Ultrastrong Translucent Glass Ceramic with Nanocrystalline

Oct 18, 2018 - ... and high translucency; whereas on the other hand, glass phase generally has negative effects on the mechanical properties of GCs du...
0 downloads 0 Views 10MB Size
Download PDF
Letter Cite This: Nano Lett. XXXX, XXX, XXX−XXX

pubs.acs.org/NanoLett

Ultrastrong Translucent Glass Ceramic with Nanocrystalline, Biomimetic Structure Le Fu,⊥,† Ling Xie,⊥,† Wenbo Fu,‡ Shuanglin Hu,‡ Zhibin Zhang,§ Klaus Leifer,† Håkan Engqvist,† and Wei Xia*,† †

Applied Materials Science, Department of Engineering Science, Uppsala University, Uppsala 751 21, Sweden Institute of Nuclear Physics and Chemistry, China Academy of Engineering Physics, Mianshan Road 64, Mianyang, Sichuan 621900, People’s Republic of China § Solid State Electronics, Department of Engineering Science, Uppsala University, Uppsala 751 21, Sweden Downloaded via UNIV OF NEW ENGLAND on October 23, 2018 at 17:16:27 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

S Supporting Information *

ABSTRACT: Transparent/translucent glass ceramics (GCs) have broad applications in biomedicine, armor, energy, and constructions. However, GCs with improved optical properties typically suffer from impaired mechanical properties, compared to traditional sintered full-ceramics. We present a method of obtaining high-strength, translucent GCs by preparing ZrO 2 −SiO 2 nanocrystalline glass ceramics (NCGCs) with a microstructure of monocrystalline ZrO2 nanoparticles (NPs), embedded in an amorphous SiO2 matrix. The ZrO2−SiO2 NCGC with a composition of 65%ZrO/35%SiO2 (molar ratio, 65Zr) achieved an average flexural strength of 1 GPa. This is one of the highest flexural strength values ever reported for GCs. ZrO2 NPs bond strongly with SiO2 matrix due to the formation of a thin (2−3 nm) amorphous Zr/Si interfacial layer between the ZrO2 NPs and SiO2 matrix. The diffusion of Si atoms into the ZrO2 NPs forms a ZrOSi superlattice. Electron tomography results show that some of the ZrO2 NPs are connected in one direction, forming in situ ZrO2 nanofibers (with length of ∼500 nm), and that the ZrO2 nanofibers are stacked in an ordered way in all three dimensions. The nanoarchitecture of the ZrO2 nanofibers mimics the architecture of mineralized collagen fibril in cortical bone. Strong interface bonding enables efficient load transfer from the SiO2 matrix to the 3D nanoarchitecture built by ZrO2 nanofibers and NPs, and the 3D nanoarchitecture carries the majority of the external load. These two factors synergistically contribute to the high strength of the 65Zr NCGC. This study deepens our fundamental understanding of the microstructure-mechanical strength relationship, which could guide the design and manufacture of other high-strength, translucent GCs. KEYWORDS: Glass ceramic, translucency, high strength, biomimetic structure, 3D nanoarchitecture, electron tomography

E

soft nature. For example, aesthetically appealing lithium disilicates dental GCs have fracture toughness and flexural strength less than half that of the full ceramics,3 which limits the application of lithium disilicates as longer bridgeworks. The mechanical property limits of GCs are far from being reached, and improvements would broaden the application of GCs. One of the major advantages of GCs over traditional full ceramics is that sufficient glass phase in GCs and free choice of embedded crystalline phase(s) offer the possibility to combine a variety of desired properties by tailoring micro- or even nanoscale structures or architectures. With this concept in mind, we prepared ZrO2−SiO2 GCs with ZrO2 NPs embedded in an amorphous SiO2 matrix, aiming for the combination of high translucency and high mechanical strength. The 65%ZrO2/ 35%SiO2 (molar ratio, 65Zr) NCGC achieved a flexural strength

ver since their accidental discovery in 1953, glass ceramics (GCs) have become one of the most studied and widely used ceramics because of their excellent combination of chemical and physical properties.1 For instance, GCs are now used in the nosecones of high-performance aircraft and missiles. Materials used in these applications must exhibit a challenging combination of properties to withstand critical conditions resulting from rain erosion and atmospheric reentry, including low dielectric constant, low coefficient of thermal expansion, low dielectric loss, and high abrasion resistance.2 No glass, metal or single crystal can simultaneously meet all of these relevant specifications. GCs consist of a glass phase and one or more embedded crystalline phases with crystallinity varying most frequently between 30% and 70%. The existence of glass phase is a double-edged sword for the properties of GCs. On the one hand, the glass phase gives GCs special properties, such as zero porosity, low thermal expansion coefficient, and high translucency; whereas on the other hand, glass phase generally has negative effects on the mechanical properties of GCs due to its © XXXX American Chemical Society

Received: August 7, 2018 Revised: October 1, 2018 Published: October 18, 2018 A

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. TEM characterization of the interfacial layer in the 65Zr NCGC. (a) BF-TEM image demonstrating an overview of the microstructure. (b) HR-TEM image showing that the NP is a monocrystalline NP. (c) STEM-HAADF image of ZrO2 NPs (bright contrast). (d) STEM-EDS map of Zr; (e) STEM-EDS map of Si; (f) EDS overlapped map showing that ZrO2 NPs are interconnected and homogeneously distributed in the network-like SiO2 matrix. (g) HR-TEM image of ZrO2 grain boundary (indicated by green arrow) and the interfaces between ZrO2 NPs and SiO2 matrix (indicated by white arrows). (h) Overlapped intensity profiles extracted from the dash lines in the STEM-HAADF image (c), Zr (d), and Si (e) maps.

