Bioceramics in the CaMgSi2O6–Li2O System: A Glass‐Ceramic Strategy for Excellent Mechanical Strength and Enhanced Bioactivity by Spontaneous Elemental Redistribution

A novel glass‐ceramic strategy for synthesizing mixed phase diopside (CaMgSi2O6)–lithium oxide (Li2O) bioceramics with excellent mechanical strength, superior biodegradation resistance, low environmental pH impact, enhanced bioactivity, and reasonable biocompatibility is developed for biomedical applications. The substitution of Li2O for MgO in CaMgSi2O6 stimulates the formation of secondary phases: CaSiO3, Li2Si2O5, SiO2, Li2SiO3, and Li2Ca2Si5O13. The evolution of CaSiO3 improves the surface hydroxyapatite (HAp) formation but lowers the mechanical strength and biological resistance, while the amorphous Li2Si2O5 phase tremendously reinforces the bioceramics by densifying the microstructure, indicating the simultaneous enhancement of bioactivity, mechanical strength, and durability. The promoted HAp formation is induced by the elemental redistribution where Mg elements are concentrated in large CaMgSi2O6 grains embedded in Li2Si2O5 amorphous matrix, which hinders the Mg2+ release and its readsorption by HAp. The cell viability is affected by Li2O substitution because of the high‐dose Li+. In the current work, Li0.25 (25 mol% Li2O) has the highest hardness (700 Hv as sintered and 197 Hv after simulated body fluid soaking), lowest weight loss (≈0.6 wt%), lowest pH variation (≈8.1), efficient HAp formation, and reasonable cell viability (70.5%), demonstrating its remarkable potential for bone implant applications due to the synergistic structural densification and biological improvement.

achievable in such systems in terms of biodegradation, bioactivity and mechanical strength. [17,18] As with Ca and Si, magnesium (Mg) is also an indispensable element for the human body, and bone regeneration in particular as Mg deficiency would stimulate the differentiation of osteoclasts and inhibit the production of osteoblasts, thus leading to osteoporosis. [19] Further, it has been proved that Mg-containing calcium silicatebased bioceramics can stimulate the apatite formation and induce a tight bond with bone tissues. [20,21] Lai et al. [22] suggested that an increase of new vessel development and blood perfusion after surgery was achieved by Mg-incorporated PLGA/TCP porous scaffolds. Further, Mg-containing thin films deposited on porous polymer scaffolds could significantly develop new bone formation and strengthen mechanical properties of newly formed bone, [23] indicating that Mg-containing bioceramics could be used as a novel material for bone tissue engineering applications.
Due to the various attractive attributes of calcium magnesium silicates as effective bioceramics, such materials from various silicate families have been selected as bone-tissue engineering materials, including the sorosilicate akermanite (Ca 2 MgSi 2 O 7 ), [24,25] orthosilicate monticellite (CaMgSiO 4 ), [26,27] and the pyroxene diopside (CaMgSi 2 O 6 ). [28][29][30] Among all the materials, diopside has superior mechanical properties, [31] and also possesses good cytocompatibility and surface apatite precipitation ability. [32,33] Consequently, diopside (CaMgSi 2 O 6 ) was chosen as the primary bioceramic material in the present study. It is worth noting that the mechanical properties of CaMgSi 2 O 6 was improved in comparison with that of CaSiO 3 , while the HAp mineralization in simulated body fluid (SBF) was compromised due to the incorporation of Mg. [34][35][36][37] Toward the repair of defects, the mechanical and biological properties of bioceramic implants should be compatible with native bone tissue and stimulate a favorable biological response. [38] One of the current techniques for addressing these issues is to introduce additional elements into main structures (doping [39][40][41][42] and/or composites [43,44] ) to modify the crystallinity, lattice parameters, and morphology of the bioceramics, further affecting their mechanical and biological properties. Table 1 summarizes the previous works on the incorporation of additional elements in CaMgSi 2 O 6 bioceramics.
