Crystallization Mechanisms and Optical Properties of Yb³⁺-Containing Glasses and Glass-Ceramics: A Brief Review

Abstract

Yb³⁺-containing glasses and glass-ceramics are attractive photonic materials because Yb³⁺ can act simultaneously as a near-infrared absorber, an energy-transfer sensitizer, a luminescent center, and a composition-dependent modifier of glass structure and crystallization. This brief review focuses on crystallization from parent glasses to glass-ceramics and examines glass-network chemistry, local Yb³⁺ coordination, phase separation, viscosity, heating rate, treatment temperature, holding time control nucleation, crystal growth, phase selection, rare-earth partitioning, transparency, and optical performance. Representative oxyfluoride, phosphate, oxyapatite, borosilicate, and aluminosilicate systems are compared using thermal analysis, X-ray diffraction, electron microscopy, vibrational spectroscopy, and optical spectroscopy. The available data show that Yb₂O₃ or YbF₃ does not have a universal effect on crystallization: low concentrations can promote fluoride-rich clustering or lower the apparent crystallization barrier, whereas higher concentrations can increase packing density, stabilize the residual glass, change the competitive phase assemblage, or suppress crystallization. Crystallization-enhanced luminescence is most consistently obtained when Yb³⁺ and the activator partition into low-phonon-energy nanocrystals while crystal size and refractive-index mismatch remain sufficiently small to preserve transparency. This review also identifies major reporting gaps, including limited quantification of crystalline fraction, partition coefficients, luminescence lifetime, quantum efficiency, and long-term thermal stability. Practical design guidelines and unresolved questions are proposed to support the rational development of transparent Yb³⁺-containing glass-ceramics for lasers, sensing, optical amplification, and related photonic applications.

1. Introduction

1.1. Materials with and Without Yb³⁺ Ions

Numerous studies have been undertaken on the formation of crystal phases and crystalline fractions in glasses, the development of competing microstructures during crystallization, and the influence of atomic-size mismatch and composition on glass-forming ability [1,2,3,4,5,6,7,8]. A recent study by Contreras Jaimes and co-workers [9] examined bioactive glasses in the SiO₂–P₂O₅–CaO–CaF₂ quaternary system, focusing on how the orthophosphate-to-silicate ratio dictates phase development. Using a combination of XRD, solid-state NMR, and high-resolution imaging, they showed that phosphate-free compositions yield xonotlite, wollastonite, and cuspidine, whereas fluorapatite emerges only when P₂O₅ exceeds 2.4 mol%. Notably, crystallization remained marginal at low phosphate levels but intensified substantially above 3 mol% P₂O₅. The phosphate-free glass exhibited volume crystallization with nearly constant activation energy, while the high-phosphate variant displayed mixed surface–volume behavior and a declining activation energy with progressive conversion. These findings underscore the necessity of linking chemical makeup, thermal response, crystallization pathway, crystal fraction, grain size, and microstructure in rational glass-ceramic design [2,3,4,5,6,7,8,10,11,12,13].

Compared to the amount of research on SiO₂-P₂O₅-CaO-CaF₂ bioactive glasses without Yb³⁺ ions, systematic studies remain limited for glasses containing Yb³⁺ compounds, including Yb³⁺ glasses in which Yb₂O₃ or YbF₃ acts as a functional constituent, although many reports have successfully identified the phases formed after heat treatment. Most studies adopt fixed concentrations for the major glass constituents and vary only the Yb³⁺-containing component or rare-earth codopant. The Yb³⁺ ion possesses a (4f)¹³(5s)²(5p)⁶ electronic configuration, in which the 4f electrons are effectively screened by the filled 5s and 5p subshells. This shielding renders the 4f→4f transitions largely insensitive to the surrounding ligand field. The ground state is ²F_7/2, and the sole excited multiplet ²F_5/2 lies approximately 10,000 cm⁻¹ higher in energy. Consequently, Yb³⁺-doped glasses and glass-ceramics display a characteristic absorption band centered near 1000 nm. Moreover, when Yb³⁺ ions are positioned in close proximity to one another or to other lanthanides (e.g., Eu³⁺), the materials exhibit cooperative luminescence and sensitized upconversion via inter-ionic energy exchange [14,15,16,17,18,19,20,21,22,23,24,25].

Upconversion—often termed anti-Stokes emission—denotes the generation of higher-energy (shorter-wavelength) photons upon excitation with lower-energy (longer-wavelength) radiation. This phenomenon is frequently observed in various Ln³⁺-doped compounds, including those containing Yb³⁺. The upconversion property underpins applications in bioanalysis, theranostics (a technical term expressing the integration of therapeutics and diagnostics), biological imaging, optical sensing, optogenetics, super-resolution microscopy, barcoding, fingerprint detection, NIR vision, photochemical manipulation, optical trapping, additive manufacturing, laser technology, and NIR-II imaging [21,22,23,24,25,26,27,28,29,30,31,32,33,34,35]. For example, regarding the application to theranostics, the following report is presented [35]. Xu et al. reviewed recent advances in lanthanide-doped upconversion nanoparticles (Ln-UCNPs) as theranostic probes, emphasizing that their anti-Stokes luminescence, near-infrared excitation, large tissue-penetration depth, low photodamage, and efficient light-conversion capability make them attractive for biomedical diagnosis and treatment [35]. In particular, the higher-energy visible or ultraviolet light produced by Ln-UCNPs from NIR irradiation can be harnessed to trigger photosensitizers and photothermal agents, facilitating photodynamic therapy as well as photothermal therapy. Through surface modification and coupling with functional molecules, they can also act as drug-delivery vehicles, optical imaging probes, biosensors, and multimodal platforms for computed tomography, magnetic resonance imaging, fluorescence imaging, and image-guided therapy. These theranostic examples clearly indicate that Yb³⁺-sensitized upconversion materials are not only important for understanding energy-transfer and luminescence mechanisms but also provide a versatile materials basis for integrated diagnosis, treatment, and biomedical monitoring.

1.2. Microstructures of Glass, Glass-Ceramics, and Ceramics

Glass-ceramics can be formed from parent glasses via controlled nucleation and crystallization, both of which are induced by appropriate heat-treatment schedules. Depending on the thermal schedule and composition, the microstructure evolves from an amorphous glass to a glass-ceramic and, ultimately, to a more fully crystallized ceramic material [2,3,4,5]. A schematic illustration of the nucleation and crystallization processes in glass during thermal treatment is shown in Figure 1 [36]. During glass-ceramic formation, the crystalline phases, crystal sizes, residual glass composition, and spatial distribution of crystals are strongly governed by the parent-glass composition, nucleating agents, heating rate, holding temperature, and holding time [2,3,4,5,6,7,8].

Figure 1. Schematic comparison of atomic arrangements in (a) an amorphous glass with randomly arranged atoms, (b) a glass-ceramic containing crystalline grains dispersed in a residual glass phase, and (c) a polycrystalline material composed of closely packed crystalline grains. The red and green spheres schematically distinguish two different atomic/ionic species and do not represent specific elements. The gray lines indicate local atomic connectivity, the blue outlines denote crystalline-grain boundaries, and the light-blue regions represent crystalline-grain regions. Reproduced with permission [36]. Copyright 2020, MDPI.

A glass is an amorphous solid lacking translational periodicity in its atomic arrangement (Figure 1a). A glass-ceramic is a two-phase material composed of one or more crystalline species dispersed in an amorphous residual phase (Figure 1b). Within this composite, the crystalline constituents—specifically their size, morphology, orientation, and abundance—dictate the resulting optical, thermal, mechanical, and chemical responses [2,3,6,7]. A fully crystalline ceramic (Figure 1c), by contrast, is essentially an assembly of crystal grains, with only a negligible glassy fraction at the grain boundaries. The ordered unit cells in each grain provide sharp XRD reflections, whereas the random orientation of grains produces a polycrystalline diffraction pattern.

1.3. Scope and Literature-Selection Strategy

This review draws on a focused selection of peer-reviewed studies published between 1959 and 2026. Web of Science, Scopus, and Google Scholar were searched using combinations of keywords including “ytterbium glass-ceramic crystallization,” “Yb₂O₃ glass crystallization,” “YbF₃ oxyfluoride glass-ceramic,” “rare-earth partitioning,” “crystallization kinetics,” “Avrami,” “DSC,” “XRD,” “TEM,” “Raman,” “XANES,” “EXAFS,” “upconversion,” and other composition-specific terms. Primary studies were considered eligible if they reported at least two of the following: parent-glass composition, thermal treatment schedule, crystallization temperature or kinetic parameters, identified crystalline phase, microstructure or grain size, evidence of rare-earth partitioning, transparency, or luminescence. Studies devoted solely to single-crystal growth or ceramic-powder processing were excluded from the core analysis unless they provided essential measurement data on Yb³⁺ energy levels or phase identification. This selection strategy ensures that this review remains focused on the glass-to-glass-ceramic transformation rather than on general ceramic processing. The final body of literature was further enriched with mechanistic investigations into phase separation, interface-controlled growth, optical refrigeration, Yb-mediated phase selection, and optical thermometry [37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61].

This review is organized around composition–processing–structure–property relationships. Table 1 integrates crystallization conditions, kinetic and structural evidence, Yb³⁺ partitioning, optical performance, transparency, thermal stability, and principal reporting gaps, while distinguishing directly measured quantities from interpretations and explicitly marking data that were not reported. This is important because enhanced emission alone does not prove that Yb³⁺ has entered a crystalline phase, and the absence of sharp XRD peaks does not exclude a small fraction of very small or weakly scattering nanocrystals.

Table 1. Comparative crystallization mechanisms, structural evidence, optical performance, and reporting gaps in representative Yb³⁺-containing glass systems. QE: quantum efficiency; PG: precursor glass; GC: glass-ceramic; UC: upconversion; n: the Avrami exponent.