of approximately 1 GPa.4 This value is quite close to the strength of yttrium partially stabilized tetragonal zirconia (900−1200 MPa),5,6 which is the strongest ZrO2-based ceramic. To our knowledge, the achieved flexural strength is one of the highest strength values of all GCs ever reported. Also, the strong ZrO2− SiO2 NCGC showed high translucency.4 To understand the origin of the high strength, we hypothesized two predominate strengthening mechanisms: strong bonding between ZrO2 NPs and the SiO2 matrix, and the three-dimensional (3D) architecture formed by interconnected ZrO2 NPs.4 The interface between inclusions and the matrix play a significant role in plastic deformation and ultimately in controlling the mechanical properties of structural ceramics,7−10 especially nanocrystalline materials because they contain a higher fraction of grain boundary volume than microcrystalline materials.11 In this work, we employed aberration-corrected scanning transmission electron microscopy (STEM) and X-ray energy dispersive spectrometry (EDS) to investigate the interfacial layer down to the atomic scale. As a

result, we provide direct evidence of a thin Zr/Si interfacial layer (2−3 nm). We also correlate the high mechanical strength with the Zr/Si interfacial layer. The 3D distribution of ZrO2 NPs could also significantly contribute to the high strength of the 65Zr NCGC. However, previous TEM observations4,12 only provided 2D information on ZrO2 NPs in the form of images, which limited the full characterization of the interconnected ZrO2 NPs that could show a complex arrangement in 3D. Thus, electron tomography in STEM-HAADF mode with nanometer resolution was applied to reveal the 3D distribution of ZrO2 NPs. This state-of-the-art microscopy technique reveals that some of the ZrO2 NPs form ZrO2 nanofibers and the ZrO2 nanofibers have a short-range ordered stacking in all three dimensions. To our knowledge, this stacking order of inclusions in ceramic nanocomposite has not been previously reported. This stacking order of ZrO 2 nanofibers resembles the architecture of cortical bone. Nature’s “wisdom” in bone resides in its structural design strategy: a complex hierarchical architecture with characteristic dimensions B

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 2. ZrO2 NPs and Zr/Si interfacial layer. (a) STEM-HAADF image of a ZrO2 NP. (b−d) Corresponding STEM-EDS maps, Zr map (b), Si map (c), and overlapped map (d), respectively. (e) Schematic illustration of ZrO2 NP and Zr/Si interfacial layer structure. (f) SEM image of the fracture surface formed from the flexural strength test. Protruded ZrO2 NPs (indicated by green arrows) and corresponding voids left by pulled out ZrO2 NPs (indicated by white arrows) can be observed.

spanning from the nanoscale to the macroscale.13,14 Each level of the hierarchy has its prevailing strengthening and toughening mechanisms, synergistically contributing to the strength and toughness of bone, especially cortical bone. At the micrometer level, the characteristic elementary unit of bone is the mineralized collagen fibril. The collagen fibrils are stacked in a particular order to form geometrically arranged fiber arrays.15,16 In some sense, the stacking of ZrO2 nanofibers in the 65Zr NCGC surprisingly replicate the stacking of mineralized collagen fibril in bone. The strengthening and toughening mechanisms of bone have been extensively studied. Thus, these strengthening and toughening mechanisms were utilized to explain the origin of the high strength of the 65Zr NCGC in terms of interface and nanoarchitecture. This study deepens our fundamental understanding of the microstructure−mechanical strength relationship in material science. The design strategy we used for strengthening ZrO2−SiO2 NCGC (i.e., strong interface between inclusions and the amorphous matrix, and tough inclusions with 3D architecture) offers a novel route toward engineering GCs with better mechanical performances. Results and Discussion. The Observation of ZrO2 NP and Zr−Si Interfacial Layer. Optical image (Figure S1a) shows that the 65Zr NCGC is highly translucent. This indicates that the 65Zr NCGC attains complete densification after sintering at 1200 °C, because the inline transmission of ceramics shows extremely strong dependence on porosity and the sintered body is almost opaque with Vp (volume porosity) = 0.1%.17 The glass transition temperature of fused quartz is approximately 1200 °C.18 During spark plasma sintering at 1200 °C, the amorphous SiO2 is in a viscous state which assists the rearrangement of ZrO2 NPs and elimination of residual pore during the late stage of sintering. The XRD pattern (Figure S1b) exhibits strong peaks at 30°, 50°, and 60°, which corresponds to the (101), (112), and (211) lattice planes of tetragonal ZrO2, respectively. A weak

