Microstructures, mechanical properties and oxidation resistance of SiBCN ceramics with the addition of MgO, ZrO₂ and SiO₂ (MZS) as sintering additives

Abstract

Nano-crystalline SiBCN ceramics were prepared by mechanical alloying (MA) plus hot pressing (HP) with the addition of MZS1 (MgO, ZrO₂ and SiO₂) and MZS2 (ZrSiO₄ and SiO₂) as the sintering additives. The effects of the two additives on the microstructure, mechanical properties and oxidation resistance of SiBCN ceramics were carefully evaluated. The addition of MZS1 and MZS2 additives in the SiBCN matrix can boost elemental diffusion and matrix densification. Owing to the effective densification caused by the sintering additives, the mechanical properties of SM1 and SM2 samples are far superior to monolithic SiBCN. The oxide layers of SM1 and SM2 samples remain relatively dense and continuous, and have a strong binding capacity with the matrix after oxidation testing at 1500 °C for 20 h. The residual excess carbon in the SiBCN matrix should be responsible for the formation of ZrC and the outermost oxide layer structure is comprised of SiO₂ and ZrC after the oxidation test. The flexural strength, Young’s modulus, fracture toughness and Vicker’s hardness of the SM2 sample are much higher than monolithic SiBCN ceramics, reaching to 394.2 ± 41.7 MPa, 152.9 ± 16.0 GPa, 5.86 ± 0.86 MPa m^1/2 and 8.3 ± 0.6 GPa, respectively.

---

1. Introduction

SiBCN ceramics are of great interest due to their combination of special structures and outstanding thermal stability up to almost 2000 °C, hence they have attracted great attention among researchers.¹ These ceramics normally have an amorphous matrix at 1100–1400 °C,² and will change to nano-ceramics with increasing temperature in an inert atmosphere.³ Recently, the methods used for fabricating SiBCN ceramics have mainly included a precursor-derived pyrolyzing route,^4,5 a mechanical alloying plus hot pressing method,^6,7 physical vapor deposition (PVD, including DC magnetron co-sputtering and RF sputtering), thermal plasma chemical vapor deposition (TPCVD) and ion implantation.^8,9 At present there is no unified opinion about the microstructure of SiBCN ceramics, but it is accepted that the polymer-derived amorphous Si–B–C–N ceramics generally consist of two phases, that is, amorphous SiC_xN_4−x (x = 1–4) and graphite-like BN(C), 1–2 nm in size.^10–12 The mechanical alloying plus hot pressing methods were studied in recent years for fabricating dense bulk SiBCN ceramics with large dimensions.¹³ Yang¹⁴ prepared amorphous SiBCN powder by mechanical alloying and subsequently applied a hot pressing technique to densify SiBCN ceramics with nano-SiC and BN(C) grains uniformly distributed in the amorphous matrix. The as-obtained ceramics prepared by mechanical alloying plus hot pressing have a similar microstructure to polymer derived SiBCN ceramics. This method has the advantage of preparing bulk ceramics with larger scale, and hence facilitating the evaluation of various mechanical properties. So far, the preparation process, microstructures and mechanical properties of SiBCN ceramics have been investigated most, whereas less is known about their sintering ability. The poor sintering ability of SiBCN ceramics has limited their application in the aerospace and aviation industries.

Some studies^15–18 have been carried out to investigate the effect of different additives on the densification behavior of ceramic based composites. The results showed that special sintering additives, such as B + C, B₄C, Al₂O₃ + Y₂O₃ and MgO + Al₂O₃ + SiO₂ (MAS) promoted the sintering process effectively. In our previous research,¹⁹ the addition of 5% mol ZrO₂ or AlN in SiBCN ceramics could contribute to reducing the volume shrinkage temperature to ∼1720 °C and 1760 °C, respectively. Meanwhile, the relative density of SiBCN ceramics incorporating ZrO₂ or AlN additives increased to 97.7% and 96.9%, respectively, while the flexural strength reached to 575.4 MPa and 415.7 MPa, respectively. The introduction of sintering additives in the SiBCN matrix gave a better performance than monolithic SiBCN ceramics, especially in mechanical properties and thermal stability. However, the addition of ZrO₂ or AlN damaged the oxidation resistance of SiBCN ceramics greatly because of the porous oxide layer covering the oxide surface, which did not retard oxygen diffusion into the inner matrix and healed surface pores and cracks ineffectively. For this reason, other sintering additives should be explored to improve the sintering ability and oxidation resistance of SiBCN ceramics. Oxide additives are commonly applied to ceramic based composites for their outstanding sintering ability at high temperature.²⁰ To prepare nano-sized ceramics with smaller grains and to enhance the mechanical properties as well as the oxidation resistance of SiBCN ceramics, we choose MgO–ZrO₂–SiO₂ (MZS1) and ZrSiO₄–SiO₂ (MZS2) as sintering additives for comparison. Therefore, it is necessary to understand the effect of sintering additives on the densification behavior and oxidation resistance of SiBCN ceramics. In the current work, the phase transition, microstructures, mechanical properties and oxidation resistance were evaluated carefully. The results would be helpful for further research and applications of SiBCN ceramics.

