduction with reducing agents to form elemental particles. At the melting temperature, the precursor salt (e.g., AgCl) is completely solvated, and then the precursor is reduced to form an atomically noble metal species. If atomic or molecular state elements are present in the glass melt, they may be in the quenched structure. During the subsequent heat treatment, these metal atoms aggregate, grow and form larger nanocrystals through atomic diffusion. That is to say, the desired metal species can be induced into the glass melt by addition of these metal species through raw materials or redox reactions. Lin et al. [57] used a metallic reducing agent (metallic Al) to reduce other metal or semiconductor oxides (for example, GeO 2 ) to their atomic state (Ge) by melt-quenching processes (process 1 in Figure 10). During subsequent heat treatment (process 2 in Figure 10), this semiconducting element (Ge) will aggregate in the atomic state, grow and form nanocrystals by atomic diffusion inside the glass. This redox reaction can be carried out in a glass matrix: at the glass transition temperature, metal Al is completely dissolved in the glass grid, forming a homogeneous "solution" at high temperature, which can be quenched to form a colorless transparent glass. Utilizing the redox equilibrium established between the reducing agent and various oxides forms crystal nuclei in the melt quenching stage. During the subsequent heat treatment stage, the redox reaction leads to the reduction in the glass network and diffusion to grow nanoparticles on the nuclei (see Figure 10). The particle size can be well controlled by the heat treatment temperature and holding time-a higher reaction temperature favors the formation of crystalline particles with larger average particle size.

Elemental Particle Precipitation
Precipitation of elemental particles in a glass matrix can be accomplished by thermal decomposition of unstable precursors or reduction with reducing agents to form elemental particles. At the melting temperature, the precursor salt (e.g., AgCl) is completely solvated, and then the precursor is reduced to form an atomically noble metal species. If atomic or molecular state elements are present in the glass melt, they may be in the quenched structure. During the subsequent heat treatment, these metal atoms aggregate, grow and form larger nanocrystals through atomic diffusion. That is to say, the desired metal species can be induced into the glass melt by addition of these metal species through raw materials or redox reactions. Lin et al. [57] used a metallic reducing agent (metallic Al) to reduce other metal or semiconductor oxides (for example, GeO2) to their atomic state (Ge) by melt-quenching processes (process 1 in Figure 10). During subsequent heat treatment (process 2 in Figure 10), this semiconducting element (Ge) will aggregate in the atomic state, grow and form nanocrystals by atomic diffusion inside the glass. This redox reaction can be carried out in a glass matrix: at the glass transition temperature, metal Al is completely dissolved in the glass grid, forming a homogeneous "solution" at high temperature, which can be quenched to form a colorless transparent glass. Utilizing the redox equilibrium established between the reducing agent and various oxides forms crystal nuclei in the melt quenching stage. During the subsequent heat treatment stage, the redox reaction leads to the reduction in the glass network and diffusion to grow nanoparticles on the nuclei (see Figure 10). The particle size can be well controlled by the heat treatment temperature and holding time-a higher reaction temperature favors the formation of crystalline particles with larger average particle size.

Powder Sintering, Co-Melting and Frozen Sorbent Processes
These three methods have the following common characteristics: stable glasses cannot be formed directly from starting materials, and the raw materials are required to be processed through these processes to form TGCs. These methods allow us to obtain a GC with the desired crystalline phases easily.
Powder sintering techniques are more efficient and versatile than solid-state reaction because a larger number of fresh particle surfaces provides abundant nucleation sites. Crystal growth during powder sintering starts at the interface and then occurs in the

Powder Sintering, Co-Melting and Frozen Sorbent Processes
These three methods have the following common characteristics: stable glasses cannot be formed directly from starting materials, and the raw materials are required to be processed through these processes to form TGCs. These methods allow us to obtain a GC with the desired crystalline phases easily.
Powder sintering techniques are more efficient and versatile than solid-state reaction because a larger number of fresh particle surfaces provides abundant nucleation sites. Crystal growth during powder sintering starts at the interface and then occurs in the mostly glassy phase. The starting glass powder can be prepared by pulverizing molten tempered glass or by conventional sol-gel processes. Various low-melting glasses can serve as matrices with sufficient chemical stability, including borate and heavy metal oxidebased glasses [67]. Co-sintering of a mixture of low melting point glass powder and stable crystalline particles can incorporate the crystalline phase into the glassy matrix to obtain GC. The crystalline particles remain stable and insoluble in the glassy phase, and the glass powder softens and, eventually, densifies after sintering. When the melting temperature of ingredients in the raw materials is very different, and part of the components has good glass-forming ability, the powders of crystalline materials can be directly sintered at a designated temperature, producing crystalline GC composites in glass. For example, oxynitride GCs with high melting temperature cannot be prepared by melt quenching, but can be synthesized by sintering crystalline powder mixtures in an inert atmosphere [68,69]. In general, the transparency of powder sintered GCs is inferior compared to that obtained by melt-quenching and heat treatment. The optical transparency of GCs is lost due to the unavoidable scattering of pores and grain boundaries during sintering.
The co-melting process is a more direct method to produce GC in which glass components and pre-synthesized NCs are eutectic and then quenched to a GC. Compared to the traditional annealing process, this method allows more or almost limitless options for crystal-glass combinations. However, there are two factors to be considered when using this method: (1) the low melting point matrix glass is usually chosen to avoid the "dissolution" of NCs during melting [70,71]; (2) to ensure superior optical transparency, the refractive index difference between the crystal and glass phases has to be minimized [72][73][74].
The "frozen sorbent (FS) method" is used to fabricate GCs containing crystalline phases of high melting temperature. Unlike the two methods described previously, this process does not require a lengthy heat treatment process to precipitate crystals but prepares GC by quenching the melt with a component outside the glass-forming region [75][76][77]. For instance, GCs containing β-Ca 2 SiO 4 crystals can be prepared by quenching the melt with the composition of 60CaO-40SiO 2 from 1550 • C to room temperature. A homogeneous melt can only be obtained at temperatures above 1630 • C, while the melt at 1550 • C consists of a liquid phase and crystalline phase. Then, this kind of liquid-solid mixture is quenched to room temperature leading to formation of a composite consisting of the target crystal [77]. The state of a melt consisting of liquid and crystalline phases in a given composition is often determined in accordance with the equilibrium phase diagram, allowing for adjustments in composite design or control. According to Nakanishi et al., the advantages of this process include [75]: (1) the ability to prepare GCs containing metastable phases; (2) the possibility to obtain novel GCs with a high temperature crystal phase by referring to the phase diagram.

