The fabrication procedure of ZrC/ZrB₂/C composites is as follows. The starting powders were ZrC (Japan New Metals, Osaka, Japan; average particle size P_s of ϕ2.0 μm, purity of 98.5%, theoretical density D_x of 6.634 Mg·m⁻³ (Powder Diffraction File, PDF: #35-0784) and amorphous B (H.C. Starck, Goslar, Germany; grade I, P_s of ϕ0.2 μm, purity ≥ 95%, D_x of 1.73 Mg·m⁻³ [15] ). ZrC and amorphous B powders were weighted into the desired ratios to fabricate ZrC/ZrB₂/C composites with the compositions of (24.97 - 64.44) vol% ZrC/(58.27 - 27.62) vol% ZrB₂/(16.76 - 7.94) vol% C; in other words, the main components of ZrC/ZrB₂ = 30/70~70/30 vol%, as shown in Table 1. They were synthesized by solid state reaction presented by the following Equation (1):

( x + y ) ⋅ ZrC + 2 y ⋅ B → x ⋅ ZrC + y ⋅ ZrB 2 + y ⋅ C (1)

Here, it should be noted that the compositions in Table 1 are displayed in vol% because when we discuss the mechanical properties of composites, their components should be presented in vol%. Therefore, “mol” in Equation (1) is transformed into “vol” in Table 1. For example, in order to fabricate ZrC/ZrB₂ = 50/50 vol% composite, x = 3.214 and y = 2.706 are calculated based on the values of dD_x(ZrC) = 6.634, D_x(ZrB₂) = 6.104 (PDF: #34-0423), and D_x(C) = 2.26 Mg·m⁻³ (PDF: #19-178)] and their molecular weights.

The ZrC and B mixed powders were dispersed into ethanol by an ultrasonic vibrating homogenizer with a 300 W output power at a frequency of 20 kHz (US-300T, Nihon Seiki, Tokyo, Japan) for 1.80 × 10³ s. The obtained slurries

⁺: calculated using D_x(ZrC) = 6.634 Mg·m⁻³ (PDF:#35-784), D_x(ZrB₂) = 6.104 Mg·m⁻³ (PDF:#34-423) and D_x(C) = 2.26 Mg·m⁻³. *: vol% presented as ZrC/ZrB₂ ratio. ^#: vol% presented as ZrC/ZrB₂/C ratio.

were dried at room temperature in air, after a small amount of Acrylic/PVA 3 mass % solution was added as a binder to the dried dispersed powders, and mixed using an alumina mortar and pestle for 1.80 × 10³ s. The mixtures were uniaxially compacted into a disk with a 20 mm diameter and a ~7 mm thickness at 75 MPa, and then cold isostatically pressed (CIP) at 245 MPa for 1.80 × 10² s.

The CIPed dense green compact, corresponding to ZrC/ZrB₂ = 50/50 vol%, with a bulk green density D_bulk(g) of 3.15 - 3.16 Mg·m⁻³ estimated from the volume and weight, was wrapped with a hexagonal BN powder (P_s: f3 µm) and put into a cylinder mold (inner diameter ϕ22 mm, outer diameter ϕ42 mm, height 30 mm) with two cylindrical plungers (diameter ϕ21.5 mm and height 30 mm) made of high-density graphite. The powder compacts were densified by PECPS (SPS-5104A, SPS SYNTEX, Yokohama, Japan) at various temperatures (1273 - 2173 K) under a uniaxial pressure of 50 MPa for 6.0 × 10² s in a vacuum (~50 Pa) with a heating rate of 1.666 K·s⁻¹d applying DC pulsed electric current (on/off interval = 12:2). Above 873 K, the heating temperature was monitored at the center position of the outside wall of the carbon mold by a monochrome pyrometer. Since this powder compact sintered at 2073 K archived the highest relative density D_r of 98.2% (details of this result will be discussed later), the powder compacts with the various ZrC/ZrB₂ compositions (i.e., 30/70-70/30 vol%, Table 1) were sintered at 2073 K/50MPa/6 × 10²s/vacuum under the same conditions described above.