2. Experimental procedures

2.1 Raw materials and sample preparation

In the current work, the raw materials used were commercially available cubic silicon (45.0 μm, 99.5% in purity, Beijing Mountain Technical Development Center, China), hexagonal boron nitride (0.6 μm, 98.0% in purity, Advanced Technology & Materials Co. Ltd., Beijing, China), and graphite powder (8.7 μm, 99.5% in purity, QingDao HuaTai Lubricant Sealing S&T Co. Ltd., China). The chemical composition was designed as Si : C : BN = 2 : 3 : 1 in molar ratio. The powder mixture was milled for 40 h under an argon atmosphere by a planetary ball mill (P4, Fritsch GmbH, Germany) with ball to powder mass ratio set as 20 : 1. The rotation speed of the main disk was set at 350 rpm and the vials at 600 rpm in reverse. The additives of MgO (0.3 μm, 98.0% in purity, WuXi ZeHui Chemical Co. Ltd., China), ZrO₂ (50.0 μm, 99.0% in purity, Shenzhen Crystal Materials Co. Ltd., China), SiO₂ (38.0 μm, 99.8% in purity, LianYunGang HuaWei Silica Powder Co. Ltd., China) were in accordance with the molar ratio of 2 : 5 : 2 and underwent ordinary milling for 24 h with ethanol solvent, then dried in a drying box at 100 °C. The as-prepared powders were designated as MZS1. Subsequently, the MZS1 powders were sintered at 1450 °C to obtain bulk ceramics but were crushed to obtain another sintering additives powder, labeled as MZS2. In order to obtain a uniform distribution mixture, the amorphous SiBCN powder and sintering additives were mixed according to the mass ratio of 9 : 1 for 5 h by high energy ball milling. The as-milled powders were loaded into graphite dies and were heated up to 1900 °C with a heating rate of 20 °C min⁻¹, then kept at the target temperatures and under constant uniaxial pressure (80 MPa) in a nitrogen atmosphere (1 bar) for 30 min. The loading process of the uniaxial pressure began at 1200 °C and ended at 1400 °C. In the cooling stage, furnace cooling was adopted and the uniaxial pressure was slowly unloaded in the following 5 min. The bulk ceramics prepared by sintering the SiBCN amorphous powder and the MZS1 additive were denoted as SM1, while SM2 represents SiBCN ceramics obtained by sintering the SiBCN amorphous powder and the MZS2 additive.

After sintering, the hot pressed SiBCN ceramics were diamond-ground to remove BN surface contamination. The resulting sample ceramics were cut into bars of 36 mm × 3 mm × 4 mm (30 mm outer span) for measuring flexural strength and Young’s modulus at room temperature by a three point bending test with a crosshead speed of 0.5 mm min⁻¹. The fracture toughness was determined using the single edge notched beam (SENB) method with a crosshead speed of 0.05 mm min⁻¹. The testing bar dimensions used were 2 mm × 4 mm × 20 mm (16 mm outer span). The depth of the notches was 2.0 mm and the width about 0.2 mm (Instron 5569, Instron Corp., USA). The Vicker’s hardness (HVS-5, Laizhou, Huayin, Testing Instrument Corp., USA) was measured on a polished sample surface with a load of 10 kg and a loading time of 10 s. For the oxidation test, the polished samples were cut into dimensions of 3 mm × 4 mm × 10 mm and the two major faces with 4 mm × 10 mm were ground parallel and polished with diamond lapping film. The surface roughness is about 0.5 μm. In order to avoid the influence of weight change caused by surface moisture, samples were placed in an oven preset at 120 °C for 10 h. Oxidation experiments were carried out in flowing air (N₂ 79%, O₂ 21%, steam content < 50 ppm) at 1100 °C, 1300 °C and 1500 °C, respectively. SM1 and SM2 samples were put into a slotted Al₂O₃ refractory brick and then placed into a tube furnace (RHTH120/600/18, Nabertherm, Germany) at the target temperatures. The heating rate was 5 °C min⁻¹ and the cooling rate was not higher than 7 °C min⁻¹. The oxidation time was set as 5 h, 10 h and 20 h at the target temperatures.

2.2 Microstructures and phase transition analysis

The phases and morphologies of the as-milled SiBCN powders, the additive powders and their mixed powders were studied using a scanning electron microscope (SEM, 30 kV, Quanta 200 FEG, FEI Co., USA), and X-ray diffraction spectroscopy (40 kV/100 mA, D/max-γB CuKα, Rigaku Corp., Japan). The structural characterization of SiBCN ceramics was carried out using the X-ray diffraction (XRD) method to obtain the X-ray diffraction spectra at 2θ = 10–90° with a scanning speed of 4° min⁻¹ and a transmission electron microscope (TEM, Tecnai. F30, 300 kV, FEI Company., USA) operating at 120 kV was employed to investigate the ceramics microstructures. Elemental composition analysis was performed using an energy dispersive spectroscopy system (EDS, Oxford instruments INCAx-act, Oxforshire, U.K.) equipped on the SEM.