Lithium (Li) has been commonly used to treat bipolar disorder, [50,51] and the psychiatric patients who took Li medication have been reported to have enhanced bone mass and lower bone turnover states. [51] This may be explained by the mechanism that Li can stimulate the osteogenesis through activating Wnt and Hh signaling pathway. [52,53] Moreover, it was found that Li doping can improve the thermal stability of calcium phosphate. [54] Table 2 summarizes the previous works on the incorporation of Li element in different types of bioceramics. Padmanabhan et al. [55] prepared lithium-substituted hydroxyapatite (Li-HAp) nanoparticles by a sonochemical synthesis process. It was reported that the incorporation of Li into HAp caused the transformation of particles from a rod shape to a needle-like structure due to the sonication effect. Further, Li-HAp showed excellent antimicrobial activity and blood compatibility, and could be used as a drug delivery carrier owing to its improved cytocompatibility. Yuan et al. [56] synthesized Li-doped calcium polyphosphate bioceramic scaffolds by a gravity sintering method. The study concluded that Li doping did not affect the crystallized phase but changed the surface morphology. Further, the Li-doped samples were found to have better degradation properties and enhanced proliferation and differentiation of osteoblasts. Wang et al. [57] synthesized Li-doped HAp by a precipitation method, where the scaffolds had better mechanical strength and Li doping was found to be beneficial to the proliferation of osteoblasts. Li-doped β-TCP was found to have two times higher compressive strength than initial β-TCP. [58]  Li Na K 2 mol% 900 °C/2 h • The crystallinity and lattice volume of diopside were changed with the most and least deviations for K and Na, respectively • The dopants altered the in vitro bioactivity of diopside in the following ranking: K-doped > Lidoped > Na-doped > undoped • The 2 mol% dopants improved the biocompatibility of diopside, where the most beneficial effect was found for Na and K • K-doping was the optimal doping for the bioactivity and cytocompatibility assessments [47] Ce 0-100 mol% 1000 °C/4 h • Addition of 25 mol% Ce had the best biomineralization performance in vitro • Less hydroxyapatite precipitates were found with further increasing Ce addition [48] Mo 0-100 mol% 1000 °C/4 h • At a lower Mo content, the mixed phase materials showed higher hardness and slower biodegradation • At a higher Mo content, mixed phase materials exhibited lower hardness and bioactivity [49] Adv. Mater. Interfaces 2023, 10,2202491 www.advmatinterfaces.de The incorporation of nanoporous lithium doping magnesium silicate into calcium sulfate hemihydrate was found to enhance the degradability, biocompatibility, vascularization, and osteogenesis. [59] As summarized above, the incorporation of Li into calcium phosphate-based bioceramics was expected as a viable way to enhance the mechanical and biological properties of bioceramics. However, few studies have been conducted on bioceramics in the diopside-lithium oxide system. Only one study synthesized Li-doped diopside through a coprecipitation method, as shown in Table 1. The result showed that the substitution of 2 mol% Li for 1 mol% Mg has improved the in vitro bioactivity and biocompatibility.
As the amounts of Li incorporated or substituted into diopside and other bioceramics were mostly limited to a maximum of 2 mol%, the present work attempts to investigate the effects of a wide range of Li content (25,50,75, and 100 mol%) on the diopside-lithium oxide (CaMgSi 2 O 6 -Li 2 O) system to determine the appropriate Li amount and thereafter optimize its mechanical and biological properties. The specimens were synthesized through a precipitation method, followed by sintering at 1000 °C for 4 h. The mineralogical, morphological, mechanical, and biological properties were investigated to comprehensively understand the factors and mechanisms affecting the properties of CaMgSi 2 O 6 -Li 2 O bioceramics system.

Crystalline Phase Identification
The phase identification of the pure diopside and CaMgSi 2 O 6 -Li 2 O bioceramics was obtained by X-ray diffraction (XRD), as demonstrated in Figure 1. In order to precisely inves-tigate the crystallized phases in the bioceramics, the as-sintered samples were ground into powder for random crystalline orientation and then measured by XRD, as shown in Figure 1a. It can be seen that a major phase of CaMgSi 2 O 6 , corresponding to ICSD code: 168107, was observed in the pure diopside. As the Li 2 O substitution increased to 25 mol%, the formation of secondary phases (CaSiO 3 , minor Li 2 Si 2 O 5 , and SiO 2 ) was detected in accordance with ICSD code: 201537, 15414, and 81382, respectively. The CaSiO 3 was identified as β-CaSiO 3 with triclinic structure which is generally formed below 1100 °C. [61] As the Li 2 O content increased to 50 and 75 mol%, the peak intensity of CaMgSi 2 O 6 phase decreased along with the increment of peak intensity in CaSiO 3 , Li 2 Si 2 O 5 , and SiO 2 phases, and two new secondary phases Li 2 [62] In order to study the crystallization behavior of the bioceramics in biological environment, the XRD patterns of the bioceramics pellets before and after SBF soaking for 28 days are compared in Figures 1b and 1c, respectively. The slight difference in peak intensity and peak position between the    [37] that the decrease of MgO in CaO-MgO-SiO 2 -P 2 O 5 bioceramics promoted the apatite formation rate and the vulnerability to aqueous solution because of better seeded HAp crystallization and lower CaO bonding energy than MgO.