System/Reference	Thermal Schedule	Phase and Size	Kinetic Evidence/Parameters	Yb Role/Optical Response	Transparency, Thermal Stability, and Interpretation
45SiO2-15Al2O3-12Na2O-21BaF2-7YF3; Er3+/Yb3+ [62]	Controlled heat treatment after melt quenching	Cubic BaYF5; ~10 nm domains and ~60–65 nm particles	No Ea or crystalline fraction reported	Preferential RE incorporation is inferred from Stark splitting. Er3+ UC occurs at 523, 541, and 653 nm with an approximately twofold enhancement; lifetime and QE were not reported.	>80% transmittance was retained in the measured range, and stability against further crystallization was reported at the sensing temperature. The optical data support a lower-phonon environment, but direct Yb/Er partition coefficients and crystalline fraction are unavailable.
20Na2O-42ZnO-28P2O5-10B2O3; Yb3+/Ho3+/Tb3+ [63]	Annealed at 420 °C; crystallized at 600 °C for 4 h	NaZnPO4; numerical size not reported	Tg = 457 °C; Tx = 537 °C; Tp = 643 °C	Yb3+ acts mainly as a sensitizer; partitioning is inferred. UC bands occur at 414, 436, 489, 546, 586, 621, and 659 nm; lifetime and QE were not reported.	A transparent GC was obtained, but thermal stability was not quantified. Emission enhancement is clear, whereas crystalline fraction, lifetime, and absolute UC QE remain unreported.
SiO2-Al2O3-CaF2; Tb3+/Yb3+ [64]	660 °C for 2 h	CaF2 nanocrystals; TEM confirmation	Ea ≈ 498 kJ mol−1; n ≈ 1.01	TbF3/YbF3 lowers the crystallization temperature by ~42 °C. Yb-sensitized emission is phase-dependent; lifetime and QE were not consistently reported.	Transparency is retained when the crystals remain nanoscale, but thermal stability was not quantified. The n value is compatible with constrained growth, although morphology and data from multiple heating rates are needed for a firm mechanism assignment.
NaF-CaF2-Al2O3-SiO2; variable Yb2O3 [65]	Heat treatment near the crystallization range	CaF2/Ca0.8Yb0.2F2.2; ~10–20 nm	Apparent Ea decreases with increasing Yb content	Yb3+ substitutes into a CaF2-related phase and produces a phase-dependent optical response; lifetime and QE were not consistently reported.	Transparency can be retained for the reported nanocrystal sizes. This is one of the strongest cases for Yb-assisted fluoride crystallization, but the relationship among crystalline fraction, Yb occupancy, emission efficiency, and long-term stability requires quantitative evaluation.
SiO2-B2O3-PbO-CdO-PbF2-CdF2; Er3+/Yb3+ [66]	683–703 K for 3–30 h	Pb4Yb3F17-related nanocrystals; self-constrained growth	DTA/XRD time series; kinetically self-constrained model	Er-Yb-F-rich clusters act as precursors, with RE enrichment in fluoride regions. Yb NIR and Er3+ Stokes/anti-Stokes emissions are observed; lifetime and QE were not reported.	A stable nanocrystallization window was reported, although short-wavelength scattering increased. Clustering plausibly links phase separation to nucleation, but quantitative crystalline fraction, RE site occupancy, and direct pre-nucleation composition maps remain unavailable.
53SiO2-36Na2O-9Y2O3-2P2O5; Yb3+/Ho3+/Tb3+ [67]	670 °C/1 h + 770 °C/40 min	Na3YSi2O7; ~100–200 nm	Ea = 580–598 kJ mol−1; crystallization index ≈ 1	RE incorporation into Y sites is inferred from luminescence. Tb3+ green emission is enhanced by ~7 times; lifetime and QE were not reported.	Higher crystallinity increases emission but reduces transmittance. Thermal stability was described qualitatively, whereas absolute transmittance at the application wavelength, lifetime, QE, and crystalline fraction were not quantified.
97(0.75NaPO3-0.25CaF2)-3Yb2O3 [68]	320 °C, then 400–450 °C	CaF2 first, then NaPO3 and Na2Ca2(P2O7)F2	Sequential phase formation from XRD/SEM	Yb3+ appears distributed between the glass and CaF2-rich regions. NIR emission is retained after CaF2 formation but decreases when phosphate-rich phases appear; lifetime and QE were not reported.	Transparency and thermal stability were not quantitatively reported. XRD and SEM detect different crystal populations, and direct evidence for Yb distribution among the three phases is still required.
15Yb2O3-25Al2O3-60SiO2 [69]	Complete crystallization at 1200 °C	Yb3Al5O12, Yb2Si2O7, and Al6Si2O13	No kinetic parameters or size distribution reported	Yb is a major crystal-forming constituent; optical emission was not the main focus.	The material has high Tg and good thermal stability but was not designed as a transparent GC after complete crystallization. The optical consequences of the multiphase assemblage remain insufficiently characterized.
Barium–zinc borosilicate and related borate glasses; variable Yb2O3 [70,71,72,73,74]	Melt-quenched glasses; thermal and structural analyses over composition-dependent Yb2O3 contents	No crystalline phase identified in the reported Yb-doped glasses	The undoped borosilicate shows composition-dependent Ea and n; Yb-doped glasses show modest Tg, Tc, and ΔT changes	Yb2O3 modifies packing, thermal response, band gap, and refractive index; lifetime and QE are not applicable or were not reported.	The parent glasses remain XRD-amorphous. These data demonstrate network modification rather than a direct nucleating effect; controlled crystallization experiments are required before assigning Yb2O3 a crystallization-promoting role.
Phase-separated fluorosilicate and oxyfluoride glasses [37,38,39,40,41,42,59]	Heat treatment within the Tg-Tp window; time- and temperature-dependent series	KZnF3/K2SiF6 and BaF2/CaF2-related nanophases; sub-10 to tens of nm	Amorphous phase separation, residual-glass depletion, and self-limited/interface-controlled growth	Local ordering and RE partitioning are system-dependent and only partly quantified. Optical enhancement can arise from changed local symmetry even without complete RE migration into the crystals.	Transparency depends on crystal size, fraction, and refractive-index mismatch. The strongest mechanistic evidence combines NMR, STEM/TEM, chemical mapping, and simulation; optical enhancement alone is insufficient to establish quantitative partitioning.
Yb3+-containing phosphate glasses and glass-ceramics [43,44,49]	Annealing near Tg and Tp across several modifier/intermediate compositions	NaYbP2O7, NaPO3, and Na5P3O10	Additives shift bulk crystallization toward surface crystallization and redirect phase selection	Yb3+ acts as an emitting ion and, in selected compositions, a crystal-forming constituent. The 2F5/2 lifetime and NIR emission bandwidth change after Yb-containing phase precipitation.	Transparency is retained only in selected compositions, and the crystallization window depends strongly on additives. Network connectivity can dominate over Yb concentration; direct Yb distribution and quantitative crystalline fraction remain limited.
Yb3+ glass-ceramics for optical refrigeration [45,46,47,48]	Controlled partial crystallization of bulk samples and fibers	Yb-containing fluoride nanocrystals in oxyfluoride matrices	Cooling performance is non-monotonic with crystallization; complete kinetic models are generally absent	Yb3+ is both the refrigerant ion and a possible participant in the precipitated phase. High or near-unity internal QE is reported for selected compositions, whereas background absorption is critical.	Transparency and cooling depend on nanocrystal size, fraction, scattering loss, OH− content, and temperature. Phase fraction and Yb partition coefficients should be linked quantitatively to cooling efficiency and long-term thermal cycling.
CaF2, KYb3F10, and NaYbF4 systems [50,51,52,53,60,61]	Melt quenching followed by composition-dependent annealing or controlled self-crystallization	CaF2, KYb3F10, NaYbF4, KZnF3, and competing Na3ScF6	Dopant- and composition-controlled phase competition; phase separation may precede crystallization	Yb3+ is a sensitizer and, in KYb3F10/NaYbF4, an essential phase-forming cation. Strong NIR or UC emission is reported; an absolute UC QE of 1.44 ± 0.02% was obtained for one KYb3F10-based system [51].	Transparent samples are obtained only under selected crystallization conditions, and thermal testing generally covers limited windows. Claims of Yb-induced crystallization require matched Yb-free controls, direct partition maps, and long-term optical-stability data.
Thermometric fluoride and oxide glass-ceramics [54,55,56,57,58]	Melt quenching plus annealing or self-crystallization	Ba4Y3F17, Ba3Gd2F12, ZnAl2O4, YOF, and GdOF	Detailed crystallization kinetics are reported only for selected systems	Yb3+ serves mainly as a sensitizer and may also enter the crystalline phase. Enhanced UC and FIR thermometry are reported; lifetime enhancement is available in selected systems, but absolute QE is usually absent.	Transparent GCs and useful sensitivities are reported, but thermal-cycling protocols vary. Pump power density, sample thickness, crystalline fraction, cycling, and absolute QE require standardization.
(60 − x)SiO2–xB2O3–20KF–20ZnF2; 1.0Yb3+/0.2Er3+ [75]	460–580 °C for 10 h	KYb2F7 at lower temperatures; transition to KYb3F10 at higher temperatures; approximately 5–30 nm	Temperature-dependent XRD and TEM/HRTEM	Yb3+ is both sensitizer and phase-forming constituent; Er3+ upconversion bands near 522, 540, and 650 nm increase with heat-treatment temperature.	The system provides direct phase-evolution evidence, but crystalline fraction and rare-earth partition coefficients were not quantified.
Yb3+-doped titanate–germanate and barium titanium borate parent glasses [76,77]	Melt-quenched, as-prepared glasses; no post-crystallization schedule in the cited datasets	No crystalline phase detected by XRD	Composition-dependent absorption, density, molar volume, Raman/FTIR, and amorphous XRD	Yb3+ modifies the glass network and near-infrared absorption; crystalline partitioning cannot be inferred.	These datasets establish parent-glass modification rather than crystallization; matched heat-treatment series are required to test nucleation effects.

Notes: T_g is the glass-transition temperature; T_x is the onset crystallization temperature; T_p is the peak crystallization temperature measured by differential scanning calorimetry (DSC) or differential thermal analysis (DTA); and T_c is the crystallization temperature as defined in the cited source. ΔT is the empirical thermal-stability or crystallization window, generally calculated as T_x − T_g. E_a is the apparent activation energy for crystallization, commonly obtained using non-isothermal Kissinger, Ozawa, Matusita, or related analyses. The Avrami exponent n gives model-dependent information on nucleation behavior and growth dimensionality.

2. Mechanistic Framework for Crystallization in Yb³⁺-Containing Glasses

Crystallization emerges from the interplay between thermodynamic driving force and atomic/ionic mobility. Upon cooling from the melt, the chemical-potential disparity between the supercooled liquid and a potential crystalline phase widens, yet the accompanying viscosity increase progressively hinders structural reconfiguration. For nucleation to proceed, the volumetric free-energy reduction must compensate for the energetic penalty associated with forming a crystal–glass interface. Growth, in turn, becomes kinetically accessible only when the relevant species can migrate through the adjacent glassy medium. As a consequence, the temperature at which nucleation peaks is typically lower than that for maximum growth—a relationship that underpins the common practice of two-step thermal treatments. In multicomponent glasses, phenomena such as liquid–liquid demixing, the clustering of network-modifying cations, or fluoride-enriched zone formation can diminish the effective interface energy and serve as heterogeneous nucleation promoters [1,2,3,4,5,6,7,8].

Yb³⁺ can influence each of these factors. Owing to its high field strength and specific coordination demands, Yb³⁺ may depolymerize one structural network while reinforcing another, alter the abundance of non-bridging oxygen species, modulate fluoride or phosphate bonding environments, and affect viscous flow. When Yb³⁺ partitions into a local region that compositionally prefigures the forthcoming crystalline phase, nucleation tends to be accelerated. Conversely, if Yb³⁺ enhances configurational entropy or stabilizes the residual amorphous framework, crystallization may be slower. The net outcome is system-dependent and must be determined within a well-defined compositional series; generalizations across silicate, phosphate, borate, and oxyfluoride systems are not warranted.

Heating rate introduces additional complexity in comparative studies. Under faster DSC or DTA ramps, the crystallization exotherm shifts upward in temperature because the system dwells for shorter durations within the window where nuclei and crystals can develop. Thus, the peak temperature is not a fixed intrinsic property. Furthermore, a given exotherm may reflect different underlying events, nucleation combined with growth, the growth of pre-existing nuclei, the concurrent formation of multiple phases, or surface-dominated crystallization. Reliable interpretation requires data collected over a range of heating rates, complemented by the phase-resolved characterization of specimens quenched before, at, and beyond each thermal signature [10,11,12,13].

2.1. Kinetic Analysis and Its Limits

The Kissinger method is widely employed to extract an apparent activation energy E_a from crystallization peaks recorded at various linear heating rates β. The relationship is given by

ln(β/T_p²) = −E_a/(RT_p) + C

(1)

where T_p denotes the exotherm maximum, R is the gas constant, and C is a constant [10]. Alternative Ozawa-type treatments use a different functional form of β and can be applied either to peak shifts or to a specified fractional conversion [11]. The term “apparent” is used advisedly, as the derived E_a may reflect contributions from transport processes, interfacial reactions, and evolving phase compositions simultaneously. A reduction in E_a upon Yb addition can be interpreted as enhanced transformability only when the compared specimens follow identical reaction routes and yield the same crystalline product.

The Johnson–Mehl–Avrami–Kolmogorov framework relates the transformed fraction α to time t through

ln[−ln(1 − α)] = n ln(t) + C′

(2)

yielding an Avrami exponent n under stipulated assumptions. In glass research, n is commonly estimated from non-isothermal variants of this formalism [12,13]. Although values of n near unity are often linked to surface-dominated growth or one-dimensional/diffusion-controlled mechanisms, morphology, pre-existing nuclei, and time-dependent nucleation can generate comparable fitted numbers. Hence, inferring a specific mechanism solely from the Avrami exponent is unjustified. Such assignments must be corroborated by microscopy observations, particle-size dependence, and the progression of XRD patterns.

Viscosity measurements offer a crucial link between thermal signatures and the underlying structural behavior. A Vogel–Fulcher–Tammann parametrization may assist in comparing the temperature sensitivity of mobility, though extrapolation beyond the measured range carries considerable uncertainty [78]. Phase-change materials further demonstrate that crystallization pathways can alternate between glass-controlled and undercooled-liquid-controlled regimes [79]. For Yb³⁺-containing glass-ceramics, the most informative design strategy integrates viscosity, nucleation rate, growth kinetics, and phase identification rather than depending on a single thermal stability criterion such as ΔT = T_x − T_g.

The crystalline LiYb(MoO₄)₂ system is retained only as a limited benchmark for phase chemistry, site occupancy, and Yb³⁺ spectroscopy. Because it is prepared by solution-mediated crystal growth or powder-based solid-state synthesis, it does not provide direct evidence for nucleation from an amorphous glass. Figure 2 summarizes its chemically feasible preparation pathway and explicitly defines the conclusions that can and cannot be drawn from this crystalline benchmark [80].