peak at 28° corresponding to (−111) plane of monoclinic ZrO2 can be observed. No peak of crystalline SiO2 is observed, indicating that the SiO2 remains amorphous after sintering. The crystallization behaviors of ZrO2 and SiO2 of the current study are in accordance with reported results in the ZrO2−SiO2 system.19 The image contrasts in bright-field TEM (BF-TEM) images are determined by a diffraction contrast. When the concentration and structure of NPs are the same, darker NPs are close to a Bragg orientation, whereas brighter NPs are not. Close to spherical shape NPs are homogeneously distributed in the matrix (Figure 1a). The size of the NPs is measured as 51.8 ± 10.8 nm using an electron tomography technique that can eliminate the projection effect of NPs along the sample thickness direction. The high-resolution TEM (HR-TEM) image (Figure 1b) shows a monocrystalline ZrO2 NP embedded in the matrix. The average lattice spacing is 2.92 Å, which can be indexed as the (101) lattice plane of tetragonal ZrO2. In the STEM-HAADF imaging mode, the contrast is proportional to the atomic number (Z), and heavier atoms appear brighter. Thus, in Figure 1c NPs with bright contrast are ZrO2 NPs. The bright contrasts overlap with each other because ZrO2 NPs are distributed in all three dimensions. The NPs are further confirmed as ZrO2 from the EDS map of Zr (Figure 1d), and they appear interconnected with each other in the 2D image. The Si map (Figure 1e) and the overlapped map (Figure 1f) show that the SiO2 matrix forms a network-like structure with the interconnected ZrO2 NPs homogeneously distributed in the matrix. This is in agreement with our previous results.4,12 Detailed analysis of the 3D distribution of ZrO2 NPs can be revealed by electron tomography, and the results will be provided in the following section. The HR-TEM image (Figure 1g) demonstrates two types of interface found in the 65Zr NCGC: one is grain boundaries between ZrO2 NPs (indicated by green arrow), which have been examined carefully in our C

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 3. Observation of the ZrOSi superlattice structure at the edge of the ZrO2 NP. (a) STEM-HAADF image of one tetragonal ZrO2 NP with corresponding Fourier transform pattern inserted. Some atomic columns (marked by a green dash square) show brighter dotlike contrast at the edge of the ZrO2 NP. (b) Integrated STEM-EDS signal intensity profiles of Zr and Si extracted from the {011} lattice planes (marked by blue bars in Figure 2b,c) showing the periodic distribution of the Zr and Si signals. (c) The crystal structure of the Si-doped tetragonal ZrO2 calculated by DFT with the Zr/Si atomic ratio of 3:1 viewed from the (010) direction. Green spheres: Zr atoms. Blue spheres: Si atoms. (d) Raman spectra of the 65Zr NCGC and amorphous SiO2. Characteristic peaks of ZrO in the tetragonal (t) ZrO2 and peaks of ZrOSi are presented.

because tetragonal and monoclinic ZrO2 are the only two crystalline phases observed in the 65Zr NCGC (Figure S1b). Moreover, from the HR-TEM images of ZrO2 NPs (Figure 1b,g), we can only observe tetragonal ZrO2 (101) planes and we do not observe any lattice fringes from the interfacial layer between ZrO2 NPs and SiO2 matrix. This also indicates that the interfacial layer is amorphous. The interfacial layer is further confirmed by EDS mapping on one ZrO2 NP (Figure 2a−d) that is located at the edge of the TEM sample. In Figure 2a, an atomically resolved STEMHAADF image shows one tetragonal ZrO2 NP that is oriented in the vicinity of the [100] zone axis. The Zr EDS map (Figure 2b) shows the core of the ZrO2 NP. Si EDS intensity is detected in the interfacial layer, which is marked using white circles in the Si map (Figure 2c) and the overlapped map (Figure 2d). In the same region (white cycle in Figure 2b), Zr EDS signals are much weaker than those of Si, indicating that the interfacial layer is Sirich and Zr-poor. This is in a good agreement with the observations obtained from the intensity profiles of Zr and Si at the interfacial layers (Figure 1h). On the basis of TEM results and the above analysis, a schematic illustration (Figure 2e) is proposed to reveal the ZrO2 NPs and the Zr/Si interfacial layer. By examining the ZrO2−SiO2 phase diagram,21 we find that the 65Zr NCGC with 35 mol % SiO2 is outside the spinodal curve at 1200 °C, indicating that both ZrO2 NPs and the Zr/Si interfacial layer are not formed through spinodal decomposition. In the model of nucleation and growth, if the system does not reach the thermal equilibrium condition, then there is a concentration gradient at the interface between the precipitated nanoparticles

previous report, showing that the grain boundaries are coherent or semicoherent and atomically sharp;4 the other one is interfaces between ZrO2 NPs and the SiO2 matrix (indicated by white arrows). A straight line is drawn along four ZrO2 NPs (dash lines in Figure 1c−e). The peak positions of the STEMHAADF signal (Figure S2a) are roughly in accordance with those of the Zr EDS signal (Figure S2b), whereas the peak positions of the Si EDS signal (Figure S2c) locate at the valleys in the STEM-HAADF and Zr EDS profiles. By plotting three intensity profiles into one figure (Figure 1h), we observe that the width of the STEM-HAADF intensity profile is approximately 2−3 nm (marked as parallel black lines) larger than the width of the Zr EDS intensity profile. In the two marked regions (Figure 1h), both STEM-HADDF intensity and the Si EDS intensity are higher than the Zr EDS intensity, which means that both Zr and Si elements contribute to the STEM-HADDF intensity in these 2−3 nm interfacial regions between the ZrO2 NPs and SiO2 matrix. Meanwhile, toward the core of the ZrO2 NPs, Si EDS intensity decreases, while Zr EDS intensity increases (Figure 1h). These observations confirm the existence of a thin interfacial layer between ZrO2 NPS and SiO2 matrix formed by the Zr/Si interdiffusion during the sintering process. For grains with spherelike shape, the volume fraction of interfaces in the nanocrystalline material can be estimated as 3Δ/d (where Δ is the average interface thickness and d is the average grain diameter).20 In our case, the average interfacial layer thickness is 2.5 nm and the average grain size is 51.8 nm. Thus, the volume fraction of interfacial layer is calculated as 14.5% (relative to the volume of ZrO2 NPs). The interfacial layer is amorphous D

