Introduction

High-strength ceramic materials typically demonstrate superior hardness, wear resistance, heat resistance, and chemical stability compared with metallic materials1,2,3. However, their inherent brittleness has historically constrained their use as structural materials. Efforts to enhance the fracture toughness of high-strength ceramics have been extensively pursued to overcome their brittleness. The fundamental involves impeding macroscopic crack propagation by controlling their microstructure in the following ways. (1) Fibrous reinforcement. In ceramic matrix composite materials, such as SiC/SiC composites used in aerospace applications4, premature failure of the matrix can be impeded by hindering crack propagation through interfacial fracture between the fibers and the matrix5. (2) Whisker or elongated-grain reinforcement. SiC whisker-dispersed Al2O3 for cutting tools6 and β-Si3N4 ceramics comprising grains with a high aspect ratio7 exhibit relatively high fracture toughness. This is attributed to crack bridging and deflection caused by crack propagation along grain boundaries8. (3) Phase transformation toughening. Crack propagation is restricted by volume expansion in the transformation zone of Y2O3-partially stabilized tetragonal ZrO2 (YSZ). This is due to the stress-induced phase transformation from the tetragonal to monoclinic phase around the tip of the cracks9,10. This results in high fracture toughness for high-strength oxide ceramics. Ceramic materials with improved fracture toughness have found practical applications in various industries. Another conceivable approach to overcome their brittleness is to improve their plastic deformability. However, attempts to improve the plastic deformability of high-strength ceramic materials through increasing dislocation activities have met limited success, because ceramics generally require higher stress for dislocation glide and multiplication than the fracture strength of polycrystals, and the number of slip systems is often insufficient to enable arbitrary deformation in polycrystals11.

In the present study, we focused on Al2O3–GdAlO3 (Gadolinium Aluminum Perovskite: GAP) oxide eutectic composites as one of the most representative eutectic ceramics12,13. We developed plastic deformability of the Al2O3–GAP fibrous eutectic composites with an interphase spacing of 170 nm, while both Al2O3 and GAP single crystals exhibited brittleness. Rather than the three conventional methods above, we introduced an approach to overcome brittleness of ceramics.

Al2O3–GAP eutectic ceramics produced by conventional unidirectional solidification from melted solution have been developed since late 20th century, as a candidate for next-generation high-temperature structural materials12,13. The eutectic composite with coarsened, continuous networks of single crystals of Al2O3 and GAP phases demonstrated excellent high-temperature creep resistance12,14. Conversely, eutectic composites with a fine anisotropic microstructure with rod-like GAP phases aligned in an Al2O3 matrix, fabricated at a high solidification rate, exhibited high bending strength up to 1800 MPa at room temperature15. Enhancement of fracture toughness has also been reported in the refined eutectic microstructure16, and the microstructural toughening mechanism has been studied17. Recently, we discovered that the Al2O3–GAP eutectic composite with submicron-scale interphase spacing can be easily produced using an electric field-assisted sintering technique, known as flash sintering18. Flash sintering (flash event) is a phenomenon in which accelerated atomic-diffusion mass transport occurs beyond a threshold condition of electric field and temperature19,20.

We successfully fabricated the Al2O3–GAP eutectic with rod-like single crystal GAP phases periodically aligned in the Al2O3 single crystal matrix at furnace temperature of 1300 °C for 40 s. The prepared Al2O3–GAP eutectic had an interphase spacing of 170 nm. Then we evaluated the mechanical response by micropillar compression test, which is a useful method to study potential plastic deformability of macroscopically brittle materials21,22,23. Micropillars of the eutectic composites with various crystalline orientations exhibited continuous plastic deformation with an engineering plastic strain of 1–5% without any crack propagation during compression at 25 °C. This starkly contrasted with the intrinsic brittleness of Al2O3 or GAP single crystals prepared from the same composites. Transmission electron microscopy (TEM) observations revealed that dislocation glide on multiple slip systems was promoted in the Al2O3–GAP eutectic material. The primary deformation mechanism was attributed to prismatic slip in the Al2O3 phase, which is the most active slip system at room temperature. Surprisingly, during plastic deformation, the basal slip system of the Al2O3 phase in the present eutectic composite was activated as the secondary slip, though it is usually inactive in Al2O3 at room temperature. The activation of the Al2O3 basal slip in the Al2O3–GAP eutectic composite is attributed to fabrication of the refined eutectic microstructure. This led to improved deformability of the ceramic materials at room temperature.

