An in-situ and ex-situ study of the microstructural evolution of a novel lithium silicate glass-ceramic during crystallization firing

Abstract

Objective.

To elucidate the compositional and microstructural developments of a novel lithium silicate glass-ceramic during its crystallization cycle.

Methods.

Blocks of a lithium silicate glass-ceramic (Obsidian®, Glidewell Laboratories) were cut into 1 mm thick plates and polished to 1 μm finish. Some of them were crystallized prior to polishing. Firstly, ex-situ compositional and microstructural characterizations of both the pre- and post-crystallized samples were performed by wavelength dispersive X-ray fluorescence, field-emission scanning electron microscopy (FE-SEM), and X-ray diffractometry (XRD). Secondly, the pre-crystallized samples were subjected to in-situ compositional and microstructural characterizations under non-isothermal heating by simultaneous thermogravimetry/differential scanning calorimetry, X-ray thermo-diffractometry, and field-emission scanning electron thermo-microscopy.

Results.

The microstructure of pre-crystallized Obsidian® consists of an abundant population of perlitic-like/dendritic lithium silicate (Li₂SiO₃) nanocrystals in a glass matrix. Upon heating, the residual glassy matrix does not crystallize into any form of SiO₂; elemental oxides do not precipitate unless over-heated above 820°C; and the Li₂SiO₃ nanocrystals do not react with the glassy matrix to form typical lithium disilicate (Li₂Si₂O₅) crystals. Nonetheless, the Li₂SiO₃ nanocrystals grow and spheroidize through the solution-reprecipitation process in the softened glass, and new lithium orthophosphate (Li₃PO₄) nanocrystals precipitate from the glass matrix.

Significance.

The identification of compositional and microstructural developments of Obsidian® indicates that, by controlling the firing conditions, it is possible to tailor its microstructure, which in turn could affect its mechanical and optical properties, and ultimately its clinical performance.

1. Introduction

Computer Aided Design/Computer Aided Machining (CAD/CAM) technology has led to the development of a large variety of glass-ceramics for monolithic restorations. The capability of producing bulk blocks with minimal defects and flaws for milling allows dental restorations to combine strength and aesthetics. Previous studies have shown flexural strength for lithia-based glass-ceramics ranging from 300 to 520 MPa [1–4] and survival rates ranging from 96% to 100% in 3 years [5–7].

Recently, a new lithium silicate glass-ceramic (Obsidian® Milling Blocks, Glidewell Laboratories, Newport Beach, USA) [8] was launched as an alternative for veneers, onlays, inlays, and single-unit crowns for posterior and anterior applications. Obsidian® is available in pre-crystallized blocks, and its composition is described as a unique combination of elemental oxides and high content of ultrafine nanometre-size lithium silicate crystals. The manufacturer reports a 96% survival rate against fracture/chipping for Obsidian® anterior and posterior crowns after 3 years in function [9], which is comparable with the results obtained from other similar glass-ceramics [5,6]. However, medium- to long-term clinical data for lithium silicate glass-ceramics in general are not yet available due to their relatively short history of use in clinical practice.

Glass-ceramic restorations require a firing cycle after milling to achieve the final crystallization. The nucleation and crystallization process that occurs during the thermal treatment defines the composition and microstructure of the final glass-ceramic products, which in turn determine their mechanical and optical properties [10,11]. Therefore, in order to optimize the mechanical and optical properties, it is paramount to understand the evolution of the composition and microstructure of these materials during crystallization firing. The vast majority of previous studies have characterized lithia-based glass-ceramics microstructure pre- and post-crystallization using ex-situ techniques [11–19]. However, it is a challenge to really understand the entire compositional and microstructural evolution of these materials relying solely on pre- and post-crystallization ex-situ analyses. There have been only a few in-situ studies [10,20–28], but all focused on a subset having a lithium disilicate composition and none investigating lithum silicate. This is because lithium disilicate glass-ceramics were developed in 1998, and since then there have been several improvements in their mechanical and optical properties [29]. As a result, they have enjoyed widespread success as a robust aesthetic restorative option [30,31].

Accordingly, the present study was aimed at elucidating the compositional and microstructural development of the novel Obsidian® lithium silicate glass-ceramic during its crystallization firing cycle. Detailed analyses were performed to characterize the pre-crystallized material as well as its final crystallized form (ex-situ analyses) and how the microstructure evolves until it reaches the final stage (in-situ analyses). To this end, a wide battery of compositional and microstructural characterization techniques were used, namely, wavelength dispersive X-ray fluorescence (WDXRF), X-ray diffractometry (XRD), field-emission scanning electron microscopy (FE-SEM), thermogravimetry (TG)/differential scanning calorimetry (DSC), X-ray thermo-diffractometry (XRTD), and field-emission scanning electron thermo-microscopy (FE-SETM).

2. Materials and Methods

2.1. Sample preparation

Pre-fabricated milling blocks of a lithium silicate glass-ceramic (Obsidian®, W14, shade A1; Glidewell Laboratories, Newport Beach, USA) were obtained and cut into plates using a low-speed diamond saw under water-cooling (IsoMet, Buehler, Lake Bluff, IL). Some of the plates had their lateral surfaces ground and polished to a 1 μm finish (termed pre-crystallized samples). Others were crystallized at 820°C following the manufacturer’s instructions, and then ground and polished using the same protocol (termed crystallized samples). The final dimensions of polished plates were 14×12×1 mm. All the samples were cleaned with ethyl alcohol in an ultrasonic bath prior to microstructural analyses.