Furthermore, to investigate the effects of carbon included in the composites on their mechanical properties, simple binary ZrC/ZrB₂ composites without C were also fabricated using PECPS of the mixed powders of Zr, B, and ZrC. Their powder characteristics were as follows: Zr (Kojundo Chemical Laboratory, Saitama, Japan; P_s of ϕ10 µm, purity of 98%, D_x of 6.505 Mg·m⁻³ (PDF: #5-0665) delivered in immersed, fine amorphous B (H.C. Starck; grade I, P_s of ϕ0.03 μm, purity ≥ 95%, D_x of 1.73 Mg·m⁻³) and the same ZrC powder as mentioned above. Here, a very fine amorphous active B powder was adopted so as to react with large Zr powder. The desired mixed powders, aiming at final compositions with ZrC/ZrB₂ = 40/60, 50/50, 60/40, 70/30 vol%, were compacted and sintered under the same conditions as the ZrC/ZrB₂/C composites.

Phase identification of powders and composites was made by X-ray diffraction (XRD; Smartlab, Rigaku, Tokyo, Japan) analysis using CuKα radiation (a wavelength of 0.15418 nm). The bulk density D_bulk of dense materials was measured by Archimedes’ method, and relative density D_r was estimated based on theoretical density D_x(Mg·m⁻³) which was calculated using the values of 6.634 for ZrC, 6.104 for ZrB₂, and 2.26 for C. Microstructural observation using a field-emission-type scanning electron microscope (FE-SEM; SU8020, Hitachi High-Technologies, Tokyo, Japan) equipped with an energy-dispersive X-ray spectroscope (EDS, JED-2300/F, JEOL, Tokyo, Japan) was performed. After phase identification, test bars for mechanical-property measurements were cut from the sintered materials with a diamond cutting blade and then their upper and lower surfaces were polished to mirror surfaces with a diamond paste (nominal P_s: ϕ1 - 3 μm); the other two sides were grinded with a diamond abrasive tool containing #400 grid-pass diamond particles.

At room temperature, various mechanical properties of composites were measured: three-point bending strength σ_b was evaluated with a crosshead speed of 8.33 × 10⁻⁶ m·s⁻¹ and an 8.0 mm span length using WC jigs. Vickers hardness H_v and fracture toughness K_IC were estimated with an applied load of 196 N for 15 s for the former and the indentation fracture method (IF) with Niihara’s equation [16] for the latter.

On the other hand, high-temperature bending strength σ_b of ZrC/ZrB₂/C composites was measured from 1273 to 2023 K in Ar using an AG-X Plus autograph (Shimadzu, Kyoto, Japan) at the National Institute for Materials (NIMS), Tsukuba, Ibaraki, Japan. Samples were heated at a rate of 0.5 K·s⁻¹ by tungsten heater elements. After the predetermined temperature was reached, three-point bending tests were carried out (a span length of 10.0 mm, a crosshead speed of 8.33 × 10⁻⁶ m·s⁻¹) by using a B₄C jig. During this test, a set of load displacements was measured to evaluate the stress-strain (σ - ε) curves as a function of testing temperature.

Hereafter, we sometimes denote the composite as the final composition, e.g., ZrC/ZrB₂/C = 43.71/43.71/12.5 7vol% = 50/50/C vol%, “50Zr-C”, here, we add “-C” in order to distinguish between three-phase ZrC/ZrB₂/C and binary-phase ZrC/ZrB₂ composites. As described in 2. Experimental procedure, a CIP-ed sample corresponding to 50ZrC-C was utilized to investigate the effects of PECPS temperature on the formation and densification of materials. The bulk density of the cylindrical green powder compacts [D_bulk(g)] was about 3.15 - 3.16 Mg·m⁻³, and its relative density [D_r(g)] was estimated to be 67.6% - 67.7%; these values were calculated using the theoretical density [D_x(g)] of 4.656 Mg·m⁻³ obtained from the D_x values of 6.634 Mg·m⁻³ of ZrC and 1.73 Mg·m⁻³ of B.