3. Results and discussion

3.1 Densification, microstructures and mechanical properties

Fig. 1 shows the XRD patterns of the SiBCN powder, the sintering additives (MZS1 and MZS2), and their mixed powders. After the SiBCN powder was milled for 40 h, the corresponding diffraction peaks of BN and C completely disappeared, the intensity of the Si peak reduced greatly and crystalline SiC was found, as indicated in Fig. 1(a). The raw materials of MgO, ZrO₂, SiO₂ powders were milled for 24 h with ethanol solvent, the diffraction peaks of the MZS1 powder correspond to the MgO, ZrO₂ and SiO₂ phases. The main peak intensity of ZrO₂ is higher than the others, which may result from the high content of ZrO₂ in the raw powders, as presented in Fig. 1(b). The MZS1 additive prepared by low energy ball milling has not changed the original phases. From the diffraction peaks of MZS2 in Fig. 1(c), the crystallization peaks are quite obvious, mainly corresponding to SiO₂ and ZrSiO₄.

Fig. 1 XRD patterns of the raw materials. (a) SiBCN powder, (b) MZS1 powder, (c) MZS2 powder, (d) MZS1 + SiBCN powder, (e) MZS2 + SiBCN powder.

The as-milled SiBCN amorphous powder and sintering additives (MZS1 and MZS2) were then mixed according to the mass ratio of 9 : 1 for 5 h by high energy ball milling, the corresponding XRD patterns of the mixtures are shown in Fig. 1(d) and (e). The result clearly shows that the mixed powers had almost non-crystalline diffraction peaks of MZS1 powder, as indicated in Fig. 1(d). Possible reasons for this are the low content of sintering aids in the mixture and the process of mechanical alloying (MA) adopted to prepare the amorphous powder. But Fig. 1(e) shows that a ZrSiO₄ phase still exists in the MZS2 + SiBCN powder after high energy ball milling for 5 h. The reason for this can be attributed to the short duration ball milling process. Mechanical alloying (MA) is an appropriate method for the preparation of various non-equilibrium materials, such as amorphous body, quasicrystal or supersaturated solid solutions.⁶ Ball milling time is an important parameter during the MA process. If the ball milling time is long enough, a completely amorphous SiBCN + MZS2 powder could be obtained. The XRD patterns of SM1 and SM2 bulk samples produced by hot pressing at 1900 °C under a pressure of 80 MPa for 30 min in N₂ (1 bar) are displayed in Fig. 2. After hot pressing sintering, obviously crystallized peaks are observed, and the main phases are SiC and BNC. With the low content of sintering aids, their crystallization peaks are not obvious in comparison with the ceramics matrix.

Fig. 2 XRD patterns of SM1 and SM2 bulk samples produced by hot-pressing at 1900 °C/80 MPa. (a) SM1 sample, (b) SM2 sample.

Fig. 3 shows the typical SEM images of pure SiBCN ceramics. Residual pores can be clearly observed in each SEM photograph, and small cracks also exist in the sample surface. The surface morphologies in Fig. 4(a) and (c) show that the pores in as-prepared SiBCN ceramics are significantly reduced, along with the residual pores becoming smaller and isolated. The ceramic particles seem to be flattened, indicating the elimination of the pores and the densification of the SiBCN ceramics with the help of additives. It is generally accepted that a high temperature and high pressure sintering process may accelerate plastic deformation, viscous flow and elemental diffusion which facilitate ceramic densification. The addition of the two sintering aids can effectively lower the porosity in SiBCN ceramics and contribute to the densification of ceramics because of liquid phase sintering mechanisms.^20–23 These results reveal that only a small amount of MZS1 or MZS2 additive could promote the sintering ability of SiBCN powder effectively, resulting from the reduction of the sintering temperature and the increased density of the bulk materials. However, a few residual pores are still remaining in the SM2 sample and this seems to indicate a better effect on densification by the MZS1 additive than the MZS2 additive, as shown in Fig. 4(d). According to the work done by F. K. van Dijen,²⁴ SiO₂, Al₂O₃ and C additives inhibit SiC grain formation and growth effectively at an appropriate proportion. However, when C was used, the second phase was free of silica, crystalline, and less weight loss during sintering was observed. Besides, the second phase is present at the triple points and no grain boundary or glassy phases are detected. R. M. German²⁵ established a model for supersolidus liquid-phase sintering of prealloyed powders in composites. The model used percolation concepts to create a rheological response that matched well with observations of rapid densification over narrow temperature or time ranges. This model could be used for a range of alloys and their oxides, which would be helpful for us to understand the sintering behavior of composites. M. F. Percio²⁶ investigated the effect of the addition of TiO₂, ZrO₂, V₂O₅ and Nb₂O₅ on the stability parameters of multi component glass. They found that the addition of any one of the oxides used as nucleants hindered the onset of crystallization and the material remained amorphous at slightly more intense heat treatments. Furthermore, the presence of ZrO₂ caused the glass to become more resistant to devitrification. A. Can²⁰ evaluated the densification and phase formation of liquid phase sintered silicon carbide (LPSSiC) with 10 wt% additives by conventional hot pressing. They showed that the densification process was greatly promoted by the addition of a small amount of aids due to liquid phase sintering. The mechanical properties of all the investigated SiBCN ceramics are shown in Fig. 5. As indicated, the SM1 specimen has higher flexural strength than SM2 because of its higher relative density. Both the SM1 and SM2 samples have better performance than monolithic SiBCN ceramics, which indicate the good effect of sintering additives on the densification behavior in the SiBCN matrix. The flexural strength and Young’s modulus were influenced greatly by the existence of porosity in the ceramic matrix. The fracture toughness and Vicker’s hardness of SM2 were 2 times and 3 times larger than pure SiBCN ceramics, respectively. The flexural strength, Young’s modulus and density of SM2 are also higher than pure SiBCN ceramics, reaching 394.2 ± 41.7 MPa, 152.9 ± 16.0 GPa and 2.69 ± 0.1 g cm⁻³, respectively. The mechanical properties of SiBCN ceramics with different additives are summarized in Table 1.