Morphology and Elemental Distribution
The scanning electron microscope (SEM) images (5000×) for sintered pure diopside and CaMgSi 2 O 6 -Li 2 O bioceramics pellets are shown in Figure 2. The surface morphology of pure diopside without Li 2 O addition was homogenous with a distribution of small pores, as shown in Figure 2a. With 25 mol% of Li 2 O substitution for MgO, a microstructure consisting of small grains embedded in a glass-like substrate was clearly observed in Figure 2b. The presence of amorphous phase was also detected by XRD and degree of crystallinity analysis in Table 3. Moreover, with the increase of Li 2 O addition, the aggregation and grain growth phenomena became more apparent, leading to an increasingly porous structure, as shown in Figure 2c-e. The size of aggregated grains was getting larger from 3 to 12 µm in length and 2 to 5 µm in width. Li et al. [35] have reported that the incorporation of MgO into CaSiO 3 gave rise to the formation of calcium-magnesium silicates, and therefore resulted in a compact surface and a decrease of total pore volume, and vice versa. However, in the present study, a fused and densified surface was observed in Li0. 25 [63] This is so-called "glass-ceramics" composed of a Li 2 Si 2 O 5 crystalline phase and an amorphous matrix, which is induced by a heterogeneous nucleation mechanism. [64][65][66] In the current study, the viscous amorphous matrix of Li 2    www.advmatinterfaces.de the Li 2 O content reached 100 mol%, the Mg signal has completely disappeared and the large grains of CaSiO 3 partially embedded in the glass phase of Li 2 Si 2 O 5 were clearly detected. As can be seen from CaMgSi 2 O 6 -Li 2 O bioceramics, it is noteworthy that Mg elements were mostly concentrated in the crystalline phase of CaMgSi 2 O 6 , while Ca elements were not only found in the crystalline phases of CaSiO 3 and CaMgSi 2 O 6 , but heavily detected in the amorphous glass phase. This phenomenon may be explained by a mechanism that Mg 2+ has higher field strength than Ca 2+ and Li + , which increases the viscosity of the glass phase, leading to a reduced diffusion of ions and ionic complexes. [68] Because of the poor ion diffusion in the regions containing concentrated Mg 2+ ions, Mg 2+ may have a strong tendency to nucleate CaMgSi 2 O 6 crystals with the localized Ca 2+ , Si 4+ , and O 2− ions. By contrast, partial Ca 2+ can be dispersed in the glass phase of Li 2 Si 2 O 5 due to its high ionic diffusivity without the influence of Mg 2+ .
The SEM images (5000×) for pure diopside and CaMgSi 2 O 6 -Li 2 O bioceramics pellets after soaking in SBF solution for 28 days are shown in Figure 4. The presence of newly formed HAp layers can be clearly observed on all the samples. For pure diopside, it can be seen that the diameter of the spherical precipitates on the sample surface varied from 1 to 2 µm. Regardless of the initial matrix structure, the diameter of the spherical precipitates on all the CaMgSi 2 O 6 -Li 2 O bioceramics samples ranged from 5 to 12 µm, which was much larger than  Figure 1c. The improved HAp formation could be attributed to the appearance of CaSiO 3 and Li 2 Si 2 O 5 caused by the Li 2 O addition, where CaSiO 3 has been reported to have better apatite mineralization but more rapid dissolution rate than CaMgSi 2 O 6 . [34,35] Furthermore, the quantitative compositions of HAp formed on pure diopside and CaMgSi 2 O 6 -Li 2 O bioceramics were measured by the point mode of FE-EPMA, as indicated by yellow arrows in Figure 4, and the results were tabulated in Table 4. This quantitative analysis mainly measured the concentrations of Ca, Mg, Si, O, and P as they are the major compositions of CaMgSi 2 O 6 -based bioceramics and HAp precipitates. Other elements, such as undetectable Li, gold coating used for enhancing conductivity, or the elements adsorbed from SBF solution, were classified as "others." It is clear that the HAp formed on CaMgSi 2 O 6 -Li 2 O bioceramics contained Ca, O, and P as its major components, corresponding to the composition of Ca 5 (PO 4 ) 3 OH. In addition to Ca, O, and P, the HAp formed on the pure diopside had a noticeable amount of Mg and Si, indicating that it readsorbed the Mg and Si elements which were released from the matrix materials into SBF solution.
In particular, the elemental mapping analysis of Li0.25 bioceramics after soaking in SBF for 28 days was performed and shown in Figure 5. It can be seen that the soaked Li0.25 bioceramics was comprised of Ca, Mg, Si, O, and P ions. The elemental mapping implied the development of a uniform microstructure with regard to an appropriate spatial distribution of elements. High concentrations of Ca and P ions were found on the surface of HAp precipitates, while high concentrations of Mg and Si ions, and a medium concentration of Ca ions were found on the surface of the Li0.25 matrix, which were consistent with their chemical compositions. Table 5 shows the bulk and apparent density of sintered pure diopside and CaMgSi 2 O 6 -Li 2 O bioceramics pellets. Following the density results, the porosity given by open pores was calculated accordingly. As it can be seen, the apparent density (D A )   [70] 2.5, [71] and 2.26 [72] g cm −3 , respectively, all of which are lower than the theoretical density of CaMgSi 2 O 6 at 3.26 g cm −3 . Further, Li0.25 had an extremely low porosity of 0.3%, indicating a dense structure which is in good agreement with the observation in SEM ( Figure 2b) (Figure 2b-e).