Figure 2. Author-created schematic illustrating the preparation pathway and interpretive scope of LiYb(MoO₄)₂ as a crystalline benchmark. (a) Formation of Li₂MoO₄ and Yb₂(MoO₄)₃ intermediates from oxide/carbonate precursors, followed by their reaction to produce LiYb(MoO₄)₂ and subsequent crystal growth using a Li₂Mo₂O₇ flux, top-seeded solution growth (TSSG), and slow cooling. A simplified polyhedral representation of the LiYb(MoO₄)₂ crystal structure is shown at the bottom. (b) Information that this crystalline benchmark can and cannot provide for interpreting glass-ceramic crystallization. The synthesis pathway was constructed by the authors using information reported in Ref. [80], and all graphical elements were created by the authors.

2.2. LiYb(MoO₄)₂ as a Structural and Spectroscopic Benchmark

LiYb(MoO₄)₂ is a tetragonal double molybdate in which Yb³⁺ occupies a stoichiometric lattice position rather than being present as a dilute dopant. This compound has drawn interest as a potential laser host, owing to the relatively uncomplicated electronic configuration of Yb³⁺, characterized by the ²F_7/2 ground multiplet and the ²F_5/2 excited multiplet. Its absorption features lie predominantly in the 0.9–1.0 μm range, with near-infrared emission centered near 1.0 μm [14,15,16,17,80,81]. The well-defined periodicity of this crystalline lattice serves as a useful reference for validating energy-level assignments and for detecting spectral modifications that arise from rare-earth ions occupying ordered sites. It does not, however, offer direct insight into the influence of Yb³⁺ on nucleation energetics or crystal-growth kinetics within an amorphous matrix.

Because LiYb(MoO₄)₂ undergoes incongruent melting at approximately 912 °C, Volkov and co-workers [80] adopted a flux-based approach using Li₂Mo₂O₇, combined with top-seeded solution growth, rather than conventional Czochralski pulling. As summarized in Figure 2a, Li₂MoO₄ and Yb₂(MoO₄)₃ were first prepared as intermediate compounds from Li₂CO₃, MoO₃, and Yb₂O₃. These intermediates were subsequently reacted to form LiYb(MoO₄)₂, which was followed by crystal growth using a Li₂Mo₂O₇ flux and slow cooling. Crystals approximately 12 mm in size were obtained by initiating growth just below the decomposition temperature. The simplified polyhedral representation at the bottom of Figure 2a illustrates the ordered crystalline structure of LiYb(MoO₄)₂. This route represents solution-mediated single-crystal growth rather than glass devitrification and is included only to establish a chemically feasible formation pathway and an ordered structural reference for Yb³⁺.

The oxidation-state balance of the precursor chemistry also requires clarification. In LiYb(MoO₄)₂, Mo is present as Mo⁶⁺ in the [MoO₄]²⁻ group; therefore, MoO₃ is the direct oxidation-state-matched precursor used in the preparation route summarized in Figure 2a [80]. If MoO₂ is employed in an alternative preparation route, the oxidation of Mo⁴⁺ to Mo⁶⁺ during heat treatment must occur and should be explicitly considered when interpreting the phase-formation mechanism. More generally, crystalline phases assigned in heat-treated glasses should be chemically compatible with the original batch composition and independently verified by phase-sensitive characterization rather than inferred from optical spectra alone.

Figure 2b clarifies the interpretive scope of crystalline LiYb(MoO₄)₂ in the present review. This crystalline benchmark provides useful reference information for assessing chemically feasible phase formation, crystalline-phase assignments, ordered Yb³⁺ site occupancy, and structural and spectroscopic characteristics. However, it does not directly demonstrate nucleation from an amorphous glass, nucleation or crystal-growth kinetics, surface versus bulk crystallization, or Yb³⁺ partitioning during glass devitrification.

2.3. Powder-Derived Reference Materials: Relevance and Limitations

Powder-derived LiYb_1−xLn_x(MoO₄)₂ has also been prepared by milling stoichiometric Li₂CO₃, MoO₃, Yb₂O₃, and Ln₂O₃, followed by staged calcination and sintering in air [82]. In this route, precursor purity, particle size, mixing, temperature, and atmosphere determine the phase purity of the final crystalline material [83,84]. These processing variables are relevant to solid-state synthesis but do not directly explain nucleation within a glass. Accordingly, general descriptions of press types, green-body mechanics, calcination, and ceramic sintering are not considered further in this review. Such a synthesis method is illustrated in Figure 3.

Figure 3. Schematic condensed preparation route for Ln³⁺-substituted LiYb(MoO₄)₂ crystals, where Ln denotes lanthanides such as Er and Eu. Green arrows indicate the processing sequence; the green region denotes the liquid milling medium; red-shaded regions indicate drying or heat treatment in air; red and black spheres represent powder particles and granules, respectively; the blue region represents uniaxial pressing; and PVB denotes the poly(vinyl butyral) binder used during granulation.

The above synthesis route was also used to prepare LiYb_1-xEr_x (MoO₄)₂ (0.01 ≤ x ≤ 1) crystals by the present authors [82], where the starting materials of Er₂O₃, Yb₂O₃, and 99.9% Li₂CO₃ and MoO₂ were weighted by an appropriate stoichiometric ratio, thoroughly mixed in an agate mortar, and transferred to a corundum crucible to sinter. The mixture powder was first pre-fired at 300 °C for 5 h and then at 500 °C for 5 h and finally heated at 800 °C for 8 h in air for preparation of the final products of the phosphors. During every step of the heat treatment, grinding and homogeneous mixing were employed.

The above powder-derived material method provides excellent crystallization formation, which is confirmed by the crystalline phase diagram measurement. XRD can verify the final crystalline phase, while microscopy and photoluminescence characterize morphology and optical response. For a glass-ceramic, however, these measurements alone cannot determine when nuclei first formed, whether crystallization occurred at the surface or throughout the volume, what fraction of the glass crystallized, or what proportion of Yb³⁺ entered the crystalline phase. The crystalline benchmark should therefore support phase and spectroscopic assignments but should not substitute for kinetic measurements or the direct analysis of Yb³⁺ partitioning.

2.4. Substitutional Chemistry and Its Relevance to Glass-Ceramics

Substitution of Ln³⁺ for Yb³⁺ in LiYb_1−xLn_x(MoO₄)₂ is facilitated by their identical charge and comparable coordination-dependent ionic radii. In LiYb_1−xEr_x(MoO₄)₂, the tetragonal structure is retained over the investigated composition range, and the lattice parameters vary approximately with Er³⁺ content, consistent with the substitution of Er³⁺ at the Yb³⁺ site [82,85,86]. Figure 4 summarizes the corresponding XRD patterns and composition-dependent unit-cell parameters.

Figure 4. (a) XRD patterns of LiYb_1–xEr_x(MoO₄)₂ (x = 0.01–1.00), with the reference pattern; (b) variation in the unit-cell parameters as a function of x. Panels (a,b) are reproduced with permission from Ref. [82]. Copyright 2024, Elsevier B.V. (c) Schematic representation of the crystal structure of LiYb(MoO₄)₂, created by the authors. The structure is illustrated by taking into account the oxygen coordination geometries of the metal sites occupied by Yb, Mo, and Li ions.

This substitutional behavior is relevant to glass-ceramics because crystal-field splitting, emission line shapes, and lifetime components may indicate whether Yb³⁺ or a codoped activator has entered an ordered crystalline environment. Nevertheless, optical evidence remains indirect: similar spectra may arise from different local coordination environments, and an approximately linear change in lattice parameters does not quantify the fraction of rare-earth ions remaining in the residual glass. Direct or quantitative methods—including EXAFS/XANES, site-selective spectroscopy, spatially resolved elemental analysis, and multicomponent lifetime analysis—are required to determine rare-earth partitioning reliably. The LiYb(MoO₄)₂ system is therefore used only as a structural and spectroscopic reference and is not generalized as a model for the thermodynamics or kinetics of glass crystallization.

3. Optical Processes Relevant to Yb³⁺-Containing Glass-Ceramics

3.1. Upconverted Emission of Yb³⁺-Doped Crystals

3.1.1. Cooperative Luminescence (CL)

A broad emission feature centered at ~497 nm (20,120 cm⁻¹) was first observed in YbPO₄ by Nakazawa and Shionoya [87] in 1970. This signal appeared under near-infrared pumping near 1000 nm at ambient temperature, with the band spanning from ~480 nm to ~520 nm on its high- and low-energy sides, respectively. The emission energy is approximately twice the excitation energy, and its intensity increases nearly quadratically with pump power. This behavior was assigned to cooperative luminescence (CL), which originates from a pair of neighboring Yb³⁺ ions through a two-photon upconversion process [18,19,87]. This cooperative process follows the reaction scheme:

Yb³⁺(²F_5/2) + Yb³⁺(²F_5/2) → Yb³⁺(²F_7/2) + Yb³⁺(²F_7/2) + CL

(3)

This mechanism is thought to proceed through an intermediate [²F_5/2 ²F_5/2] state, which arises from electronic coupling between two neighboring Yb³⁺ ions—one in the ²F_5/2 excited level and the other also in the ²F_5/2 state.

Since its initial observation in YbPO₄, a broad cooperative-luminescence band near 500 nm has been reported in numerous Yb³⁺-containing crystals, glasses, ceramics, and solutions [18,19,20,87,88]. In Yb³⁺-doped phosphate glasses, the probability of cooperative luminescence increases as the average Yb³⁺–Yb³⁺ separation decreases, with pronounced cooperative emission reported when the separation is shorter than approximately 0.4 nm [88]. As schematically illustrated in Figure 5a, the shorter inter-ionic distance enhances electronic coupling between neighboring excited Yb³⁺ ions, whereas a larger separation suppresses the cooperative process. The resulting emission is characterized by a broad band centered near 500 nm. Narrow emission features associated with intentionally or unintentionally incorporated rare-earth ions may be superimposed on this band, as illustrated schematically in Figure 5b. Representative Tm³⁺-related emissions occur near 470, 474, and 480 nm, whereas Er³⁺-related emissions have been reported near 521, 525, and 528 nm [87,88].

Figure 5. Author-created schematic illustrating the influence of Yb³⁺–Yb³⁺ separation and rare-earth impurities on cooperative luminescence. (a) Under 975 nm near-infrared excitation, a short Yb³⁺–Yb³⁺ separation (R ≲ 0.4 nm) promotes strong electronic coupling between neighboring excited Yb³⁺ ions and enhances cooperative luminescence near 500 nm, whereas a larger separation weakens the electronic coupling and suppresses the cooperative process. The asterisk (*) in “Yb³⁺” denotes an Yb³⁺ ion in its excited state. (b) Schematic spectral representation of the broad Yb³⁺ cooperative-luminescence band centered near 500 nm, with representative narrow Tm³⁺- and Er³⁺-related emission features superimposed. The spectrum is provided for conceptual illustration and does not represent digitized experimental data. The scientific concepts and representative wavelength assignments are based on Refs. [87,88], and all graphical elements were created by the authors.

A simplified model for the CL mechanism is shown in Figure 6 [18,19,87,88]. Two neighboring Yb³⁺ ions, Yb³⁺(1) and Yb³⁺(2), are simultaneously excited to the ²F_5/2 state under pumping at 10,590 cm⁻¹. The donor Yb³⁺(1) relaxes non-radiatively to the ground state and passes its energy to the adjacent Yb³⁺(2), which is already in the excited ²F_5/2 state, promoting it to an intermediate Yb³⁺–Yb³⁺ dimer state denoted as [²F_5/2 ²F_5/2] at approximately 21,180 cm⁻¹ (= 2 × 10,590 cm⁻¹). This cross-relaxation-type energy-transfer process is followed by radiative emission from the dimer state, producing CL.

Figure 6. Schematic energy-level diagram illustrating cooperative luminescence (CL) from a Yb³⁺ dimer with a short Yb³⁺–Yb³⁺ separation. Under 944 nm excitation, cross-relaxation (CR) occurs between two neighboring Yb³⁺ ions, denoted Yb³⁺(1) and Yb³⁺(2), and CL is assumed to originate from Yb³⁺(2). Labels such as (1) denote crystal-field-split Stark substates of the ²F_7/2 ground state, whereas labels such as (2′) denote substates of the ²F_5/2 excited state. The corresponding energy gaps are labeled in cm⁻¹.