Fig. 3 SEM surface and fracture images of pure SiBCN ceramics hot-pressed at 1900 °C/80 MPa. (a) Surface morphology, (b) fracture morphology.

Fig. 4 SEM surface and fracture images of SM1 and SM2 ceramics hot-pressed at 1900 °C/80 MPa. (a) and (b) SM1 sample, (c) and (d) SM2 sample.

Fig. 5 Mechanical properties of all the investigated SiBCN ceramics.

Table 1 The effect of different additives on the densities and mechanical properties of SiBCN ceramics

Sample	Flexural strength/MPa	Young’s modulus/GPa	Fracture toughness/MPa m^1/2	Vicker’s hardness/GPa	Density/g cm^−3
Pure SiBCN	331.1 ± 40.5	139.4	2.80 ± 0.90	2.7 ± 0.4	2.60 ± 0.01
SM1	470.4 ± 71.1	134.9	5.10 ± 0.62	7.3 ± 0.5	2.78 ± 0.01
SM2	394.2 ± 41.7	152.9	5.86 ± 0.86	8.3 ± 0.6	2.69 ± 0.01

---

Fig. 6 and 7 present the TEM microstructures of the SM1 and SM2 ceramics prepared at 1900 °C/80 MPa/N₂. Nano-sized BNC and SiC grains have a relatively uniform distribution in the matrix without abnormal growth. The SiC grain sizes in SM1 and SM2 vary from 100 nm to 400 nm, while the grain sizes of the BNC phase are much smaller than SiC and it is difficult to show BNC nano-crystals by TEM due to their small size. Additionally, lots of light-and-dark strips can be found in many SiC grains, implying the existence of numerous stacking faults, as indicated in Fig. 7(b). The BNC phase is mainly distributed between SiC grains or in their junction regions with no fixed shape. There is no Si₃N₄ appearing in SM1 and SM2 ceramics, which is a little different from organic polymer derived SiBCN ceramics. This may be related to the designed chemical composition, ball milling process and applied sintering techniques, requiring more research to identify. Although the microstructures of the SiC and BNC phases and their phase interface require further characterization, it is commonly accepted that the prepared SiBCN ceramics are composed of β-SiC, graphite like BNC and a small amount of α-SiC with a grain size of 100 nm or so.¹⁴ The ZrO₂ in the MZS1 powder could be observed in the SiBCN matrix by TEM in Fig. 6, which matched with the XRD results. The selected area electron diffraction (SAD) patterns from areas A and B exhibit [01̄1̄] and [121] indicative of β-SiC and α-SiC structures. Other phases such as ZrC and ZrO₂ are also observed by TEM in Fig. 6(e) and (f). However, we could not find the existence of MgO and SiO₂ due to their relatively low content in the MZS1 additive. Fig. 7(b) shows the detailed features of BNC, including its size, morphology and distribution. The ZrSiO₄ and SiO₂ phases in the MZS2 additive are hardly seen in SM2 ceramics by TEM, as shown in Fig. 7. According to the literature,²⁷ ZrSiO₄ begins to decompose into ZrO₂ and SiO₂ in the range of 1450–1600 °C depending on the purity and particle size of the additive. At temperatures higher than 1650 °C the reaction may accelerate greatly. ZrSiO₄ in the SM2 sample may decompose at such a high sintering temperature, which matches the XRD result. Besides, the ZrO₂ may react with the excess carbon to form ZrC. That’s why we can find the presence of ZrC in the SM1 and SM2 samples by TEM. In Ya. V. Lipatov’s research,²⁸ the use of ZrSiO₄ does not cause phase crystallization during fiber melt spinning, and the fiber chemical composition is almost identical to the matrix. Therefore, the decomposition of ZrSiO₄ into ZrO₂ and SiO₂ will benefit the phase formation instead of ZrSiO₄ itself. All the factors shown above state clearly that the addition of MZS1 and MZS2 additives in SiBCN ceramics could contribute to matrix densification because of liquid phase sintering mechanisms.

Fig. 6 Structural features of the as-prepared SM1 ceramics hot-pressed at 1900 °C/80 MPa. (a) Bright field image showing the morphologies of SM1, (c)–(f) SAD patterns of corresponding regions A–D in (b).