Hardness Measurement
Hardness results of sintered pure diopside and mixed phase CaMgSi 2 O 6 -Li 2 O bioceramics before and after immersing in SBF solution are shown in Figure 6. Each sample was indented at three different locations, and then the averages and standard deviations were calculated. It was expected that the addition of Li 2 O would affect the hardness property through different   Table 5 and Figure 6a, it was found that the hardness was directly proportional to the apparent density and inversely proportional to the porosity. The pure diopside had the highest apparent density, but the high porosity of 45.8% yielded a fragile structure and a low hardness. With the incorporation of Li 2 O, Li0.25 had a similar porosity but a higher apparent density compared to Li0.50, leading to a superior hardness of Li0.25 to that of Li0.50. As the increment of Li 2 O from 50 to 75 and 100 mol%, the apparent densities of Li0.50, Li0.75, and L1.00 remained at ≈2.7 g cm −3 , while the porosity was increased from 0.7% to 21.3% and 41.1% with a remarkable drop in hardness from 521 down to 416 and 318 Hv. Based on the XRD results (Figure 1), the secondary phases (CaSiO 3 and Li 2 Si 2 O 5 ) occurred in Li0.25, suggesting that the highest hardness measured in Li0.25 was associated to the synergistic effect of CaMgSi 2 O 6 , CaSiO 3 , and Li 2 Si 2 O 5 . In addition, an amorphous phase of 4.46% was able to increase the bulk density of Li0.25 and reduce its porosity substantially. As elucidated in SEM and density measurement, the Li 2 O substitution varied the porosity and the density of Li0.25, inducing an aggregated and densified crystal structure as the reason of its outstanding hardness. With further increasing the Li 2 O quantity, though the amorphous phase increased to 8.37-15.44%, the Mg deficiency and the abundant CaSiO 3 and Li 2 Si 2 O 5 phase triggered progressive aggregation and grain growth, followed by a lowered density and a porous morphology, as shown in Figure 2c- Figure 6b), which was due to the biodegradation behavior during soaking process. Among all the soaked samples, Li0.25 retained the highest hardness of 197 Hv, which is much higher than other samples' hardness ranging from 4 to 12 Hv, revealing a strong resistance to biodegradation. In comparison with dense (>93%) TCP bioceramics synthesized by novel ultrafast high-temperature sintering, the CaMgSi 2 O 6 -Li 2 O specimens have much higher hardness than dense TCP (maximum: 350 Hv). [75] In addition, compared to the most commonly used Ti-based alloys for bone implant applications, the hardness of as-sintered Li0.25 at 700 Hv is 1.7 times that of Ti-6Al-4 V alloys at 416 Hv, while the hardness of soaked Li0.25 at 197 Hv is comparable to that of Ti-25Nb-3Zr-3Mo-2Sn alloys at 202 Hv. [76] The decreased hardness as a function of soaking time may allow Li0.25 to be more favorable for bone repair than Ti-based alloys as the cortical bone has a relatively low hardness around 40-60 Hv. [77] The deteriorated hardness of Li0.50, Li0.75, and Li1.00 after SBF soaking could be ascribed to the escalating CaSiO 3 formation as CaSiO 3 has been reported to have a higher dissolution rate than diopside. [34,35]

Dissolution and Precipitation Behaviors in SBF Solution
The dissolution and precipitation behaviors of pure diopside and CaMgSi 2 O 6 -Li 2 O bioceramics were evaluated by the weight loss of the samples as well as the variations of the pH values and ion concentrations of the SBF solution after soaking in SBF for certain periods. To detect the degradability of pure diopside and CaMgSi 2 O 6 -Li 2 O bioceramics, the variations of sample weight during the immersion in SBF solution were measured and shown in Figure 7. It is clear that pure diopside and Li0.25 had smaller percentage of weight loss between 0% and 2%, while bioceramics with the higher concentrations of Li 2 O substitution had higher percentage of weight loss around 4% after 28 days of immersion time.

www.advmatinterfaces.de
The mass variation is related to the biodegradation of the sample (weight loss) and the HAp formation on the sample surface (weight gain). From the results of overall weight loss, it can be suggested that the biodegradation of all the samples was predominant over HAp formation (especially at higher concentrations of Li 2 O) as the sample weights were reduced. For bone tissue repair applications, bioceramics with suitable biological degradation are needed. Thus, Li0.25 sample was considered to be favorable for repairing bone defects due to its slowest degradation rate, which could be attributed to the compact structure, as observed in Figure 2b. The samples with Li 2 O addition more than 25 mol% revealed the accelerated biodegradation properties in accordance with the decreased hardness (Figure 6b), which could be resulted from the growing secondary phase of CaSiO 3 with higher dissolution rate. A similar result was obtained by Du et al. [78] that the weight loss of pure CaSiO 3 was more severe than that of Mg-incorporated CaSiO 3 , which could be due to the stronger MgO bonding than the CaO bonding. [37] Compared to another CaSi-based bioceramic: baghdadite (Ca 3 ZrSi 2 O 9 ), where the network of ZrCa bonds stabilizes the crystal structure and enables baghdadite a 70% higher compressive strength than CaMgSi 2 O 6 , the biodegradation rates of CaMgSi 2 O 6 -Li 2 O bioceramics in the current work were much lower than that of baghdadite (weight loss: 4-10%). [79] The pH values in the SBF solution following immersion of bioceramics samples are shown in Figure 8a. The upward trend of pH values for all the samples during 28 days of soaking indicates a sustained ion exchange of Ca 2+ , Mg 2+ , and Li + with H + and H 3 O + in the SBF solution, followed by the formation of amorphous silica-rich colloid layer and the subsequent formation of HAp. [49,80] In the first 7 days, all the CaMgSi 2 O 6 -Li 2 O bioceramics, especially those with higher levels of Li 2 O addition, had significantly higher pH values than pure diopside. After 1 week of soaking, the rapidly rising trends in pH value for all the samples tended to level off, indicating that the ion exchange between the samples and the SBF solution gradu-ally reached saturation. After 28 days of soaking, Li0.25 had a similar pH value to pure diopside at around 8.0, while the pH values of Li0.50, Li0.75, and Li1.00 were about 8.3, implying that more ions were exchanged with higher Li 2 O substitution. As mentioned above, the ion exchange between the cations of bioceramics (Ca 2+ , Mg 2+ , and Li + ) and the cations of SBF (H + and H 3 O + ) is a prerequisite of superficial HAp formation. In other words, the higher pH values stood for the more frequent ion exchange and better silica-rich layer formation for HAp deposition. This observation is consistent with XRD and SEM results that the increase of Li 2 O content would stimulate the generation of CaSiO 3 and thus improve the HAp formation. Du et al. [78] obtained an analogous outcome that the pH value of soaking medium provided by pure CaSiO 3 was higher than Mg-incorporated CaSiO 3 . Porosity may have negligible effect on the pH value as the comparably high pH values were given by Li0.50, Li0.75, and Li1.00 with different porosities. Since a weak alkaline microenvironment would improve osteoblast production and hinder osteoclast differentiation, [81] the Li0.25 sample is considered to be more suitable for implant applications because of its outstanding hardness ( Figure 6), low degradation rate (Figure 7), reasonable HAp formation, and minor influence on the environmental pH value.
The changes of ion concentration in SBF solution during immersion are shown in Figure 8b-f. In this test, the ion concentration of SBF was dominated by three reactions: 1) ion exchange to form silica-rich layer, followed by an increase in pH value, and an increase in the concentration of Ca 2+ , Mg 2+ , and Li + ; 2) adsorption of ions by the silica-rich layer to form HAp, followed by a decrease in the concentration of ions, especially in Ca 2+ and P 5+ ; 3) dissolution of bioceramics and HAp, followed by an increase in the concentration of each ion. Thus, it can be summarized that ion exchange and material dissolution led to an increase in ion concentration, while the formation of HAp caused a decrease in ion concentration.
First, the interaction mechanism between pure diopside and SBF solution is discussed as the followings. Pure diopside mainly exchanged Ca 2+ and Mg 2+ for H + and H 3 O + in SBF, forming a silica-rich layer on the surface. As can be seen in Figure 8c, the Mg 2+ concentration of pure diopside kept increasing, representing a greater driving force for ion exchange of Mg 2+ and material dissolution than for adsorption. In Figure 8b, the Ca 2+ concentration of pure diopside at day 14 was slightly lower than that of fresh SBF, indicating that more Ca 2+ ions were adsorbed by the silica-rich layer than they were released, and HAp was being formed at this stage. By day 28, the Ca 2+ concentration has risen, suggesting that the adsorption of Ca 2+ has reached saturation; then the ion exchange and material dissolution have outpaced the adsorption. The change of P 5+ concentration in Figure 8f also revealed that P 5+ had not yet been fully adsorbed at day 14, confirming that HAp was still being formed by Ca 2+ and P 5+ at this stage. By day 28, P 5+ ions were completely consumed, indicating the saturation of ion adsorption and the completion of HAp formation.
Second, the interaction mechanism of Li0.25 is discussed. As can be seen from Figure 8b, the Ca 2+ concentration of Li0.25 was slightly higher than that of fresh SBF at day 14 and then remained stable until day 28, suggesting a two-way

www.advmatinterfaces.de
balance between the release and adsorption of Ca 2+ , with a continuous HAp formation. Based on Figure 8f, P 5+ ions have not been completely consumed after 28 days, verifying an ongoing ion adsorption and HAp formation throughout the immersion. Figure 8c shows that the Mg 2+ concentration of Li0.25 was slightly lower than that of fresh SBF at both day 14 and 28,

www.advmatinterfaces.de
indicating that the adsorption rate of Mg 2+ was higher than the release rate. In Figure 8d, Li0.25 displayed a steady release of Li + over 28 days of immersion.