In order to investigate the crystalline phase change during PECPS, the ZrC/B mixed powder compacts were sintered at various temperatures from 1273 K to 2173 K for 6.0 × 10² s in a vacuum (~50 Pa) and many corresponding test pieces were prepared. XRD patterns of sintered samples with Si powder as an internal standard which had been added to determine their lattice parameters precisely are shown in Figure 1, in addition to that of the starting mixture. At a glance, it is difficult to recognize the formation of ZrB₂ at 1273 K. However, in the XRD patterns measured at 1373 K, a small diffraction peak of the ZrB₂ phase is observed around 2θ = 42˚ and there are only two phases: ZrC (PDF: #35-0784) and ZrB₂ (PDF: #34-0423) up to 2173 K; graphite C and B₄C phases are not recognized, which might be due to a much smaller diffraction peak compared with those of Zr-based compounds. The formation of graphite C is confirmed by the SEM-EDX observation, which will be described later.

A precise Rietveld analysis was performed on these XRD patterns; the crystallite sizes (X_s) of ZrC and ZrB₂, and the lattice parameters a of cubic ZrC, a and c of hexagonal ZrB₂ phases in the materials obtained at room temperature to 2173 K were investigated. Table 2 shows their results: as far as X_s is concerned, the particle size P_s of ZrC of starting material was around ϕ2.0 µm, however, its X_s was around f100 nm, indicating starting ZrC was a polycrystal powder. Each ZrC particle was attacked by amorphous active B particles around 1273 K, then the contact area on the surface of ZrC changed into a very small ZrB₂ (X_s ~f2.0

*Si was used as an external standard. ZrC (PDF#35-0784), cubic: a = 0.46930 nm, ZrB₂ (PDF: #34-423), hexagonal: a = 0.316870, c = 0.353002 nm.

nm) + C. The X_s of ZrB₂ grew gradually from f2.0 nm at 1273 K to f47 nm at 2173 K. On the other hand, the X_s of ZrC decreased gradually from ϕ100 nm (R.T.) to f56 nm (2173 K), as it had been eroded by active B. As the lattice parameters of both ZrC and ZrB₂ were almost constant and agreed with the reported PDF data, they did not form a solid solution during active PECPS, as reported in an equilibrium phase diagram of ZrC-ZrB₂ [17].

The microstructures of the composites fabricated at various temperatures were observed and their bulk densities were measured. Figure 2 shows SEM photographs of the samples sintered at 1373 K to 2173 K for 6.0·10² s. Their D_r values are also inserted into the SEM photographs. Between 1373 K and 1773 K, the microstructures changed little: mainly large f2.0 µm ZrC and fine f0.2 µm B particles are recognized with nearly the same D_r of 62%. This value is smaller than that (67.6% - 67.7%) of the starting powder compact. The difference must be due to the crystalline phase change from (ZrC/B) to (ZrC/ZrB₂/C), followed by the changes in D_x (4.656 Mg·m⁻³) and D_x (5.853 Mg·m⁻³) for the former and the latter, respectively. However, from 1973 K the sample begins to be densified into 77.2%. At 2023 K, the SEM photograph of the composite indicates a change in microstructural morphology: relatively finer grains agglomerated with each other instead of remaining isolated particles, suggesting the dynamic reconstructive reaction of (ZrC + 2·B → ZrB₂ + C) and resulting in the formation of ZrB₂ accompanied by much higher densification around 87%. When the composite was heated to 2073 K, sintering reached the final stage with a D_r of 98.2%, and grain growth to several microns was observed. Upon further heating to 2173 K, the D_r value was the same, suggesting almost the completion of densification at 2073 K, and the grain sizes (G_s) of ZrC and ZrB₂ were also unchanged. Here, the volume change from the mixture of ZrC and B to the composite of ZrC/ZrB₂/C