Results

Compression test of the eutectic micropillars

Table 1 summarizes the geometries of the eutectic micropillars (e1–e20, E1–E4) along with the yield stress, fracture stress, and engineering plastic strain that were calculated from the engineering stress–strain curves. The yield stresses of the plastically deformed micropillars were determined to be 0.1% proof stress (σ0.1). Some micropillars were compressed until fracture occurred, whereas the compression of others were intentionally stopped and micropillars were unloaded before fracture. Engineering plastic strain values in the fractured micropillars denoted by “(f)” (Table 1) represent the strain observed until fracture. In contrast, in the unloaded micropillars, the reported values denote the strain observed until unloading. Notably, most micropillars exhibited engineering plastic strains of 1–5%, a significant departure from the brittle fracture that is typically observed in high-strength ceramic materials at room temperature. Figure 1a displays the representative stress–strain curves obtained from the eutectic micropillars, accompanied by a scanning electron microscopy (SEM) image of the eutectic micropillar before compression. The image illustrates bright rod-like, single crystal GAP phases aligned in a dark, single crystal Al2O3 matrix phase. The angle θ between the compression axis and the GAP rod growth direction (equivalent to the crystal growth direction) was considered. Because our previous study confirmed that the crystal growth direction was [0001]Al2O318, θ is also regarded as the angle between the compression axis and [0001]Al2O3. The stress–strain curves in Pillars e1 (θ = 4.7°), e15 (θ = 32.1°), and E3 (θ = 37.6°) are shown in Fig. 1a. Pillars e15 and E3 were intentionally stopped and unloaded at engineering plastic strains of 3.0% and 1.4%, respectively. Pillar e1 exhibited brittle fracture at 17.7 GPa, whereas Pillar e15 yielded at 9.8 GPa (as indicated by the red arrow), followed by plastic deformation with strain hardening. Pillar E3 demonstrated a deformation behavior similar to Pillar e15 with a yield stress of 7.5 GPa and subsequent plastic deformation of 1.4% with strain hardening. The yield stresses in the micropillars did not show a remarkable difference with varying sizes. This suggests a negligible size effect on the deformation behavior of these micropillars. The SEM images of Pillars e1 and e15 after the compression test are shown in Fig. 1b, c respectively. Pillar e1 exhibited vertical cracking within individual phases but not along the interfaces. In addition, in several specimens that failed after plastic deformation, interphase cracks were partly observed. Supplementary Fig. 4 exhibits Pillar e8 and Pillar e10 after compression. These specimens exhibited plastic deformation and then fractured. The fractured micropillars showed not only cracks across the interfaces but also those along some of the interfaces. In contrast, Pillar e15 exhibited plastic deformation without any cracks or interphase delamination. Overall, the eutectic micropillars deformed by plastic deformation of the individual phases, and interfacial fracture did not occur until just before the fracture of the micropillars. Figure 1d presents SEM images of Pillar E3 before and after the compression test, revealing more pronounced plastic deformation due to the larger specimen size. Importantly, the micropillar underwent smooth deformation without obvious slip traces or interphase delamination after the compression test. This indicates that plastic deformation progressed not through persistent and localized slip bands but through dislocation activity across a large part of the micropillar. Figure 1e depicts the θ dependency of the deformation behavior of the eutectic micropillars. Fracture stresses of the micropillars exhibiting brittle fracture are plotted with blue triangles, while σ0.1 values of micropillars demonstrating plastic deformation are plotted with red circles (Pillar e3, e7–e19) or red squares (Pillar E1–E4). Micropillars with relatively small θ values (<15°) generally exhibited brittle fracture at approximately 20 GPa. In contrast, micropillars with larger θ values displayed plastic deformation with yield stresses of ≤17 GPa. The yield stress tended to decrease with increasing θ from 15° to 40° and leveled off at 7–9 GPa beyond 40°. This trend suggests that the plasticity of the eutectic micropillars follows Schmid’s law, where plastic yielding occurs when the critical resolved shear stress (CRSS) is reached in a dominant slip system. The steep drop in yield stress with increasing θ from 15° to 40° indicates that the dominating slip direction and/or slip plane normal could be orthogonal to the crystal growth direction. The CRSS was not reached before the occurrence of fracture at approximately 20 GPa in the low angular range of θ < 15°. Notably, micropillars with θ = 90° showed little engineering plastic strain, approximately 1% or less, as demonstrated in Supplementary Fig. 6.

Table 1 Geometry of the micropillars comprising the Al2O3–GAP eutectic composite, including pillar ID, θ, top diameter \(\phi\), height h, aspect ratio h/\(\phi\), taper angle ψ, σ0.1, fracture stress, and engineering plastic strain
Full size table
Fig. 1: Deformation behavior of the Al2O3-GAP eutectic micropillars.
figure 1

a Stress–strain curves obtained from the compression test on Pillars e1, e15, and E3. Pillar e1, which showed brittle fracture, is colored blue, while Pillar e15 and E3, which showed plastic deformation, are colored red. The arrows indicate the yield points. SEM images of (b) Pillar e1 and (c) Pillar e15 after compression. d SEM image of Pillar E3 before (left) and after (right) the compression test. Outline of the pillar is traced by white dashed line. e The relationship between θ and the fracture stress of the micropillars showing brittle fracture (plotted by blue triangles) and σ0.1 of the micropillars that display plastic deformation (plotted by red circles or squares). Schematics of the eutectic micropillar at θ = 0°, 45° and 90° are depicted at the bottom. Source data are provided as a Source Data file.