Li0.50, Li0.75, and Li1.00 were grouped into the same category for discussion. As can be seen from Figure 8b, the Ca 2+ concentration of these three samples increased continuously from ≈100 to ≈265 ppm throughout the 28-day immersion period, representing that the ion exchange and material dissolution were much more driven than ion adsorption and thus HAp growth had been completed. Figure 8f shows that the P 5+ ions were fully consumed by day 14, further confirming a complete formation of HAp. Based on Figure 8c, these three samples were identical to Li0.25 in that the adsorption rate of Mg 2+ was slightly higher than the release rate. Figure 8d illustrates that the Li + concentration increased with the increase of soaking period and the amount of Li 2 O. Figure 8e shows the Si 4+ concentrations of all samples, and it can be seen that all the samples continued to release Si 4+ . Huang et al. [80] stated that the dissolution of silicon ions was dependent on not only the chemical composition, but the crystal structure of silicates, especially the bonding types. Since the Si content of materials remained constant here, the concentration of Si 4+ in SBF solution should be predominantly related to its incorporation in crystalline or amorphous structures. Among the structures present in our materials, CaMgSi 2 O 6 , CaSiO 3 , and minor phase Li 2 SiO 3 are inosilicates, where two oxygen atoms of each silicate tetrahedron are shared with adjacent tetrahedra to form single chains of connected silicate tetrahedra. The Li 2 Si 2 O 5 phase induced by the substitution of Li 2 O is a phyllosilicate, where sheet-like silicates composed of six-membered silica rings occur in a chair structure. [82] The SiO 2 phase is a tectosilicate, where silicate tetrahedra are connected in the form of 3D framework. In addition, the minor secondary phase Li 2 Ca 2 Si 5 O 13 is a complex silicate, containing four-and fivemembered rings of silicate tetrahedra to form double chains. [74] Therefore, it can be deduced that it is more difficult for Si 4+ to be dissolved from Li 2 Si 2 O 5 , SiO 2 , and Li 2 Ca 2 Si 5 O 13 than from CaMgSi 2 O 6 , CaSiO 3 , and Li 2 SiO 3 because of the stronger SiO covalent bonds. This may have resulted in the Si 4+ concentrations of CaMgSi 2 O 6 -Li 2 O bioceramics becoming lower than that of pure diopside at the end of immersion period. Further, the concentration of Si 4+ in Li0.25 was significantly lower than in the other samples, which could be due to its dense structure.
The change in the concentration of Mg 2+ ions in the SBF solution is particularly worthy of closer scrutiny. As shown in Figure 8c, the amount of Mg 2+ released by pure diopside throughout the soaking period was much higher than that released by CaMgSi 2 O 6 -Li 2 O bioceramics. The elemental analysis of precipitated HAp in Table 4 shows that the HAp formed on pure diopside contained 8-13 times more Mg 2+ ions than the HAp formed on CaMgSi 2 O 6 -Li 2 O bioceramics, indicating that pure diopside adsorbed considerable Mg 2+ from the surrounding environment when forming HAp. In contrast to pure diopside, mixed phase CaMgSi 2 O 6 -Li 2 O bioceramics adsorbed only a trace amount of Mg 2+ , corresponding to the slight decrease of Mg 2+ concentration shown in Figure 8c. Moreover, after 28 days of immersion, Mg 2+ ions remained undetected after the HAp had fully grown on the surface of CaMgSi 2 O 6 -Li 2 O bioceramics. The low levels of Mg 2+ detected in both SBF and HAp of CaMgSi 2 O 6 -Li 2 O bioceramics implied that Mg 2+ ions were neither released from the materials nor then readsorbed by the HAp. Thus, it can be inferred that the addition of Li 2 O suppressed the release of Mg 2+ , thereby reducing the adsorption of Mg 2+ by HAp, which can be further confirmed by the elemental mapping in Figure 5, where high Mg signals were detected in the matrix of CaMgSi 2 O 6 -Li 2 O bioceramics instead of in the HAp. The reason for the restricted release of Mg 2+ may be explained as the followings. As clarified in the elemental mapping in Figure 3, the amorphous phase of Li 2 Si 2 O 5 started to form after the substitution of Li 2 O, and the viscosity increased in the region where Mg 2+ ions were present, resulting in a decreased ion diffusion. Hence, Mg 2+ began to react with the surrounding Ca 2+ , Si 4+ , and O 2− to form large grains of CaMgSi 2 O 6 , which were more difficult to dissolve in SBF and release ions than the small grains of CaMgSi 2 O 6 . In addition, the embedment of CaMgSi 2 O 6 in the amorphous phase of Li 2 Si 2 O 5 further reduced the dissolution rate, leading to Mg 2+ being trapped in the large grains of CaMgSi 2 O 6 .
As an indicator of silica-rich layer formation, the relationship between the pH value and ion concentration can be explored in depth here. According to Figure 8b- O bioceramics demonstrated the formation of a large quantity of silica-rich layer, laying the foundation for the following substantial HAp formation. However, the ion release of Ca 2+ reached about 160 ppm, which was much higher than that of Li + (11 to 33 ppm), coupled with the fact that Li + is a monovalent ion and can exchange half as many H + ions as Ca 2+ /Mg 2+ , indicating that Ca 2+ was more dominant than Li + in the formation of silica-rich layer. It is believed that the high Ca 2+ and Li + concentrations were contributed by CaSiO 3 and Li 2 Si 2 O 5 phases, which were more likely released from the vulnerable amorphous matrix than from the solid crystals, as shown in Figure 3. Since Li0.25 had a dense structure and relatively small amount of CaSiO 3 , its overall ion exchange and material dissolution were lower than other CaMgSi 2 O 6 -Li 2 O bioceramics. Moreover, the pH value of Li0.25 was close to that of pure diopside, suggesting a comparable overall ion exchange and similar ability of silica-rich layer formation.