Sol-Gel Process
The sol-gel method is a liquid phase process in which compounds containing high chemically active components are cured by solution, sol and gel and then heat-treated to form oxides or other solid compounds. Using inorganic substances or metal alkoxides as precursors, these raw materials are uniformly mixed in the liquid phase, and catalytic hydrolysis and condensation reactions are carried out to form a stable transparent sol system in the solution. The sol is further polycondensed and polymerized by aging to form a gel with a three-dimensional spatial network structure. Finally, molecular and even nanostructured materials are prepared from the gel by drying, sintering and solidifying. This process allows the formation of various glasses with special compositions (such as highsilica glasses) that cannot be achieved by ordinary fusion quenching methods under mild synthesis conditions. In many instances, these compounds can be replaced with common inorganic metal salts (such as nitrates and chlorides) and organic chelating agents, rather than expensive and volatile metal alkoxides or organic compounds. After a long period of careful gel aging and high temperature sintering densification process, a monolithic sol-gel glass is finally obtained [78,79].
Since the raw materials used in the sol-gel process are first dispersed in solvent to form a low-viscosity solution, molecular-level homogeneity can be achieved in a very short time. The glasses prepared through a sol-gel route are found to have advantages over the routine melted-quenching method, such as better homogeneity, higher level purity and lower stoichiometric losses [80][81][82]. Another significant advantage is the high flexibility of the composition of the prepared materials and the ease of application of the large substrate area [83,84]. Glasses containing nanoparticles can be transparent when the size of the nanoparticles is sufficiently smaller than the wavelength of light. Various monolithic transparent fully densified glasses containing nanoparticles have been prepared by the sol-gel method [82]. Zhou et al. [85] synthesized a Er 3+ doped SiO 2 -CaF 2 TGC by the sol-gel method. With an increasing heating temperature, the CaF 2 nano-crystallites gradually grow up. The CaF 2 crystallites in the sample heat-treated at 900 • C are 10-20 nm in size, distributed homogeneously among the amorphous silica matrix. The optical transmittance spectrum shows that the transmission of the sample is around 80% in the visible region.
The sol-gel method can also prepare various GCs more directly without the crystallization process by heat treatment. This is achieved by dispersing pre-synthesized NCs in a liquid sol, and the subsequent gelation process leads to the formation of solid gels containing dispersed NCs. This process allows for an almost limitless choice of crystals, as long as they remain stable during the sol and gel process. Introducing functional NCs into amorphous gels can greatly expand the applications of conventional silica and glass materials. Quantum dots (QDs) and different types of inorganic nanoparticles can be incorporated into silica gels through this sol-gel process without a crystallization precipitation process [86,87]. However, compared with ordinary GCs prepared by conventional methods, sol-gel composites are rarely studied in the field of photocatalysis.

Laser Induced Crystallization Techniques
Since the laser was invented in 1960, it has been contributing to the development of modern chemistry-particularly in molecular spectroscopy and photochemistry. Glass laser-induced crystallization is a new type of glass material crystallization technology, which can be used to precipitate crystals in a localized fashion either on the surface or the interior of the glasses. Actually, the precipitation of crystals in glass is not easy, since most oxide glasses, e.g., silicate glasses, are stable against crystallization at temperatures up to 1000 • C. In addition, optical glasses are transparent in the visible and near-infrared range (NIR), and the thermal effect induced by most continuous wave (CW) lasers operating in the visible/near-infrared range (NIR) is too weak to initialize crystallization of the glass. By incorporating active metal ion centers in the glass, the absorption of the metal ion centers induces significant localized heating, leading to crystalline precipitation inside glass or on the glass surface.
An oxide glass with the composition (mol%) of 10RO-10Sm 2 O 3 -80TeO 2 (R = Mg or Ba) was obtained by Komatsu et al. [88] for the first CW laser induced crystallization. After irradiation of a CW YAG:Nd laser at 1064 nm for 30 s, a dot pattern with a diameter of about 50-150 lm can be generated, which induced the change of refractive index. This laser crystallization method was later extended to "write" different crystalline patterns on glass surfaces, using rare earth ions such as Sm 3+ and Dy 3+ as the absorbers for the NIR laser [89][90][91]. In the recent decade, different lasers have been used to craft crystalline patterns in/inside glasses of diverse compositions and to fabricate optical waveguide by taking advantage of the induced change in refractive index [92].
Compared with the CW laser, pulse lasers offer higher peak power and better spatial resolution in glass microfabrication. The structure of selected points in glasses can be modified by laser irradiation [93]. Focusing intensive laser pulses into the bulk of a transparent material is a novel technique that has been used to modify the irradiated area. Depending on the laser conditions, the induced changes are different. Parameters such as laser power and scanning speed have an effect on the size and density of the crystal spot. Sufficiently high laser pulse intensities can produce a local redistribution of the glass structure and major permanent modifications of the microstructure and its properties. The result of this laser treatment is a change in the physical and chemical properties of the irradiated area, such as transparency, refractive index, etc. With the rapid development of ultrafast laser technology in recent decades, nanosecond (ns), picosecond (ps) and femtosecond (fs) lasers have been extensively used in microstructure fabrication. These lasers, especially the fs laser, enable direct fabrication of microstructure in a submicrometer scale, facilitating a wide range of applications in diverse areas. Ultrafast laser direct writing-based glass crystal optics has been widely reported to realize 3D nonlinear photonic crystals [94,95], second harmonic generation [96,97], optical waveguides [98,99] and perovskite quantum dots in the glass [100,101]. Femtosecond lasers are more universal to material micropatterning, including the crystallization of glass. Compared to other pulsed laser systems, fs lasers offer unprecedented peak power density, and the use of the fs laser for microfabrication exerts minimal thermal effect on the materials, thus providing high spatial resolution. More importantly, the fs laser allows the precipitation of 3D crystalline patterns or microstructures in a transparent glass, which is not easily achieved by most other types of lasers. The first SHG crystal precipitated from glass using an fs laser was originated by Miura et al. [102] in a parent glass of 47.5BaO-5Al 2 O 3 -47.5B 2 O 3 .
In a word, the most important thing for the design and preparation of PGCs is to properly control the transparency and crystallinity of GCs. On the one hand, it is difficult for most GCs to obtain high transparency due to the difference of refractive index between nanocrystals and the glass matrix. On the other hand, the excessive growth and coarsening of crystals will lead to the prominent Mie scattering effect. In addition, it is highly desirable to obtain surface crystallization of PGCs by various means of controlled crystallization. Low photocatalytic activity limits the wide application of GCs, especially under visible light or solar irradiation, and thus, a great deal of effort still needs to be invested in the development of PGC materials with improved properties. It has been experimentally proved that higher crystallinity of PGC materials possesses better photocatalytic activity. However, higher crystallinity of PGC materials means overgrowth and coarsening of the crystals, which will cause the glass to lose its transparency. Furthermore, the higher crystallinity of PGC materials easily results in macro and micro defects such as cracks, vacancies, dangling bonds at the interface between precipitated crystals and residual glass matrix, which evidently decreases not only the transparency but also photocatalytic efficiency. From the transparency and crystallinity (photocatalytic activity) points of view, there is strong conflict between them; thereby, how to balance the crystallinity (photocatalytic activity) and transparency is very critical to preparation and widely application of transparent photocatalytic glass ceramics.