Full size image

Compression test of single crystal micropillars

The deformation behavior of Al2O3 or GAP single crystals was investigated to elucidate the plastic deformability observed in the Al2O3–GAP eutectic micropillars. Here, the crystal structure of Al2O3 is corundum (space group R\(\bar{3}\)C, 167), while that of GAP is orthorhombic perovskite (space group Pbnm, 62). Typical deformation behaviors of the micropillars of Al2O3 and GAP single crystals are depicted in Fig. 2a, b, respectively. The inverse pole figures (IPF) in Fig. 2 illustrate crystal orientations parallel to the compression axis of the Al2O3 and GAP single crystal micropillars. The engineering stress–strain curves, SEM images after compression tests, and the corresponding compression axes of Al2O3 micropillars i–iii in the IPF are presented in Fig. 2a. Compression axes were selected within approximately 16°–90° from the [0001] direction indicated by the pink-colored region in the IPF. The eutectic micropillars exhibited plastic deformation when the orientation of the Al2O3 phase for the compression axis are in this angular range. In contrast to them, all Al2O3 micropillars exhibited brittle fracture without plastic deformation, as demonstrated by the stress–strain curves in Fig. 2a. The micropillars i–iii showed cleavage on specific planes, as indicated by white arrows in the SEM images. The observed cleavage planes were as follows: {1\(\bar{1}\)02} rhombohedral plane on Pillar i, {1\(\bar{1}\)00} prismatic plane on Pillar ii, and {11\(\bar{2}\)0} prismatic plane on Pillar iii. Rhombohedral cleavage in Pillar i aligned with the cleavage behavior observed in the in situ micropillar compression test of an Al2O3 single crystal, where the slip on {1\(\bar{1}\)02} < 0\(\bar{1}\)11> occurred in [0001]-oriented micropillars24. The prismatic cleavage was probably associated with the {11\(\bar{2}\)0} < 1\(\bar{1}\)00> prismatic slip, which is a preferred slip system in Al2O3 at temperatures below 600 °C25. Cleavage on a basal plane associated with the (0001) < 11\(\bar{2}\)0> basal slip was not observed in the present study; typically, the basal slip system is activated at temperatures above 600 °C25. The cleavage fracture along the specific planes must be formed due to rapid shear propagation caused by dislocation activity on the cleavage plane at the fracture stress. The similar shear failure response has also been observed in other oxide ceramics such as yttria-stabilized tetragonal zirconia (YSZ), in which dislocation activities have been shown to trigger cleavage fracture on the glide plane26.

Fig. 2: Deformation behavior of the Al2O3 or GAP single crystal micropillars.
figure 2

a IPF of the crystal orientation parallel to the compression axis of the micropillars comprising the Al2O3 single crystal. X plots are colored according to the cleavage planes. The pink region corresponds to the orientation tilted by 16° to 90° from [0001]Al2O3. Stress–strain curves on micropillar compression tests and the SEM images after compression of micropillars indexed as i, ii, and iii in the IPF are attached. White arrows in SEM images indicate the cleavage plane. b IPF of the crystal orientation parallel to the compression axis of the micropillars comprising the GAP single crystal. The crystal orientation is expressed by pseudo-cubic (pc) lattice defined in Methods. The shape of the plots corresponds to the deformation behavior. Stress–strain curves on micropillar compression tests and the SEM images after compression of micropillars indexed as iv, v, and vi in the IPF are attached. White arrows in SEM images indicate the cleavage plane. The crystal orientations of the micropillars are also shown in Supplementary Table 2. Source data are provided as a Source Data file.

Full size image

The engineering stress–strain curves, SEM images after compression tests, and the corresponding compression axes of the GAP micropillars iv–vi in the IPF are presented in Fig. 2b. In this figure, the compression orientation of each GAP micropillar is defined in a pseudo-cubic lattice and plotted in a standard triangle. The shape of the plots in the IPF corresponds to the compressive response of each micropillar. Regardless of the compression axes, all GAP micropillars exhibited brittle fracture. Some micropillars compressed in nearly [001]pc directions including Pillar v showed cleavage on {110}pc, which is an active slip plane in SrTiO3 cubic perovskite27. This result suggests that the cleavage fracture was caused by rapid shear propagation with dislocation activity, although the slip system of GAP has not been clarified. Micropillars compressed in nearly [101]pc directions (marked by triangles) exhibited a small strain burst lower than 0.5%, as indicated by brown arrows in the stress–strain curve of Pillar vi. This strain burst was probably caused by cracking without plastic deformation, because the elastic slope remained the same before and after the strain burst.

The deformation behavior in the micropillars of Al2O3 or GAP single crystal markedly differed from that observed in the eutectic micropillars. During elastic deformation with a compression stress of >10 GPa, all the single crystal micropillars exhibited brittle fracture, regardless of the compression axes.