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters and matrix due to interdiffusion.22 Thereby, we explain the formation of the Zr/Si interfacial layer as a result of nucleation and growth of ZrO2 NPs. The driving force underlying the precipitation and growth of particles from the surrounding matrix consists of two parts: the decrease of the interfacial free energy and the solute concentration gradient at the interface.23 The concentration gradient at the interface can be modeled by Fick’s first law of diffusion, as a function of the particle radius. An interfacial layer always exists between the precipitated particle and the surrounding matrix when the system is at the nonequilibrium condition.24 In our studied case, the formation of such an interfacial layer is due to the concentration gradient at the interfacial region, and such concentration gradients of Zr and Si were characterized by the STEM-EDS technique. In thin film technology, the adhesion between film and substrate primarily depends on the interface layer, and the formation of the diffusion interface layer results in strong bonding between film and substrate.25 Thus, similarly, the Zr/Si interfacial layer in the 65Zr NCGC provides strong bonding between the ZrO2 NPs and SiO2 matrix. For particulate composite materials, the maximum stress that the material can sustain depends on the effectiveness of stress transfer between the reinforcing particulate and matrix.26,27 For our ZrO2−SiO2 NCGCs, applied stress can be effectively transferred from the SiO2 matrix to ZrO2 NPs through the interfacial layer, which significantly enhances the macroscopic mechanical strength of the 65Zr NCGC. Protruded ZrO2 nanoparticles (indicated by green arrows) and near-spherical shaped voids (indicated by white arrows) can be found on the fracture surface formed during the flexural strength test (Figure 2f). ZrO2 NPs are pulled out from the matrix, leading to interfacial fracture mode. As one of the strengthening mechanisms of ceramic composites, pull out contributes to the strength of ceramic composites by consuming fracture energy.27,28 The Formation of the ZrOSi Superlattice Inside ZrO2 NPs. STEM-HAADF images (Figure 3a and Figure S3a) show one ZrO2 NP that was oriented in the vicinity of the [100] zone axis. From the Fourier transform pattern (inset in Figure 3a and Figure S3b), the reflection points can be indexed by tetragonal ZrO2 lattice planes, indicating that the NP is a monocrystalline tetragonal ZrO2 NP. The atomically resolved STEM-HAADF images enable us to perform more detailed structural analyses at the atomic level. In particular, we observe that some of the atomic columns show brighter dotlike contrasts than other atomic columns inside the imaged ZrO2 NP. These brighter atomic columns are concentrated in an area of approximately 5 × 5 nm and located near to the edge of ZrO2 NPs (yellow circle marked area in Figure 2a, green rectangular marked area in Figure 3a). The regions that show brighter dotlike contrasts also have tetragonal structure (Figure 3a). We only observe atomic columns with brighter contrasts at the edge of the ZrO2 NP. The contrast generated using the STEM-HAADF imaging technique is more sensitive to atomic numbers in materials. Thus, these brighter atomic columns are not chemically identical with the other parts of the ZrO2 NP. Because these brighter regions are located at the edge of ZrO2 NPs, we hypothesize that a small amount of Si diffused into the ZrO2 NPs, forming a ZrOSi superlattice at the edge of the ZrO2 NP. To prove this hypothesis, a STEM-EDS technique was applied to examine the elemental distribution of Si and Zr in the regions that show brighter dotlike contrasts (marked by a yellow dash cycles in Figure 2a−d). In the STEM-HAADF image (Figure 2a), the marked area shows tetragonal lattices that are oriented to the