Modification Strategies of PGCs
Given the various limitations of photocatalysts in powder or thin film form, researchers have proposed glass-based materials, which could be a good option and can play a vital role due to easy control of properties, low-cost scale-up possibilities and lower maintenance [103]. However, the inefficient utilization of visible light and the rapid recombination of the photogenerated electrons and holes are often two major limiting factors which inhibit the improvement of photocatalytic activity. Generally, different strategies were employed to enhance its photocatalytic activity, including chemical etching, doping with rare earth metals, non-metallic or metal dopants, etc. [104][105][106] Table 1 summarizes and compares the photocatalytic properties of various reported GCs.

Chemical Etching
For photocatalytic applications of glass and GCs, only surface crystallization is desirable. Generally, surface/bulk crystallization is governed by crystallization kinetics. To achieve surface crystallization, some of the advanced techniques such as ultrasonic treatment and laser-based heating can be selected, which are either cumbersome or expensive [128,129]. Thus far, numerous studies have shown that fluorination treatment is beneficial to improve the photocatalytic performance of Bi-based compounds [130][131][132][133]. The fluorination enhances the interfacial charge transfer dynamics for better surface adsorption. In addition, fluorination can also reduce the grain size and change the band gap.
Sharma et al. [112] Figure 11. Enhanced visible light-driven photocatalytic activity was observed in HF etched GCs compared to the as-quenched BBO glass, and the degradation rate of Rh 6G by BBO HF3 glass samples can reach up to 85%. The photocatalytic activity increases with the increase in HF concentration, which is due to the formation of the BiF 3 phase and the sharp increase in surface roughness value by the etching of high-concentration HF solution. Contact angle of the as-quenched glass was 90.2 • , which could be reduced to 20.02 • with the increase in the HF concentration in the etching solution. This hydrophilic nature of the etched glass samples is favorable for the photocatalytic performance because the water droplets can cover a larger portion of the surface for wiping out the pollutants. Such photo-induced hydrophilicity and catalytic properties make translucent GCs promising for self-cleaning and waste water treatment applications. Simply via surface fluorination of SrO-Bi 2 O 3 -B 2 O 3 (SBBO) GC, by using HF solution, Singh et al. [114] obtained the coating of cubic BiO 0.51 F 1.98 over SBBO GC similarly. Figure 12 shows a focused SEM image of as-quenched SBBO GC and fluorinated SBBO GCs. SBBO-HF-2 contains small spheres with diameters in the 40-80 nm range, including cubic clusters in addition to small spherical particles, but more fluorination will reduce spheres with diameters between 50 and 100 nm. Under the maximum fluorination, large cubes with a size of 150 nm to 250 nm are displayed, with a small number of agglomerated particles between 10 and 20 nm distributed between these cubes. With the continuous fluorination of concentrated HF solution, the crystal structure of the coating tends towards the cubic α-BiF 3 structure from the cubic BiO 0.51 F 1.98 . The structures of cubic BiO 0.51 F 1.98 and cubic α-BiF 3 have completely different morphologies, with the former having compact spheres and the other being densely packed cubes, respectively. With HF fluorination, the glass goes from transparent to translucent and then again to transparent; the unique process is due to the reduction in the refractive index difference between the growing crystal and the residual glass. The high transparency after fluorination may be due to the controlled homogeneous nucleation and growth of cubic α-BiF 3 crystals. The research group of Edgar D. Zanotto [134] expressed that, with further fluorination, the high transparency of the cubic crystallized glass is due to the simultaneous change in the composition of the glass matrix and crystals during the crystallization process, so that the respective refractive indices do not change considerably and remain almost similar to glass.  (Figure 12f). The fluorinated SBBO GC had more adsorption of MB than as-quenched SBBO GC, which was due to the increased surface roughness. The relatively rapid drop in the C t /C o curve and increasing photocatalytic reaction rate from SBBO to SBBO-HF-2 are due to the reduction in the particle size and bandgap E1 of BiO 0.51 F 1.98 along with structural transformation to cubic α-BiF 3 . However, after further fluorination, the photocatalytic reaction rates of SBBO-HF-3 and SBBO-HF-4 decreased, respectively. Due to the increased relative hydrophobic properties of SBO-HF-3 and SBO-HF-4 GCs, further fluorination resulted in a relatively slow decrease in Ct/Co and photocatalytic reaction rate. to the increased surface roughness. The relatively rapid drop in the Ct/Co curve and increasing photocatalytic reaction rate from SBBO to SBBO-HF-2 are due to the reduction in the particle size and bandgap E1 of BiO0.51F1.98 along with structural transformation to cubic α-BiF3. However, after further fluorination, the photocatalytic reaction rates of SBBO-HF-3 and SBBO-HF-4 decreased, respectively. Due to the increased relative hydrophobic properties of SBO-HF-3 and SBO-HF-4 GCs, further fluorination resulted in a relatively slow decrease in Ct/Co and photocatalytic reaction rate.  In the same way, a similar effect can be obtained by etching the surface with HCl. Yoshida et al. [109] proposed a unique and novel method to achieve surface crystallization. The fabrication principle is very simple, that is, using acid to etch glass at room temperature to achieve surface crystallization. The results show that the etched sample has 16 times higher activity of H 2 generation than that of the non-etched one. TiO 2 nano-particles at the surface of the crystallized glass were precipitated by chemical etching, and the specific surface area improved. Sharma et al. [113] also achieved crystallization of the BiOCl structure over the 2Bi 2 O 3 -B 2 O 3 (BBO) glass surface by HCl chemical etching and studied its photocatalytic response. Scanning electron microscopy revealed that large numbers of elongated plates or nanosheets grew on the glass' surface due to etching. These nanosheets cross each other to form a network-like porous hierarchical structure, similar to the morphology of flowers ( Figure 13). During the etching process, Bi in the BBO composition was dissolved in HCl by following the Thomson-Freundlich formulation. In HCl solution, the Bi 3+ ion reacts with the Cl − ion and, hence, creates BiCl 3 , which immediately hydrolyzes with H 2 O (Equations (4) and (5)) and creates the polarized structure of BiOCl bonded with strong forces in a layered arrangement of "-(Cl-Bi-O-Bi-Cl)-": In the same way, a similar effect can be obtained by etching the surface with HCl. Yoshida et al. [109] proposed a unique and novel method to achieve surface crystallization. The fabrication principle is very simple, that is, using acid to etch glass at room temperature to achieve surface crystallization. The results show that the etched sample has 16 times higher activity of H2 generation than that of the non-etched one. TiO2 nano-particles at the surface of the crystallized glass were precipitated by chemical etching, and the spe- graded by a BiOCl crystallized sample within 225 min, while maximum 90% degradation was achieved in a BiF3 crystallized sample, even after 300 min (Figure 16). It is due more to the narrow band gap values of BiOCl than those of BiF3 and also due to the presence of an internal electric field in the BiOCl layered structure, which prevents electron-hole pair re-combinations. The superior self-cleaning properties of BiOCl crystallized glasses were mainly due to the enhanced hydrophobicity of BiOCl crystallized glass compared with BiF3 crystallized glass.  This arrangement stacks layer by layer along the c axis with weak interlayer van der Waals interactions [135]. The layer by layer and alternate stacking sequence of "-(Cl-Bi-O-Bi-Cl)-" along the (001) plane lead to semicircular plates/petals and automatically assemble as the flower-or cabbage-like structures. Increasing the concentration of HCl for etching of BBO creates pores as well as cracks, which increases the surface roughness, but excessive etchings with HCl decreased the transparency of the BBO, and its color changed from yellow to white, respectively (Figure 13l). Therefore, the etching process needs to be optimized to obtain better transparency and superior photocatalytic performance. A typical schematic diagram for the degradation of pollutants by HCl-etched BBO GCs under UV light irradiation is shown in Figure 14. Due to the absorption of photons with energy greater than or equal to the energy of the material, an electron is excited from the valence band to the conduction band, thereby forming an electron pair and a positive hole. These photoinduced positive holes and electrons undergo a series of reactions and finally produce hydroxyl (·OH) radicals and superoxide radicals (·O 2 − ). An HCl-etched BBO GC is irradiated with ultraviolet rays; with the help of superoxide ions as well as hydroxyl radicals, the harmful pollutants are reduced to harmless products. Photocatalytic experiments on BBO glass samples etched with different concentrations of HCl showed that the degradation rate of methylene blue by BBO HCl3 glass was the highest, reaching almost 100% within 180 min. With the further increase in HCl concentration, the penetration depth in the glass sample increases, and more and more nucleation sites are formed, which further promotes the growth of BiOCl. BiOCl grows and is densely arranged along the (110) plane to form the flower-like hierarchical structure assembled with petals, which is conducive to absorbing the best solar radiation and reducing annihilation rate of electron and hole so as to obtain the best kinetic rate of photocatalytic reaction.
Singh et al. [115] also used HCl chemical etching to obtain the BiOCl layered structure over the SrO-Bi 2 O 3 -B 2 O 3 (SBBO) TGC. The schematic diagram is shown in Figure 15 [116]. Using four different concentrations of acid (3, 6, 9 and 12 M), different amounts of BiOCl and BiF 3 surface crystals were obtained. As reported by Sharma [113], the morphology of the glass samples changed from plate-shaped particles to flower-shaped particles with the increasing concentration of etching acid. Additionally, the transparency was decreased with an increase in the concentration of acid used for etching owing to the increase in glass surface roughness. Compared with BiF 3 surface-crystallized glass, BiOCl surface-crystallized glasses have more efficient photocatalytic performance. Under ultraviolet irradiation (UV), methylene blue was completely degraded by a BiOCl crystallized sample within 225 min, while maximum 90% degradation was achieved in a BiF 3 crystallized sample, even after 300 min ( Figure 16). It is due more to the narrow band gap values of BiOCl than those of BiF 3 and also due to the presence of an internal electric field in the BiOCl layered structure, which prevents electron-hole pair re-combinations. The superior self-cleaning properties of BiOCl crystallized glasses were mainly due to the enhanced hydrophobicity of BiOCl crystallized glass compared with BiF 3 crystallized glass.
crystallized glass, BiOCl surface-crystallized glasses have more efficient photocatalytic performance. Under ultraviolet irradiation (UV), methylene blue was completely degraded by a BiOCl crystallized sample within 225 min, while maximum 90% degradation was achieved in a BiF3 crystallized sample, even after 300 min (Figure 16). It is due more to the narrow band gap values of BiOCl than those of BiF3 and also due to the presence of an internal electric field in the BiOCl layered structure, which prevents electron-hole pair re-combinations. The superior self-cleaning properties of BiOCl crystallized glasses were mainly due to the enhanced hydrophobicity of BiOCl crystallized glass compared with BiF3 crystallized glass.