3.1.2. Yb³⁺-Sensitized Rare-Earth Emission

In addition to the broad cooperative-luminescence band near 500 nm, narrow upconversion-emission features may appear when trivalent lanthanide ions such as Er³⁺, Ho³⁺, Tb³⁺, or Tm³⁺ are intentionally or unintentionally codoped into Yb³⁺-containing compounds, as schematically illustrated in Figure 7 [18,19,20,21,22,23,24,25,31,32,33,34,63,82,87,88]. Each upconversion emission typically exhibits a power-law relationship with the excitation power, I ∝ Pⁿ, where P denotes the near-infrared pump intensity and the exponent n corresponds to the number of excitation quanta required to populate the emitting level [18,19,63]. Values of n between 1 and 2 usually indicate two-photon processes affected by saturation or cross-relaxation, whereas values approaching 3 indicate three-owhoton processes. The observed emissions are generally attributed to cooperative sensitization or to stepwise energy migration from excited Yb³⁺ pairs to Ln³⁺ acceptor ions. The reduction in CL in codoped systems suggests that energy transfer to Ln³⁺ ions competes efficiently with radiative CL from Yb³⁺ pairs [18,19,20,21,22,23,24,25,31,32,33,34].

Figure 7. Upconversion spectrum of LiYb_0.98Tm_0.01Er_0.01(MoO₄)₂ under 975 nm excitation, showing rare-earth emission superimposed on part of the broad Yb³⁺ cooperative-luminescence band. Reproduced with permission from Ref. [82]. Copyright 2024, Elsevier B.V.

3.2. Downconverted Yb³⁺ Emission Band

Yb³⁺ has the outer-electron configuration 4f¹³5s²5p⁶, with a ²F_7/2 ground state and only one excited multiplet, ²F_5/2 [14,15,16,17]. The shielding provided by the filled 5s² and 5p⁶ subshells renders the 4f electrons largely insensitive to the host matrix, except for the Stark splitting that emerges in non-centrosymmetric crystal fields. At low temperature, the ²F_5/2 → ²F_7/2 emission and ²F_7/2 → ²F_5/2 absorption spectra can show several sharp Stark components [14,82,89]. At room temperature, however, vibronic coupling and thermal broadening often merge these components into a broad near-infrared emission band around 1000 nm, as illustrated for LiYb(MoO₄)₂-based crystals in Figure 8 [82,89,90].

Figure 8. Cooperative-luminescence spectrum from LiYb_0.695Eu_0.30Ho_0.005(MoO₄)₂ compared on an energy scale with its near-infrared Yb^{3+ 2}F_5/2 → ²F_7/2 emission. A weak Ho³⁺ band is also present.

3.3. Stokes and Anti-Stokes Luminescence Bands in Er³⁺/Yb³⁺-Doped Oxyfluoride Glass-Ceramics

Representative Stokes and anti-Stokes emissions from Er³⁺/Yb³⁺-doped oxyfluoride glass-ceramics are shown in Figure 9. The spectra were obtained from NaGdF₄-containing precursor glasses and glass-ceramics under 974 nm excitation [91]. In the visible region, the characteristic Er³⁺ upconversion emissions are assigned to the ²H_11/2, ⁴S_3/2 → ⁴I_15/2 green transitions and the ⁴F_9/2 → ⁴I_15/2 red transition. In the near-infrared region, the emission features are associated mainly with the Yb^{3+ 2}F_5/2 → ²F_7/2 transition near 1.0 μm and the Er^{3+ 4}I_13/2 → ⁴I_15/2 transition around 1.5 μm. Compared with the precursor glass, the heat-treated glass-ceramic exhibits enhanced and better-resolved emission features, which supports the incorporation of Er³⁺ and Yb³⁺ ions into the low-phonon-energy NaGdF₄ nanocrystals.

Figure 9. Representative upconversion and near-infrared emissions of Er³⁺/Yb³⁺-doped NaGdF₄-containing oxyfluoride glass-ceramics. (a) Visible upconversion emission spectra of Er³⁺-doped and Er³⁺/Yb³⁺-codoped 70Si7Gd precursor glasses and glass-ceramics heat-treated at 550 °C for 80 h under 974 nm excitation. (b) Near-infrared emission spectra of the Er³⁺/Yb³⁺-codoped precursor glass and corresponding glass-ceramic under 974 nm excitation, showing the Yb^{3+ 2}F_5/2 → ²F_7/2 emission near 1.0 μm and the Er^{3+ 4}I_13/2 → ⁴I_15/2 emission around 1.5 μm. Adapted from Ref. [91] under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0). The original panels were reformatted and rearranged for presentation without alteration of the underlying spectral data.

Analogous simultaneous Stokes and anti-Stokes luminescence was reported for Er³⁺/Yb³⁺-codoped Pb₄Yb₃F₁₇-containing oxyfluoride nanoglass-ceramics under 940 nm excitation [66]. Although the corresponding experimental spectrum is not reproduced here, the observation of Er³⁺ emissions under the preferential excitation of Yb³⁺ provides evidence for efficient Yb³⁺ → Er³⁺ energy transfer in this system.

The observation of Er³⁺ emission under excitation near the Yb³⁺ absorption band indicates efficient Yb³⁺ → Er³⁺ energy transfer. The initial resonant energy-transfer step can be expressed as ⁴I_15/2 (Er³⁺)+ ²F_5/2 (Yb³⁺)→⁴I_11/2 (Er³⁺)+ ²F_7/2 (Yb³⁺). A subsequent energy-transfer-upconversion step, ⁴I_11/2 (Er³⁺)+ ²F_5/2 (Yb³⁺)→⁴F_7/2 (Er³⁺)+ ²F_7/2 (Yb³⁺), populates the higher-lying Er³⁺ manifold. Multiphonon relaxation from ⁴F_7/2 then populates the ²H_11/2, ⁴S_3/2, and ⁴F_9/2 levels, giving rise to the characteristic green and red upconversion emissions. These principal excitation, energy-transfer, relaxation, and emission pathways are summarized in the author-created energy-level scheme shown in Figure 10.

Figure 10. Author-created simplified energy-level scheme illustrating Yb³⁺-sensitized Er³⁺ upconversion under excitation within the Yb³⁺ absorption band. The diagram shows Yb³⁺ excitation, Yb³⁺ → Er³⁺ energy transfer to the ⁴I_11/2 level, sequential energy-transfer upconversion to the ⁴F_7/2 and higher-lying Er³⁺ levels, multiphonon relaxation, and the principal visible Er³⁺ emissions. The near-resonant Er³⁺–Er³⁺ cross-relaxation process ⁴S_3/2 + ⁴I_9/2 → ⁴F_9/2 + ⁴F_9/2 increases the population of the red-emitting ⁴F_9/2 level [41].

Figure 10 presents an author-created simplified energy-level scheme for Yb³⁺-sensitized Er³⁺ upconversion in oxyfluoride glasses and glass-ceramics. The relevant Er³⁺ levels are located at approximately 10,000–10,500 cm⁻¹ for 4I_11/2, 12,500 cm⁻¹ for ⁴I_9/2, 15,200 cm⁻¹ for ⁴F_9/2, 18,300 cm⁻¹ for ⁴S_3/2, and 20,400 cm⁻¹ for ⁴F_7/2. In addition to sequential Yb³⁺ → Er³⁺ energy transfer, Er³⁺–Er³⁺ cross-relaxation can influence the relative intensities of the green and red emissions. In particular, the near-resonant process ⁴S_3/2(Er³⁺) + ⁴I_9/2(Er³⁺) → ⁴F_9/2(Er³⁺) + ⁴F_9/2(Er³⁺) increases the population of the red-emitting ⁴F_9/2 level. This process can therefore contribute to an increase in the red-to-green intensity ratio, R_r/g.

In the Pb₄Yb₃F₁₇-containing oxyfluoride nanoglass-ceramic investigated in Ref. [66], a weak upconversion emission was additionally observed at approximately 480 nm and was assigned to the Er^{3+ 4}F_7/2 → ⁴I_15/2 transition. The ⁴F_7/2 level can be populated through a two-step Yb³⁺-sensitized process: an initial Yb³⁺ → Er³⁺ energy-transfer step populates the Er^{3+ 4}I_11/2 level, followed by a second energy-transfer-upconversion step to ⁴F_7/2. The emission near 407 nm was assigned to the ²H_9/2 → ⁴I_15/2 transition. Population of ²H_9/2 may involve an additional Yb³⁺-sensitized upconversion step to a higher-lying Er³⁺ level, such as ²P_3/2, followed by multiphonon relaxation to ²H_9/2. These proposed pathways are included in the generalized mechanism shown in Figure 10.

4. Design and Characterization of Yb³⁺-Containing Glass-Ceramics

4.1. Nucleation, Phase Fraction, and Optical Transparency

The objective is to achieve a controlled population of crystals dispersed in a continuous glassy matrix—not merely to detect an XRD signal. Scattering behavior is governed by four interrelated factors: crystalline fraction, grain-size distribution, spatial uniformity, and refractive-index mismatch. When crystallites are significantly smaller than the wavelength of light, high transparency can be retained; however, aggregation or a wide size distribution may still introduce optical losses. Consequently, heat-treatment protocols must be optimized with respect to at least two performance metrics—crystallization and transparency—rather than against a single parameter such as peak intensity [92,93,94,95,96,97,98,99].

Rare-earth partitioning introduces a third consideration. Luminescence may be enhanced when Yb³⁺ and an activator ion are incorporated into a low-phonon fluoride nanocrystal, yet the high local concentration within such a confined phase can also promote cross-relaxation or clustering. Meanwhile, the composition of the residual glass evolves as crystallization proceeds, altering its viscosity, refractive index, and rare-earth solubility over the course of the same thermal schedule. These coupled effects explain why the most prolonged or highest-temperature anneal is rarely the optimum choice.

The crystalline fraction should ideally be quantified via Rietveld refinement using an internal standard or through a calibrated complementary technique. The Scherrer estimate derived from XRD peak broadening represents the coherent domain size, which does not necessarily correspond to the actual particle diameter observed in SEM. Reporting both quantities—along with the instrument-broadening correction and a size distribution extracted from imaging—helps prevent apparent discrepancies from being misinterpreted as evidence of a novel physical effect.

4.2. Capabilities and Limitations of Characterization Methods

DSC/DTA identifies thermal events and supports kinetic analysis, but overlapping exotherms and baseline choices can obscure multiple phases. XRD is phase-specific for sufficiently ordered and abundant crystals, yet very small, strained, or low-fraction nanocrystals may produce peaks too broad or weak for confident detection. TEM with selected-area electron diffraction can reveal such nanocrystals and distinguish surface from volume populations, although its field of view may not represent the bulk. Raman spectroscopy is sensitive to short-range network reorganization and can detect vibrational signatures unavailable from XRD, but band overlap and fluorescence backgrounds complicate quantification. A combined XRD–TEM–Raman strategy has directly demonstrated heat-treatment-induced NaYF₄ nanocrystals and associated network changes in Ho³⁺/Yb³⁺ oxyfluoride glass-ceramics [100].

XANES can test oxidation state and near-edge coordination changes, whereas EXAFS yields average neighbor identities, distances, and coordination numbers around Yb. These methods are particularly valuable when Yb-rich clusters or nanophases are below the XRD detection limit; EXAFS has resolved the local environment and phosphate association of Yb³⁺ in phase-separated silica-based preforms [101]. Their limitation is ensemble averaging, so coexisting Yb sites may require model comparison or complementary spectroscopy. Photoluminescence, lifetime, and quantum-yield measurements probe the functional consequence of site changes, but optical enhancement alone is not proof of crystal incorporation. The strongest mechanistic claim is obtained when thermal, diffraction, imaging, local-structure, and optical evidence converge on the same sequence.

5. Representative Ln³⁺-Yb³⁺ Codoped Glass-Ceramics

5.1. Er³⁺- Yb³⁺ Codoped Fluoro-Aluminosilicate Glass-Ceramics

In 45SiO₂-15Al₂O₃-12Na₂O-21BaF₂-7YF₃-0.5ErF₃-1YbF₃ glass codoped with Er³⁺ and Yb³⁺, controlled heat treatment produces cubic BaYF₅ nanocrystals [62]. That study is instructive because it connects phase formation, microstructure, transmission, and upconversion. Yb³⁺ absorbs near 980 nm and sensitizes Er³⁺, while the fluoride nanophase provides a lower-phonon environment than the oxide-rich matrix.