[100] zone axis. In Figure 2c, an increase of the Si signal is observed in the marked area, which indicates that the brighter atomic columns contain Si. By drawing an intensity profile on the Si and Zr EDS maps (blue lines in Figure 2b,c) along the (011) lattice plane direction, a periodic intensity distribution of both Zr and Si is observed (Figure 3b). The spacing between the two nearest EDS intensity peaks is approximately 5.96 ± 0.7 Å (Figure 3b), which corresponds to two-times the (011) lattice plane spacing of tetragonal ZrO2. The peak positions of the Si EDS intensity profile shift by 2.98 Å relative to the peaks observed from the Zr EDS intensity profile. This means that only the second nearest Zr atoms are replaced by Si atoms on the {011} lattice planes. The atomic concentration ratio of Zr to Si in the brighter dotlike contrasts region (marked by yellow dash cycles in Figure 2a−d) is quantified approximately as 3:1 by STEM-EDS. This means that one Zr atom in every two tetragonal ZrO2 unit cells is replaced by one Si atom. The STEM-EDS analysis results confirm that Si atoms diffuse into the edge of the ZrO2 NPs and replace some of the Zr atoms, forming a ZrOSi superlattice. On the basis of the STEM-EDS analysis, an atomic model of the ZrOSi superlattice for the regions that appear brighter was proposed (Figure 3c and Figure S4a), and first-principles calculations using density functional theory (DFT) were performed to gain more fundamental insight about the Zr− O−Si superlattice. In the proposed ZrOSi superlattice model, one Zr atom in every two tetragonal ZrO2 units cell is replaced by one Si atom. From the DFT result, the spacing between the two nearest Si atomic columns is 6.09 Å after geometry relaxation (Figure 3c), which is in a good agreement with the Si EDS peak intensity spacing (5.99 Å) measured from the STEM-EDS analysis (Figure 3b). A 3D visualization of the structure is shown in Figure S4a. Although the calculated lattice spacing of Si-doped ZrO2 is 2.838 Å (d/2, Figure 3c), which is slightly smaller than that of the reference pure ZrO2 (2.987 Å, Figure S4b), the crystal structure of Si-doped ZrO2 (Zr/Si atomic ratio = 3:1) still maintains its tetragonal structure without much distortion (Figure 3c and Figure S4a). It is possible that the crystallographic growth direction in the Zr OSi super lattice is slightly different from the pure tetragonal ZrO2. The observation of parallel lines ({011} lattice planes in Figure 3a) is because the ZrO2 NP is not aligned to the zone axis with respect to the electron beam. The ZrOSi super lattice is occasionally well aligned to the zone axis, which results in dots like contrast (Figure 3a). Moreover, relative to the pure ZrO2 and SiO2 crystalline phases, the Si-doped ZrO2 structure has a very low formation energy of −3.22 eV per Si, which indicates that it would be energy favorable to form such a ZrOSi superlattice structure when ZrO2 and SiO2 coexist. This explains the formation of the ZrOSi superlattice in the regions that show bright dotlike contrast. The crystal structure of Si-doped ZrO2 with Zr/Si atomic ratio of 1:1 is also calculated and showed in Figure S4c. The original tetragonal ZrO2 crystal structure is distorted when Zr atoms in every other ZrO2 {011} lattice planes are completely replaced by Si atoms, which is not consistent with STEM results (Figure 3a). Meanwhile, the structure has a formation energy of −2.86 eV per Si, which is higher than that of the structure with a Zr/Si atomic ratio of 3:1 (−3.22 eV). The formation of the ZrOSi bond is confirmed by Raman (Figure 3d) and infrared (IR) (Figure S5) tests. Amorphous SiO2 shows a broad peak at around 400 cm−1 (Figure 3d). We observed characteristic peaks for tetragonal ZrO2 at 264, 313, 457, 600, and 643 cm−1 (Figure 3d).29,30 The E

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 4. Observation of orderly stacked ZrO2 nanofibers. (a,b) Two slices with a spacing of approximately 38 nm along the Z-axis direction (Movie S1). In slice A, parallel ZrO2 nanofibers (marked by white solid lines) and individual ZrO2 NPs (marked by white solid cycles) can be observed. In slice B, new ZrO2 nanofibers (indicated by yellow dash lines) or ZrO2 NP (indicated by yellow dash cycles) appear in the middle of the projections of ZrO2 nanofibers or individual ZrO2 NP presented in slice A. (c) Snapshot taken from Movie S2 giving an overview of the 3D spatial distribution of ZrO2 NPs. (d) Snapshots taken from Movie S3 showing near parallel ZrO2 nanofibers stacked in the Y-axis direction (indicated by yellow and white dash lines in Snapshot A). The nanofibers are connected by ZrO2 NPs (marked by orange arrows) in the Y-axis direction. (e) Snapshot taken from Movie S4 demonstrating that of ZrO2 NPs in nanofibers are connected by grain boundaries (marked by white arrows). The ZrO2 nanofibers (marked by yellow dash lines) also show a stacking order in the Z-axis direction.

peaks at 357, 432, 952, and 991 cm−1 are assigned to the vibration of ZrOSi.30 The ZrOSi vibration signals originate from two sources: the Zr/Si interfacial layer around ZrO2 NPs (Figure 2b−e) and the ZrOSi superlattice at the edge of ZrO2 NPs (Figure 3a). To our knowledge, such a tetragonal ZrOSi superlattice inside of ZrO2 NPs has not been reported elsewhere. The origin of such a ZrOSi superlattice can be attributed to the small amount of Si diffused into the edge of ZrO2 NPs. We did not observe such a ZrO Si superlattice structure in the core region of ZrO2 NPs. Much remains to be learned about the effects of this ZrOSi superlattice on the transformability of ZrO2 NPs and mechanical properties of ZrO2−SiO2 NCGCs. Three-Dimensional Stacking Order of ZrO2 Nanofibers and NPs Analyzed by Electron Tomography. As mentioned above,

two factors could significantly influence the mechanical strength of the 65Zr NCGC. One is the Zr/Si interfacial layer, which has been discussed above; the other one is the 3D distribution of ZrO2 NPs, which can not be revealed by the above 2D TEM images due to the projection effect of overlapped NPs along the sample thickness direction. This motivates us to obtain a real 3D view of the ZrO2 NPs utilizing an electron tomography technique. A movie file (Movie S1) playing reconstructed slices along the Z-axis direction is provided in the Supporting Information. Two slices (Figure 4a,b) from the reconstructed tomogram are presented to illustrate the distribution of the ZrO2 NPs. The colored contrast observed in the slices corresponds to ZrO2 NPs. In slice A, ZrO2 nanofibers (marked by white solid lines) composed of connected ZrO2 NPs can observed. The ZrO2 nanofibers have a length of approximately 500 nm. These F

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 5. Hierarchical structure of bone (left panel) and our ZrO2−SiO2 NCGC (right panel). At the length-scale of approximately 100 μm, compact bone is composed of osteons. Osteons have a lamellar structure. Each individual lamella consists of fibers arranged in geometrical patterns at the lengthscale of approximately 5 μm. The fibers comprise several mineralized collagen fibrils. At the length-scale of hundreds of nanometers, ZrO2 NPs are randomly distributed in 3D, showing a long-range disordered structure. At the finer length-scale of approximately 100 nm, ZrO2 NPs form nanofibers and the nanofibers are orderly stacked, forming a short-range 3D nanoarchitecture.