2.2. Ex-situ analyses

The methods described below were used to characterize the pre-crystallized and crystallized samples.

2.2.1. Wavelength Dispersive X-Ray Fluorescence (WDXRF)

The chemical elements present in the pre-crystallized samples and the concentration (wt.%) of their corresponding elemental oxides were determined by WDXRF using a sequential spectrometer (S8 Tiger, Bruker AXS, Germany) equipped with an Rh X-ray tube. The measurements were carried out in vacuo over the energy range 0.5–42 keV with three analyzer crystals (XS-55, PET, and LiF (200)) and two detectors (proportional flow counter and scintillation counter).

2.2.2. X-Ray Diffractometry (XRD)

XRD was used to determine the crystalline phases present in the pre-crystallized and crystallized samples. The measurements were performed using a high-resolution diffractometer (D8 Advance, Bruker AXS, Germany) equipped with a Cu X-ray tube, operating over the 2θ range 15–45/60° with a Ge primary monochromator and an ultra-fast linear detector. The XRD patterns collected were first indexed with the help of the PDF2 database to determine the crystalline phase(s) present, and then refined by the Pawley method to calculate the lattice parameters of the silicate phase(s).

2.2.3. Field Emission-Scanning Electron Microscopy (FE-SEM)

FE-SEM observations were done on the pre-crystallized and crystallized samples using a dual-beam environmental microscope (Quanta 3D FEG, FEI, The Netherlands) equipped with a field-emission electron gun, operating at 20 keV in low-vacuum mode (at 80 Pa pressure) with the (gaseous analytical) backscattered electron detector (BSED). The micrographs captured were analysed by image analysis software to determine the degree of crystallinity (defined as the fraction of crystallized area) and the morphology of the crystals.

2.3. In-situ analyses

The microstructural development upon heating of the Obsidian® milling blocks was examined using a wide battery of characterization techniques and various heating schedules. Some samples were subjected to general non-isothermal heat treatments in air (up to 1100°C) to thus obtain a broad picture of the thermal evolution. Other samples, however, were sintered under the exact firing cycle recommended by the manufacturer to thus reproduce clinical practice. Specifically, the microstructural development upon heating of the pre-crystallized samples was examined using the characterization techniques described below.

2.3.1. Thermogravimetry (TG)/Differential Scanning Calorimetry (DSC)

TG/DSC analyses were performed to assess the thermal stability of the material during crystallization. The TG/DSC curves were obtained using a calorimetric thermobalance (STA 449 F3 Jupiter, Netzsch, Bruker, Germany) equipped with a SiC furnace, operating under flowing pure air (at 50 ml/min) in the temperature range 40–1100°C at a heating rate of 10°C/min.

2.3.2. X-Ray Thermo-diffractometry (XRTD)

The crystallization sequence of the glass-ceramic was investigated by XRTD. The XRTD patterns were acquired using the same diffractometer as before but equipped with a furnace-type heating chamber (RP-Furnace, MRI, Germany) [27,32], operating over the 2θ range 15–45° and collecting XRTD patterns each 5°C in the temperature range 30–900°C at an effective heating rate of ~0.9°C/min. The ex-situ sample (after the in-situ XRTD analysis) also had the XRD pattern measured and the FE-SEM observations performed in the same conditions as the pre-crystallized and crystallized samples. The ex-situ sample was also cross-sectioned, and the distribution of crystals throughout its thickness was observed by FE-SEM.

Two sets of complementary XRTD patterns were also acquired at some selected temperatures, implementing in the diffractometer heating chamber the firing cycle recommended by the manufacturer together with a slight modification of this.

2.3.3. Field-Emission Scanning Electron Thermo-Microscopy (FE-SETM)

FE-SETM observations were done in-situ at various temperatures in the range 400–875°C using the same microscope as before but equipped with a hot-stage [27], operating at 20 keV in an environmental mode (at 200–400 Pa pressure) with the gaseous secondary electron detector (GSED).

3. Results

3.1. Microstructural characterization of the pre-crystallized milling blocks

Fig. 1 shows the WDXRF spectrum, including the corresponding peak assignations, of the pre-crystallized Obsidian® milling blocks. Despite WDXRF not measuring Li, it was used to determine the chemical composition of these blocks because the manufacturer indicates that Obsidian® consists of an innovative combination of over 20 unique elemental oxides, but only provides the concentration range for seven oxides (i.e., SiO₂, Li₂O, GeO₂, K₂O, P₂O₅, Al₂O₃, and ZrO₂) and gives no information on the rest of them (not even what these oxides are) [33]. Indeed, the WDXRD indicated that the Obsidian® milling blocks are chemically very complex, containing 23 different oxides (i.e., 22 detected by WDXRF, plus Li₂O). Table 1 compares the chemical composition provided by the manufacturer with that measured here by WDXRD when the percentage of Li₂O is fixed to 15 wt.% (i.e., right in the middle of its concentration range provided by the manufacturer). As expected [33], the combination of SiO₂, Li₂O, GeO₂, K₂O, P₂O₅, Al₂O₃, and ZrO₂ accounts for ~91.5 wt.% of the composition, with the sum of CeO₂, Y₂O₃, Ta₂O₅, Na₂O, HfO₂, CaO, MgO, SO₃, Fe₂O₃, Sm₂O₃, TiO₂, Er₂O₃, BaO, V₂O₅, Nd₂O₅, NiO, and NaCl accounting for the remaining ~8.5 wt.%. Therefore, the Obsidian® milling blocks were obtained from glass ingots formulated with SiO₂ as the main network-forming oxide plus a multitude of intermediates and network-modifiers that act as stabilizers, nucleating agents, fining agents, fluxes, and colourants.