Rare Earth Element Doping
Rare earth elements consist of 15 special lanthanide elements in the Periodic Table of Elements and 2 other elements, Sc and Y, closely related to lanthanide elements. Rare earth elements have abundant energy levels and 4f electron transition characteristics, which can

Rare Earth Element Doping
Rare earth elements consist of 15 special lanthanide elements in the Periodic Table of Elements and 2 other elements, Sc and Y, closely related to lanthanide elements. Rare earth elements have abundant energy levels and 4f electron transition characteristics, which can

Rare Earth Element Doping
Rare earth elements consist of 15 special lanthanide elements in the Periodic Table  of Elements and 2 other elements, Sc and Y, closely related to lanthanide elements. Rare earth elements have abundant energy levels and 4f electron transition characteristics, which can easily form poly-electron configurations to enhance the optical absorption of material and accelerate the efficiency for the separation of photo-generated electron-hole pairs. Lanthanide-based upconverting GCs contain luminescent fluorides with low phonon energy (i.e., SrF 2 , CaF 2 and NaYF 4 ), which can facilitate amplified stimulated emission at low pump power and are highly attractive for upconversion materials or laser applications. The GCs will have a higher refractive index and less photon energy with the addition of the yttrium ion as a nucleation agent [137]. Milewska et al. [138] obtained borate-bismuth GCs containing SrF 2 nanocrystals doped with Eu 3+ ions and found that the luminescence intensity increased after annealing in the samples containing 20 mol% SrF 2 . Through the doping process, some of the Eu 3+ ions were located in the SrF 2 crystalline structure, which would lead to a reduction in the multi-phonon relaxation probability and improvement in the emission efficiency, thereby effectively enhancing the luminescence performance of GCs. Similarly, trivalent erbium ions doped Bi 2 O 3 -B 2 O 3 TGCs containing CaF 2 prepared by Jamalaiah et al. [139] showed a significant improvement in the 1.53 µm emission of Er 3+ : 4I 13/2 →4I1 5/2 transition after the formation of Bi 3 B 5 O 12 and CaF 2 in the matrix. The upconversion's emission efficiency in Yb 3+ /Er 3+ co-doped boron bismuth oxyfluoride GCs was greatly enhanced with the precipitation of CaF 2 crystals [140].In addition, Panyata al. et al. [141] improved the microstructure of Bi 2 GeO 5 GC by Er 2 O 3 incorporation, making it have a desired crystal dimension.
In recent years, lanthanide-doped photocatalysts have also been extensively studied, continuously absorbing low-energy photons and emitting higher-energy photons, making it possible to use low-energy light, such as near-infrared (NIR) light. The NIR light-responsive upconversion photocatalysts improve the utilization efficiency of sunlight and also have better penetration compared with short-wavelength light and maintain photoactivity, even when the photocatalyst surface is shaded. NIR photocatalysts can be divided into two types: lanthanide direct doping and core-shell structures. Direct doping of lanthanide ions into semiconductors can improve the utilization of light energy and promote the separation efficiency of electron-hole pairs, but the upconversion luminescence efficiency of such NIR photocatalysts is not high enough, which limits its photocatalytic activity [142]. The core-shell structure-type NIR photocatalyst is combined with a light-emitting substrate and a semiconductor, wherein the photoactive center doped with lanthanide ions is the core, and the semiconductor is the surface coating. NIR photocatalysts can capture nearinfrared photons and emit upconversion luminescence, which can perform photocatalysis by themselves under NIR light irradiation. Li et al. [122] synthesized a novel core-shell structure (Er 3+ /Yb 3+ -CaO-Al 2 O 3 -SiO 2 -CaF 2 )/TiO 2 NIR photocatalyst by solid-phase reaction for the first time. GC is the host material for upconversion luminescence and the bridge connecting TiO 2 . Compared with pure TiO 2 , the photocatalyst easily collects NIR photons and enhances the photocatalytic activity under NIR and UV-vis-NIR radiation. As shown in Figure 17, based on solid-phase reaction combined with chemical etching, they synthesized BiOCl nanosheets on upconversion SrO-Bi 2 O 3 -B 2 O 3 (SBBO) GC to synthesize NIR PGC [123]. First, rare earth Er 3+ /Yb 3+ doped SrF 2 luminescent nanocrystals were crystallized in SBBO GC particles, and then a new NIR GC photocatalyst coated with BiOCl nanosheets was synthesized by a simple in situ etching method. The PGC also has a coreshell structure, GC powder containing optically active centers Er 3+ , Yb 3+ and Y 3+ -doped SrF 2 nanocrystal, as the core and BiOCl as the surface coating. NIR photocatalysts can efficiently collect NIR photons and emit strong ultraviolet (379 nm), violet (408 nm), green (523 and 540 nm) and red (656 nm) light, which can be used by themselves for photocatalysis. The organic pollutants were degraded under NIR radiation for 90 min, and high removal rates of 66% (norfloxacin) and 79% (methyl orange) were achieved. Meanwhile, they obtained an SBBO upconversion GC that can remove and utilize Cl ions through optimization experiments [124]. Among them, abundant porous structures are continuously formed during the Cl removal process, which further provides channels for the deep migration of Cl − and H + to its internal Bi source. Under high chloride ion concentration (1300-13,000 mg/L), the Cl − removal rate of the best samples all reached more than 96%. The terminal BiOCl-based precipitates are directly collected as NIR photocatalysts with abundant defect heterostructures, excellent upconversion luminescence and electron energy transport properties. Under UV-Vis-NIR, Vis-NIR and NIR irradiation, the removal rates of norfloxacin (NOR) reached 98%, 73% and 36%, respectively. Similarly, co-worker Zhang et al. [126] also developed a narrow-bandgap Bi 1.95 Yb 0.04 Er 0.01 V 2 O 8 bulk upconversion PGC by Yb 3+ /Er 3+ co-doping. Precipitated crystal Bi 1.95 Yb 0.04 Er 0.01 V 2 O 8 in volume, this GC also grew BiOCl nanosheets in situ on the catalyst surface. The GC produced