Fig. 7 Structural features of the as-prepared SM2 ceramics hot-pressed at 1900 °C/80 MPa. (a) Bright field image showing the morphologies of SM2 sample, (b) SiC and BNC phases, (e) SAD pattern of region A in (c) and (f) SAD pattern of region B in (d).

3.2 Oxidation resistance

In order to study the oxidation behavior of the SM1 and SM2 ceramics in flowing air, SM1 and SM2 samples were put into an air furnace at different target temperatures (1100 °C, 1300 °C and 1500 °C) for 5 h, 10 h and 15 h. The phase transition, microstructures, and morphologies of all investigated samples were carefully studied. From the macro morphologies of SM1 and SM2 samples after the oxidation test at 1100 °C for 10 h (not shown), it can be clearly seen that most regions of the samples keep their integrity and no visible macro-cracks are found.

In Fig. 8, the images show the surface morphologies of SiBCN ceramics after oxidation tests at 1100 °C with different oxidation times (5 h, 10 h and 20 h). The surface morphology in Fig. 8(a) shows a few cracks staggered with each other. When the oxidation temperature increases to 1100 °C for 5 h, the SM1 ceramic has a thin oxide layer and only a few micro cracks propagate in local areas. This indicates that the oxidation phenomenon of SM1 is not obvious at low temperature for 5 h. With the oxidation time increasing from 10 h to 20 h, SM1 ceramic samples suffer a more severe oxidation treatment, and may significantly increase the thickness of the oxide layer. However, the oxide layer still remains relatively intact and homogeneous, as displayed in Fig. 8(c). However, small pores are observed on the oxide surface of pure SiBCN ceramics under the same oxidation conditions. The SEM investigations of the oxidized exterior and cross-section microstructure of the SM1 samples oxidized at 1100 °C, 1300 °C and 1500 °C for 5 h are shown in Fig. 9. A rough and porous surface microstructure could be observed after oxidation tests, as shown in Fig. 9(a). The EDS spectrum inserted in Fig. 9(a) indicates that B₂O₃ still exists in local areas of the oxide layer in comparison with the image in Fig. 8(a). When the oxidation temperature reaches 1300 °C, lots of bubbles emerge on the oxide surface of the SM1 sample. At an oxidation temperature of 1500 °C, the bubbles disappear and cracks are found on the sample surface. The EDS results of SM1 in Fig. 9(b) and (c) show that Si and O elements are dominant on the oxide surface. The thickness of the oxide layers is changed at elevated oxidation temperature. Besides that, no distinctive features of a bonding interface between the oxide layer and the SiBCN matrix are discovered, according to cross-section images of the SM1 sample. However, the images illustrate a typical structure of the SM1 sample after oxidation tests for 5 h at various temperatures. A relatively dense oxide layer with uniform thickness adhering to the ceramic matrix can be seen in the pictures, as shown in Fig. 9(b), (d) and (f).

Fig. 8 Surface morphologies of SiBCN ceramics after oxidation tests at 1100 °C at different oxidation times. (a) SM1 5 h, (b) SM1 10 h, (c) SM1 20 h, (d) pure SiBCN 20 h.

Fig. 9 Surface and fracture morphologies of SM1 ceramics after oxidation tests for 5 h at different temperatures. (a) and (b) 1100 °C, (c) and (d) 1300 °C, (e) and (f) 1500 °C.

Similar to the SM1 ceramics, the surface morphologies of SM2 remain smooth, transparent and relatively dense at 1100 °C, as shown in Fig. 10(a). However, lots of micro cracks emerge and propagate on the oxide layer and the thickness of oxide layer increases as the oxidation temperature increases from 1100 °C to 1500 °C. Fig. 10(d)–(f) show the fracture morphologies of SM2 at 1100 °C, 1300 °C and 1500 °C for 10 h, respectively. At 1100 °C, the oxide layer is thinner with a clear boundary between the SiBCN matrix and the oxide layer. The oxygen distribution shows a continuous downward trend from the outermost oxide layer to the internal matrix in both samples. The element distribution in the vicinity of the interface illustrates that the oxygen content in the oxide layer is significantly higher than that of the matrix. The oxide layer remains relatively dense and continuous, and has a strong binding capacity with the matrix, with a size of 5–15 μm. The main chemical reactions of SM1 and SM2 during the oxidation test are listed below.

2SiC (s) + 3O₂ (g) → 2SiO₂ (l) + 2CO (g) (1)

C (s) + O₂ (g) → CO₂ (g) (2)

2C (s) + O₂ (g) → 2CO (g) (3)

2BN (s) + 3/2O₂ (g) → B₂O₃ (g) + N₂ (g) (4)

ZrSiO₄ (s) → SiO₂ (l) + ZrO₂ (l) (5)

ZrO₂ (s) + 3C (s) → ZrC (s) + 2CO (g) (6)

2ZrC (s) + 3O₂ (s) → 2ZrO₂ (s) + 2CO (g) (7)

Fig. 10 Surface and fracture morphologies of SM2 ceramics after oxidation tests for 10 h at different temperatures. (a) and (d) 1100 °C, (b) and (e) 1300 °C, (c) and (f) 1500 °C.