Figure 1.

WDXRF spectrum the Obsidian® milling blocks. Peak assignations are included. The numbers in the figure denote the various combinations of analyzer crystals/detectors/filter used, namely: XS-55/proportional flow counter/none (❶), PET/proportional flow counter/none (❷), PET/proportional flow counter/none (❸), LiF (200)/proportional flow counter/none (❹), LiF (200)/scintillation counter/none (❺), and LiF (200)/scintillation counter/Cu (❻) without sample.

Table 1.

Chemical composition of the pre-crystallized Obsidian® milling blocks.

Compound (wt.%)
Manufacturer’ specification	Measured by WDXRF
SiO2 50–58	SiO2 59.75
Li2O 10–20	Li2O 15.00
GeO2 1–10	GeO2 0.18
K2O 2–6	K2O 4.28
P2O5 2–4	P2O5 2.64
Al2O3 2–4	Al2O3 3.09
ZrO2 2–4	ZrO2 6.56
Others 0–10	CeO2 0.95
	Y2O3 0.12
	Ta2O5 0.12
	Na2O 0.51
	HfO2 0.07
	CaO 1.04
	MgO 0.27
	SO3 0.02
	Fe2O3 0.03
	Sm2O3 3.52
	TiO2 0.78
	Er2O3 0.5
	BaO 0.28
	V2O5 0.13
	Nb2O5 0.10
	NiO 0.04
	NaCl 0.02

Fig. 2 shows a set of representative FE-SEM images of the Obsidian® milling blocks taken at different magnifications with the BSED on unetched, polished specimens. It can be seen that they have the typical microstructure of glass-ceramics, with crystals embedded in a glass matrix without pores. This is because the Obsidian® milling blocks, as is the case with other milling blocks available in the dental market, are fabricated in two stages – first by cooling the glass melt to form dense ingots, and then heat treating the resulting glass ingots to induce the nucleation and pre-crystallization required for the subsequent preparation of dental prostheses by CAD/CAM methods. However, what is surprising is the high degree of crystallinity since, in principle, the crystals account for as much as ~85–90 vol.% of the microstructure (Figs. 2A–B). Apparently the crystals have micrometre sizes in the range ~1–10 μm (Figs. 2B–C), and most of them exhibit a rod-like shape, although some exhibit a snowflake-like shape. However, detailed FE-SEM observations with the BSED on specimens first polished and then etched with 4.9% HF for 30 s, such as the ones shown in Fig. 3, reveal that these crystals are much more complex than at first appears. Certainly, etching with HF brings out more contrast in the microstructure (Fig. 3A), revealing that what in principle seemed to be micrometre single-crystals are actually perlitic-like or dendritic-like structures formed by finer crystallites with glass in-between (Figs. 3B–C). The degree of crystallinity is then lower than 85–90%, but is nonetheless very high. Interestingly, the FE-SEM images (Figs. 2 and 3) show only two different compositional contrasts, thus indicating that the microstructure of the Obsidian® milling blocks is formed exclusively by a single type of crystals with light chemical elements (darker contrast) plus residual glass with heavier elements (lighter contrast).

Figure 2.

Representative FE-SEM micrographs of the Obsidian® milling blocks taken at magnifications of (A) 1000x, (B) 2500x, and (C) 8000x. Imaging was done with the BSED on polished specimens.

Figure 3.

Representative FE-SEM micrographs of the Obsidian® milling blocks taken at magnifications of (A) 10000x, and (B-C) 42000x. Imaging was done with the BSED on (A) a polished and partially-etched (HF for 30 s) specimen and on (B-C) polished and etched (HF for 30 s) specimens. The images in (B) and (C) show zones of the microstructure with rod-like crystals and snowflake-like crystals, respectively.

Fig. 4 shows the XRD pattern, including the corresponding peak assignations (Fig. 4A) and the Pawley refinement (Fig. 4B), of the Obsidian® milling blocks. It can be seen that the XRD pattern exhibits well-defined, intense diffraction peaks together with a very weak, broad hump in the low angle region (~17.5–32.5° 2θ). This is indeed the XRD pattern expected for glass-ceramics with little glassy phase, which corroborates the direct observations of FE-SEM (Figs. 2A–B). The indexing of the XRD pattern (Fig. 4A) indicates that these crystals are all lithium metasilicate (Li₂SiO₃), which is surprising given that the Obsidian® milling blocks were formulated with 23 different oxides. According to the Pawley refinement of the XRD pattern (Fig. 4B), performed with the orthorhombic space group Cmc2₁ of Li₂SiO₃, the lattice parameters of the Li₂SiO₃ crystals are a=9.374(9) Å, b=5.407(9) Å, and c=4.673(6) Å. Within the errors, these lattice parameters are similar to the average of the values reported in the PDF2 database for Li₂SiO₃ ($\overline{a}=9.38(2)\AA$, $\overline{b}=5.396(3)\AA$, and $\overline{c}=4.664(6)\AA$, when its PDFs cards are referred to the space group Cmc2₁). This rules out that the Li₂SiO₃ crystals contain spurious cations in solid solution, and therefore suggests that the other 21 oxides (i.e., all except LiO₂ and SiO₂) used in the formulation of the glass ingots would be subsumed into the glassy matrix. This is consistent with the two compositional contrasts observed by FE-SEM (Figs. 2 and 3).