TEM observation of the deformed eutectic micropillars

The microstructure of the Al2O3–GAP eutectic micropillars after plastic deformation was examined using TEM to elucidate the plastic deformation process. TEM images of the longitudinal cross-section of Pillar E4, including the compression axis are shown in Fig. 3. Figure 3a shows a bright-field image of the entire specimen and a schematic of the deformed eutectic microstructure. The TEM specimen was encased in Pt protection. This suggests that the micropillar’s surface was preserved within the TEM specimen. The pillar’s top is oriented upward in the figure. The bright and dark phases represent the Al2O3 and GAP phases, respectively. Noteworthy observations from the deformed microstructure include the following: (i) the eutectic microstructure displayed continuously curved interfaces across the right to left edges of the micropillar, (ii) both Al2O3 and GAP phases exhibited cooperative deformation without interphase fracture or delamination, and (iii) numerous ununiform contrasts, likely induced by dislocations, were observed in the Al2O3 phase. After plastic deformation, a similar microstructure was also observed in other Al2O3–GAP eutectic micropillars.

Fig. 3: Longitudinal cross-section of Pillar E4 after compression.
figure 3

a Montage bright-field TEM image of the longitudinal cross-section of Pillar E4, including the compression axis and schematic. Direction of the pillar top and the pillar bottom is presented. The yellow lines in the schematic represent the traces of the Al2O3–GAP interface in the curved area. b Magnified image of the area in the orange square in (a). c–e SAED patterns with a zone axis of [010]GAP taken from the areas indicated by yellow circles in (b). The yellow lines trace the 00lGAP systematic reflections. The rotation angle of the Al2O3-GAP interface between the area corresponding to (c) and (d, e) is presented in (b).

Full size image

The gradual and continuous curvature of the eutectic interfaces was present in the top and middle parts of the micropillar, as indicated by the yellow lines in the schematic in Fig. 3a. The rotation angle of the curved interface observed in the middle part of the micropillar was 10°. Strain contours were intensively observed in the deformed region with curvature but were limited in the less deformed region near the bottom of the pillar. These contours were stronger in the Al2O3 phase than in the GAP phase in the bright-field image. This suggests the presence of considerable dislocations in the Al2O3 phases, although individual dislocations could not be resolved in this cross-section. The magnified view in Fig. 3b provides a closer look at the deformed area, and Fig. 3c–e demonstrate selected area electron diffraction (SAED) patterns from three portions near the same Al2O3–GAP interface from top-right to bottom-left as indicated by the yellow circles in Fig. 3b. The zone axis in Fig. 3c–e was [010]GAP and the orientation relationship in this micropillar was characterized as [0001]Al2O3//[\(\bar{1}\)02]GAP with respect to the crystal growth direction. Such orientation of GAP phases was also observed in directionally-solidified Al2O3–GAP eutectic28. The continuous counterclockwise rotation up to 10° of the yellow lines tracing the 00lGAP systematic reflections in Fig. 3c–e corresponds to the rotated interfaces in the bright-field image in Fig. 3b. This continuous crystal rotation across the micropillar is associated with the geometrical curvature of the Al2O3–GAP interface observed in Fig. 3a, b. This continuous crystal rotation must be associated with primary crystallographic slip activities under constraints from the indenter tip and the Al2O3–GAP substrate, as discussed later.

The TEM images in the cross-section perpendicular to the crystal growth direction picked up from Pillar E3 are presented in Fig. 4. Figure 4a shows the bright-field TEM image of the specimen. In addition, the SEM image in Fig. 4b indicates a portion of the cross-sectional TEM specimen. The diffraction pattern acquired from the entire TEM specimen in Fig. 4c indicates the zone axis of [0001]Al2O3//[\(\bar{1}\)02]GAP. Figure 4d–f represents the dark-field images obtained by g = 03\(\bar{3}\)0, 30\(\bar{3}\)0, and 3\(\bar{3}\)00 in Al2O3, respectively, taken from the area inside the white square in Fig. 4a. The bright contrasts in Fig. 4d–f indicate the appearance and disappearance of dislocations under different excitation conditions. The g⋅b analysis indicates that each of these dislocation segments has one of the three different Burgers vectors of b1 = 1/3[\(\bar{2}\)110], b2 = 1/3[1\(\bar{2}\)10], and b3 = 1/3[11\(\bar{2}\)0], as depicted with different colors in Fig. 4g. In addition, their curved morphology suggests that these dislocations lie on (0001)Al2O3 as the slip plane. Each dislocation had both ends at individual GAP rods without a dislocation (Orowan) loop surrounding the GAP phase. In contrast, evident dislocations could not be observed in the GAP phases. For continuous plastic deformation without interface fracture, the dislocations on the plane perpendicular to the crystal growth direction should transmit through the GAP rods without leaving Orowan loops and glide across the entire micropillar. It is possible that dislocations in the GAP rods remained near the Al2O3–GAP interfaces, and the contrasts were covered by the fringes of the interfaces. As additionally demonstrated in Supplementary Fig. 5, bright diffraction contrasts are observed within the GAP phase but concentrated in the vicinities of interfaces. These contrasts might be attributed to such dislocations which have glided within the GAP phase. The Al2O3–GAP interface can be a barrier to dislocation motion, due to lattice mismatch and discontinuity of slip systems between phases29,30.