From the results of XRD ( Figure 1c) and SEM (Figure 4) analyses, it can be observed that all the CaMgSi 2 O 6 -Li 2 O bioceramics were found to have a greater extent of HAp formation than pure diopside. It can be further confirmed that Li0.50, Li0.75, and Li1.00 had superior silica-rich layer and subsequent HAp formation to pure diopside based on the variations of pH value (Figure 8a) and the consumption of P 5+ ions (Figure 8f). In addition, the high Ca 2+ levels found in the SBF of Li0.50, Li0.75, and Li1.00 may increase the supersaturating concentration of SBF, which would facilitate the apatite formation on the surface of the bioceramics. [83] While Li0.25 had the similar capacity of silica-rich layer formation as pure diopside, the high remaining P 5+ ions in SBF implied that Li0.25 should have had the worse HAp formation than pure diopside as fewer P 5+ ions were adsorbed to form HAp, which is inconsistent with the results shown in XRD and SEM. This inconsistency could www.advmatinterfaces.de be due to the fact that the Mg 2+ and Si 4+ released from pure diopside into SBF have been readsorbed by the HAp formed on the surface of pure diopside and hence reduced the crystallization of HAp. Hafezi-Ardakani et al. [84] have stated that Mg 2+ ions may be able to enter the forming HAp nuclei and further inhibited the HAp evolution as Mg 2+ ions are not readily accommodated in the HAp structure. In the present study, all the CaMgSi 2 O 6 -Li 2 O bioceramics blocked the release of Mg 2+ , and only trace amounts of Mg 2+ were adsorbed by HAp, as shown in Table 4, which significantly improved the HAp growth. Therefore, Li0.25 eliminated the negative effect of Mg 2+ on HAp growth by means of low-level Li 2 O addition, allowing for much more efficient HAp formation with a lower P 5+ consumption than pure diopside. It can be expected that extending the soaking time for Li0.25 would allow the complete utilization of P 5+ ions and the further increase in the amount of HAp precipitate. Moreover, Li0.25 had the lowest overall ion release among all the samples due to its dense structure, which is in good agreement with the result of weight loss.

Cell Viability
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay is an effective method to assess the cytotoxicity of materials by measuring the concentration of mitochondrial dehydrogenase, which shows cell proliferation and hence reflects the biocompatibility of materials. In the current work, cell viability was studied with pure diopside and CaMgSi 2 O 6 -Li 2 O bioceramics samples by MTT test. Their optical densities (24 h) with incubated D1 cells were compared to the positive control (with sodium dodecyl sulfate [SDS] addition) and negative controls (with only cells and with Al 2 O 3 addition), as shown in Figure 9. From the result, it can be seen that cells were more viable in the negative controls and pure diopside than in the positive control and CaMgSi 2 O 6 -Li 2  In this study, Li0.25 is more favorable for implant applications due to its excellent hardness (as-sintered: 700 Hv; SBF-soaked: 197 Hv), low degradation rate (≈0.6 wt%), and minor impact on the environmental pH value. In addition, Li0.25 eliminated the negative effect of Mg 2+ on HAp growth using low-level Li 2 O substitution, permitting much more efficient HAp formation with a lower P 5+ consumption than pure diopside and hence resulting in a superior bioactivity. A more detailed composition adjustment (such as low concentration of ion substitution) can be considered in the future work to optimize the biocompatibility of CaMgSi 2 O 6 -Li 2 O bioceramics as the presence of high-dose Li + suppressed the proliferation of mouse bone marrow cells (D1).

Experimental Section
Samples Preparation: Diopside and mixed phase CaMgSi 2 O 6 -Li 2 O bioceramics were synthesized by a precipitation method based on a procedure described in detail elsewhere. [48,49,85] For pure diopside, 0.01 mol of calcium chloride (CaCl 2 , 95%, Fisher Scientific, UK), 0.01 mol of magnesium chloride hexahydrate (MgCl 2 ·6H 2 O, 95%, EMSURE, Germany), and 0.02 mol of tetraethyl orthosilicate (TEOS, 98%, Seedchem, Australia) as reactants were dissolved in 200 mL of ethanol (C 2 H 5 OH, 95%) to produce 0.01 mol of CaMgSi 2 O 6 stoichiometry solution. The solution was magnetically stirred at 80 °C for 2 h. Then ammonium hydroxide (NH 4 OH, 28%, Nihon shiyaku reagent, Japan)  Materials Characterization: The mineralogy of the pure diopside and mixed phase bioceramics was analyzed by a powder X-ray diffractometer (D2 Phaser, Bruker, USA) with Cu Kα radiation (λ = 1.54184 Å) at 30 kV and 10 mA with diffraction angles (2θ): 10-70°, step size: 0.01° 2θ and step speed: 0.5 s/step. The phase identification and degree-ofcrystallinity analysis based on the diffraction patterns were processed by Match! software (Crystal Impact, Germany). The surface morphology and chemical composition of the samples were observed using a FE-EPMA (JXA-8530F, JEOL, Japan) which consisted of a SEM and a wavelength dispersive spectrometer. FE-EPMA was operated at 15 kV with an analytical depth of ≈700 nm, where mapping mode was used to analyze the elemental distribution and point mode was used to precisely determine the elemental concentration. The hardness was measured using a micro Vickers hardness tester (FM-110, Future-Tech, Japan) with a force of 1 kg over a loading time of 10 s under room temperature.