The XRD patterns of the SM1 and SM2 samples show the phase transition on the oxide layers, also giving support to the analysis above. After heating up to 1100 °C with a holding time of 5 h, SiO₂, BNC and ZrC can be found in the oxide layer in Fig. 11(a). Formation of a BN(O)–SiO₂ double layer²⁹ during oxidation has been reported by Baldus et al. However, the BN(O)–SiO₂ double layer could not be found on the oxide surface of all investigated SiBCN ceramics. SiC and BNC were oxidized to form SiO₂ and B₂O₃ at 900 °C. But B₂O₃ was evaporated rapidly at 1100 °C, as a result, we could not find any diffraction peaks of B₂O₃ in the oxide layers of both SM1 and SM2 samples. According to our previous analysis and other’s work,²⁷ we believe that ZrO₂ will react with the excessive carbon to form ZrC because of carbothermal reduction reactions during the oxidation test. And the remaining ZrC in the oxide layer of the SM2 sample may result from the reaction of carbon and ZrO₂ produced by the decomposition of ZrSiO₄. The formation process of ZrC on the oxide surface can be summarized as follows: the MZS1 additive is composed of MgO, ZrO₂ and SiO₂, while the MZS2 powder consists of SiO₂ and ZrSiO₄ phases. For the SM1 sample, ZrO₂ would react with excess C to form thermally stable ZrC on the oxide surface. For the SM2 sample, the ZrSiO₄ phases would decompose to ZrO₂ and SiO_2, as a result, ZrO₂ reacts with C to shape ZrC on the oxide surface. However, according to B. J. Tang’s work,³⁰ they found that no trace of ZrC was observed on the oxide surface after oxidation tests at 1500 °C. The low content of C in the ZrSiO₄–SiBCN(O) amorphous coating resulted in the above phenomenon. So the residual excess carbon in the SiBCN matrix after sintering should be responsible for the formation of ZrC, as indicated in Fig. 11. In Fig. 11(a) and (b), the content of SiO₂ in the oxide layer decreases while ZrC rises rapidly when the holding time increases from 5 h to 20 h after the oxidation test. According to M. K. Cinibulk’s³¹ investigation of oxidized polymer-derived SiBCN fibers, they found that the oxidized SiBCN fibers contained three distinct concentric layers, each increasing in oxygen concentration from the core to the outer surface, which is a little different from our results shown in Fig. 9 and 10. In our current work, we find that the oxide layer can be divided into two parts according to the SEM images and XRD patterns. The first oxidation layer next to the core consisted of a mixture of nano-SiC, ZrC and turbostratic BNC, which evolved into a more oxygen-rich glass with SiO₂, ZrC and turbostratic BN grains dispersed throughout nearer the surface. However, only SiO₂ and ZrC exist in the outermost layer at 1500 °C while B₂O₃ evaporates completely, which is different from previous research.³⁰ One different point between SM1 and SM2 after the oxidation test is that the diffraction intensities of SiO₂ on the SM1 surface oxide layer are higher than on SM2 when the oxidation time reaches 20 h at 1300 °C. Fig. 11(c) shows the influence of temperature on the oxide layer of the SM2 sample with a holding time of 10 h at various target temperatures. The content of SiO₂ maintains an upward trend when the temperature rises from 1100 °C to 1500 °C, and the ZrC phase is also formed at 1100 °C but reaches its maximum content at 1500 °C. This result is in accordance with the oxide layer surface and fracture morphologies of the SM2 samples. Elke Butchereit³² prepared three precursor-derived ceramics, which were oxidized at 1300 °C and 1500 °C for times up to 100 h. They revealed that all types of scale thicknesses measured were smaller than those expected from pure SiC or Si₃N₄. As a result, SiBCN ceramics have a better oxidation resistance than monolithic SiC and Si₃N₄ prepared by the same sintering conditions at high oxidation temperature.

Fig. 11 XRD patterns of the SM1 and SM2 ceramics after heat treatment in air at different temperatures with various oxidation times. (a) SM1 after heat treatment in air at 1300 °C, (b) SM2 after heat treatment in air at 1300 °C, (c) SM2 after heat treatment for 10 h.