was considered. In the case of the composition of ZrC/ZrB₂/C = 50/50/C vol%, a volume change ratio DV/V_o, here, DV is a volume change after phase change, V_o is a starting volume, is about - 2.02%, which is calculated from the values of D_x(ZrC) = 6.634, D_x(amorphous B) = 1.73, D_x(ZrB₂) = 6.104, and D_x(C) = 1.26 Mg·m⁻³. This DV/V_o = −2.02% means that it is difficult to crush up or densify the ZrC/ZrB₂/C composite up to the theoretical density by an ordinal uniaxial pressure of 50 MPa during PECPS. Hereafter, based on these results, all composites were sintered at 2073 K. The microstructures of ZrC/ZrB₂/C composites fabricated at 2073 K were then observed. Figure 3 upper stand shows SEM photographs of their fractured surfaces, and D_r values are also presented. The D_r reaches the top value of 98.7% around the 60ZrC-C composite, i.e., ZrC/ZrB₂/C = 60(53.81)/40(35.87)/(10.32) vol% as shown in Table 1.

Here we have to mention the fabrication of simple binary ZrC/ZrB₂ as abbreviated as 40ZrC to 70ZrC composites. The XRD patterns of materials sintered at 2073K showed that the composites consisted of only two components of ZrC and ZrB₂, indicating that the solid-state reaction between Zr and B was completed at 2073 K. The 40ZrC-70ZrC composites showed D_r values higher than 99.9%; their SEM photographs are displayed in Figure 4. Both 50ZrC and 60ZrC composites show full densified microstructures with well-developed grains. If we compare the 99.9% value with the D_r of ZrC/ZrB₂/C composites, this difference might be explained in terms of the lower melting point T_m of Zr than ZrC: i.e., T_m (Zr) = 2128 K is 1405 K lower than T_m (ZrC) = 3533 K, nevertheless, there is much higher volume reduction, DV/V_o, i.e., the DV/V_o values are estimated to be from −20.66% (40 vol%ZrC) to −11.52% (70 vol%ZrC). Therefore, these high

densities might be achieved by plastic deformation of soft Zr at only 55 K lower temperature than T_m (Zr) = 2128 K during PECPS.

Then, the mechanical properties of ZrC/ZrB₂/C composites, such as three-point bending strength σ_b, Vickers hardness H_v, and fracture toughness K_IC, were measured at room temperature. Figure 5 displays these properties as a function of ZrC volume content vs. ZrB₂. The σ_b increases from 460 (30ZrC-C) to 555 MPa (60ZrC-C) and decreases gradually to 510 MPa (70 volZrC-C). The H_v values start from 13.2 to 19.2 GPa around 60ZrC-C and then pitch downward to 13.4 GPa. The K_IC takes also a constant value of 5.0 ± 0.25 MPa·m^1/2. It should be noted that all the highest values of σ_b and H_v are attained at 60ZrC-C composite. In general, the strength of inorganic composites depends much on their composition, relative density, grain and pore sizes, and distribution, and even on the preparation process. On the other hand, the hardness H_v of a composite obeys the mixing rule that H_v (composite) is the summation of H_v from each component. As each component’s H_v values of ZrC and ZrB₂ are 25.48 and 21.56 GPa, respectively [18], H_v of ZrC/ZrB₂/C composite could increase with increasing ZrC content. However, in the present composites, H_v (70 vol% ZrC-C) decreases rapidly from 19.2 to 13.4 GPa. This might be explained in terms of 1) grain growth from f2~3 µm to f3~4 µm and 2) a small decrease in relative density from 98.7% (60ZrC-C) to 98.2% (70ZrC-C) and; these factors are strongly related to the reduction in the H_v value. The K_IC depends on the microstructure, especially grain size G_s of composites having the same composition and density. These factors might explain why the 60ZrC-C composite has the highest K_IC value among ZrC-C composites. Furthermore, in order to study the effect of graphite addition, the mechanical properties of binary ZrC/ZrB₂ composites were measured as a counterpart. Table 3 1) and 2) present their properties at room and high temperatures, respectively. As a whole, σ_b, H_v, and K_IC at R.T., i.e., the highest values of σ_b (474 MPa), H_v (18.5 GPa), and K_IC (4.45 MPa·m^1/2),

are not so high as compared with those for ZrC-C composites in Figure 5. Based on common sense about the mechanical properties of composites, as graphite has low σ_b, H_v, and K_IC, three-phase composites should have lower mechanical properties. However, the experimental results show the opposite. This could be explained by the differences in preparation routes, as follows:

1) As the binary composites (ZrC/ZrB₂) were prepared from the mixture of (Zr, B, ZrC), the binding force between ZrB₂ and ZrC grains might be weak, because a strong solid-state reaction could be introduced only between Zr and B, forming ZrB₂ grains.

2) On the contrary, in the three-phase composites (ZrC/ZrB₂/C), the binding force between ZrB₂ and ZrC through their interface/grain boundary should be strong because ZrB₂ grains were formed from ZrC by attacking B particles and separated from ZrC during PECPS. Nevertheless, there exist soft C grains among them.

Next, we address the main purpose of the present study: to evaluate high-temperature σ_b and toughness. Figure 6 shows σ_b of all ZrC-C composites as a function of measurement temperatures from 1273 to 2023 K. The σ_b values previously reported on the simple ZrC/ZrB₂ = 10/90 vol% composite [14] are also plotted as a reference. Overall, it is clear that the σ_b values at 1273 K of all ZrC-C composites are higher than those at ambient temperature, and at 1873 K some composites reveal the highest σ_b values in a series of compositions, i.e., the 30 and 60ZrC-C composites showed σ_b of 922 and 962 MPa, respectively. These values are much higher than that (301 MPa) reported previously [13], and our previous high-temperature σ_b value (803.5 MPa) attained for B₄C/ZrB₂ = 60/40 vol%

composites [9].

Among many ultra-high temperature ceramics (UHTC), such as boride, carbides, and nitrides, there is little information on three-point σ_b above 900 MPa at 1873 K. In addition, only a few reports [12] [14] describe the fabrication processes and mechanical properties of simple ZrC/ZB₂ composites at both room and elevated temperatures. However, no detailed study has been published up to now.

Many points of view have been taken to explain why 60ZrC-C composite shows such high s_b. After much consideration, it might be explained mainly in three ways. First, the high content of ZrC (60 vol% vs. ZrB₂), of which T_m (3533 K) [18] and Young’s modulus E (500 GPa) [18] are 220 K and 152 GPa higher than those (T_m: 3313 K, E: 348 GPa) of ZrB₂ [18], respectively; these two factors have much influence on strength. Second, the binding force between ZrC and ZrB₂ grains must be strong, as mentioned before. In addition, the dislocation could form at the interface between cubic ZrC and hexagonal ZrB₂ grains. Of course, the latter originates from the ZrC-B contact area. This dislocation, especially edge dislocation, might play an important role in high-temperature strength. Third, the 60ZrC-C composite consists of smaller grains, in addition to relatively high D_r (98.7%), as shown in Figure 7.

On the other hand, it is difficult to explain why the 30ZrC-C composite revealed a high σ_b (922 MPa) at 1873 K. One reason might be a large amount of graphite (as shown in Table 1) at the grain boundary between carbide and boride. This graphite can connect the grains of ZrC and ZrB₂ strongly, with the inverse temperature dependence of the strength of graphite C: the higher the temperature, the higher the strength. Another reason is their smaller G_s, irrespective of the lower D_r. Therefore, the best composition for the three-phase composites could be determined by the balance among the interface area of ZrC/ZrB₂, the

content of ZrC, and the amount of graphite C and its distribution.

Here, we evaluated the high-temperature σ_b of binary ZrC/ZrB₂ composites and compared them with those of three-phase ZrC/ZrB₂/C composites. The σ_b values, as shown in Table 3(b), are lower than expected. Of course, these simple composites revealed elastic deformation up to 1873 K and plastic deformation at 2023 K, the same as ZrC/ZrB₂/C composites, the latter of which will be described. These phenomena could support the additional effect of graphite superficially. At first, it was considered that if graphite is contained in the composites, their mechanical properties will be reduced by soft graphite. However, when we compared these mechanical properties at room and high temperatures, we found that graphite addition can bring good mechanical properties.