Fig. 4: Cross-section perpendicular to the crystal-growth direction of Pillar E3 after compression.
figure 4

a Bright-field TEM image of the specimen picked up from Pillar E3 after compression. b SEM image of Pillar E3 after compression. The white dashed line indicates the approximate location of the TEM specimen. The left and right side of the TEM specimen presented as “L” and “R” in (a) correspond to those in (b). c The diffraction pattern corresponding to (a), with a zone axis of [0001]Al2O3//[\(\bar{1}\)02]GAP. d–f Dark-field TEM images in the area indicated by white squares in (a). d g = 03\(\bar{3}\)0 Al2O3, e g = 30\(\bar{3}\)0 Al2O3, and f g = 3\(\bar{3}\)00 Al2O3. The area in (d–f) is on the back side in (b). g Schematic of the TEM image in (d–f) with observed dislocations. Dislocations are colored by the Burgers vectors characterized from g⋅b analysis with (d–f). The orientation of Al2O3 phase is shown as coordinate axis.

Full size image

Evaluation of fracture toughness by indentation tests

We conducted micro-Vickers indentation fracture tests on the Al2O3 and GAP single phases as well as the Al2O3–GAP eutectic region to evaluate the effect of the eutectic microstructure on the fracture toughness, as summarized in Supplementary Note 2. The crack paths were straight forward in both single phases, while slight crack deflections occurred at the Al2O3–GAP interfaces in the eutectic microstructure, as shown in Supplementary Fig. 2a–c. In addition, the crack length introduced in the eutectic materials by the micro-Vickers indentation tests was significantly shorter than those on the Al2O3 and GAP single phases. The fracture toughness determined by indentation fracture method of the Al2O3–GAP eutectic composite was 2.25 ± 0.28 MPa⋅m1/2, which was definitely higher than 1.60 ± 0.05 MPa⋅m1/2 and 1.07 ± 0.18 MPa⋅m1/2 in the Al2O3 and GAP single phases, respectively, as presented in Supplementary Table 1. In general, the relatively high fracture toughness in anisotropic composites has often been attributed to crack deflection, bridging and branching8,13. In fact, crack deflection was slightly observed in the Al2O3-GAP eutectic in Supplementary Fig. 2c. However, plastic deformability could also contribute to the improved fracture toughness in the present materials. Therefore, the contributions of crack deflection and plastic deformation need to be carefully separated in the future study.

Discussion

When the crystal growth direction was tilted by more than 15° from the compression axis, the Al2O3–GAP eutectic micropillars showed significant plastic deformation. In contrast, the Al2O3 or GAP single crystal micropillars showed brittle fracture without plastic deformation, which is generally observed in high-strength ceramics. In other words, the eutectic composite with refined microstructure consisting of brittle phases could exhibit plastic deformability under the compression axis with a significant degree of freedom. Pillar size effect was not observed in a range of 1–2 µm diameter. This is probably because the microstructure length scale, i.e. the interface spacing of 170 nm, was significantly smaller than the pillar size.

The TEM observations of the deformed eutectic micropillars revealed that both the Al2O3 and GAP phases cooperatively showed continuous deformation with gradual crystal rotation without any crack or interphase delamination, as shown in Fig. 3. It is also noteworthy that the (0001) < 11\(\bar{2}\)0 > Al2O3 basal slip was activated in the eutectic micropillars, as shown in Fig. 4, although the basal slip of Al2O3 is generally inactive at room temperature25,31,32.

The plastic deformation of the Al2O3–GAP micropillars was probably limited by the deformation of the harder Al2O3 phase (Supplementary Fig. 1). Furthermore, the perovskite GAP phase probably has more slip systems than the corundum Al2O3 phase like {1\(\bar{1}\)0}c < 110>c and {001}c < 110>c as known in other perovskite oxides27.

Crystal rotation occurred in the middle part of the eutectic micropillar. Hence, the crystal growth direction deviated from the compression axis, as shown in Fig. 3. This crystal rotation can be explained by Taylor’s model33,34 for plastic deformation of the crystal under compression. According to this model, when each specimen end is constrained by compression plates, the slip plane normal rotates toward the loading axis. Applying this model to the eutectic micropillars, the dislocation slip with the slip plane normal orthogonal to the crystal growth direction, which corresponds to [0001]Al2O3, was activated as the primary slip. In Al2O3, the {11\(\bar{2}\)0} < 1\(\bar{1}\)00> prismatic slip has been reported as the primary slip system at temperatures below 600 °C, above which the (0001) < 11\(\bar{2}\)0 > Al2O3 basal slip becomes the easiest slip system25,31. Furthermore, the (0001) < 11\(\bar{2}\)0 > Al2O3 basal slip was also activated in the eutectic micropillars, as shown in Fig. 4. These results indicate that primary prismatic slip and secondary basal slip are required for plastic deformation of the eutectic micropillar.

A similar deformation behavior has been reported for the micropillar compression of WC single crystals at 600 °C35,36; the single-crystalline hexagonal WC micropillar with the c axis tilted by 10° from the compression axis exhibited continuous deformation with gradual crystal rotation, with primary prismatic slip in the entire pillar and secondary basal slip near the pillar top. It was proposed that the continuous crystal rotation in the WC micropillar was caused by the glide of the prismatic dislocation, and the secondary basal dislocation. The former generated most of the strain, and the latter accommodated the stress concentration built up at the pillar top due to the constraint from the indenter. This would be the similar behavior for the Al2O3–GAP micropillar that deformed under compression.

