For the preparation of ZrB₂-SiC composites, the raw materials included silicon carbide (SiC, 99.9% purity, D₅₀ = 1.35 µm), scandium oxide (Sc₂O₃, 99.9% purity, D₅₀ = 13.5 µm), yttrium oxide (Y₂O₃, 99.9% purity, D₅₀ = 5.70 µm) and lanthanum oxide (La₂O₃, 99.9% purity, D₅₀ = 4.64 µm). All materials were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. The particle sizes and morphologies of SiC, Sc₂O₃, Y₂O₃ and La₂O₃ are presented in Fig. S1 and Fig. S2. The preparation of ZrB₂ powders was based on the previous research [38].

The synthesis method of ZrB₂-SiC-based composites is illustrated in Step 1 of Fig. 1, primarily involving powder mixing and sintering process. In particular, an appropriate amount of ZrB₂, 20 vol% SiC and 5 vol% rare earth oxides powders were loaded into a ball milling jar. The 5 vol% rare earth oxide doping concentration was selected based on previous studies [39,40]. Using ZrO₂ balls and anhydrous ethanol as the grinding media, the mixture was ball‑milled at 300 rpm for 6 h. Subsequently, the mixtures underwent solvent removal at 55 °C for 10 h, and then the dried mixture was ground in an agate mortar for 30 min to obtain homogeneous mixed powders. The mixed powders were transferred to graphite mold and densified at 1800 °C for 2 h under 30 MPa. Upon completion, the load was removed and the furnace was cooled to ambient temperature. The bulk density was measured by the Archimedes drainage method, and the porosity was calculated based on the measured density [41].

The hot-pressed samples were cut into 4 mm × 4 mm × 3 mm blocks, subsequently polished with sandpaper, and then ultrasonically cleaned in ethanol. The isothermal static oxidation test was performed as shown in step 2 of Fig. 1, conducted in a high temperature tube furnace at 1773 K (Hefei Kejing Material Technology Co. Ltd) in air. The samples were taken out every 10 min to cool down and be weighed, then returned for further oxidation. The SEM cross-section method was employed to determine the oxide layer thickness, which involves observing the cross-section of oxidized samples to measure the thickness from the oxide layer to the matrix interface.

The ablation resistance was performed using a customized plasma platform (Multiplaz 3500 generator), as shown in step 3 of Fig. 1. The ablation samples had a diameter of 20 mm and a thickness is 5 mm. The movement of the sample during the ablation process was controlled by a mechanism consisting of a reciprocating motor and a connecting rod. Each full rotation of the motor drove the specimen to complete one back-and-forth horizontal movement, corresponding to one ablation cycle. The duration for which the specimen remained at the ablation end represented the single loading time. The plasma was operated at 6 A/160 V power parameters with H₂O. The pressure of the compressed air supplied at 0.4 MPa and the flow rate of 2 m³/h. Both the plasma and airflow nozzles had an inner diameter of 2 mm, positioned at a distance of 10 mm from the specimen surface, with the generated jet directed perpendicular to the specimen surface. Throughout ablation, the temperature at the center of the specimen surface was measured using two infrared pyrometers: the Endurance E3ML (323−1273 K, ± 0.3% accuracy, 20 ms response time) and the E1RH (1273−3473 K, ± 0.5% accuracy, 10 ms response time). The ablation flame was maintained at 2273 K and each cycle was 20 s, with 5 cycles repeated. The ablation rate was quantified by the change in central thickness and mass after ablation, yielding the linear ablation rate (LAR) and mass ablation rate (MAR), calculated via the equations below:

where l₀, l₁ and m₀, m₁ denote the initial and final thicknesses and masses, respectively, with t representing the total ablation time.

The phase analysis was performed via X-ray diffraction (XRD, X’Pert pro, PANalytical, Netherlands). The microstructure of ZrB₂-SiC composites was examined with a scanning electron microscope (SEM, SU8020, Hitachi, Japan) integrated with an energy dispersive spectroscopy (EDS), the operating voltage of EDS was 15 kV, with an acquisition time of ≥60 s and a dead time controlled within 20%. The particle distribution was measured using a laser particle size analyzer (Mastersizer 3000, Malvern, UK). Raman spectroscopy testing was conducted on a confocal Raman microscope (inVia Reflex, Renishaw, UK) employing a 532 nm laser source. Vickers hardness (Hv) measurements of the samples were carried out with a sclerometer (Wilson, VH3100, Switzerland). For each test, a minimum of five indentations were made under a 4.9 N load and held for 15 s.