Figure 4.

XRD pattern of the Obsidian® milling blocks showing (A) the peak assignations with the phase identification and (B) the plotted output from the corresponding Pawley refinement.

3.2. Microstructural evolution under non-isothermal heating

Fig. 5 shows the TG/DSC curves of the Obsidian® milling blocks registered from 40 to 1100°C in flowing air. As expected, it can be seen in the TG curve (Fig. 5A) that there is no mass variation (neither gain nor loss) in the entire temperature interval investigated. Firstly, there is no weight gain (i.e., < 0.06% if any at all) attributable to oxidation (which is the only possible process of mass gain) simply because the Obsidian® milling blocks are indeed oxide glass-ceramics. Secondly, there is no weight loss attributable to the formation and release of gases because there are no volatile oxides free in the microstructure (note that Li₂O formed Li₂SiO₃, and that the other 21 oxides used to fabricate the glass ingots are incorporated into the network of the SiO₂ glass). Regarding the DSC curve, it can be seen (Fig. 5B) that there is a subtle endothermic peak at ~485°C (T_g), which is attributable to a glass transition. Indeed, this temperature dictates the onset of a progressive endothermic event whose peak occurs at ~980°C (T_m,1), attributable to a gradual melting, and more specifically to the progressive softening of the glassy matrix up to the formation of a liquid. Concurrently there is a series of exothermic events, all attributable to crystallizations. The first exhibits a weak, broad exothermic peak at ~750°C (T_C,1). The others are less evident because they exhibit very weak exothermic peaks at ~810 (T_C,2), 820 (T_C,3), 910 (T_C,4), and 930°C (T_C,5). The weakness of these exothermic peaks indicates that the new crystals formed are minor phases. Finally, there is another endothermic peak at ~1005°C (T_m,2), attributable to a sudden, not gradual, melting of a crystalline phase. In sum, TG indicates that the Obsidian® milling blocks preserve their elemental chemical composition during the heat treatments, whereas DSC indicates that they are glass-ceramics with a high thermal stability and resistance to devitrification. This last is very distinctive and unusual because other lithia-based glass-ceramics do exhibit DSC (or DTA) curves with intense exothermic peaks attributable to more extensive crystallizations [2,26,27,34–37].

Figure 5.

Curves of (A) TG and (B) DSC of the Obsidian® milling blocks measured as a function of the temperature in the range 40–1100°C. Peak assignations are included in (B).

Fig. 6 shows the three-dimensional plot of the XRTD patterns collected in-situ during the non-isothermal heating of the Obsidian® milling blocks from 30 to 900°C in air. It can be seen that these glass-ceramics have a high thermal stability, exhibiting only a few, but relevant, changes of crystallinity at high temperatures. This observation is consistent with the deductions made by DSC (Fig. 5B). In particular, the XRTD patterns look essentially the same from room temperature up to ~750°C, above which temperature the diffraction peaks of Li₂SiO₃ undergo a continuous increase of absolute intensity. At ~810–820°C additional diffraction peaks appear, and no longer disappear, in the XRTD patterns, although they are very weak and broad compared to the diffraction peaks of Li₂SiO₃. This indicates (i) that Li₂SiO₃ is the only crystalline phase in these glass-ceramics up to ~810–820°C, and (ii) that above these temperatures the Li₂SiO₃ crystals coexist with other permanent crystals which are much less abundant and much smaller.

Figure 6.

XRTD patterns (angular interval 15–45° 2θ) of the Obsidian® milling blocks collected in-situ as a function of the temperature (T) in the range 30–900°C, in air.