strong green (526 and 546 nm) and red (658 nm) emissions under 980 nm excitation. Extending the absorption edge of the GC photocatalyst to 620 nm, the NIR photons were efficiently harvested, and most of the upconversion emissions during the photocatalysis process were utilized. A high norfloxacin degradation rate of 55.2% was obtained over 90 min of NIR light irradiation (Figure 18a). Interestingly, the photocatalytic activity of the recycled samples was enhanced again, with the degradation efficiency increased from 48.1% to 55.6% after the HCl etching experiment, which indicated the photocatalyst could be easily regenerated by HCl etching again. Moreover, Figure 18b revealed the photocatalyst could photodegrade low-concentration antibiotics in landfill leachate wastewater, which provided a novel strategy for the practical wastewater treatment. Finally, the photocatalytic mechanism of the photocatalyst is proposed (Figure 18c (Figure 18a). Interestingly, the photocatalytic activity of the recycled samples was enhanced again, with the degradation efficiency increased from 48.1% to 55.6% after the HCl etching experiment, which indicated the photocatalyst could be easily regenerated by HCl etching again. Moreover, Figure 18b revealed the photocatalyst could photodegrade low-concentration antibiotics in landfill leachate wastewater, which provided a novel strategy for the practical wastewater treatment. Finally, the photocatalytic mechanism of the photocatalyst is proposed (Figure 18c). The BiOCl nanosheet is grown in situ on the surface of the bulk GC, and the lanthanide-doped upconversion luminescence matrices (Bi1.95Yb0.04Er0.01V2O8) crystallized in the GC harvest NIR photons to produce upconversion emissions. The surface coating of BiOCl has a wide absorption edge of 620 nm. Direct absorption or fluorescence resonance energy transfer directly uses photons with wavelengths of 652, 792 and <620 nm and excites electrons from the valence band (VB) of BiOCl to the conduction band (CB). Then, the electrons of CB BiOCl reduce O2 to •O2 − , and the holes of VB BiOCl oxidize OH − to •OH, resulting in the photodegradation of organic pollutants by the active species.  Compared with conventional binary oxides, perovskite oxides generally exhibit better separation of charge carriers due to higher lattice distortion [143]. Li et al. [127] introduced SiTiO 3 and Sr 2 Bi 4 Ti 4 O 15 ferroelectric perovskite substrates doped with Yb 3+ /Er 3+ rare earth ions in GC and prepared efficient and stable NIR photocatalysts by a facile in situ etching method. The results show exposure of Sr 2 Bi 4 Ti 4 O 15 , SrTiO 3 and BiOCl in the surface coating of the core-shell structured PGC and constructing a Z-type heterojunction. The heterojunction with built-in electric fields can significantly promote the separation of the charge carrier while trapping the NIR light for photocatalytic reactions. Under NIR illumination for 90 min, the degradation rate of NOR reached 86%, and the PGC can also be used for photodegradation of low-concentration antibiotic-containing wastewater. The SrTiO 3 /SrBi 4 Ti 4 O 15 @BiOCl heterojunction was generated during GC formation, which contributed to the generation of •O 2 − and •OH radicals in the catalytic reaction, which played crucial roles in the degradation process.
However, unlike BiOCl, which has very limited visible light responsive properties, BiOBr exhibits a suitable band gap, good visible-light-driven photocatalytic activity and high chemical stability. Therefore, as shown in Figure 19, Li et al. [125] used the same method to grow BiOBr nanosheets on the surface of SrF 2 -Bi 2 O 3 -B 2 O 3 /Yb 3+ , Tb 3+ (SBBF) on-converted GC. During the in situ etching process, B 3+ , Yb 3+ and Tb 3+ ions in GC can be doped into the BiOBr layer. The lanthanide-doped upconversion luminescence agent (SrF 2 /Yb 3+ , Tb 3+ ) in the core of SBBF/ BiOBr-10 harvests NIR photons to then emit red (623 nm), yellow (586 nm), green (546 nm), blue (489 nm), violet (438 and 415 nm) and UV (381 nm) light emissions. The core-shell structured GC@BiOBr heterojunction photocatalyst is beneficial to increase charge transport, reduce the recombination rate of excited carriers and efficiently collect near-infrared photons for UV and visible upconversion emission, which can be utilized in photocatalytic processes. In SBBF/BiOBr-10, the excitation e − is promoted to be transferred from the VB to the CB of BiOBr with the energy transfer mechanism, then e − is captured by O 2 to produce •O 2 − free radicals. The PGC showed a high degradation rate of NOR (56%) under NIR illumination for 180 min, and its photocatalytic activity was greatly enhanced under UV-vis-NIR compared with pure BiOBr. Moreover, by re-etching, the photocatalyst can be easily regenerated and made suitable for large-scale fabrication, which will provide a new strategy for the development of efficient photocatalysts. Compared with conventional binary oxides, perovskite oxides generally exhibit better separation of charge carriers due to higher lattice distortion [143]. Li et al. [127] introduced SiTiO3 and Sr2Bi4Ti4O15 ferroelectric perovskite substrates doped with Yb 3+ /Er 3+ rare earth ions in GC and prepared efficient and stable NIR photocatalysts by a facile in situ etching method. The results show exposure of Sr2Bi4Ti4O15, SrTiO3 and BiOCl in the surface coating of the core-shell structured PGC and constructing a Z-type heterojunction. The heterojunction with built-in electric fields can significantly promote the separation of the charge carrier while trapping the NIR light for photocatalytic reactions. Under NIR illumination for 90 min, the degradation rate of NOR reached 86%, and the PGC can also be used for photodegradation of low-concentration antibiotic-containing wastewater. The SrTiO3/SrBi4Ti4O15@BiOCl heterojunction was generated during GC formation, which con- degradation rate of NOR (56%) under NIR illumination for 180 min, and its photocataly activity was greatly enhanced under UV-vis-NIR compared with pure BiOBr. Moreov by re-etching, the photocatalyst can be easily regenerated and made suitable for larg scale fabrication, which will provide a new strategy for the development of efficient ph tocatalysts.