Hot press sintering was employed to fabricate ZrB₂-SiC-based composites, with the preparation process illustrated in step 1 of Fig. 1, and the constituent volume fractions are provided in Fig. 2a. The relative density of the sintered blocks is presented in Fig. 2b, demonstrating that all samples achieve high densification. At a sintering temperature of 1800 °C, although no macroscopic liquid phase forms in the ZrB₂-SiC system, a liquid-solid coexistence may occur. This two-phase region plays a crucial role in both sintering densification and oxidation resistance. Following the addition of rare earth oxides, these oxides react with impurities on the SiC surface during sintering to form a eutectic liquid phase with a lower melting point. This liquid phase may also accumulate B₂O₃ present on the surface of ZrB₂ particles, which can significantly enhance particle fluidity and diffusion coefficients during sintering, thereby promoting material densification and improving the overall density of the composites [39]. Fig. 2c displays the hardness comparisons before and after doping with rare earth oxides, indicating that incorporation of rare earth oxides preserved the mechanical integrity of the composites. XRD analysis of the sintered samples is presented in Fig. 2d. The primary phases are ZrB₂ and SiC, with no diffraction peaks of other substances detected. The attenuated SiC peak is attributed to the dominant ZrB₂ signal. In addition, the absence of rare earth oxides characteristic peaks may be due to the low doping concentration. The microstructure of the polished surface and cross-sections of the four samples are presented in Fig. 3. All specimens possess a high-density microstructure with no noticeable defects. The fracture surfaces of the composites reveal that the fracture mode is a mixture of intergranular and transgranular fractures. SEM and EDS mapping results indicate that SiC exhibits a relatively uniform distribution in the matrix (Fig. S3).

The oxidation behavior of ZS20, ZS20S5, ZS20Y5, and ZS20L5 samples was investigated using isothermal oxidation at 1773 K. The isothermal static oxidation test is performed as shown in step 2 of Fig. 1. The dependence of weight gain on oxidation time is shown in Fig. 4a, b. As the oxidation time progressed, all four composites exhibited a continuous weight increase. In the initial 140 min, ZS20 exhibited comparatively slow oxidation weight gain, lower than that of ZS20S5, ZS20Y, and ZS20L5. The oxidation-induced weight gain of the four samples was analyzed at different temperatures (Fig. S4), and the activation energy for their initial oxidation stage was calculated using the Arrhenius equation [42]. The incorporation of rare earth oxides reduced the oxidation activation energy of the composites (Fig. S5), leading to accelerated oxidation kinetics for the doped specimens relative to ZS20 during the initial oxidation stage. The oxidation weight gain of ZS20 increased rapidly and soon exceeded that of the composites containing rare earth oxides after 140 min, indicating a significant decline in the oxidation resistance of ZS20 during prolonged high temperature exposure. In contrast, ZS20S5, ZS20Y5, and ZS20L5 samples maintained relatively stable oxidation weight gain, demonstrating excellent long-term oxidation resistance, with ZS20L5 exhibiting the optimal performance. This was further confirmed by the fitted plot of (ΔW/A)² versus oxidation time presented in Fig. S6.