The anisotropy of the yield stress in Fig. 1e can be approximately explained by assuming that the {11\(\bar{2}\)0} < 1\(\bar{1}\)00> was the primary slip system in the Al2O3 phase. If the eutectic micropillar yields when the CRSS of the Al2O3 prism slip τCRSS is reached, the theoretical yield stress σy can be estimated as below:

$$\frac{{E}_{{{{\rm{Al}}}}2{{{\rm{O}}}}3}+{E}_{{{{\rm{GAP}}}}}}{2{E}_{{{{\rm{Al}}}}2{{{\rm{O}}}}3}}\left(\frac{{\tau }_{{{{\rm{CRSS}}}}}}{{M}_{\max }}\right)\le {\sigma }_{{{{\rm{y}}}}}\le \frac{{\tau }_{{{{\rm{CRSS}}}}}}{{M}_{\min }}$$
(1)

The lower limit was calculated using the equal-strain model with compressive orientations on the mean line of the standard stereographic triangle, while the upper limit was calculated using the equal-stress model with compressive orientations on the <1\(\bar{1}\)00> zone. The Schmid factors of the {11\(\bar{2}\)0} < 1\(\bar{1}\)00> prismatic slip in these cases are described as Mmax and Mmin, respectively. EAl2O3 = 388 GPa and EGAP = 326 GPa are the Young’s moduli of Al2O3 and GAP phases obtained by nanoindentation (Supplementary Fig. 1). For θ < 40°, Eq. (1) yields τCRSS = 0.5–1.5 GPa, which is considerably lower than the CRSS for prism slip at room temperature estimated from the value obtained from bulk compression test at >200 °C in the previous report (~4 GPa)25. However, the obtained CRSS values increased to 3–5 GPa for θ > 40°. This result suggests that the secondary basal slip must lower the CRSS of the primary prismatic slip. This is also consistent with the low engineering plastic strain to failure with a cleavage on prismatic planes at θ = 90° (Also see Supplementary Fig. 6).

The origin of the activation of the Al2O3 basal slip system in the eutectic composite remains an open question. First, the effect of flash processing should be examined, because that may introduce excessive lattice defects such as microcracks or dislocations37,38. In our previous study18, however, TEM observation revealed that neither microcracks nor excessive dislocations existed in the as-fabricated Al2O3–GAP eutectic composite produced using a flash event. In addition, Al2O3–GAP eutectic micropillars fabricated by a conventional melt-grown method without electric field also exhibited plastic deformation without any failure up to an engineering plastic strain of 3%, as shown in Supplementary Fig. 3. This result indicates that the plastic deformability of the Al2O3–GAP eutectic composite was not due to the flash processing, but the refined eutectic microstructure.

The effect of the Al2O3–GAP interface is a very important issue to consider the mechanical response of the refined eutectic microstructure. The possible roles of the Al2O3–GAP interface are as follows; (1) sources and barriers for dislocations, (2) crack paths, and (3) origins of residual stress. It is possible that the Al2O3–GAP interfaces may be involved in triggering the dislocation activity. Several previous reports have investigated the mechanical response of lamellar composites of ductile metal phase and brittle intermetallic or ceramic phase39,40,41. The brittle phase in the lamellar composites exhibited cooperative plastic deformation with the metallic phase without cracks, when the interlamellar spacing was thinned to nanoscale. The plastic deformation of the brittle phase was attributed to the motion of dislocations nucleated from the heterointerfaces39,42,43,44,45,46. A very recent study also showed that La2O3 ceramic phase bonded with Mo phase by orderly-bonded interface exhibited significant ductility at room temperature. The improved ductility was explained in terms of activated dislocations propagated through the metal-ceramic interface from Mo phase47. The heterointerfaces in the Al2O3–GAP eutectic micropillars might also have acted as dislocation sources. However, in several micropillars that fractured after the deformation, fracture cracks partly propagated at the heterointerfaces, as shown in Supplementary Fig. 4. This result suggests that the Al2O3–GAP interfaces may also limit their plastic deformability at some extent. On the other hand, for the residual stress generated at the interfaces, it does not appear that sufficient internal residual stress is generated to activate dislocation motion. The reported residual stress in Al2O3 phase of the directionally-solidified Al2O3–GAP eutectic material with Chinese-script microstructure measured by Raman shift is 200–300 MPa in compression at room temperature48. The residual stress is likely insufficient for the remarkable reduction in yield strength at GPa orders in the Al2O3–GAP eutectic. Actually, dislocations were not observed by TEM observation of the as-fabricated Al2O3–GAP eutectic composite produced using a flash event. Further investigations are required to reveal the multiple roles of interfaces as sources and barriers for dislocations, crack paths, and origins of residual stress in more detail.