XRD establishes the crystalline phase, and transmission above 80% indicates that the heat-treated microstructure remains optically homogeneous on the wavelength scale (Figure 11). The transmittance spectra of both as-produced and heat-treated GC samples (measured over 275–1000 nm) provide complementary information to the Raman data. In Figure 11b, absorption features are observed at approximately 378, 487, 523, 541, 653, and 976 nm. The first five of these correspond to Er³⁺ transitions from the ⁴I_15/2 ground state to the ⁴G_11/2, ⁴F_7/2, ²H_11/2, ⁴S_3/2, ⁴F_9/2 excited levels, while the 976 nm line arises from the Yb^{3+ 2}F_7/2 → ²F_5/2 transition. Notably, no absorption feature attributable to the BaYF₅ crystalline phase is detected. This absence is consistent with the small dimensions of the precipitated BaYF₅ nanocrystals—approximately 10 nm as shown in Figure 12B—which are substantially smaller than the wavelengths of visible light. This size regime explains why both the as-produced and heat-treated GC samples maintain high transmittance (>80%). By contrast, the presence of larger BaYF₅ crystallites would be expected to introduce significant scattering losses and thereby reduce transparency.

Figure 11. (a) XRD patterns of as-produced and heat-treated glass-ceramics with the cubic BaYF₅ reference; (b) corresponding transmission spectra. Reproduced with permission from Ref. [62]. Copyright 2021, John Wiley and Sons.

Figure 12. SEM micrographs of the as-produced (A) and heat-treated (B) Er³⁺–Yb³⁺ fluoro-aluminosilicate glass-ceramics. Reproduced with permission from Ref. [62]. Copyright 2021, John Wiley and Sons.

Electron microscopy showed approximately 25 nm domains in the as-produced material and particles near 60–65 nm after heat treatment, together with smaller newly precipitated domains (Figure 12) [62]. The XRD coherent-domain estimate and the image-based particle size are similar in scale but should not be treated as identical quantities. The coexistence of sizes suggests that nucleation continued while earlier particles grew.

This observation is mechanistically more informative than a single average size. Continued nucleation broadens the population and can preserve transparency if both populations remain nanoscale; excessive growth would increase scattering. Rare-earth partitioning into BaYF₅ was inferred from the enhanced emission and crystal-field features, but a quantitative Yb³⁺ or Er³⁺ partition coefficient was not reported.

The case therefore supports a coupled design rule: annealing should maximize the fraction of rare-earth ions in the desired nanophase before particle coarsening compromises transmission. Quantitative phase fraction, lifetime, and absolute quantum efficiency would be needed to compare this optimum with other glass families.

UC Luminescent Behavior

One advantage of controlled glass crystallization is that it can place rare-earth ions into low-phonon-energy crystalline environments while maintaining the mechanical stability and processability of the glass matrix [92,93,94,95,96,97,98,99]. In Er³⁺–Yb³⁺ codoped systems, Yb³⁺ acts as a sensitizer under 940–980 nm excitation and transfers energy to Er³⁺, leading to population of the ⁴I₁₁/₂, ⁴F_7/2, ²H_11/2, ⁴S_3/2, and ⁴F_9/2 levels. The resulting green and red emissions are strongly affected by crystallite size, rare-earth partitioning, phonon energy, and cross-relaxation pathways [18,19,20,34,62,92,93,94,95,96,97].

Figure 13A presents the upconversion (UC) emission spectra of Er³⁺–Yb³⁺ codoped GC samples—both as-produced and heat-treated—under 980 nm excitation, recorded over the 500–700 nm range. Three characteristic emission peaks appear at 523 nm (²H_11/2 → ⁴I_15/2), 541 nm (⁴S_3/2 → ⁴I_15/2), and 653 nm (⁴F_9/2 → ⁴I_15/2). A notable feature is the distinct Stark splitting observed in both samples, particularly the resolved doublet at ~541 nm and ~549 nm within the ⁴S_3/2 → ⁴I_15/2 band. This splitting originates from the crystal-field-induced lifting of degeneracy in the ⁴I₁₅/₂ ground state, given that the emitting level is the lowest relaxed component of the ⁴S_3/2 manifold even at room temperature.

Figure 13. (A) Upconversion spectra of as-produced and heat-treated glass-ceramics; (B) pump-power dependence for the heat-treated sample; (C) proposed Er³⁺–Yb³⁺ energy-transfer pathways under 980 nm excitation. Reproduced with permission from Ref. [62]. Copyright 2021, John Wiley and Sons.

The heat-treated GC exhibits approximately twice the UC intensity of its as-produced counterpart. This enhancement is attributed to the preferential incorporation of Er³⁺ ions into the BaYF₅ fluoride nanocrystals, where the lower phonon energy environment suppresses non-radiative multiphonon relaxation. The UC emission intensity (IUC) generally scales with excitation power (P) according to IUC ∝ Pⁿ, where n denotes the number of 980 nm photons required to populate the emitting states (²H_11/2, ⁴S_3/2, or ⁴F_9/2). Because n ≥ 2, the process corresponds to anti-Stokes emission. Figure 13B shows the logarithmic plot of IUC versus P; the fitted slopes are 1.95 for the 541 nm band and 1.88 for the 653 nm band, confirming that a two-photon mechanism dominates the UC process.

The underlying energy-transfer scheme is illustrated in Figure 13C. Upon 980 nm excitation, Yb³⁺ ions are initially promoted from the ²F_7/2 ground state to the ²F_5/2 excited level via ground-state absorption (GSA). These excited Yb³⁺ ions then relax non-radiatively, transferring energy to adjacent Er³⁺ ions and promoting them from ⁴I_15/2 to ⁴I_11/2 through a resonant ET1 process: ⁴I_15/2 (Er³⁺) + ²F_5/2 (Yb³⁺) → ⁴I_11/2 (Er³⁺) + ²F_7/2 (Yb³⁺). In parallel, Er³⁺ ions can also undergo direct GSA from ⁴I₁₅/₂ to ⁴I₁₁/₂. A subsequent ET2 step—⁴I_11/2 (Er³⁺) + ²F_5/2 (Yb³⁺) → ⁴F_7/2 (Er³⁺) + ²F_7/2 (Yb³⁺)—drives the system to the ⁴F_7/2 level. From there, non-radiative relaxation populates the ²H_11/2 and ⁴S_3/2 emissive states, which then decay radiatively to ⁴I_15/2, producing the 523 and 541 nm emissions. The 653 nm red emission (⁴F_9/2 → ⁴I_15/2) can be populated via two pathways: (i) non-radiative relaxation from ⁴S_3/2, or (ii) an ET3 process—⁴I_13/2 (Er³⁺) + ²F_5/2 (Yb³⁺) → ⁴F_9/2 (Er³⁺) + ²F_7/2 (Yb³⁺)—following non-radiative decay from ⁴I_11/2.

To summarize the UC behavior: the heat-treated GC shows a twofold enhancement over the as-produced GC, attributable to the more effective partitioning of both Er³⁺ and Yb³⁺ into the low-phonon BaYF₅ environment. Notably, the heat-treated material remains stable against further crystallization, ensuring stable UC performance at operating temperatures, and exhibits promising temperature-sensing characteristics.

Attribution of the upconverted 523, 541, and 653 nm emissions to the two-photon excitation process under 980 nm LD excitation is also explained as follows. The 980 nm LD irradiation to the heat-treated GC sample of 45SiO₂-15Al₂O₃ -12Na₂O-21BaF₂-7YF₃-0.5ErF₃-1YbF₃ gives rise to the excitation of not only the Yb³⁺ of YbF₃ but also the Er³⁺ of ErF₃ because the two ions have a gap energy of ~10,206 cm⁻¹ (corresponding to 980 nm). For instance, the energy separation between the ⁴I_11/2 and ⁴I_15/2 levels of Er³⁺ is approximately 10,000 cm⁻¹, which matches the energy of a 980 nm photon and thus allows excitation from the ground state to the ⁴I_11/2 manifold. From this intermediate level, a second 980 nm photon can promote the ion further to the ⁴F_7/2 state, as the ⁴F_7/2-⁴I_11/2 gap (~11,000 cm⁻¹) is also resonant with the excitation wavelength. Consequently, Er³⁺ can be raised to the ⁴F_7/2 level through a two-photon absorption sequence. Subsequent non-radiative multiphonon relaxation then populates the lower-lying ²H_11/2, ⁴S_3/2, and ⁴F_9/2 emitting states, which give rise to the 523, 541, and 653 nm emissions, respectively (Figure 13C).

An alternative two-photon pathway involves the initial excitation of Yb³⁺ from ²F_7/2 to ²F_5/2 by a 980 nm photon. The excited Yb³⁺ ion subsequently undergoes spontaneous ²F_5/2→²F_7/2 emission at ~980 nm, and this emitted photon can then be reabsorbed by an Er³⁺ ion to promote it from ⁴I_11/2 to ⁴F_7/2. This means that two 980 nm photons are used to excite into the ⁴F_7/2 state, followed by the generation of the 523, 541, and 653 nm emissions finally.

5.2. Tb³⁺–Ho³⁺–Yb³⁺-Codoped Glass-Ceramics

Phosphate glasses with compositions (in mol %) consisting of seven constituents of 20Na₂O-42ZnO-28P₂O₅-10B₂O₃-2.0Yb₂O₃-0.15Ho₂O₃-xTb₄O₇ (with x = 0.6, 0.8, 1.0, 1.2) were melted, annealed, and subsequently heat-treated at 600 °C to form a transparent NaZnPO₄-containing glass-ceramic (GC₆₀₀) [63]. The reported thermal markers were T_g ≈ 457 °C, onset T_x ≈ 537 °C, and a crystallization peak near 643 °C. These values define a processing window, but they do not alone identify whether nucleation began at T_g or quantify the crystalline fraction. The reported XRD results distinguish the predominantly amorphous precursor glass (PG) from the NaZnPO₄-containing GC₆₀₀ product [63]; the corresponding heat-treatment-induced phase evolution is summarized schematically in Figure 14. Yb³⁺ supplies near-infrared sensitization, while Ho³⁺ and Tb³⁺ provide multiple visible emitting levels. The system therefore tests how crystallization changes a multi-ion transfer network, but concentration quenching and cross-relaxation must be separated from the effect of the crystalline host.

Figure 14. Author-created schematic summarizing the heat-treatment-induced conversion of the 20Na₂O-42ZnO-28P₂O₅-10B₂O₃-2.0Yb₂O₃-0.15Ho₂O₃-1.0Tb₄O₇ precursor glass (PG) into the NaZnPO₄-containing glass-ceramic (GC₆₀₀). The thermal parameters and phase assignment are summarized from Refs. [38,63].

The crystallization behavior and upconversion optical properties of NaZnPO₄:Yb³⁺/Tb³⁺/Ho³⁺ nanocrystals embedded in a glass matrix have been investigated in multicomponent systems [63]. These studies are important because Tb³⁺ and Ho³⁺ introduce additional green, red, and near-infrared emitting levels, while Yb³⁺ provides efficient sensitization under near-infrared excitation. Such multi-ion systems allow emission color tuning through energy transfer, cooperative sensitization, and cross-relaxation processes [18,19,20,31,32,33,34,63,67].

5.2.1. Crystallization and Upconversion Spectra

The PG samples with nominal composition 20Na₂O-42ZnO-28P₂O₅-10B₂O₃-2.0Yb₂O₃-0.15Ho₂O₃-xTb₄O₇ with x = 0.6, 0.8, 1.0 and 1.2 mol% were excited by a 980 nm laser. The principal upconversion emission bands and their transition assignments are summarized in Figure 15a. Five intense bands are observed at 489, 546, 586, 621, and 659 nm. The first four are assigned to Tb³⁺ transitions (⁵D₄→⁷F₆, ⁷F₅, ⁷F₄, and ⁷F₃), while the 659 nm band corresponds to the ⁵F₅→⁵I₈ transition of Ho³⁺. Two weaker violet emissions at 414 and 436 nm are also attributable to Tb³⁺ [63].

Figure 15. Author-created summary of the upconversion behavior of Yb³⁺/Tb³⁺/Ho³⁺-doped precursor glasses and the corresponding GC₆₀₀ glass-ceramic under 980 nm excitation: (a) principal Tb³⁺ and Ho³⁺ emission bands and their transition assignments; (b) qualitative effects of Tb₄O₇ content and crystallization, showing the maximum overall UC response at x = 1.0 mol%, concentration quenching at x = 1.2 mol%, weakening of the Ho³⁺ 659 nm emission with increasing x, and enhanced emission following the conversion of PG into GC₆₀₀. The wavelengths, assignments, and qualitative trends are summarized from Ref. [63].