Fig. 4c shows the oxidation kinetics of the four composites [43]. For the ZS20 sample, there was a linear relationship between the oxidation degree (α) and oxidation time at the initial stage. However, the oxidation degree increased rapidly in the later stages, indicating the failure of the oxidation protection layer. In contrast, the samples doped with rare earth oxides exhibited a linear relationship between α and oxidation time throughout the oxidation process, indicating that a relatively stable oxide barrier capable of effectively restricting oxygen ingress, resulting in a continuous oxidation at a constant rate. The oxidation rate constant (k) of ZS20S5, ZS20Y5, and ZS20L5 samples was calculated using the Jander equation [44], with the k value of ZS20L5 being 0.0573, indicating its superior oxidation resistance. Fig. 4d presents the porosities of the four composites after oxidation, revealing that the samples containing rare earth oxides were significantly less porous ZS20. Specifically, the porosity of ZS20L5 after oxidation was only 1.8%, further confirming that La₂O₃ addition effectively improves the oxidation resistance of ZrB₂-SiC composite. This improvement is likely due to the better oxygen-blocking properties of the protective layer formed after the incorporation of La₂O₃. Fig. 4e demonstrates the XRD results of the oxidized composites. The main phases of ZS20 after oxidation are ZrSiO₄ and m-ZrO₂, while those of ZS20S5 are m-ZrO₂, t-ZrO₂, and Sc₂Si₂O₇, suggesting that Sc₂O₃ doping effectively suppresses the transformation of t-ZrO₂ to m-ZrO₂. For ZS20Y5, the primary oxidized phases are m-ZrO₂ and Y₂Si₂O₇, while the primary phases of ZS20L5 after oxidation are m-ZrO₂ and La₂Si₂O₇, with no detectable characteristic peaks of ZrSiO₄. The XRD patterns after oxidation clearly show the formation of Sc₂Si₂O₇, Y₂Si₂O₇, and La₂Si₂O₇. This can be attributed to the reaction between rare earth oxides and SiO₂ at high temperatures during the oxidation process, resulting in the synthesis of silicate phases with good crystallinity. The generated Re₂Si₂O₇ continuously accumulated in the oxide layer, exceeding the XRD detection limit. Moreover, ZrO₂ formed by ZrB₂ oxidation exhibits relatively poor crystallinity, resulting in peak intensity insufficient to mask the Re₂Si₂O₇ signal. The thermodynamics of the reactions involved in high temperature oxidation are shown in Fig. 4f. In the case of the ZS20 sample, ZrO₂ generated during high temperature oxidation reacts with SiO₂ to form ZrSiO₄, which is also thermodynamically favorable. For the ZS20S5, ZS20Y5 and ZS20L5 samples, the formation of rare earth silicates is thermodynamically favored, as evidenced by the more negative Gibbs free energy (ΔG) values for the reactions between rare earth oxides and SiO₂ at elevated temperatures. This thermodynamic preference aligns with the phase identification results from XRD (Fig. 4e).

To further reveal the mechanism by which the addition of rare earth oxides enhances oxidation resistance, density functional theory (DFT) was performed to calculate the oxygen adsorption energy at metal sites on the (001) surfaces of ZrSiO₄, Sc₂Si₂O₇, Y₂Si₂O₇, and La₂Si₂O₇. As shown in Fig. 4g, e, ZrSiO₄ exhibits the lowest oxygen adsorption energy, indicating a higher propensity for oxygen adsorption. Subsequently, Sc₂Si₂O₇, La₂Si₂O₇, and Y₂Si₂O₇ exhibit progressively higher adsorption energies, suggesting that the rare earth silicates formed after the addition rare earth oxide exhibit a stronger resistance to oxygen adsorption. Consequently, during the initial oxidation stage, the formation of ZrO₂ and SiO₂ is limited due to low oxidation degrees, resulting in minimal generation of ZrSiO₄. As oxidation time prolongs, the amount of ZrSiO₄ generated increases steadily, correlating with a marked deterioration in the oxidation resistance. In contrast, the samples doped with rare earth oxides exhibit sustained oxidation resistance because the generated rare earth silicates provide superior oxygen resistance.