The bulk density and apparent density of the sintered pure diopside and mixed phase CaMgSi 2 O 6 -Li 2 O bioceramics were measured using Archimedes technique in deionized water. The density measurement was performed as the following. First, the samples were dried in an oven at 110 °C for 24 h to remove free water, then naturally cooled down to room temperature in an electronic moisture-proof box at a constant zero humidity for 24 h. The weight of the dried samples (denoted as m1) was measured by a precision analytical balance (AS 220.R2 PLUS, Radwag, Poland) with 1 mg of accuracy. Second, the samples were saturated with deionized water under a vacuum condition for 2 h. The saturated samples were then suspended in deionized water and the immersed mass of the samples (denoted as m2) was recorded. Third, the samples were taken out from deionized water and the remaining water on the sample surface was carefully removed. The weight of the saturated samples in air was measured and denoted as m3. According to Archimedes principle, bulk density was the ratio of the material mass without free water to the macroscopic material volume, which was occupied by solid material, open pores, and closed pores. The calculation of bulk density was expressed by the following equation.
Apparent density was the ratio of the material mass to the material volume including solid material and closed pores, which was calculated as the equation described below.
where ρ 0 was the density of deionized water at 25 °C. Based on the bulk density and apparent density, the porosity given by the open pores was calculated as the following.
In Vitro Biodegradation Analysis: In vitro biodegradation tests were performed by soaking pure diopside and mixed phase CaMgSi 2 O 6 -Li 2 O bioceramics pellets in a SBF solution at 37 °C. The SBF solution had a similar ion concentration with human blood plasma, as proposed by Kokubo et al. [86] For the evaluation of degradation, each pellet was immersed in 40 mL SBF solution with pH 7.4 at 37 °C for 7, 14, 21, and 28 days. After the set soaking time, each pellet was washed by deionized water and then completely dried, and the final weight loss was measured and expressed in percentage (wt%) as below: where W L was the weight loss of pellet after SBF soaking, W 0 was the initial weight of pellet, and W F was the final weight of pellet after soaking.
The pH values of the SBF solution before and after immersion were measured by a pH meter (pH 510, Eutech, Singapore) and the ion concentrations of Ca 2+ , Mg 2+ , Li + , Si 4+ , and P 5+ in the SBF solution were analyzed using an inductively coupled plasma-mass spectrometer (ELEMENT XR, Thermo fisher scientific, USA). The initial concentrations of Ca 2+ , Mg 2+ , Li + , Si 4+ , and P 5+ ions in fresh SBF were 92.6, 43.9, 0, 0, and 40 ppm, respectively.
Cell Viability Assay: The cell viability was evaluated using the liquid extracts of bioceramic samples and MTT (98%, Alfa Aesar, USA) assay. Dulbecco's modified Eagle's medium (DMEM, Gibco, USA) containing 10% fetal bovine serum (Sigma-Aldrich, USA) and 0.5% penicillinstreptomycin (Sigma-Aldrich, USA) was used in the cell culture and the extraction of sample substances. MTT solution was prepared at the concentration of 5 mg mL −1 by dissolving MTT in sterile and filtered phosphate buffered saline (PBS). The samples for preparing liquid extracts were pure diopside and mixed phase bioceramics with four levels of Li 2 O substitution (25-100 mol%). A group with the addition of SDS (99%, J.T Baker, USA) was prepared as positive control. A group containing only cells and a group with the addition of aluminum oxide (Al 2 O 3 , 99.5%, Sigma-Aldrich, USA) were prepared as negative controls.
First of all, mouse bone marrow stromal D1 cells were seeded at a density of 2 × 10 5 cells/well in 96-well plates and cultured in 100 µL of DMEM at 37 °C in an incubator under 5% CO 2 atmosphere for 24 h. Second, liquid extracts of all the test groups were prepared in DMEM at a concentration of 0.1 g mL −1 (except SDS group at 0.2 mg mL −1 ) at 37 °C for 24 h, followed by a filtration of extracts with a 0.25 µm filter. After 24 h of incubation, the initial DMEM was removed and then 100 µL of the filtered extract was added to each well. The D1 cells were cultured in the extracts in the incubator for 24 h for the following cytotoxicity test. After 24 h of cell culturing in the presence of extracts, the osteoblast cell viability was evaluated by MTT assay. The extracts were removed and 100 µL of MTT solution (5 mg mL −1 in PBS) was added to the cultures in each well, and the cells were incubated at 37 °C in 5% CO 2 for 2 h. Then the MTT solution was removed and 100 µL of dimethyl sulfoxide (99.5%, Sigma-Aldrich, USA) was added to each well to dissolve purple crystals of formazan. The optical density of each well at 570 nm of wavelength was measured by an ELISA microplate reader (Model 550, Bio-Rad, USA). The cell viability was calculated using: where OD s was the optical density values of sample groups and OD c was the optical density value of control group containing only cells. The quantitative data were expressed as means ± standard deviation of six repeats per experiment.