Fig. 7 shows the two-dimensional contour plot generated from the XRTD patterns in Fig. 6. It can be seen after the corresponding phase identification (Fig. 7A) that the only change in the XRTD patterns from room temperature up to ~750°C is gradual shifting of diffraction peaks of Li₂SiO₃ towards higher diffraction angles. Consequently, the Obsidian® milling blocks retain their state of crystallinity up to ~750°C, accompanied only by the expected thermal expansion of the unit cell of the Li₂SiO₃ crystals. At ~750°C Li₂SiO₃ continues to be the only crystalline phase present, but its diffraction peaks (Fig. 7B) start to undergo (i) a marked increase of intensity, (ii) a notable decrease of broadening, (iii) an abrupt selective splitting into two (only for the multiplets of diffraction peaks labelled as ❹ and ❺), and (iv) a sudden shift towards either lower or higher diffraction angles. The first two observations are indicative of growth of the Li₂SiO₃ crystals coupled with increase of their inherent scattering power, whereas the last two observations reflect a complex modification of the crystal lattice of Li₂SiO₃ not ascribable to pre-melting effects (i.e., to configurational changes observed in other crystalline silicates ~200°C below the congruent melting point) [38–40]. Finally, at ~810–820°C additional diffraction peaks appear, attributable to lithium orthophosphate (Li₃PO₄) and cerium dioxide (CeO₂). These are permanent crystalline phases that no longer disappear, at least for heat treatments up to 900°C. Moreover, the detailed analysis of the XRTD patterns at and above 810–820°C indicates that (i) Li₂SiO₃ is the major phase while both Li₃PO₄ and CeO₂ are minor phases, (ii) the Li₂SiO₃ and Li₃PO₄ crystals are randomly oriented while the CeO₂ crystals exhibit preferred orientation along their <100> directions, (iii) the Li₃PO₄ and CeO₂ crystals are significantly smaller than the Li₂SiO₃ crystals, (iv) the Li₃PO₄ crystals formed at ~810–820°C and did not grow, and (v) CeO₂ continues to precipitate more and more as the temperature increases. Interestingly, greater precipitation of CeO₂ crystals occurred above 880°C, which was accompanied by a sudden contraction of the crystal lattice of Li₂SiO₃, as inferred from the shift of its diffraction peaks towards higher diffraction angles. This last, which is consistent with an earlier study on other dental-grade milling blocks available on the market [27], suggests that CeO₂ precipitated, at least partially, from Li₂SiO₃, and, consequently, that some Ce was necessarily dissolved in solid solution within the Li₂SiO₃ host.

Figure 7.

Two-dimensional intensity contour extracted from the XRTD patterns of the Obsidian® milling blocks, plotted (A) over the entire measured temperature range (30–900°C) and angular interval (15–45° 2θ), and (B) over the higher-temperature interval for more specific angular intervals (as indicated by the numbers in the figures). Line assignations are included (LMS=lithium metasilicate, Li₂SiO₃; LOP=lithium orthophosphate, Li₃PO₄; and cerium oxide, CeO₂).

Clearly, the evolution upon heating of the Obsidian® milling blocks is different from that observed in other lithia-based glass-ceramics [12,16,21,22,25], including the also commercially available IPS e.max CAD milling blocks [27], namely: (i) the glassy matrix did not crystallize as any form of SiO₂, (ii) the Li₂SiO₃ crystals did not react with the glassy matrix to form Li₂Si₂O₅ crystals, and (iii) ZrO₂ crystals did not precipitate from the glassy matrix despite the large ZrO₂ content (i.e., ~6.56 wt.%) used to fabricate the glass ingots. All of this is the result of the great thermal stability of the residual glassy matrix, chemically designed to contain a multitude of stabilizer oxides, and of the high crystallinity achieved during the pre-crystallization heat treatment. Regarding the specific comparison with the IPS e.max CAD material, the Obsidian® milling blocks consists of a higher Li₂O:SiO₂ ratio, a broader range of stabilizer oxides, a higher degree of crystallinity, and different types of crystals present in the pre-crystallized state. As a result, the subsequent firing produces very different crystallization pathways and microstructural developments in the two glass-ceramic materials. Whatever the case, the fact is that (i) the Obsidian® milling blocks stand out within the current portfolio of dental-grade glass-ceramics due to their unusual resistance to devitrification and phase separation upon heating, and that (ii) strictly speaking, the dental materials fabricated from them are not Li₂SiO₃ glass-ceramics reinforced with ZrO₂ particulates, rather Li₂SiO₃ glass-ceramics with the glass matrix being a glassy network reinforced with Zr⁴⁺ and many other cations.

Figs. 8 and 9 show the XRD pattern and a set of representative FE-SEM images, respectively, of the dental materials resulting from the non-isothermal heat treatment of the Obsidian® milling blocks up to 900°C. It can be seen from the XRD pattern (Fig. 8A) that the only crystalline phases present are Li₂SiO₃ (major), Li₃PO₄ (minor), and CeO₂ (minor, and textured). These are also the three phases identified by XRTD at 900°C (Fig. 7A), which rules out the occurrence of both reversible and additional transformations during the cooling stage. Interestingly, the comparison with the XRD pattern of the Obsidian® milling blocks (Fig. 4) reveals that the diffraction peaks of Li₂SiO₃ are now much sharper, indicating that the Li₂SiO₃ crystals grew during the non-isothermal heat treatment up to 900°C. Also, according to the Pawley refinement of the XRD pattern (Fig. 8B), performed with the orthorhombic space groups Cmc2₁ of Li₂SiO₃ and Pnmb of Li₃PO₄ and the cubic space group $Fm\overline{3}m$ of CeO₂, the lattice parameters of the Li₂SiO₃ crystals are a=9.402(9) Å, b=5.409(5) Å, and c=4.654(5) Å. These lattice parameters are slightly different (a is ~0.30% larger, b is ~0.04% smaller, and c is ~0.41% smaller) than those determined for the Li₂SiO₃ crystals in the Obsidian® milling blocks, which indicates that during the heat treatment the crystal lattice of Li₂SiO₃ was somewhat modified (not attributable to pre-melting effects [38–40]). More importantly, the microstructural observations performed by FE-SEM on unetched specimens with the BSED show that the microstructure changed markedly. In particular, it can be seen in the FE-SEM images (Figs. 9A–C) that there are now four different types of crystals embedded in the glassy matrix, namely: (i) fine crystals of Li₂SiO₃ (darker crystals), (ii) tiny nano-crystals of Li₂SiO₃ (darker crystals), (iii) nano-crystals of CeO₂ (lighter crystals), and (iv) tiny nano-crystals of Li₃PO₄ (grey crystals). Interestingly, the FE-SEM microstructural observations of the (intentionally broken) fracture surfaces, such as those shown in Fig. 10, indicate that the Li₂SiO₃ and Li₃PO₄ crystals are uniformly distributed everywhere (Figs. 10A–B), whereas the CeO₂ crystals are located only in the outermost part of the surface (first 1–3 μm in depth; Fig. 10A). This last suggests that the surrounding air atmosphere might play a fundamental role in their formation. Finally, it can also be seen from Figs. 9 and 10 that the fine crystals of Li₂SiO₃ are equiaxed or only moderately elongated, and in addition that they are single-crystals whose size is much smaller than that of the perlitic-like/dendritic-like structures observed in the microstructure of the Obsidian® milling blocks (Figs. 2B–C and 3A) but much larger than that of their crystalline constituents. This indicates that the microstructural evolution of the Obsidian® milling blocks necessarily involved the growth and spheroidization of the small crystallites existing within the perlitic-like/dendritic-like structures.