Among the four nominal Tb₄O₇ contents investigated, the strongest UC emission is observed for the sample with x = 1.0 mol%. Beyond this level, concentration quenching sets in and the Tb³⁺-related emission intensity declines. Concurrently, the Ho³⁺ emission intensity progressively decreases as the Tb³⁺ content rises from 0.6 to 1.2 mol%, which suggests a Ho³⁺ → Tb³⁺ energy-transfer pathway [63]. These qualitative composition-dependent trends are summarized in Figure 15b.

In comparison with the PG precursor, the GC₆₀₀ sample derived from the same composition (x = 1.0 mol%) exhibits a substantially enhanced UC emission, as summarized in Figure 15b. This enhancement was attributed to the crystallization of NaZnPO₄ nanocrystals with low phonon energy, which effectively suppresses non-radiative multiphonon relaxation and thereby promotes radiative transitions [63].

5.2.2. Two-Photon Excitation Process for the UC Emission

The log–log-plotted UC emission intensities (IUC) vs. pump power (P) in the GC₆₀₀ sample suggest that the slopes are 1.57 for the 546 nm green band (Tb³⁺: ⁵D₄ → ⁷F₅) and 1.42 for the 659 nm red band (Ho³⁺: ⁵F₅ → ⁵I₈). The log–log plot indicates that IUC is proportional to Pⁿ with n = 2, where n stands for the number of photons necessary to generate the UC emission, suggesting that a two-photon energy-transfer process with two 980 nm photons is involved to populate the ⁵D₄ state of Tb³⁺ and also to populate the ⁵F₅ state of Ho³⁺. The proposed Yb³⁺-sensitized pathways that populate the Tb^{3+ 5}D₄ and Ho^{3+ 5}F₅ emitting states are summarized in the author-created energy-level scheme shown in Figure 16.

Figure 16. Author-created energy-level scheme illustrating Yb³⁺-sensitized Tb³⁺/Ho³⁺ upconversion under 980 nm excitation, including ground-state absorption (GSA), energy transfer (ET), energy-transfer upconversion (ETU), excited-state absorption (ESA), cross-relaxation (CR), non-radiative relaxation, and the principal Tb³⁺ and Ho³⁺ emissions. The proposed pathways and emission assignments are summarized from Ref. [63].

6. Comparative Analysis of Crystallization Mechanisms Across Yb³⁺-Containing Glass Families

6.1. CaF₂ Nanocrystallization in Oxyfluoride Glasses

A glass of composition 47.4SiO₂-19Al₂O₃-28.4CaF₂-2TbF₃-3.2YbF₃ (mol%) was fabricated by melting a homogeneous powder mixture of the constituent oxides and fluorides in a covered Pt-Rh crucible at 1400 °C for 1 h in air [64]. The resulting melt was cast onto a steel plate, covered with another plate, and left to cool gradually to ambient temperature.

The as-quenched product was fully amorphous, as expected for this glass-forming system. Subjecting it to a thermal treatment at 660 °C for 2 h triggered crystallization, producing a glass-ceramic in which CaF₂ nanocrystals were detected by XRD [64]. The corresponding transformation from the amorphous precursor to the CaF₂-containing glass-ceramic is summarized schematically in Figure 17. Undoped versions of the glass remained amorphous after quenching and could also be crystallized upon heating but only at temperatures some 42 °C higher than their Tb/Yb-containing counterpart. This difference demonstrates that the introduction of TbF₃ and YbF₃ effectively promotes crystallization at lower temperatures.

Figure 17. Author-created schematic summarizing the heat-treatment-induced formation of CaF₂ nanocrystals in Tb³⁺/Yb³⁺-doped oxyfluoride glass at 660 °C for 2 h. The reported apparent activation energy, E_a ≈ 498 kJ mol⁻¹, and Avrami exponent, n ≈ 1.01, indicate diffusion-controlled crystal growth. The kinetic parameters, phase assignment, and crystallization-temperature trend are summarized from Refs. [39,64].

The crystallization kinetics were investigated by DSC, which yielded an apparent activation energy of ~498 kJ/mol and an Avrami exponent of n = 1.01. These values point toward a diffusion-governed growth regime, with crystallites adopting needle- and plate-like morphologies of limited length. Figure 17 summarizes both the reported kinetic descriptors and the heat-treatment-induced phase evolution, while transmission electron microscopy provided independent confirmation of the CaF₂ nanocrystalline phase within the glass-ceramic matrix [64].

6.2. Cluster-Assisted Formation of Multicomponent Fluoride Phases

6.2.1. Ca–Yb Fluoride Solid Solutions

The incorporation of Yb₂O₃ into NaF-CaF₂-Al₂O₃-SiO₂ glass does more than simply introduce an optically active ion. The trivalent cation can interact with fluoride-rich regions, maintain charge balance via local structural modifications, and be incorporated into a CaF₂-derived lattice. This dual function—acting as a network modifier prior to crystallization and as a lattice constituent afterward—accounts for the observed dependence of the apparent activation energy and the resulting phase assemblage on Yb content [65].

Within the composition series examined, heat treatment yielded CaF₂-related crystalline phases along with a Ca_0.8Yb_0.2F_2.2 solid solution, with grain sizes in the range of 10–20 nm [65]. The apparent activation energy decreased monotonically with increasing Yb³⁺ loading. Because the crystalline product incorporates Yb, this trend is consistent with the idea that Yb-rich fluoride environments serve as precursors to the solid-solution phase.

This finding should not be overinterpreted as evidence that Yb³⁺ universally accelerates crystallization. Higher Yb content also modifies network connectivity, charge compensation mechanisms, and liquidus relationships. Disentangling catalytic nucleation from a shift in equilibrium phase stability requires compositionally matched series, quantitative phase-fraction analysis, and the local structural characterization of Yb prior to crystallization.

A related but lead-free example is provided by Yb³⁺/Er³⁺-codoped fluoro-borosilicate glasses, in which the partial replacement of SiO₂ by B₂O₃ modifies the oxide–fluoride network and improves the selectivity of rare-earth-fluoride nanocrystallization [75]. In this system, Yb³⁺ acts not only as an optical sensitizer but also as a component of the precipitated KYb₂F₇ and KYb₃F₁₀ phases. The resulting phase evolution therefore provides direct evidence that rare-earth concentration, glass-network composition, and heat-treatment temperature jointly determine the identity of the nanocrystalline phase.

6.2.2. Yb³⁺-Induced Rare-Earth-Fluoride Nanocrystallization in Fluoro-Borosilicate Glasses

An open-access example that demonstrates the coupled effects of glass composition, Yb³⁺ addition, and heat-treatment temperature was reported for nominal (60 − x)SiO₂–xB₂O₃–20KF–20ZnF₂ glasses (x = 0–40 mol%) [75]. YbF₃ and ErF₃ were introduced to provide 1.0 mol% Yb³⁺ and 0.2 mol% Er³⁺, respectively. Partial substitution of B₂O₃ for SiO₂ reduced the glass-preparation temperature and altered the competition between the fluoride phases precipitated during subsequent heat treatment.

The nominal 40SiO₂–20B₂O₃–20KF–20ZnF₂ composition, denoted as 40Si–20B, exhibited a particularly clear temperature-dependent phase evolution. As shown in Figure 18, the precursor glass displayed only a broad amorphous-scattering feature. After heat treatment at 460 and 500 °C for 10 h, the diffraction peaks were assigned predominantly to KYb₂F₇. Additional KYb₃F₁₀ reflections appeared after treatment at 540 °C, whereas KYb₃F₁₀ became the dominant detectable phase after treatment at 580 °C [75]. Thus, increasing the treatment temperature changed not only the extent of crystallization but also the identity of the Yb-containing fluoride phase.

Figure 18. XRD patterns of the 1.0 mol% Yb³⁺/0.2 mol% Er³⁺-codoped 40Si–20B precursor glass and nanocrystallized glass-ceramics heat-treated at 460–580 °C for 10 h, showing the temperature-dependent evolution from KYb₂F₇ to KYb₃F₁₀. Adapted from Ref. [75] under the Creative Commons Attribution 4.0 International License (CC BY 4.0). Original Figure 2c of Ref. [75] was reformatted for presentation, and its panel designation was omitted; the underlying diffraction data were not altered.

Transmission electron microscopy of the sample treated at 580 °C showed nanocrystals approximately 5–30 nm in diameter that were dispersed within the residual glass matrix [75]. The reported lattice-fringe spacing of approximately 0.204 nm was assigned to the (440) plane of KYb₃F₁₀, providing microscopic support for the XRD phase assignment.

These results illustrate a dopant-associated crystallization pathway in which Yb³⁺ participates directly in the precipitated fluoride phases. Nevertheless, the simultaneous detection of a Yb-containing crystalline phase and enhanced luminescence does not establish the fraction of the total Yb³⁺ population incorporated into the nanocrystals. Quantitative phase analysis, spatially resolved compositional mapping, and site-selective spectroscopy would still be required to determine the partitioning of Yb³⁺ and Er³⁺ between the nanocrystals and the residual glass.

6.2.3. Spectroscopic Consequences of Network Modification and Nanocrystallization

A complementary view of the sensitivity of Yb³⁺ to the surrounding glass network is provided by the xTiO₂–(60−x)GeO₂–30BaO–9.5Ga₂O₃–0.5Yb₂O₃ system, where x ranges from 0 to 50 mol% [76]. XRD confirmed that all investigated compositions remained amorphous. Figure 19 shows the characteristic Yb³⁺ absorption associated with the ²F_7/2 → ²F_5/2 transition. A resolved feature near 976 nm is observed, while the band shape and spectral distribution vary with the GeO₂/TiO₂ ratio.

Figure 19. Absorption spectra of Yb³⁺-doped titanate–germanate glasses with different GeO₂/TiO₂ molar ratios, including the characteristic Yb^{3+ 2}F_7/2 → ²F_5/2 absorption near 976 nm. Adapted from Ref. [76] under the Creative Commons Attribution 4.0 International License (CC BY 4.0); original Figure 3a was reformatted for presentation, and its panel designation was omitted; the underlying spectral data were not altered.

The composition-dependent absorption profiles indicate that modification of the titanate–germanate network changes the distribution of local ligand fields and Stark components experienced by Yb³⁺. However, because all these samples remain XRD-amorphous, the spectral changes should not be interpreted as evidence of Yb³⁺ incorporation into a crystalline phase. Absorption spectroscopy is sensitive to the average Yb³⁺ environment, but it is not intrinsically phase-selective. Claims of Yb³⁺ partitioning into nanocrystals therefore require complementary diffraction, microscopy, lifetime, or site-selective spectroscopic evidence.

The spectroscopic consequences of controlled nanocrystallization can be evaluated more directly in the Yb³⁺/Er³⁺-codoped 40Si–20B fluoro-borosilicate system discussed above. Under 980 nm excitation, the precursor glass exhibits only weak upconversion emission, whereas the heat-treated samples show distinct Er³⁺ emission bands near 522, 540, and 650 nm, corresponding to the ²H_11/2 → ⁴I_15/2, ⁴S_3/2 → ⁴I_15/2, and ⁴F_9/2 → ⁴I_15/2 transitions, respectively [75].

As shown in Figure 20, the upconversion-emission intensity increases as the heat-treatment temperature rises from 460 to 580 °C. This enhancement occurs concurrently with the phase evolution from KYb₂F₇ to KYb₃F₁₀ identified by XRD. The low-phonon-energy fluoride environment reduces multiphonon relaxation and facilitates Yb³⁺-to-Er³⁺ energy transfer, providing a chemically plausible explanation for the enhanced green and red emissions.

Figure 20. Upconversion emission spectra of the 1.0 mol% Yb³⁺/0.2 mol% Er³⁺-codoped 40Si–20B precursor glass and nanocrystallized glass-ceramics heat-treated at 460–580 °C for 10 h under 980 nm excitation. Adapted from Ref. [75] under the Creative Commons Attribution 4.0 International License (CC BY 4.0). Original Figure 3a was reformatted for presentation, and its panel designation was omitted; the underlying emission data were not altered.