The surface microstructure and elemental distribution of oxidized ZS20 are shown in Fig. 5a-d. After oxidation, ZS20 was converted into ZrO₂, B₂O₃, and SiO₂, which subsequently reacted at high temperatures to form a protective layer composed of ZrSiO₄ and borosilicate glass phases. The oxidized surface of ZS20 is distributed inhomogeneous protuberances and holes, which may be attributed to the volume expansion of oxides and the escape of the low melting point B₂O₃ and CO gases [45,46]. The oxidized surface of ZS20S5 is relatively flat compared with ZS20 (Fig. 5e-g), and the results of EDS mapping and point scanning reveal that the primary component of the glassy phase is SiO₂. The granular phases correspond to ZrO₂, while the surface is coated with B₂O₃. Sc³⁺ is mainly distributed within the ZrO₂ aggregates, and a portion of Sc³⁺ may have dissolved into the ZrO₂ lattice at high temperatures, which induces the formation of t-ZrO₂. This can also be explained by the fact that the oxidized ZS20S5 sample contains a higher content of t-ZrO₂, and the content of t-ZrO₂ is regulated by the solid solution of rare earth elements in the ZrO₂ lattice [47,48]. The volumetric changes during the phase transition result in the formation of protrusions. Furthermore, Sc₂O₃ can react with SiO₂ to form Sc₂Si₂O₇ at high temperatures, and the distribution of Sc₂Si₂O₇ in B₂O₃ can effectively increase the glass viscosity, thereby reducing the B₂O₃ volatilization [49]. As presented in Fig. 5h-j, the oxidized surface of ZS20Y5 exhibits a uniform distribution of ZrO₂ particles, accompanied by discontinuous borosilicate glass phases, with no obvious glassy protective layer observed. This suggests that Y₂O₃ doping may decrease the fluidity of the glass phase, thereby preventing the formation of a continuous protective film, as corroborated by Fig. 5i, j. As presented in Fig. 5k-m, ZS20L5 exhibits a flatter and denser oxidized surface compared with other samples, with the glass phase primarily composed of B-Si-O-La system. The area uncovered by the glass phase is mainly ZrO₂ particles, and the borosilicate glass phase is uniformly filled between the ZrO₂ particles, which effectively blocks the diffusion of oxygen. This also explains why Y₂Si₂O₇ has the highest oxygen adsorption energy compared to La₂Si₂O₇, the oxidation resistance of ZS20Y5 is inferior to that of ZS20L5. Oxidation cross-sections in Fig. S7 demonstrate that incorporating rare earth oxides decreases the oxidation layer depth compared to ZS20. Among these samples, ZS20L5 exhibits the densest oxide layer and the shallowest oxidation depth. In summary, the addition of La₂O₃ results in the optimal comprehensive effect of the glass phase, which is most conducive to improving ZrB₂-SiC oxidation resistance.

To further evaluate the service performance of the composites under operating conditions, cyclic ablation testing was performed via air plasma flame. The setup of the ablation experiment is shown in Fig. 6a, where the temperature was set at 2273 K, the duration of a single ablation cycle was 20 s, and the experiment was repeated for 5 cycles. Fig. 6b presents the macro images of the four composites after ablation. Specifically, ZS20 cracked after 4 ablation cycles, ZS20Y5 cracked after 5 ablation cycles, while ZS20S5 and ZS20L5 remained intact after the experiment. This indicates that rare earth oxide additions improve the thermal stability of the composites under cyclic ablation conditions. Fig. 6c shows the surface temperature variations of the composites during ablation test, with rapid heating within the first 10 s. Notably, the ZS20, ZS20S5, ZS20Y5, and ZS20L5 samples reached peak temperatures of 1302 K, 1345 K, 1388 K, and 1335 K, respectively. This indicates the possible existence of an effective thermal barrier effect. During ablation, the oxide layer formed on the material surface (such as porous ZrO₂ and silicate glass) typically exhibits a high spectral emissivity (ε) at elevated temperatures. This enables the samples to efficiently re-radiate the heat absorbed via convection and conduction to the surrounding environment. Furthermore, the relatively short duration of single ablation cycle also contributes to the low surface temperature of the composite. The slight increase in the surface temperature of the composites may be attributed to rare earth oxide doping, which inhibits the volatilization of B₂O₃ and SiO₂ [50], thereby reducing heat loss through volatilization and resulting in a slight surface temperature increase.