Figure 8.

XRD pattern of the glass-ceramic material resulting from the non-isothermal heating up to 900°C of the Obsidian® milling blocks showing (A) the peak assignations with the phase identification and (B) the plotted output from the corresponding Pawley refinement. The vertical lines in (A) mark the positions of the diffraction peaks of Li₂SiO₃ in the XRD pattern of the Obsidian® milling blocks, included here to facilitate the observation of the shift of the diffraction peaks after the heat treatment.

Figure 9.

Representative FE-SEM micrographs of the glass-ceramic material resulting from the non-isothermal heating up to 900°C of the Obsidian® milling blocks taken at magnifications of (A) 5000x, (B) 25000x, and (C) 35000x. Imaging was done with the BSED on polished specimens.

Figure 10.

Representative FE-SEM micrographs of the fracture surface of the glass-ceramic material resulting from the non-isothermal heating up to 900°C of the Obsidian® milling blocks taken at magnification of 35000x (A) close to the surface and (B) in the interior. Imaging was done with the BSED on unetched specimens.

3.3. Microstructural evolution during the commercial firing cycle

Figs. 11 and 12 show the XRD pattern and a set of representative FE-SEM images, respectively, of the dental materials obtained when the Obsidian® milling blocks are fired under the firing cycle recommended by the manufacturer, which is presented in Fig. 13. It can be seen (Fig. 11A) that the XRD pattern exhibits only diffraction peaks of Li₂SiO₃ and Li₃PO₄, together with a very weak, broad hump in the low angle region (~17.5–32.5° 2θ). The diffraction peaks of Li₂SiO₃ are sharper and much more intense than the diffraction peaks of Li₃PO₄, indicating that the Li₂SiO₃ crystals are larger (ultrafine size) and much more abundant (major phase) than the Li₃PO₄ crystals (nanometre size; minor phase). Certainly, it can be seen in the FE-SEM images (Fig. 12) that these dental materials are glass-ceramics whose microstructure consists of two types of small crystals embedded in a glassy matrix, namely: (i) abundant Li₂SiO₃ crystals of ultrafine size (< 500 nm) and rounded shape, and (ii) tiny Li₃PO₄ crystals of a nanometre size (< 100 nm). Also, these Li₂SiO₃ crystals are crystallographically different (not attributable to pre-melting effects [38–40]) from those in the Obsidian® milling blocks because the Pawley refinement of the XRD pattern (Fig. 11B), performed with the orthorhombic space groups Cmc2₁ of Li₂SiO₃ and Pnmb of Li₃PO₄, reveals that the lattice parameters of Li₂SiO₃ changed slightly, namely: a=9.393(9) Å (0.29% larger), b=5.403(7) Å (0.33% smaller), and c=4.651(5) Å (0.45% smaller). Interestingly, surface CeO₂ crystals did not form during the firing cycle, which is due to the lower temperature (820°C vs 900°C) and shorter duration (heating ramps of 90°/min and 40°C/min vs ~0.9°C/min) of the firing cycle in relation to the non-isothermal heat treatment. Indeed, complementary XRTD studies indicate that the formation of surface CeO₂ crystals is very sensitive to the temperature. Certainly, Fig. 14 shows two sets of XRTD patterns collected in-situ at selected temperatures when the Obsidian® milling blocks were fired (i) under the exact firing cycle recommended by the manufacturer (Fig. 14A) and (ii) under a slight modification of this cycle (Fig. 14B), demonstrating that a slight increase of only 15°C in the temperature of the firing cycle (i.e., 835°C instead of 820°C) causes surface precipitation of CeO₂ crystals.

Figure 11.

XRD pattern of the glass-ceramic material obtained when the Obsidian® milling blocks were subjected to the firing cycle indicated by the manufacturer, showing (A) the peak assignations with the phase identification and (B) the plotted output from the corresponding Pawley refinement. The vertical lines in (A) mark the positions of the diffraction peaks of Li₂SiO₃ in the XRD pattern of the Obsidian® milling blocks, included here to facilitate the observation of the shift of the diffraction peaks after the firing cycle.

Figure 12.

Representative FE-SEM micrographs of the glass-ceramic material obtained when the Obsidian® milling blocks were subjected to the firing cycle indicated by the manufacturer, taken at magnifications of (A) 5000x, (B) 20000x, and (C) 35000x. Imaging was done with the BSED on polished specimens.