From the above analysis and chemical reaction equations, the oxidation process of SiBCN ceramics can be divided into three stages. The weight variation of the samples as a function of time at different temperatures are presented in Fig. 12. This will be beneficial to understanding the oxidation mechanisms. In the initial stage of oxidation (∼1100 °C), BNC would be oxidized to form B₂O₃ on the oxide surface but was evaporated rapidly, and a small amount of SiC reacted with O to give SiO₂. At the same time, ZrSiO₄ in the SM2 matrix decomposed to SiO₂ and ZrO₂, along with ZrO₂ being reduced to ZrC by excess carbon. ZrO₂ in the MZS1 additive also made a contribution to the formation of ZrC in the SM1 sample. During oxidation tests below 1100 °C, the formed B₂O₃ might spread on the oxide surface and evaporate during heating, besides, SiO₂ spread on the surface and gradually formed an oxide layer. However, before the end of B₂O₃ volatilization, the residual B₂O₃ on the oxide surface and B₂O₃ formed inside on the interface would react with SiO₂ to form borosilicate glass.³³ The formation of borosilicate glass would increase the melting point itself.³⁴ The viscosity of borosilicate is lower than that of fused silica.³⁵ The continuous diffusion of O could greatly thread the existing borosilicate layer easily.³⁶ Thereafter, the borosilicate structure on the oxide layer would be destroyed completely with the volatilization of B₂O₃ at elevated temperatures up to 1300 °C, which is very different from B. Lu’s work.³³ In comparison with SiC and Si₃N₄ ceramics, the oxidation behavior of SiBCN ceramics is more significant at the beginning of oxidation before 1100 °C. The reasons for this should be that it is more difficult to form a homogenous and continuous oxide layer on SiBCN ceramics because the loss of boron, carbon and nitrogen occurred simultaneously. Therefore, at this stage, for all the investigated samples, significant weight loss took place, as presented in Fig. 12. At a relatively low temperature, the formation of SiO₂ could not effectively fill pores on oxide layer caused by gas evaporation, as shown in reaction eqn (1)–(4). This is consistent with the surface morphologies of the SM1 and SM2 samples. So the SM1 and SM2 samples will be oxidized below 1100 °C with a relatively high oxidation rate, but the oxide layer is thin enough and keeps a good connection with the SiBCN matrix. When the oxide temperature jumps to 1300 °C, the viscosity of SiO₂ is relatively higher but it still could not form a continuous and dense oxide layer on the surface of the SM1 and SM2 ceramics, which resulted in continuous oxidation happening in the matrix. However, the weight loss was alleviated effectively compared with the former situation. The contents of SiO₂ and ZrC increase quickly along with the consumption of excess carbon and ZrO₂ in this oxidation stage. When the oxide temperature climbs to 1500 °C, SiO₂ with moderate viscosity will form a relatively continuous and integral oxide layer on the surface of the samples, and firmly combines with the matrix, which protects the internal materials from being oxidized.^37–39 In this stage of oxidation, in summary, pure SiC and Si₃N₄ ceramics exhibit a superior performance to monolithic SiBCN, SM1 and SM2 samples because of gas evaporation according to reactions (2)–(7). From Fig. 12, it seems that the weight loss phenomenon of SM1 and SM2 is more violent than pure SiBCN ceramics. This may be caused by reactions (6) and (7) due to gas evaporation. As the oxidation time increases at over 1500 °C, the oxidation resistance of SiBCN was obviously better than that of SiC and Si₃N₄ ceramics because special structures were formed on the surface of SiBCN which were responsible for the excellent oxidation resistance.^17,34 However, the formation of a large amount of ZrC uniformly distributed on the surface can lead to crack formation and expansion. The existence of ZrC at local regions will result in an incomplete structure of the oxide layer, and it can be predicted that a continuous oxidation behavior will take place at a low oxidation rate accompanying the entire process of oxidation, as shown in Fig. 9 and 10. The emergence and propagation of cracks on the oxide layer may due to the thermal mismatch of ZrC and SiO₂ in this layer. Furthermore, the excess carbon in the SiBCN matrix will continue to react with ZrO₂ to form ZrC at elevated temperatures for various oxidation times, while the reaction rate climbs rapidly with the elevation of temperature over 1500 °C. It is foreseeable that ZrC will eventually disappear, resulting from the depletion of excess carbon and oxidation of ZrC with the increasing oxidation temperature, and the oxide layer will be composed mainly of SiO₂ and ZrO₂. In our experiments, the formation of the oxide layer structure of SiBCN ceramics was not clear because of the complex phase composition. We still need to do more work on detailed structures of the SiBCN oxide layer, and FIB, TEM and HRTEM could be helpful.

Fig. 12 The weight variation of the samples as a function of time at different temperatures. (a) Pure SiBCN sample, (b) SM1 sample, (c) SM2 sample.

4. Conclusions

This paper reports a study on the preparation and characterization of SiBCN ceramics with two different additives of MZS1 and MZS2. For SiBCN ceramics with the addition of different additives, SM1 has higher flexural strength than SM2 because of its relatively high density. However, the Young’s modulus, fracture toughness and Vicker’s hardness of the SM2 samples were better than those of the SM1 samples. Both SM1 and SM2 have better mechanical properties than pure SiBCN ceramics. The fracture toughness and Vicker’s hardness of SM2 are 2 times and 3 times larger than pure SiBCN ceramics, respectively. MZS1 and MZS2 additives in the SiBCN ceramics contribute to matrix densification owing to liquid phase sintering mechanisms. The oxidation process of SiBCN ceramics can be divided into three stages. In the initial stage, the oxidation process is much more significant and the weight loss maintains a sustained downward trend. Due to the relatively homogenous and dense oxide layer covering the oxide surface, the weight loss rates of the SM1 and SM2 samples are retarded effectively. Sintering aids in SiBCN ceramics contribute to the oxidation resistance and the oxide layer remains at the size of 5–15 μm up to 1500 °C for 20 h in flowing air. The oxide layer of the SM1 and SM2 samples remains dense and continuous, and has a strong binding capacity with the matrix at 1500 °C for a 20 h holding time. Because of the gas evaporation provided by reactions (6) and (7), the weight loss rates of SM1 and SM2 are more dramatic compared with pure SiBCN ceramics at various oxidation times of all investigated temperatures. The residual excess carbon in the SiBCN matrix should be responsible for the formation of ZrC during the oxidation tests. The structure of the outermost oxide layer is comprised of SiO₂ and ZrC. In summary, preparing SiBCN ceramics with MZS1 and MZS2 additives has the potential to improve their mechanical properties and oxidation resistance, which may lead to the preparation of dense ceramics.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (NSFC, Grant number 51072041, 50902031and 51021002).