NCGC is quantitatively measured as 60 ± 3% from electron tomography results. This measured value (60 ± 3%) is consistent with the calculated value (58.7%, Table S1) of ZrO2 volume fraction, indicating that almost all of Zr element from the amorphous raw powder crystallize to from ZrO2 NPs during sintering process. One snapshot from Movie S4 is displayed in Figure 4e in which the ZrO2 nanofibers are also stacked in the Z-axis direction (outlined by yellow dash lines). Also, the ZrO2 nanofibers are composed of ZrO2 NPs connected by grain boundaries (marked by white arrows). The ZrO2 NPs have irregular and complex morphologies (Figure 4e). In Figure 4a,b, the visualization area is approximately 700 nm × 700 nm in which there are approximately 80 NPs. And, the visualization volume of Movie S2 is 216 nm (X) × 216 nm (Y) × 104 nm in which there are hundreds of NPs. The sol−gel raw powder shows high homogeneity.4 Furthermore, during the SPS process, the pressure was homogeneously applied and the temperature gradient was negligible since the diameter of the sintered specimen is only 20 mm. The 65Zr NCGC should show high homogeneity in terms of microstructure. Thus, the observed ZrO2 nanofibers and the stacking order of the ZrO2 nanofibers can represent the structural characteristics of the 65Zr NCGC. However, the formation mechanisms of the ZrO2 nanofibers and the short-range stacking order are unclear. Microstructure−Mechanical Strength Relation. From the results of our previous studies,4,12 the average flexural strengths of the 35Zr, 45Zr, 55Zr, and 65Zr NCGCs are 242, 533, 737, and 1014 MPa, respectively (Table S1). The corresponding nominal volume fractions of ZrO2 are 28.1%, 39.7%, 47.7%, and 58.7%, respectively (Table S1). The flexural strength shows good positive correlation with ZrO2 volume fraction, increasing with the increase of ZrO2 volume fraction (Figure S7). As schematically demonstrated in Figure S7 (inset a), in the 35Zr NCGC most of the ZrO2 NPs (28.1 vol %) are isolated and embedded in the SiO2 matrix (61.9 vol %).12 Whereas in the 65Zr NCGC, as shown in the above electron tomography results

ZrO2 nanofibers are in situ formed during the sintering process. Also, they are close to being arranged in parallel with a center-tocenter spacing of about 75 ± 5 nm. Individual ZrO2 NPs (indicated by white solid cycles) can also be observed, and they are equally spaced. The center-to-center distance between these individual ZrO2 NPs is approximately 80 nm. In slice B (Figure 4b), the ZrO2 nanofibers presented in slice A disappear. New ZrO2 nanofibers (marked by yellow dash lines) start appearing between the projections (outlined by white solid lines) of ZrO2 nanofibers presented in Slice A. The individual ZrO2 NPs also have this stacking order, that is, the ZrO2 NPs in slice B (outlined by yellow dash cycles) locate at the middle of the projections (marked by white solid cycles) of the ZrO2 NPs presented in slice A. Thus, both the ZrO2 nanofibers and individual ZrO2 NPs are not randomly distributed in 3D space but rather in an ordered form. The ZrO2 nanofibers and individual NPs have an ABAB··· stacking sequence in the X−Z plane, as schematically shown in Figure S6. Movie S2 and Figure 4c give an overview of the 3D spatial distribution of ZrO2 NPs, from which it can be observed that ZrO2 NPs are randomly distributed without obvious ordering. To closely observe the arrangement of ZrO2 NPs, three snapshots from a trimmed volume (Movie S3) are presented in Figure 4d. On the bottom part of the snapshot A (Figure 4d), near parallel ZrO2 nanofibers stacked in the Y-axis direction can be clearly observed (marked with yellow dash lines). On the top part of snapshot A, ZrO2 nanofibers (indicated by white dash lines) can also be found. However, the orientation of those two arrays of ZrO2 nanofibers (yellow and white dash lines) is oblique. The misorientation of the ZrO2 nanofiber arrays breaks the ABAB··· stacking order in the long-range. Thus, ZrO2 nanofiber arrays are only orderly stacked in a short-range. We also notice that there are ZrO2 NPs (marked by orange arrows in Snapshot B,C, Figure 4c) sitting between the arrays of ZrO2 nanofibers and connecting the two nearest arrays of ZrO2 nanofibers. The volume fraction of ZrO2 NP in the 65Zr G

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

constructed by the orderly stacked ZrO2 nanofibers. The high flexural strength of the 65Zr NCGC originates from synergistic strengthening effects of the thin Zr/Si interfacial layer and 3D nanoarchitecture. The synergistic strengthening approaches could be extended to the design and manufacture of other highstrength, translucent GC materials.