Finally, we would like to emphasize that a micropillar compression test on Al2O3–8 mol% Y2O3-stabilized ZrO2 (8YSZ) fibrous eutectic composite fabricated using a flash event demonstrated the transferability of the current approach. The Al2O3–8YSZ eutectic consisted of Al2O3 as the matrix phase and rod-like 8YSZ phase aligned within the matrix phase with the interphase spacing of about 150 nm, as well as the Al2O3–GAP eutectic composite. Despite brittleness of the constituent Al2O3 phase, the eutectic micropillar exhibited significant plastic deformation up to 7% engineering plastic strain without failure (Supplementary Fig. 7). This result shows that the enhancement of the plastic deformability due to the refined eutectic microstructure is not a phenomenon unique to the Al2O3–GAP, but is universally transferrable, at least in the Al2O3–YSZ system.

In summary, we conducted micropillar compression tests on the Al2O3–GAP eutectic composite with the refined rod-like GAP phase aligned in the Al2O3 matrix. The micropillars of the Al2O3–GAP eutectic showed continuous plastic deformation with an engineering plastic strain of 1–5% under the compression axis with a significant degree of freedom. In contrast, the micropillars of the Al2O3 or GAP single crystals exhibited brittle fracture with compression stress larger than 10 GPa, regardless of the crystal orientation to the compression axis. The Al2O3–GAP eutectic micropillars showed plastic deformation at the yield stress of lower than 17 GPa, where the angle between the crystal growth direction and the compression axis was larger than 15°. In contrast, the micropillars with smaller angles showed a brittle fracture at a fracture stress of approximately 20 GPa. TEM observation of the deformed eutectic micropillars revealed that the plastic strain was produced by the geometrically gradual and continuous rotation of the Al2O3 and GAP crystals without delamination. This crystal rotation suggests that the plastic strain was primarily attributed to the {11\(\bar{2}\)0} < 1\(\bar{1}\)00> prismatic slip in the Al2O3 phase. In addition, the dislocation activity on the (0001) < 11\(\bar{2}\)0 > Al2O3 basal slip system, which is known to be inactive at room temperature, was observed from the cross-section perpendicular to the crystal growth direction. These results indicate that in high-strength ceramic materials, plastic deformability can be induced by fabricating an aligned eutectic microstructure with submicron-scale interphase spacing. Future research needs to be focused on several key ideas, including the mechanisms, transferability, and practical applications. In particular, it will be crucial to investigate the role of eutectic interfaces as dislocation sources, examine the impacts of eutectic systems and microstructures (e.g. morphologies and dimensions), and develop the strategies to extend these findings to macroscopic scales. These comprehensive studies would make a new guiding principle for improving the reliability of high-strength structural ceramics.

Methods

Fabrication of eutectic composite specimens

Commercially available raw powders of Al2O3 (TM-DAR; Taimei Chemical, purity >4 N) and Gd2O3 (Kanto Chemical, purity >3 N) with eutectic composition (at the Al2O3:Gd2O3 = 77:23 molar ratio) were mixed by ball milling in ethanol for 24 h. A rectangular-shaped powder compact with a cross-section of 1.4 × 4.0 mm and a length of 23.5 mm was calcined at 1450 °C for 2 h in air. Pt wires were attached to both ends of the calcined specimen using Pt paste. The specimen was then heated to 1300 °C in air, while suspended by the Pt wires in a box furnace. Subsequently, a sinusoidal voltage of 1000 V⋅cm−1 at the frequency of 1 kHz was applied to the specimen from an external AC power supply with a current limit of 110 mA for 40 s. The specimen was partially melted by a flash event and accompanying Joule heating, and then abruptly cooled by turning off the external power supply. The specimen’s appearance, referred to as the flash-treated specimen, is shown in Supplementary Fig. 8a in the Supplementary information. The partially colored region corresponds to the melted region. An Al2O3–GAP eutectic microstructure with an average interphase spacing of 170 nm was formed in the melted area. The detailed material preparation method is also described in the previous report18. For a supplemental experiment, an Al2O3–8YSZ eutectic composite was also fabricated in the same way as the Al2O3–GAP composites. The result of the experiment is shown in Supplementary Fig. 7.

After polishing the surface with diamond slurry, the microstructure of the flash-treated specimen was observed by SEM (JEOL JSM-7000F). Supplementary Fig. 8b demonstrates the SEM image in the melted region, where the dark and bright regions correspond to the Al2O3 and GAP phases, respectively. The microstructure exhibited fibrous eutectic structures with rod-like GAP phases aligned in the Al2O3 matrix, with an average interphase spacing of approximately 170 nm. Additionally, coarsened crystals of Al2O3 and GAP with a grain size of around 5–10 µm were partially observed in the melted region, as shown in Supplementary Fig. 8c, d.