The XRD analysis (Fig. 6d) reveals that the composite surfaces are primarily consist of ZrB₂ and m-ZrO₂, with trace amounts of SiO₂. As illustrated in Fig. 6e, the incorporation of rare earth oxides leads to a notable reduction in the linear ablation rate. Fig. 6f shows that the mass ablation rate of sample ZS20L5 is close to zero, suggesting minimal mass change. During ablation, the material undergoes oxidation, volatilization, and mechanical erosion concurrently. While oxidation tends to increase both mass and thickness, volatilization and erosion contribute to their reduction. The mass ablation rate of ZS20 composite is closer to 0 compared with ZS20S5 and ZS20Y5, which may be due to the rapid volatilization or erosion of the oxides generated during the ablation, as supported by the higher linear ablation rate of ZS20 (Fig. 6e). The reason for the difference in ablation rate is that the surface of ZS20 formed an oxide layer mainly composed of ZrO₂ and low-viscosity borosilicate glass phase after ablation. The porous structure exhibits low strength and weak adhesion, making it highly susceptible to erosion under the shear action of high-speed airflow. Consequently, the material undergoes rapid recession in the thickness direction, leading to an extremely high linear ablation rate. However, the relatively low mass ablation rate does not indicate the absence of material loss, but rather the offset between intense oxidation weight gain and severe weight loss from volatilization/erosion. The addition of Y₂O₃ promotes the formation of Y₂Si₂O₇ and a glass phase with higher viscosity, which exhibits stronger anchoring to the ZrO₂ framework, resulting in significantly enhanced resistance to mechanical erosion by gas flow compared to ZS20. Consequently, as the oxidation proceeds, most oxidation products are retained on the sample surface, resulting in a more negative mass ablation rate. By comparing the linear ablation and mass ablation rates, ZS20L5 demonstrates superior ablation resistance. The reason why both the linear ablation rate and mass ablation rate of ZS20L5 are close to 0 lies in the possible formation of a glass phase with superior fluidity and oxygen barrier performance, which results in smaller oxidation-induced weight gain and mass change caused by gas flow erosion. Fig. S8 compares the ablation rates of ZS20 and ZS20L5 composites with other typical UHTCs reported in the literature. It can be seen that the ablation resistance of ZS20L5 composites is comparable to that of most UHTCs under the same similar conditions. Fig. 6g shows the Raman spectra of the ablation center region, where the samples containing rare earth oxides display more intense Raman peaks corresponding to B₂O₃ and SiO₂ [51,52]. This implies that the addition of rare earth oxides effectively inhibits the glass phase volatilization during ablation, thereby improving the ablation resistance.

The micrographs and elemental maps of ZS20 ablation center are displayed in Fig. 7a-d. The microstructure reveals a porous surface characterized by a ZrO₂-wrapped borosilicate glass skeleton, abundant with holes and bubbles. This morphology primarily results from the high-temperature liberation of low melting point B₂O₃ and gaseous CO/CO₂ products from the oxide layer. Fig. 7e-h shows the microstructure of ZS20S5 ablation center reveals a significant amount of plate like glass phases, with the main component being the B-Si-O-Sc system. The areas uncovered by the glassy phase are primarily aggregated ZrO₂ particles, with some cracks visible on the surface. Fig. 7i-l shows the microstructure of ZS20Y5 ablation center. Similar to ZS20S5, the ablation center of ZS20Y5 also contains a significant amount of plate like glass phases, whose main component is the B-Si-O-Y system. However, the ablated surface of ZS20Y5 exhibits more cracks, and no obvious glass phase is observed between the cracks. Fig. 7m-p shows the microstructure of ZS20L5 ablation center. Only a small amount of plate like glass phase exists in the ablation center of ZS20L5, with B₂O₃ as the main component. The ablated surface of ZS20L5 has few cracks, and these cracks are well filled with glass phase. In summary, the glass phase formed at high temperatures of samples doped with Sc₂O₃ and Y₂O₃ exhibit inferior fluidity, which promotes the rapid formation of the glass phase during the ablation, leading to the crack formation due to the volume shrinkage of ZrO₂. The absence of glass phase filling allows the cracks become the diffusion channel for oxygen. The glass phase formed in the La₂O₃-doped sample exhibits superior fluidity at high temperatures, which can effectively fill the cracks, seal pores, improve the density of the oxide film, and block the diffusion of oxygen [53].

Fig. 8 demonstrates the schematic diagram of the ablation mechanism across four specimens. At high temperature aerobic environment, the composites undergo oxidization to form B₂O₃, SiO₂, and ZrO₂. However, the holes left by the volatilization of B₂O₃ and gases, along with the consumption of SiO₂, result in a degradation of the ablation performance. The ZS20S5 and ZS20Y5 exhibit better ablation resistance compared to ZS20, which is attributed to the formation of plate like glass phase on the surface that provides a more effective oxygen blocking effect. Nevertheless, the cracks present on the surface still act as diffusion channels for oxygen. In contrast, ZS20L5 exhibits superior ablation performance due to the enhanced flowability of its high-temperature glass phase, which forms a dense, stable and continuous protective barrier, effectively inhibiting oxygen diffusion.