and the inset (Figure S7, inset b), most of the ZrO2 NPs (58.7 vol %) are interconnected, forming a complex 3D nanoarchitecture with ZrO2 nanofibers orderly stacked in shortrange. As schematically demonstrated in Figure 5, the nanoarchitecture of the 65Zr NCGC is analogous to cortical bone. Owing to the similarities, the strengthening mechanisms of bone can be utilized to elucidate the origin of the high strength of the 65Zr NCGC. Bone is a highly hierarchical composite material.31 At the nanometer scale, nature organizes a brick-and-mortar-like architecture in bone with hard bricks of nanosized mineral crystals sandwiched with soft mortar of collagen matrix.32 The stiffness and fracture strain of bone depends on the amount and precise arrangement of mineral deposited in the collagen matrix.32,33 From a simple rule of mixture, one expects an increase in stiffness (but also brittleness) with increasing mineral density and a staggered arrangement of mineral particles is by far superior to a strictly parallel arrangement in terms of mechanical performance.33 In the ZrO2−SiO2 NCGCs, the tough ZrO2 NPs resemble mineral crystals, whereas the soft amorphous SiO2 matrix resembles collagen matrix. Thus, the increase of the ZrO2 volume fraction in ZrO2−SiO2 NCGCs enhances flexural strength, which is somehow similar to the increase of bone strength with the increase of mineral density. The stacking order of the ZrO2 nanofiber arrays resembles that of mineralized collagen fibrils in cortical bone (Figure 5). At the microscale level of the hierarchy, crack bridging and crack deflection are the primary strengthening mechanisms of cortical bone.31 Here, it is reasonable to speculate that the ZrO2 nanofibers have similar crack bridging and crack deflection effects. From another perspective, the 65Zr NCGC can be viewed as fiber-reinforced ceramic composite due to the formation of ZrO2 nanofibers. For ceramic composite composed of reinforcement and matrix, the difference between the Young’s modulus of the reinforcement and the matrix gives rise to different stresses in the reinforcement and in the matrix when they are deformed to the same strain.27 If the Young’s modulus of the reinforcement is much greater than that of the matrix, then the reinforcement carries more load than the matrix.27 In the developed ZrO2− SiO2 NCGCs, the Young’s modulus of tetragonal ZrO2 (approximately 210 GPa5) is much larger than that of amorphous SiO2 (approximately 70 GPa34). Meanwhile, the Zr/Si interfacial layer enables load transfer from the SiO2 matrix to the ZrO2 3D nanoarchitecture. Thus, both compressive stress on the upper sample surface and tensile stress on the lower sample surface are applied and dissipated on ZrO2 nanofibers or ZrO2 NPs during the piston-on-three-ball test (Figure S8). It is the ZrO2 3D nanoarchitecture that carries the majority of the external load, rather than the amorphous SiO2 matrix. The ZrO2 3D nanoarchitecture acts as hard bricks for primary load bearing, and the SiO2 matrix as soft mortar for load distribution and energy dissipation. This mimics bone’s load-bearing mechanism.32,35 As a result, the 65Zr NCGC achieved a flexural strength of as high as 1014 MPa (Table S1). In conclusion, our results demonstrate that ZrO2 NPs in the 65%ZrO2/35%SiO2 (molar ratio, 65Zr) NCGC a thin (2−3 nm) amorphous Zr/Si interfacial layer allows the ZrO2 NPs to bond strongly with the SiO2 matrix. ZrOSi superlattice forms at the edge of ZrO2 NPs due to the diffusion of Si atoms into ZrO2 lattice. DFT calculations show that the formation of the ZrOSi superlattice with a Zr/Si atomic ratio of 3:1 is energetically favorable. STEM-HAADF electron tomography results provide direct evidence of a 3D nanoarchitecture

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.8b03220.

■

Four movies showing the 3D distribution of ZrO2 NPs (ZIP) Additional details on the materials and methods. Optical image and XRD pattern, STEM-EDS mappings, STEM image and FFT, STEM-EDS signal intensity, ZrOSi superlattice and pure tetragonal ZrO2 crystal structures calculated by DFT, IR spectrum, flexural strength versus ZrO2 volume fraction, schematic diagram of the stress situation during piston-on-three balls test. Table showing composition, sample notation, ZrO2 volume fraction, and flexural strength of different ZrO2−SiO2 NCGCs (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Le Fu: 0000-0001-6812-1827 Shuanglin Hu: 0000-0001-9729-5500 Zhibin Zhang: 0000-0003-0244-8565 Author Contributions ⊥

L.F. and L.X. contributed to this work equally and are co-first authors. W.X and H.E conceived and directed the research; L.F carried out the previous studies and wrote the manuscript; L.X performed the TEM and electron tomography experiments and analyzed the results; W.B.F and S.L.H did DFT calculation; Z.B carried out the Raman experiment; K.L analyzed TEM results. All authors contributed to interpreting the data and editing the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank Rui Sun for help with the FT-IR test and acknowledge the support of Carl Tryggers Stiftelse and the Chinese Scholarship Council (CSC). S.L.H is grateful for support from the Science Challenge Project (No. TZ2018004) and the President foundation of the China Academy of Engineering Physics (Grant No. YZJJLX2016004).