For a two-phase composite, the best composition is 40ZrC. This suggests that the mechanical properties might much depend on the ZrB₂ content; because 1) the difference in the thermal expansion coefficient a between ZrC (7.06 × 10⁻⁶ K⁻¹) [18] and ZrB₂ (6.98 × 10⁻⁶ K⁻¹) [18] is small; and 2) the binding force between base ZrB₂ grains could control the mechanical properties of composites, then the 40ZrC (60 vol% ZrB₂) composite revealed high σ_b at elevated temperatures. However, the reason why the σ_b values of both two- and three-phase composites tend to drop at 2023 K except for 70ZrC-C composite still remained.

At this stage, we paid much attention to the fracture modes, i.e., elastic and plastic deformations. The stress-strain (σ - ε) curves of all composites were measured at various temperatures. Figure 7 displays representative σ - ε curves measured at (a) 293, (b) 1873, and (c) 2023 K. In (a) and (b), all σ - εcurves exhibit elastic deformation. However, in (c) composites fractured after plastic deformation. From a practical viewpoint, applicable composites must be used in the elastic deformation field. Then it should be noted that the top σ_b values are attained within the elastic deformation field. Figure 7(d) focuses on the 60ZrC-C composite, which has the highest σ_b, its σ - εcurves are summarized as a parameter of measurement temperatures. It is clear that the fracture mode changes from elastic to plastic deformation between 1873 K and 2023 K. The microstructural changes in the elastically fractured surfaces of the composites of 30ZrC-C, 40ZrC-C, and 60vZrC-C at room temperature (297 K) and at high temperature (1873 K) were compared. As mentioned before, the upper and lower three SEM images in Figure 3 are the rupture cross-sections of these ZrC-C composites fractured at 297 K and 1873 K, respectively. We find little change in grain size, pore distribution, or dense texture. Furthermore, at the same temperature, i.e., along with the horizontal line, there is no drastic change in microstructure among them.

Next, by focusing on the 60ZrC-C composite, the microstructural change between elastically fractured at 1873 K and plastically fractured at 2023 K was precisely compared using SEM+EDS elemental mapping. Here we have to mention what we call “carbon contamination in SEM”, which is sometimes caused by the oil-vacuum pump system. However, as the present SEM + EDS system adopted a rotary + turbo molecular pump system, the vacuum space chamber is almost free from carbon contamination. This new concept is widely known and common knowledge to many researchers who treat this advanced SEM equipped with a “rotary + turbo molecular pump system” as a vacuum pump. The upper images in Figure 8 show various SEM photographs of 1873 K- fractured surfaces, such as secondary electron image SEM with an elemental concentration distribution, SEM + Zr, SEM + C, and SEM + B. The lower images show 2023 K-fractured surfaces. Although ZrC and ZrB₂ grains are discriminated by dense-colored areas, which indicate high concentrations of corresponding elements, it is very difficult to find the graphite C among the matrix. In Figure 8, by pointing to C, ZrC, and ZrB₂ grains, we tried to distinguish microstructures between 1873 K- and 2023 K-fractured surfaces, but there was little difference. Therefore, we might speculate that plastic deformation, followed by fracture, occurred along the hexagonal (001) slip plane of graphite or edge dislocation. And generally speaking, based on a sintering temperature of 2073 K, sintered materials might begin to deform plastically around 50 - 100 K below the sintering temperature, i.e., 2023 K - 1973 K. This phenomenon introduced by plastic deformation under 50 - 100 K below the sintering temperature is often used for densifying

poorly sinterable inorganic material by hot isostatic pressing (HIP). This is based on an empirical rule: at these temperatures, inorganic material will deform plastically into dense materials. If the present rule is also applicable to ZrC/ZrB₂-based composites, the plastic deformation of all composites could be explained.