Figure 13.

Firing cycle recommended by the manufacturer for the crystallization of the glass-ceramic material prepared from the Obsidian® milling blocks, prior to its placement in the mouth. The different segments are separated by the vertical dotted lines. The heat-treatment atmosphere is also indicated. In total, the firing cycle lasts for ~24 min.

Figure 14.

XRTD patterns (angular interval 15–45° 2θ) collected in-situ at selected temperatures when the Obsidian® milling blocks were subjected to (A) the firing cycle recommended by the manufacturer and (B) a slight modification of this cycle (everything equal except soaking at 835°C instead of 820°C). The XRTD patterns were acquired at room temperature before the heat treatments (i.e., at 30°C, the bottom pattern), then at the conclusion of the drying segment (i.e., after 3 min at 400°C), next after the conclusion of the soaking phase at maximum temperature (i.e., after 10 min at 820°C (A) or at 835°C (B)), then at the end of the stage of controlled cooling (i.e., at 680°C), and finally once the dental pieces had cooled naturally to room temperature (i.e., at 30°C).

To demonstrate the effect of heat treatment temperature on the microstructural evolution of Obsidian®, Fig. 15 shows the FE-SEM micrographs of the pre-crystallized Obsidian® milling blocks (Fig. 15A) and following the firing cycle indicated by the manufacturer (Fig. 15B). For comparison, the microstructure of the same material subjected to non-isothermal heat-treatment up to 900°C (i.e., 80°C higher than the recommended heat-treatment temperature) is also included (Fig. 15C).

Figure 15.

Representative FE-SEM micrographs of (A) the Obsidian® milling blocks, and the glass-ceramic materials obtained when the Obsidian® milling blocks were subjected to (B) the firing cycle indicated by the manufacturer and (C) a non-isothermal heat treatment up to 900°C, taken at magnifications of 10000x. Imaging was done with the BSED on (A) polished and etched (HF for 30 s) specimens and (B-C) polished specimens.

To further elucidate the microstructural evolution of Obsidian®, Fig. 16 shows the FE-SETM micrographs of the Obsidian® milling blocks taken in-situ at 400°C (Fig. 16A), 825°C (Fig. 16B), 850°C (Fig. 16C), and 875°C (Fig. 16D). Imaging was done on the same area (to monitor the microstructural evolution) with the GSED on a polished specimen. The temperatures registered by FE-SETM are not directly comparable with those registered by XRTD or by the conventional furnace due to the different geometries of the heating chambers used, but it is clear that the FE-SETM observations show the growth and spheroidization of the small crystallites existing within the perlitic-like/dendritic-like structures.

Figure 16.

Representative FE-SETM micrographs of the Obsidian® milling blocks taken in-situ at a magnification of 5000x at (A) 400°C, (B) 825°C, (C) 850°C, and (D) 875°C. Imaging was done on the same area (to monitor the microstructural evolution) with the GSED on a polished specimen. The temperatures registered by FE-SETM are not directly comparable with those registered by XRTD or by the conventional furnace due to the different geometries of the heating chambers used [27].

4. Discussion

The present study demonstrates that the Obsidian® milling blocks are glass-ceramics with great thermal stability (i.e., with high resistance to devitrification and phase separation) whose microstructural evolution upon heating is fascinating. On the one hand, they undergo only merely marginal crystallization (i.e., precipitation of Li₃PO₄ nanocrystals, and of CeO₂ nanocrystals at the surface if overheated), but on the other hand, they undergo an intense microstructural change. Specifically, the FE-SEM images in Fig. 15 show that, during the heat treatments, their microstructure evolves from containing micrometre-sized, perlitic-like/dendritic-like Li₂SiO₃ crystallites (Fig. 15A) to ultrafine, rounded Li₂SiO₃ single-crystals (Fig. 15B) whose size and aspect ratio both increase with increasing firing temperature (Fig. 15C). It is proposed that this microstructural evolution occurs as follows.

Initially, glass ingots are prepared using a combination of 23 different oxides (Fig. 1 and Table 1), which results in a glass with SiO₂ as the main network-forming oxide plus a multitude of intermediate and network-modifier oxides. Subsequently, these glass ingots are subjected to a pre-crystallization heat treatment during which the tiny Li₂SiO₃ nanocrystals nucleate and crystallize abundantly within the glass forming submicrometre-sized perlitic-like/dendritic-like structures (Figs. 2–4). The pre-crystallization consumes a great part, but not all, of the LiO₂ available and part of the SiO₂, so that the unconsumed SiO₂ together with the other oxides added form a glassy matrix distributed between g and within the perlitic-like/dendritic-like structures. During the later firing cycle (also termed crystallization cycle), there are no essential changes in the microstructure upon heating up to ~750°C. Above ~750°C, the already-softened glassy matrix (Fig. 5) has a viscosity that allows mass transport (i.e., the diffusion of Li, Si, and O) between the Li₂SiO₃ nanocrystals within the perlitic-like/dendritic-like structures. In this scenario, the tiny Li₂SiO₃ nanocrystals grow and spheroidize through solution-reprecipitation, in a process akin to what occurs during the liquid-phase sintering of polycrystalline ceramics [41–43]. As can be seen from Fig. 16, the direct observations of the microstructural evolution of the Obsidian® milling blocks taken in-situ by FE-SETM at different temperatures confirm the reasonableness of this proposed microstructural development mechanism.