The combined XRD and luminescence results establish a correlation between rare-earth-fluoride nanocrystallization and upconversion enhancement. Nevertheless, emission intensity is also affected by pump absorption, sample thickness, scattering, crystalline fraction, and the spatial distribution of the activator. Consequently, the stronger emission should not, by itself, be treated as a quantitative measurement of the fraction of Er³⁺ or Yb³⁺ incorporated into the crystalline phase.

6.2.4. Yb³⁺-Modified ZnAl₂O₄ Nanocrystals

In 58SiO₂-16[(1 − x%)Al₂O₃ + x%Yb₂O₃]-15ZnO-7Li₂O-4TiO₂-0.2Sb₂O₃ glasses, ZnAl₂O₄ nanocrystals were already detectable in the Yb-free composition after quenching, and heat treatment at 720 °C further developed the phase [102]. AsYb₂O₃ replaced Al₂O₃, Yb³⁺ entered the spinel-related nanocrystals, and the reported symmetry changed from $\mathrm{F}\mathrm{d}\overline{3}m$ toward $\mathrm{F}\overline{4}3m$.

This system is a counterexample to a simple nucleating-agent narrative: crystallization occurs without Yb³⁺, while Yb substitution modifies the structure of an existing crystallization pathway. The relevant role of Yb³⁺ is therefore phase modification and site substitution rather than the initiation of nucleation. A composition series measuring nucleation density and phase fraction would be needed to determine whether Yb also changes the rate.

6.3. What Promotes Heterogeneous Nucleation?

A constituent can promote nucleation by creating a compositionally distinct region, lowering the crystal–glass interfacial energy, increasing the thermodynamic driving force for a particular phase, or changing mobility. These mechanisms must be distinguished from merely observing that a phase appears after the constituent is added. In a multicomponent glass, the effective nucleating species may be a cluster or separated nanophase rather than the nominal oxide itself.

The SiO₂-Li₂O-K₂O-ZnO-P₂O₅ series provides a useful control [103]. Increasing P₂O₅ changes the balance between surface and volume crystallization and produces phosphate-rich precursors that favor fine-grained lithium disilicate microstructures. The effect is associated with glass-network reorganization, phase separation, and interfacial energetics.

For Yb³⁺ systems, the available studies collectively suggest three recurring routes: (i) Yb/F-rich clustering before a Yb-bearing fluoride phase; (ii) Yb³⁺ substitution into a phase that already nucleates in the host; and (iii) Yb-induced network changes that alter viscosity or phase competition without placing most Yb³⁺ in the crystals. Distinguishing these routes requires in situ thermal/diffraction measurements, TEM, Raman spectroscopy, and element-specific local-structure probes [2,3,4,5,6,7,8,10,11,12,13,94,97,98,99].

6.4. Additional Mechanistic Evidence from Recent Yb³⁺-Containing Glass-Ceramics

Several mechanistic studies beyond the core case studies sharpen the interpretation of Yb-containing systems. Lin et al. showed that amorphous phase separation can predetermine fluoride-rich regions that later crystallize, while Bhattacharyya et al. and Bocker et al. demonstrated that growth may become self-limited or interface-controlled when crystallization changes the composition and viscosity of the surrounding glass [37,38,39]. Zhao et al. connected fluoride-rich domains to nanocrystal formation using solid-state NMR, STEM-EDS, and molecular-dynamics simulation [40]. These studies support phase separation as a plausible precursor, but they also show that phase-separated domains should not be equated with crystals unless diffraction or lattice imaging confirms the long-range order.

Ali et al. found that enhanced Er³⁺/Yb³⁺ upconversion after CaF₂ precipitation was associated with increased structural ordering and changed local symmetry rather than requiring complete migration of the rare-earth ions into the fluoride crystals [41]. A related Yb³⁺/CaF₂ germanotellurite study detected 8–10 nm nanocrystallites, but the amount and site occupancy of Yb³⁺ remained only partly constrained [42]. Collectively, these reports reinforce the need to combine optical spectra with direct chemical mapping or element-selective local-structure measurements.

Yb-only phosphate systems provide a useful control because sensitizer-to-activator transfer is absent. Sen et al. showed that Al₂O₃/Y₂O₃ additions strengthen the phosphate network, slightly slows crystallization, and favor surface crystallization, while precipitation of a Yb-containing phosphate phase changes the ²F_5/2 lifetime and emission bandwidth [43]. Hongisto et al. obtained transparent NaYbP₂O₇-containing glass-ceramics only in selected NaF-free compositions and showed that NaF redirects bulk crystallization toward surface crystallization [44]. Their later composition study demonstrated that Al₂O₃, TiO₂, or ZnO shifts the phase assemblage toward Na₅P₃O₁₀ and suppresses Yb-containing phosphate phases [49]. These results indicate that network connectivity and modifier/intermediate balance can dominate over Yb concentration in determining phase selection.

Optical-refrigeration studies add quantitative criteria that are often absent from upconversion reports. de Lima Filho et al. found that the cooling figure of merit depends non-monotonically on the degree of crystallization [45]. Fiber and bulk studies subsequently linked performance to photoluminescence quantum efficiency, background absorption, OH⁻ content, scattering, temperature, and Yb³⁺ concentration [46,47,48]. These reports are particularly valuable because they measure absolute or near-absolute optical quantities; however, crystalline fraction and Yb partition coefficients should be linked more systematically to the measured cooling response.

Recent fluorosilicate and oxyfluoroborosilicate studies show that Yb³⁺ can be an active phase-forming constituent. Jamalaiah reported CaF₂-containing Yb³⁺ glass-ceramics with useful 981 nm emission [50], while Fang et al. and Li et al. demonstrated controllable KYb₃F₁₀ precipitation and enhanced activator emission [51,53]. Long et al. used phase-separated fluoride-rich regions to alter activator distribution before crystallization [52]. The competitive formation of NaYbF₄ and Na₃ScF₆ further shows that the Na/F ratio and competing-cation chemistry determine whether Yb enters the selected fluoride phase [60]. Xiang et al. extended this strategy by using B₂O₃ to lower the processing temperature and regulate KZnF₃/KYb₃F₁₀ nanocrystallization [61]. These studies provide stronger evidence for Yb-induced phase selection than studies based only on DSC shifts, although such claims still require matched Yb-free controls and direct compositional mapping.

Ba₄Y₃F₁₇, Ba₃Gd₂F₁₂, ZnAl₂O₄, YOF, and GdOF glass-ceramics demonstrate that improved thermometric response can arise through lower phonon energy, changed site symmetry, crystallization-enhanced lifetime, or modified thermal coupling [54,55,56,57,58]. These studies report useful phase assignments and sensitivities, but application-level comparison remains difficult because excitation density, sample thickness, crystalline fraction, thermal cycling, and absolute quantum yield are not standardized. Pump-power-dependent measurements are especially important because the apparent fluorescence-intensity ratio and sensitivity may change when the upconversion pathway approaches saturation [57].

Finally, Blanc et al. demonstrated that nanoparticle composition can evolve before easily detectable crystallization [59]. This observation explains why early amorphous segregation may change the Yb³⁺ environment without producing sharp XRD peaks and provides a mechanistic link between glass-network heterogeneity and later phase selection. Taken together, the additional literature strengthens three conclusions: pre-nucleation segregation must be distinguished from crystallization; Yb³⁺ partitioning requires direct evidence; and optical enhancement should be evaluated together with crystalline fraction, transparency, lifetime, and absolute efficiency.

7. Composition Dependence: When Does Yb₂O₃ Promote or Suppress Crystallization?

Claims about the effect of Yb₂O₃ are often based on different base compositions, replacement schemes, and thermal protocols. A lower T_p in one glass and an amorphous XRD pattern in another are not contradictory unless the host network and phase field are comparable. The strongest evidence comes from a composition series in which Yb₂O₃ replaces a specified component while all other variables are fixed.

7.1. Yb-Free Barium–Zinc Borosilicate Kinetic Baseline

The crystallization kinetics of a barium–zinc borosilicate glass with the molar composition 40BaO-20ZnO-30B₂O₃-10SiO₂ were examined using the Kissinger, Ozawa, and Augis–Bennett formalisms under non-isothermal conditions to determine the activation energy (E_c) and Avrami exponent (n) [10,11,12,13,70]. These non-isothermal approaches are commonly employed to elucidate crystallization mechanisms, distinguish between surface- and volume-dominated crystallization, and assess the relative crystallizability of different glass compositions [10,11,12,13].

For the first exothermic event, the local activation energy E_c(χ) fell from ~700 to ~500 kJ/mol as the crystallized fraction (χ) increased from 0.1 to 0.9. By contrast, the second exothermic peak exhibited a slight increase in E_c(χ), from ~490 to ~570 kJ/mol, over the same χ range. The corresponding Avrami exponent n(χ) rose from ~1 to ~1.4 for the first peak, while it declined from ~1.7 to ~1.4 for the second peak. SEM observations of sintered samples revealed that crystallization begins at the particle surfaces, with lamellar crystallites growing inward at elevated temperatures, accompanied by quasi-spherical nano-sized crystallites. The dependence of the local activation energy on the crystallized fraction suggests that the overall process involves multiple kinetic steps during sintering and crystallization.

7.2. Yb₂O₃-Doped Barium Titanium Borate Glasses

The composition dependence of the parent-glass structure can be illustrated using the open-access Yb₂O³-doped barium titanium borate system reported by Marzouk et al. [77]. The investigated compositions were 54.250B₂O₃–8.320TiO₂–37.430BaO for the undoped glass and related compositions containing 0.258, 0.769, and 1.275 mol% Yb₂O₃.

The XRD patterns shown in Figure 21 contain broad amorphous-scattering features without sharp Bragg reflections, confirming that all four as-prepared samples remained non-crystalline. The structural and physical trends discussed for this series therefore describe the modification of the parent-glass network rather than the development of a glass-ceramic phase.

Figure 21. XRD spectra of Yb³⁺-doped barium titanium borate glasses containing 0–1.275 mol% Yb₂O₃; the broad amorphous-scattering features confirm the absence of detectable long-range crystalline order. Reproduced from Ref. [77] under the Creative Commons Attribution 4.0 International License (CC BY 4.0).

As shown in Figure 22, the measured density increased from 3.652 to 3.761 g cm⁻³ as the Yb₂O₃ content increased. The molar volume changed non-monotonically: it initially decreased from 27.875 to 27.784 cm³ mol⁻¹ and subsequently increased to approximately 28.06 cm³ mol⁻¹ at the highest Yb₂O₃ concentration [77]. Consequently, the density increase should not be interpreted simply as uniform network compaction; both the larger molar mass of Yb₂O₃ and composition-dependent changes in network packing contribute to the observed behavior.

Figure 22. Density and molar volume of Yb³⁺-doped barium titanium borate glasses as functions of Yb₂O₃ content. Reproduced from Ref. [77] under the Creative Commons Attribution 4.0 International License (CC BY 4.0).

Raman and FTIR measurements showed that Yb₂O₃ addition modified the distributions of BO₃ and BO₄ structural units. A characteristic broad Yb³⁺ absorption band near 977 nm, assigned to the ²F_7/2 → ²F_5/2 transition, became more intense with increasing Yb₂O₃ content. Over the same composition range, the optical band gap decreased from approximately 3.198 to 3.115 eV, whereas the calculated linear refractive index increased from approximately 2.346 to 2.367 [77].

These results confirm that Yb₂O₃ acts as an optically and structurally active component even when no crystalline phase is detected. However, the amorphous XRD patterns do not establish whether Yb₂O₃ promotes or suppresses subsequent crystallization. Resolving that question would require a matched heat-treatment series supported by DSC, quantitative XRD, and electron-microscopy measurements. The present results should therefore be interpreted as evidence for the composition-dependent modification of the precursor-glass network rather than direct evidence of Yb-induced nucleation.

8. Multiphase Crystallization and Phase Competition

8.1. Phosphate–Fluoride Glass

Yb³⁺-doped phosphate glasses with compositions (100 − x)(0.75NaPO₃-0.25CaF₂)-xYb₂O₃, with x = 3.0, were synthesized by a standard melt-quenching technique followed by a two-step thermal treatment [68]. The glass was first heat-treated at 320 °C, above the T_g (glass-transition temperature), and then heated at temperatures between 400 and 440 °C, corresponding to the crystallization range [68]. This system is useful for examining multiphase crystallization in phosphate–fluoride glasses containing Yb³⁺ [3,10,11,12,13,68].