Visible Light Active Photocatalysis
Most photocatalytic materials, including TiO2, only respond to UV light irradiatio with limited functionality and small scale. While it is necessary to develop highly efficie stable visible light active photocatalysts for the purification of large-scale wastewater air pollution, such as pharmaceutical and chemical wastewater, printing and dyeing te tile wastewater, pesticides, NOx and SOx, etc. Consequently, with the goal of maximizi light absorption, maximizing surface charge transfer and minimizing recombination achieving efficient charge separation and maximizing surface redox reactions, conside ble efforts have been devoted to the development of photocatalyst materials with i proved performance. Kumar et al. [110] obtained TiO2 surface crystallized BaO-TiO2-B2 (BTBO) GCs with visible light catalytic activity. Due to the existence of TiO2 crystals crystalline glass, the positive charge on the glass surface has a high electrostatic affin for negatively charged phosphor-peptides and proteins in the cell membrane, which conducive to better adhesion to bacterial cells. The antibacterial rate of the crystalliz BaO-TiO2-B2O3 glass reached 98.3% under the condition of 2 h of sun irradiation (Figu 20a). Under solar irradiation, electrons move from the valence band to the conducti band of TiO2 in the crystalline glass, then excited electrons are generated on the surface TiO2 crystalline glass, triggering a redox reaction (Figure 20c). A redox reaction can pr duce different reactive oxygen species (ROS) and induce the peroxidation of microb cells, leading to cell dissolution. ROS produced by hydrolysis enters into cells and creases the oxidative stress inside the cells, thus further enhancing the intracellular RO production. Intracellular ROS intensity of bacterial cells was almost three times higher the presence of crystallized glass than in control and as-quenched glass under light (F ure 20d). A Cu-doped TeO2-based GC containing TiO2 and TiTe3O8 visible light acti crystals was prepared by Kushwaha et al. [118] using an optimized crystallizati method. The visible light photocatalytic activity of the crystallized xCaCu3Ti4O12 − (100− TeO2 (x = 0.25 mol% to 3 mol%) glass was investigated for estrogenic pharmaceutical p lutants, and the degradation rate was observed as 168.56 min −1 ·m −2 (x = 3 mol%). This cry tallized glass plate also behaved at a higher visible light photocatalytic H2 production ra (135 μmol h −1 g −1 ). The catalytic efficiency of this visible light-driven multifunctional ph tocatalyst depends on the active concentration of the photoactive crystallites present the glass matrix. Iida et al. [119] added Al2O3 to iron-containing soda lime silicate glass