The driving force for grain growth and spheroidization is the reduction in total Gibbs free energy due to the decrease of both the surface energy of the crystals and the interfacial energy at the crystal-glass interfaces. Coarsening then becomes more evident with increasing firing temperature because the larger Li₂SiO₃ nanocrystals grow progressively at the expense of the smaller ones (thus maintaining the volume fraction of Li₂SiO₃ crystals fairly constant). This process, termed Ostwald ripening, is accompanied by the precipitation of very small Li₃PO₄ nanocrystals from the glass matrix (Figs. 7–12), which otherwise remains thermally very stable. Once formed, Li₃PO₄ constitutes a permanent crystalline phase, and the fact that the tiny Li₃PO₄ nanocrystals are distributed everywhere in the glassy matrix (Figs. 9, 10, and 12) suggests that their formation most likely did not consume Li from the Li₂SiO₃ nanocrystals but rather the residual LiO₂ from the glass network. As commonly observed in liquid-phase-sintered ceramics [41–43], coarsening of the Li₂SiO₃ nanocrystals would also come accompanied by a subtle modification of their crystal structure (Figs. 7, 8, and 11) because small amounts of other cations dissolved in the glass would also reprecipitate together with Li, Si, and O, thus forming solid solutions with slightly modified lattice parameters. As occurred in particular for CeO₂ (Figs. 7–10), precipitation of other oxides would take place when overheated once the solubility limit of these cations in the crystal lattice of Li₂SiO₃ is exceeded, a process that nonetheless only occurs at the surface of the dental restorations because it would require a minimal uptake of oxygen from the surrounding atmosphere (indeed, imperceptible under TG).

It is interesting to note that the glassy matrix exhibits high thermal stability. This derives from the multitude of stabilizer cations in its network, which provides the Obsidian® milling blocks with a set of features that distinguish them from other milling blocks available on the market. First, the glass is resistant to both devitrification and phase separation, and therefore, if not overheated, it neither crystallizes as any form of SiO₂ nor precipitates particulate elemental oxides. This last is the case including for ZrO₂ even though the chemical formulation of the glass ingots contained an important concentration of this oxide. Second, the glass is very inert, and therefore does not react chemically with Li₂SiO₃ to form the Li₂Si₂O₅ crystals that typically develop in other dental-grade glass-ceramics in-situ during the firing cycles. Consequently, the dental restorations prepared from the Obsidian® milling blocks have to be catalogued as being Li₂SiO₃ glass-ceramics with glass of cationically-reinforced glassy network.

The results of the present study are scientifically interesting, but may also be of great interest for clinical practice. In particular, they demonstrate that, by choosing appropriately the temperature of the firing cycle, it is feasible to tailor the microstructure of dental restorations prepared from the Obsidian® milling blocks. Although in all cases it is a glassy matrix with similar degrees of crystallinity, the crystalline component of these glass-ceramics can nonetheless be judiciously tuned from micrometre-sized perlitic-like/dendritic-like structures composed of tiny Li₂SiO₃ nanocrystals and rounded Li₂SiO₃ single-crystals with an average size that is modifiable within the ultrafine and coarse scales plus tiny Li₃PO₄ nanocrystals. Future studies are thus needed to elucidate to what extent this microstructural repertoire enables the optimization of the aesthetic and mechanical/tribological properties of these dental materials, and consequently the performance and lifespan of the corresponding dental restorations. Although this pending investigation is vital to the province of the clinical practice, it is subordinated to first understand the crystallization pathway and microstructural development, and thus, lies beyond the scope of the present study.

5. Conclusions

The microstructural development of the Obsidian® milling blocks during crystallization firing has been elucidated using a wide battery of in-situ and ex-situ characterization techniques. Based on the experimental results and analyses, the following conclusions can be drawn:

1. The Obsidian® milling blocks are glass-ceramics with a complex chemical composition, obtained from the pre-crystallization of glass ingots formulated with more than 23 different elemental oxides.

2. The Obsidian® milling blocks are highly pre-crystallized glass-ceramics whose microstructure consists of submicrometre-sized perlitic-like/dendritic-like structures formed by Li₂SiO₃ nanocrystals immersed in a glass matrix. The structure of the residual glass is constituted by SiO₂ as the main network-forming oxide, plus a multitude of intermediate and network-modifier oxides.

3. The Obsidian® milling blocks possess very high thermal stability during the firing cycles, exhibiting great resistance to devitrification and phase separation. In particular, the glassy matrix does not crystallize, elemental oxides do not precipitate (if not over-heated above the crystallization temperature), and the Li₂SiO₃ crystals do not react with the glass matrix. Thus, the corresponding dental restorations placed in the mouth are Li₂SiO₃ glass-ceramics with glass of cationically-reinforced glassy network.

4. The microstructural development of the Obsidian® milling blocks involves the spheroidization and growth through solution-reprecipitation processes in the softened glass of the original Li₂SiO₃ nanocrystals within the perlitic-like/dendritic-like structures, plus the precipitation of new Li₃PO₄ nanocrystals from the glass matrix. This opens up new avenues for designing and optimizing the composition and microstructure of these dental glass-ceramics with much improved clinical performance and lifespan.