The temperature-dependent phase evolution reported for the Yb³⁺-doped phosphate–fluoride glass is summarized schematically in Figure 23. Experimental XRD results reported in Ref. [68] showed that the as-prepared glass exhibited broad amorphous-scattering features without detectable sharp Bragg reflections. Heat treatment at 400 °C, close to the onset crystallization temperature (T_x), resulted in the preferential precipitation of CaF₂. When the heat-treatment temperature was increased to approximately 450 °C, near the reported crystallization-peak temperature (T_p = 452 °C), additional reflections assigned to NaPO₃ and Na₂Ca₂(P₂O₇)F₂ appeared, indicating multiphase crystallization [68]. A similar crystallization sequence has been reported for Er³⁺-doped glasses in the same system, in which CaF₂ precipitated before the phosphate-containing phases [104]. SEM observations further supported the formation of crystalline domains. The lower traces in Figure 23 are qualitative diffraction signatures intended to illustrate the reported phase evolution and should not be interpreted as experimental XRD patterns.

Figure 23. Author-generated schematic of the temperature-dependent crystallization pathway in Yb³⁺-doped phosphate–fluoride glass. (a) The as-prepared glass exhibits an amorphous structure; (b) CaF₂ precipitates preferentially after heat treatment at 400 °C; and (c) additional NaPO₃ and Na₂Ca₂(P₂O₇)F₂ phases emerge near 450 °C, resulting in multiphase crystallization. The lower traces are qualitative diffraction signatures rather than experimental XRD patterns. The phase evolution is summarized based on Refs. [43,68].

SEM observations indicate that crystallization occurs throughout the glass volume when heat treatment is performed near T_x, and the resulting crystals are approximately spherical. Elemental mapping identifies these crystals as CaF₂, consistent with the XRD analysis. Increasing the heat-treatment temperature increases the number of crystals while reducing their average size. Yb³⁺ remains broadly distributed throughout the material, suggesting that part of the Yb³⁺ population is incorporated into the CaF₂ crystals while the remainder stays in the residual glass matrix. Phosphate-rich crystals were not observed in SEM images of the sample treated at 450 °C, although they were detected by XRD, suggesting preferential surface crystallization. The emission spectra further indicate that CaF₂ precipitation has little effect on the Yb³⁺ near-infrared ²F_5/2 → ²F_7/2 emission, whereas formation of NaPO₃ and Na₂Ca₂(P₂O₇)F₂ substantially reduces the emission intensity.

8.2. Ytterbium Aluminosilicate Glass

The 15Yb₂O₃-25Al₂O₃-60SiO₂ system represents the opposite composition limit: Yb₂O₃ is a major constituent rather than a dilute additive [69,85,97,105,106,107]. The as-prepared bulk glass combines a high T_g (847 °C), low room-temperature thermal conductivity, and substantial mechanical strength. At 1200 °C, it crystallizes into Yb₃Al₅O₁₂, Yb₂Si₂O₇, and Al₆Si₂O₁₃.

The formation of two Yb-bearing phases demonstrates that Yb³⁺ can be incorporated directly into the crystalline assemblage when its concentration is high. Phase selection is governed by the Yb₂O₃–Al₂O₃–SiO₂ stoichiometry and heat treatment, not by a generic “Yb nucleating-agent” effect. Competing phases also mean that a single optical or mechanical trend cannot be assigned to one crystal without quantitative phase analysis.

Rare-earth aluminosilicate glasses are of interest for plasma-resistant components because rare-earth fluorides formed at exposed surfaces can have low volatility [69,105]. This application is relevant to the stability of the material but is secondary to the crystallization mechanism. The present review therefore retains only the composition–phase–property connection and removes the extended semiconductor-processing description.

The phosphate–fluoride and aluminosilicate cases define two ends of a design spectrum. At low Yb content, Yb³⁺ may partition only partly into an early fluoride nanophase; at high content, several Yb-rich oxide phases can consume most of the glass. Between these limits, controlling phase competition is more important than maximizing a nominal nucleation rate.

8.3. Integrated Roles of Yb³⁺ in Crystallization and Luminescence

The preceding examples demonstrate that Yb³⁺ cannot be assigned a single universal function in glasses and glass-ceramics. Depending on the parent-glass composition, Yb³⁺ concentration, local coordination environment, codopants, and heat-treatment schedule, it may act as a glass-network modifier, participate in nucleation precursors, serve as an optical sensitizer, or function as an emitting center and a component of Yb-rich clusters. These four composition-dependent roles and their relationships with crystallization and luminescence are summarized in Figure 24.

Figure 24. Schematic summary of the four major roles of Yb³⁺ in crystallization and luminescence of glasses and glass-ceramics: network modifier, participant in nucleation precursors, sensitizer, and activator/cluster center. The same Yb³⁺ component can therefore influence both phase formation and optical response, depending on glass composition, local coordination, clustering, and heat-treatment schedule.

As a glass-network modifier, Yb³⁺ can alter the connectivity and packing of the parent-glass network. Its relatively high field strength and coordination requirements may change the population of non-bridging oxygen atoms, modify the association of fluoride or phosphate species, and influence viscosity, density, and the thermal stability window. However, the direction and magnitude of these changes depend on the host network. Yb₂O₃ may depolymerize one glass network while strengthening or stabilizing another; therefore, it should not be assumed to exert the same structural effect in silicate, phosphate, borate, and oxyfluoride glasses [43,44,49,65,87,88].

Yb³⁺ may also participate in the formation of nucleation precursors when it becomes enriched in modifier-rich, fluoride-rich, or phosphate-rich regions. If the local composition and coordination of these regions resemble those of the emerging crystalline phase, the effective crystal–glass interfacial barrier may be reduced, and nucleation may become more favorable. This behavior is particularly relevant when Yb³⁺ can enter CaF₂-related solid solutions or form Yb-containing phases such as Pb₄Yb₃F₁₇, Yb₃Al₅O₁₂, or Yb₂Si₂O₇. Conversely, when Yb³⁺ stabilizes the residual glass, increases configurational disorder, or is incompatible with the initially precipitating phase, it may slow crystallization or redirect the process toward competing crystalline phases. Yb³⁺ should therefore be described as a composition-dependent participant in nucleation rather than as a universal nucleating agent [37,38,39,40,41,42,43,44,50,51,52,53,59,60,61,64,65,66,67,68,69,102].

The third role is optical sensitization. Yb³⁺ efficiently absorbs near-infrared radiation in the approximately 900–1000 nm range through the ²F_7/2 → ²F_5/2 transition and can subsequently transfer excitation energy to neighboring activator ions such as Er³⁺, Ho³⁺, or Tb³⁺. Controlled incorporation of Yb³⁺ and the activator into a low-phonon-energy crystalline phase may reduce non-radiative relaxation and enhance upconversion emission. Nevertheless, stronger steady-state emission does not by itself prove that Yb³⁺ has entered the nanocrystals, because heat treatment can also change pump absorption, scattering, activator–sensitizer distances, and the composition of the residual glass [18,19,20,41,51,52,53,54,55,56,57,58,62,63,64,65,66,67].

Finally, Yb³⁺ can itself act as an optically active center through the ²F_5/2 → ²F_7/2 near-infrared emission. At higher local concentrations, neighboring Yb³⁺ ions may form pairs or clusters that support cooperative luminescence. Such aggregation can be beneficial when cooperative emission is the desired function, but it may also enhance energy migration, cross-relaxation, and concentration quenching. The optimal Yb³⁺ concentration is therefore determined by a balance among pump absorption, rare-earth partitioning, inter-ionic distance, multiphonon relaxation, and clustering [18,19,20,50,51,52,53,87,88,89,90].

The four roles shown in Figure 24 are not independent. Network modification can create compositionally distinct precursor regions; these regions may subsequently nucleate Yb-containing nanocrystals; crystallization then redistributes Yb³⁺ and codoped activators between the crystalline phases and the residual glass; and this redistribution changes both energy-transfer efficiency and concentration-quenching behavior. Accordingly, composition and heat treatment must be optimized jointly. A reliable interpretation requires thermal analysis, phase identification, microscopy, local-structure measurements, and quantitative optical characterization to follow the same Yb³⁺ population from the parent glass to the final glass-ceramic.

9. Conclusions

The surveyed evidence supports a conditional, rather than universal, picture of Yb³⁺-controlled crystallization. Yb³⁺ can promote fluoride-rich precursor formation, enter CaF₂-related or Pb₄Yb₃F₁₇ nanocrystals, substitute into spinel- or oxyapatite-related lattices, or remain largely in the residual glass while modifying its network and viscosity. Which role dominates is determined by the parent composition, Yb³⁺ concentration, charge-compensation chemistry, and thermal schedule.

9.1. Practical Design Guidelines

First, define the target phase and the intended Yb³⁺ site before selecting a heat treatment. A fluoride nanophase designed to host Yb³⁺/Er³⁺ requires different precursor chemistry from an oxide glass intended to crystallize Yb₃Al₅O₁₂. Phase diagrams and chemically plausible oxidation states should constrain the proposed pathway.

Second, separate nucleation from growth experimentally. Use multiple heating rates to estimate apparent kinetics, then arrest samples across each thermal event and correlate DSC/DTA with XRD, TEM, and Raman measurements. An Avrami exponent or a ΔT value should be treated as supporting evidence rather than a stand-alone mechanism.

Third, optimize a multi-objective response. Crystal fraction and luminescence should be evaluated together with transmission, size distribution, and thermal stability. The best glass-ceramic is commonly obtained before the strongest XRD peak or longest anneal, because coarsening and secondary phases can reduce optical performance.

Fourth, determine where Yb³⁺ resides. Steady-state emission enhancement is not sufficient proof of incorporation. EXAFS/XANES, site-selective spectroscopy, lifetime components, elemental mapping, or another direct/quantitative method should be used to distinguish crystal-hosted Yb³⁺ from clusters and residual-glass sites [98,100,101].

Fifth, report negative and non-monotonic results. A composition in which Yb₂O₃ suppresses a phase, shifts the crystallization peak upward, or preserves an amorphous structure is mechanistically valuable. Such data reveal the boundary between nucleation promotion, network stabilization, and phase competition.

For codoped systems, optical optimization also requires control of Yb-Yb and Yb-Ln distances. Increasing Yb³⁺ can not only improve pump absorption but also strengthen clustering, cooperative luminescence, cross-relaxation, or concentration quenching. Composition and heat treatment must therefore be designed jointly.

A minimum comparative dataset should include the complete parent-glass composition, replacement basis, T_g/T_x/T_p at specified heating rates, nucleation and growth schedules, phase identity, crystalline fraction, image-based size distribution, Yb³⁺ partitioning, transmission, lifetime, absolute quantum efficiency, and stability after repeated thermal exposure.

9.2. Unresolved Questions and Reporting Priorities

The largest unresolved question is the identity of the earliest Yb-associated precursor. Fluoride-rich clusters, phosphate-rich domains, and modifier-related non-bridging-oxygen environments are frequently inferred but rarely observed before XRD-detectable crystals form. In situ total scattering, synchrotron X-ray absorption, small-angle scattering, and high-resolution microscopy on time-resolved samples could close this gap.

A second gap is comparability. Activation energies obtained by different non-isothermal methods are often compared without confirmation of identical transformed fractions or phase pathways. Community reporting of raw thermal curves, fitting intervals, uncertainty, and phase evolution would make cross-study comparisons more defensible.

A third gap is quantitative structure–optical correlation. Many studies report relative intensity enhancement but omit absorbed pump power, lifetime, quantum efficiency, and crystalline fraction. Without these quantities, it is difficult to determine whether crystallization improved intrinsic radiative efficiency or merely changed absorption and scattering.

Finally, environmental and manufacturing constraints should enter materials selection early. Lead- and cadmium-containing oxyfluoride systems offer clear mechanistic examples but present volatility and toxicity concerns. Lead-free compositions, scalable heat treatments, reproducible nanocrystal distributions, and long-term optical stability are necessary for translation.

In summary, Yb³⁺ is best viewed as a composition-dependent participant in a coupled glass-network, crystallization, and energy-transfer system. Treating it only as an optical dopant or universally as a nucleating agent obscures the central mechanism. Progress will come from measurements that follow the same Yb population from the parent glass, through the precursor stage, into the final crystal and residual matrix.

The revised framework turns the available case studies into testable design rules while making the limitations explicit. It also explains why nominally conflicting effects of Yb₂O₃ can coexist across glass families: the host network determines whether Yb³⁺ lowers an interfacial barrier, alters mobility, enters the product phase, or stabilizes the amorphous state.