Micropillar compression experiment

Cylindrical micropillars with a top diameter of 1 µm and a height of 2–3 µm were fabricated in the fibrous eutectic microstructure of the flash-treated specimen, which are referred to as Pillar e1–e20. They were fabricated by focused ion beam (FIB, JEOL JIB-4700F) technique at an acceleration voltage of 30 kV and beam currents of 10 nA to 10 pA after Pt deposition with a thickness of approximately 2 nm by Pt coater (JEOL JEC-3000FC) to protect the specimen surface from damage by ion beam. The angle between compression axis of micropillars and the GAP rod growth direction (corresponding to the crystal growth direction) observed from the pillar surface was defined as θ to each micropillar fabricated from the Al2O3–GAP eutectic. Because our previous study confirmed that the crystal growth direction was oriented in the [0001] Al2O3 direction18, θ was also regarded as the angle between the compression axis and [0001]Al2O3. Micropillars with different θ values, ranging from nearly 0° to 90° were fabricated to observe the anisotropy of the deformation behavior. Larger micropillars with a top diameter of 2 µm and a height of 5–6 µm were also fabricated in the anisotropic eutectic microstructure in the same way to perform TEM observation in the microstructure after compression. They are referred to as Pillar E1–E4. For comparison, micropillars comprising Al2O3 or GAP single crystals with a top diameter of 1 µm and the height of 2–3 µm were fabricated from the Al2O3 or GAP coarsened crystals, after crystallographic orientation analysis using electron backscatter diffraction (EBSD), as shown in Supplementary Figs. 8e, f. GAP generally has an orthorhombic lattice structure49. However, due to the small difference in the lattice parameters (\(\sqrt{2}a\) = 7.426 Å, \(\sqrt{2}b\) = 7.497 Å, c = 7.444 Å)50, it is difficult for the conventional EBSD technique to distinguish the three orthogonal directions. When considering the plastic deformation of orthorhombic perovskites, the lattice structure is often approximated by pseudo-cubic lattice in the literature51,52. In the present study, the GAP phase was assumed to be a pseudo-cubic lattice with a lattice parameter of 7.4 Å. The correspondence between orthorhombic axes and pseudo-cubic axes was as follows: [100]pc = [1\(\bar{1}\)0], [010]pc = [110], [010]pc = [001], where the subscript “pc” represents the orientation index in the pseudo-cubic lattice unit cell. The fabricated micropillars were slightly tapered with a taper angle of < 5°. Uniaxial compression tests were performed on the micropillars using a nanoindenter (Bruker TI-980 Triboindenter) with a 30 µm diameter diamond flat-end tip in a displacement-rate control with an initial strain rate of 10−3 s−1 at room temperature (25 °C). Each compression test was started after the drift rate of the tip fell below 0.5 nm∙s−1. Some micropillars exhibited abrupt load drop to 0 N during compression test, which means rapid and discontinuous increase of displacement in the micropillars. This point was regarded as fracture of the micropillars.

The raw load–displacement data were corrected in the stress–strain analysis to eliminate the influence of elastic compliance in both the diamond tip and base materials. These values were calculated as Sneddon’s compliance (Si, i: tip or base)53:

$${S}_{i}=\frac{\sqrt{\pi }\left(1-{v}_{i}^{2}\right)}{2{E}_{i}\sqrt{{A}_{i}}}$$
(2)

where vi and Ei are Poisson’s ratio and Young’s modulus of the polycrystal, and Ai is the contact area with the micropillar. In the present study, vtip = 0.07, vbase = 0.22 for Al2O3, 0.25 for GAP, 0.26 for Al2O3-GAP eutectic, Etip = 1140 GPa and Ebase = 400 GPa for Al2O3, 300 GPa for GAP, 373 GPa for Al2O3-GAP eutectic, respectively54,55,56,57. Ai were calculated for each micropillar using the ϕ, h and ψ values presented in Table 1 and Supplementary Table 2. The displacement of micropillar (Δh) was corrected from the compliances, measured load (P) and total displacement (Δhtot) by the below equation:

$$\Delta h=\Delta {h}_{{tot}}-\left({S}_{{{{\rm{tip}}}}}+{S}_{{{{\rm{base}}}}}\right)P$$
(3)

Eventually, the engineering strain (\(\epsilon\)) was obtained by the below relation:

$$\epsilon=\frac{\Delta h}{h}$$
(4)

The engineering stress (σ) was obtained from the below relation:

$$\sigma=\frac{P}{{A}_{{{{\rm{top}}}}}}$$
(5)

where Atop is the surface area of the pillar top before compression. Corrected stress–strain curves of the Pillars e1–e20 and E1–E4 are shown in Supplementary Fig. 9 in the Supplementary information. The same correction procedure is also described in our previous report26.

Yield stresses of the plastically deformed micropillars are defined as 0.1% proof stress (σ0.1), whereas metallic materials often use 0.2% proof stress. Because of the relatively small engineering plastic strain observed (5% at most), the present study opted for σ0.1.

The micropillars were observed by SEM before and after compression. Specifically, the eutectic micropillar with a top diameter of 2 µm after compression was subjected to further examination using TEM. After compression, the micropillars were thinned using FIB (FEI Helios 650) into two different cross sections: one parallel and another perpendicular to the crystal growth direction. During FIB thinning, acceleration voltages ranging from 2 to 30 kV and beam currents between 9 pA and 47 nA were used. Subsequently, the cross sections were lifted using a tungsten probe, affixed to a silicon mesh (Hitachi NANOMESH) via Pt deposition, and finally milled to a thickness of less than 100 nm. Then, conventional TEM (JEOL JEM-2100HC) observations were performed on the thinned specimens.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.