Visible Light Active Photocatalysis
Most photocatalytic materials, including TiO 2 , only respond to UV light irradiation, with limited functionality and small scale. While it is necessary to develop highly efficient stable visible light active photocatalysts for the purification of large-scale wastewater or air pollution, such as pharmaceutical and chemical wastewater, printing and dyeing textile wastewater, pesticides, NOx and SOx, etc. Consequently, with the goal of maximizing light absorption, maximizing surface charge transfer and minimizing recombination by achieving efficient charge separation and maximizing surface redox reactions, considerable efforts have been devoted to the development of photocatalyst materials with improved performance. Kumar et al. [110] obtained TiO 2 surface crystallized BaO-TiO 2 -B 2 O 3 (BTBO) GCs with visible light catalytic activity. Due to the existence of TiO 2 crystals in crystalline glass, the positive charge on the glass surface has a high electrostatic affinity for negatively charged phosphor-peptides and proteins in the cell membrane, which is conducive to better adhesion to bacterial cells. The antibacterial rate of the crystallized BaO-TiO 2 -B 2 O 3 glass reached 98.3% under the condition of 2 h of sun irradiation (Figure 20a). Under solar irradiation, electrons move from the valence band to the conduction band of TiO 2 in the crystalline glass, then excited electrons are generated on the surface of TiO 2 crystalline glass, triggering a redox reaction (Figure 20c). A redox reaction can produce different reactive oxygen species (ROS) and induce the peroxidation of microbial cells, leading to cell dissolution. ROS produced by hydrolysis enters into cells and increases the oxidative stress inside the cells, thus further enhancing the intracellular ROS production. Intracellular ROS intensity of bacterial cells was almost three times higher in the presence of crystallized glass than in control and as-quenched glass under light (Figure 20d). A Cu-doped TeO 2 -based GC containing TiO 2 and TiTe 3 O 8 visible light active crystals was prepared by Kushwaha et al. [118] using an optimized crystallization method. The visible light photocatalytic activity of the crystallized xCaCu 3 Ti 4 O 12 − (100−x) TeO 2 (x = 0.25 mol% to 3 mol%) glass was investigated for estrogenic pharmaceutical pollutants, and the degradation rate was observed as 168.56 min −1 ·m −2 (x = 3 mol%). This crystallized glass plate also behaved at a higher visible light photocatalytic H 2 production rate (135 µmol h −1 g −1 ). The catalytic efficiency of this visible light-driven multifunctional photocatalyst depends on the active concentration of the photoactive crystallites present in the glass matrix. Iida [120] combined the highly photoactive metastable βand γ-Bi 2 O 3 phases with BiFeO 3 in one GC material. The best photocatalytic performance for the degradation of the RY160 dye was exhibited with the sample containing equal amounts of Fe 2 O 3 and Bi 2 O 3 due to the formation of a γ-Bi 2 O 3 /BiFeO 3 heterojunction. As shown in Figure 21a, Kadam et al. [121] proposed a novel approach to generate bismuth sulfide (Bi 2 S 3 ) QDs in silicate glass. The photocatalytic hydrogen (H 2 ) production under solar light was performed and the maximum H 2 generation, i.e., 6418.8 mmol h −1 g −1 , was achieved for the sample containing 7 mg Bi 2 S 3 quantum dots (Figure 21b). It is noteworthy that the enhanced H 2 evolution rate has been obtained for glass nanocomposites having increased concentration and density of Bi 2 S 3 QDs. Another beauty of this photocatalytic material is that it can be reused several times without a reduction in its activity. To achieve visible light-driven photocatalysis, common methods investigated for performance improvement are morphology optimization, doping, using sensitizers and/or co-catalysts and using different types of heterojunctions [144]. However, the improvement in visible light-active PGCs still lacks systematic exploration.  [120] combined the highly photoactive metastable β-and γ-Bi2O3 phases with BiFeO3 in one GC material. The best photocatalytic performance for the degradation of the RY160 dye was exhibited with the sample containing equal amounts of Fe2O3 and Bi2O3 due to the formation of a γ-Bi2O3/BiFeO3 heterojunction. As shown in Figure 21a, Kadam et al. [121] proposed a novel approach to generate bismuth sulfide (Bi2S3) QDs in silicate glass. The photocatalytic hydrogen (H2) production under solar light was performed and the maximum H2 generation, i.e., 6418.8 mmol h −1 g −1 , was achieved for the sample containing 7 mg Bi2S3 quantum dots (Figure 21b). It is noteworthy that the enhanced H2 evolution rate has been obtained for glass nanocomposites having increased concentration and density of Bi2S3 QDs. Another beauty of this photocatalytic material is that it can be reused several times without a reduction in its activity. To achieve visible light-driven photocatalysis, common methods investigated for performance improvement are morphology optimization, doping, using sensitizers and/or cocatalysts and using different types of heterojunctions [144]. However, the improvement in visible light-active PGCs still lacks systematic exploration.  as-quenched and crystallized experimental group by relative DCFH fluorescence intensity. Reprinted with permission from Ref. [110]. This section elaborates the modification strategies of PGCs to improve their photocatalytic activity by chemical etching and rare earth doping, as well as the requirements for visible light photocatalytic materials ( Figure 22). Chemical etching makes the glass surface form a network porous hierarchy, which is conducive to better capture light, leading to more efficient photocatalysis. Moreover, the increase in roughness caused by etching is more conducive to the adsorption of pollutants, and the decrease in grain size is conducive to transparency, while the increase in transparency will lead to the activation of more photocatalytic crystals. Doping rare earth elements greatly improves the upconversion emission efficiency of GCs. Through the energy conversion mechanism of direct absorption (DA), emission reabsorption (ER) and Förster resonance energy transfer (FRET), a photocatalysis response is generated in near-infrared. NIR light has a higher penetration depth through various media, especially biological tissues and wastewater. Under UV-vis-NIR light irradiation, the NIR driven degradation of organic pollutants can be achieved, and the photocatalytic efficiency under UV-vis-NIR light irradiation can be improved. At the same time, rare earth elements doped into the lattice will lead to the contraction and reduction in the crystal size. At present, there are few GCs that can realize This section elaborates the modification strategies of PGCs to improve their photocatalytic activity by chemical etching and rare earth doping, as well as the requirements for visible light photocatalytic materials ( Figure 22). Chemical etching makes the glass surface form a network porous hierarchy, which is conducive to better capture light, leading to more efficient photocatalysis. Moreover, the increase in roughness caused by etching is more conducive to the adsorption of pollutants, and the decrease in grain size is conducive to transparency, while the increase in transparency will lead to the activation of more photocatalytic crystals. Doping rare earth elements greatly improves the upconversion emission efficiency of GCs. Through the energy conversion mechanism of direct absorption (DA), emission reabsorption (ER) and Förster resonance energy transfer (FRET), a photocatalysis response is generated in near-infrared. NIR light has a higher penetration depth through various media, especially biological tissues and wastewater. Under UV-vis-NIR light irradiation, the NIR driven degradation of organic pollutants can be achieved, and the photocatalytic efficiency under UV-vis-NIR light irradiation can be improved. At the same time, rare earth elements doped into the lattice will lead to the contraction and reduction in the crystal size. At present, there are few GCs that can realize visible light photocatalysis, which is very demanding for the crystallinity and transparency of GCs, requiring a higher concentration of visible light-active microcrystals to precipitate from the glass matrix. Photocatalytic microcrystals absorb light to the maximum extent so as to improve effective charge separation efficiency and maximize surface redox reaction, thus maximizing the photocatalytic efficiency of PGCs for various application fields.
Catalysts 2022, 12, x FOR PEER REVIEW 33 of 40 visible light photocatalysis, which is very demanding for the crystallinity and transparency of GCs, requiring a higher concentration of visible light-active microcrystals to precipitate from the glass matrix. Photocatalytic microcrystals absorb light to the maximum extent so as to improve effective charge separation efficiency and maximize surface redox reaction, thus maximizing the photocatalytic efficiency of PGCs for various application fields. Figure 22. Summary of modification engineering for PGCs.