■

REFERENCES

(1) Zanotto, E. D. Am. Ceram. Soc. Bull. 2011, 89, 609−612. (2) Glass-Ceramic Technology, 2nd ed.; Hoeland, W., Beall, G. H., Eds.; The American Ceramic Society/Wiley: New York, 2010. (3) Yilmaz, H.; Aydin, C.; Gul, B. E. J. Prosthet. Dent. 2007, 98, 120− 128. (4) Fu, L.; Engqvist, H.; Xia, W. J. Eur. Ceram. Soc. 2017, 37, 4067− 4081. (5) Piconi, G.M. C. Compr. Biomater. 1999, 20, 1−25. (6) Denry, I.; Kelly, J. R. Dent. Mater. 2008, 24, 299−307. H

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters (7) Buban, Y.I. J. P.; Matsunaga, K.; Chen, J.; Shibata, N.; Ching, W. Y.; Yamamoto, T. Science (80- 2006, 311, 1757−1762. (8) Jiang, S.; Wang, H.; Wu, Y.; Liu, X.; Chen, H.; Yao, M.; Gault, B.; Ponge, D.; Raabe, D.; Hirata, A.; Chen, M.; Wang, Y.; Lu, Z. Nature 2017, 544, 460−464. (9) Lu, K.; Lu, L.; Suresh, S. Science (Washington, DC, U. S.) 2009, 324, 349−352. (10) Yu, Z.; Cantwell, P. R.; Gao, Q.; Yin, D.; Zhang, Y.; Zhou, N.; Rohrer, G. S.; Widom, M.; Luo, J.; Harmer, M. P. Science (Washington, DC, U. S.) 2017, 358, 97−101. (11) Kumar, K. S.; Van Swygenhoven, H.; Suresh, S. Acta Mater. 2003, 51, 5743−5774. (12) Fu, L.; Wu, C.; Grandfield, K.; Unosson, E.; Chang, J.; Engqvist, H.; Xia, W. J. Eur. Ceram. Soc. 2016, 36, 3487−3494. (13) Reznikov, N.; Shahar, R.; Weiner, S. Acta Biomater. 2014, 10, 3815−3826. (14) Reznikov, N.; Bilton, M.; Lari, L.; Stevens, M. M.; Kröger, R. Science (Washington, DC, U. S.) 2018, 360, eaao2189. (15) Reznikov, N.; Shahar, R.; Weiner, S. Bone 2014, 59, 93−104. (16) Reznikov, N.; Chase, H.; Brumfeld, V.; Shahar, R.; Weiner, S. Bone 2015, 71, 189−195. (17) Tsukuma, K.; Yamashita, I.; Kusunose, T. J. Am. Ceram. Soc. 2008, 91, 813−818. (18) Bucaro, J. A.; Dardy, H. D. J. Appl. Phys. 1974, 45, 5324−5329. (19) Aguilar, D. H.; Torres-Gonzalez, L. C.; Torres-Martinez, L. M.; Lopez, T.; Quintana, P. J. Solid State Chem. 2001, 158, 349−357. (20) Meyers, M. A.; Mishra, A.; Benson, D. J. Prog. Mater. Sci. 2006, 51, 427−556. (21) Kim, H.; McIntyre, P. C. J. Appl. Phys. 2002, 92, 5094−5102. (22) Kwon, S. G.; Hyeon, T. Small 2011, 7, 2685−2702. (23) Xie, L.; Jarolimek, K.; Kocevski, V.; Rusz, J.; Zeman, M.; van Swaaij, R.A.C.M.M.; Leifer, K. Nanoscale 2017, 9, 6703−6710. (24) Thanh, N. T. K.; Maclean, N.; Mahiddine, S. Chem. Rev. 2014, 114, 7610−7630. (25) Pulker, H. K. Thin Solid Films 1982, 89, 191−204. (26) Fu, S. Y.; Feng, X. Q.; Lauke, B.; Mai, Y. W. Composites, Part B 2008, 39, 933−961. (27) Xia, K.; Langdon, T. G. J. Mater. Sci. 1994, 29, 5219−5231. (28) Ohji, T.; Jeong, Y.-K.; Choa, Y.-H.; Niihara, K. J. Am. Ceram. Soc. 1998, 81, 1453−1460. (29) Vasanthavel, S.; Derby, B.; Kannan, S. Dalt. Trans. 2017, 46, 6884−6893. (30) Wang, Z.; Xu, Q.; Xu, M.; Wang, S.; You, J. RSC Adv. 2015, 5, 11658−11666. (31) Launey, M. E.; Buehler, M. J.; Ritchie, R. O. On the Mechanistic Origins of Toughness in Bone. Annu. Rev. Mater. Res. 2010, 40, 25−53. (32) Rho, J. Y.; Kuhn-Spearing, L.; Zioupos, P. Med. Eng. Phys. 1998, 20, 92−102. (33) Jäger, I.; Fratzl, P. Biophys. J. 2000, 79, 1737−1746. (34) Khan, A.; Philip, J.; Hess, P. J. Appl. Phys. 2004, 95, 1667−1672. (35) Barthelat, F.; Yin, Z.; Buehler, M. J. Nat. Rev. Mater. 2016, 1, 1− 16.

I

DOI: 10.1021/acs.nanolett.8b03220 Nano Lett. XXXX, XXX, XXX−XXX