Conclusions
In conclusion, this review highlights the basic theories, design strategies, crystallization methods and modification investigations of GC photocatalysts. Obviously, GCs have proven to be one of the candidate materials suitable for designing and fabricating advanced photocatalysts for various applications. PGCs have the following advantages: (i) photocatalytic crystals embedded in the matrix without any leaching problems, (ii) easy to manufacture in different shapes and sizes for applications and (iii) high transparency to maximize photocatalytic activity. PGCs have broad application prospects in water purification, bacterial disinfection, self-cleaning and hydrogen evolution. Whether it is surface crystallization, bulk crystallization or a combination of the two, GCs with desired properties can be designed by controlling crystallization. Maintaining high transparency enables the activation of photocatalytic crystals on the inner-surface of TGCs during the photocatalytic process, resulting in highly efficient photocatalytic performance. Transparency can be achieved by controlling glass composition and crystal size, rate crystallization, crystal distribution, and the like. In particular, the transparency of TGCs by congruent crystallization with almost identical composition of crystalline and glass matrix is appreciable. Similar to the wet chemical process, different types of crystals can be precipitated into the glassy phase by different processes. Among solid-phase glasses, direct reaction of the solvent phase is the most popular approach for synthesizing NCs in glasses. Furthermore, different strategies to enhance their photocatalytic activity, including chemical etching, rare earth metals doping, non-metallic or metal dopants, etc., have also been proposed. These modification methods are conducive to improving the microstructure, the size of crystal, so as to obtain more efficient photocatalysis.

Conclusions
In conclusion, this review highlights the basic theories, design strategies, crystallization methods and modification investigations of GC photocatalysts. Obviously, GCs have proven to be one of the candidate materials suitable for designing and fabricating advanced photocatalysts for various applications. PGCs have the following advantages: (i) photocatalytic crystals embedded in the matrix without any leaching problems, (ii) easy to manufacture in different shapes and sizes for applications and (iii) high transparency to maximize photocatalytic activity. PGCs have broad application prospects in water purification, bacterial disinfection, self-cleaning and hydrogen evolution. Whether it is surface crystallization, bulk crystallization or a combination of the two, GCs with desired properties can be designed by controlling crystallization. Maintaining high transparency enables the activation of photocatalytic crystals on the inner-surface of TGCs during the photocatalytic process, resulting in highly efficient photocatalytic performance. Transparency can be achieved by controlling glass composition and crystal size, rate crystallization, crystal distribution, and the like. In particular, the transparency of TGCs by congruent crystallization with almost identical composition of crystalline and glass matrix is appreciable. Similar to the wet chemical process, different types of crystals can be precipitated into the glassy phase by different processes. Among solid-phase glasses, direct reaction of the solvent phase is the most popular approach for synthesizing NCs in glasses. Furthermore, different strategies to enhance their photocatalytic activity, including chemical etching, rare earth metals doping, non-metallic or metal dopants, etc., have also been proposed. These modification methods are conducive to improving the microstructure, the size of crystal, so as to obtain more efficient photocatalysis.

Future Prospects
Up to now, although research on GCs has made preliminary progress in recent years, there are still several challenges in rationally preparing efficient PGCs that are suitable for various applications, controlling the transparency and surface crystallization of GCs and improving the photocatalytic activity under visible light. Due to the refractive index difference between nanocrystals and the glass matrix and the Mie scattering effect caused by oversized grains, it is difficult to obtain high transparency in most GCs. Studies have shown that PGCs with high photocatalytic activity required high crystallinity. However, the higher crystallinity is bound to significantly reduce the transparency and even cause defects that reduce the photocatalytic efficiency. Therefore, there is a long way to go in balancing transparency and crystallinity (photocatalytic activity) in the preparation and application of PGCs.
Furthermore, although conventional GCs have received extensive attention due to their natural advantages, such as low cost, ease of manufacture, transparency, chemical and mechanical stability, the chemical stability and recyclability of PGC materials have a large gap to real time application in fields of energy conversion and environmental remediation because of different applied conditions. It has been shown that the feasibility of precipitation or fixation of photocatalyst nanocrystals in GCs to avoid disadvantages of photocatalytic nanocrystals applied in form of powder and film, such as difficulty in recovery of nanopowder, secondary pollution, low adhesion between photocatalytic coating and substrate material, short service life and less recyclability, is limited. However, these PGC materials are usually assessed using only model organic contaminants. How PGC materials perform (stability, recyclability, efficiency, etc.) in the real case of decontamination, especially in extreme conditions, has not been explored systematically. Furthermore, it is well known that some parent glass systems for preparing PGC such as borate, phosphate and tellurite glass systems, have relatively weak chemical durability due to their hydroscopic properties; thereby, it is vitally important for real time application to choose suitable parent glass system and photocatalytic active crystals.
Additionally, the function of currently reported PGCs is relatively simple; investigation and exploration on the multifunctional PGCs with high photocatalytic performance, the broad photocatalytic response from UV and visible to near infrared (NIR) ranges, and high transparency is of great significance. The crystalline grains (quantum dots, nanoparticles containing rare earth ions such as Yb, Nd, Tm, Er, Ho, Pr, etc.) with light conversion luminescence and the grains (Ti, Zn, Bi and other compounds) with photocatalytic function need to be precipitated simultaneously in GCs. Co-precipitation of light conversion luminescent crystals into PGC can improve the optical response range of PGC and energy transfer between luminescent grains and photocatalytic active grains, thus increasing the photocatalytic performance of PGC materials. Thereby, precipitation of crystallites in PGCs to broaden the range of light absorption to the NIR region and convert to UV and visible emission is desirable. The visible light and near-infrared light can be fully utilized efficiently for photocatalysis by up-conversion of near-infrared light. At the same time, to ensure the effective activation of photocatalytic grains precipitated on the surface of PGCs by incident sunlight and upconversion luminescence, designing the spatial distribution and controlling the nucleation and growth process of conversion luminescence grains and photocatalytic grains can solve the problem of translucence and even opacity of PGCs caused by crystallization. Finally, PGC materials with highly efficient antibacterial, disinfectant activities and air cleaning properties indoors will be strongly focused on, as COVID-19 and another infectious virus are spreading across the world. These surveys remain in their infancy, and there are still numerous research gaps worth exploring in the near future.

Conflicts of Interest:
The authors declare no conflict of interest.