References

1. R. Riedel, A. Kienzle, W. Dressler, L. Ruwisch, J. Bill and F. Aldinger, Nature, 1996, 382, 796.
2. M. A. Schiavon, G. D. Soraru and I. V. P. Yoshida, J. Non-Cryst. Solids, 2004, 348, 156.
3. R. Kumar, Y. Cai, P. Gerstel, G. Rixecker and F. Aldinger, J. Mater. Sci., 2006, 41, 7088.
4. Z. Zhang, F. Zeng, J. Han, Y. Luo and C. Xu, J. Mater. Sci., 2011, 46, 5940.
5. G. R. Kunmar, J. Vivekanandan, A. Mahudeswaran and P. S. Vijayanand, Iran. Polym. J., 2013, 22, 923.
6. Z. Yang, Y. Zhou, D. Jia and Q. Meng, Mater. Sci. Eng., A, 2008, 489, 187.
7. P. Zhang, D. Jia, Z. Yang, X. Duan and Y. Zhou, J. Mater. Sci., 2012, 47, 7291.
8. V. M. Vishnyakov, A. P. Ehiasarian, V. V. Vishnyakov, P. Hovsepian and J. S. Colligon, Surf. Coat. Technol., 2011, 206, 149.
9. A. Vijayakumar, R. M. Todi, V. O. Todi and K. B. Sundaram, J. Electrochem. Soc., 2007, 154, 875.
10. J. Haug, P. Lamparter, M. Weinmann and F. Aldinger, Chem. Mater., 2004, 16, 83.
11. J. Schuhmacher, F. Berger, M. Weinmann, J. Bill, F. Aldinger and K. Muller, Appl. Organomet. Chem., 2001, 15, 809.
12. N. Liao, W. Xue, H. Zhou and M. Zhang, RSC Adv., 2013, 3, 14458.
13. P. Zhang, D. Jia, Z. Yang, X. Duan and Y. Zhou, Ceram. Int., 2013, 39, 1963.
14. Z. Yang, D. Jia, Y. Zhou and J. Zhang, Mater. Sci. Eng., A, 2008, 448, 241.
15. J. M. Bind and J. V. Biggers, J. Am. Ceram. Soc., 1975, 58, 304.
16. D. H. Kim and C. H. Kim, J. Am. Ceram. Soc., 1990, 73, 1431.
17. B. Lu and Y. Zhang, Mater. Lett., 2014, 137, 483.
18. J. Han, Y. Song, X. Liu and F. Wang, RSC Adv., 2014, 4, 51995.
19. P. Zhang, D. Jia, Z. Yang, X. Duan and Y. Zhou, J. Alloys Compd., 2012, 537, 346.
20. A. Can, M. Herrmann, D. S. Mclachlan, I. Sigalas and J. Adler, J. Eur. Ceram. Soc., 2006, 26, 1707.
21. S. N. Perevislow, Glass Ceram., 2013, 70, 265.
22. F. Yan, F. Chen, Q. Shen and L. Zhang, Key Eng. Mater., 2007, 336, 1062.
23. R. M. German, P. Suri and S. J. Park, J. Mater. Sci., 2009, 44, 1.
24. F. K. van Dijen and E. Mayer, J. Eur. Ceram. Soc., 1996, 16, 413.
25. R. M. German, Metall. Mater. Trans. A, 1995, 26, 279.
26. M. F. Percio, S. D. D. Campos, R. Schneider and E. A. D. Campos, J. Non-Cryst. Solids, 2015, 411, 125.
27. A. Kaiser, M. Lobert and R. Telle, J. Eur. Ceram. Soc., 2008, 28, 2199.
28. Ya. V. Lipatov, S. I. Gutnikov, M. S. Manylov, E. S. Zhukovskaya and B. I. Lazoryak, Mater. Des., 2015, 73, 60.
29. H. P. Baldus and G. Passing, Mater. Res. Soc. Symp. Proc., 1994, 346, 617.
30. B. Tang, Z. Feng, S. Hu and Y. Zhang, Appl. Surf. Sci., 2015, 331, 490.
31. M. K. Cinibulk and T. A. Parthasarathy, J. Am. Ceram. Soc., 2001, 84, 2197.
32. E. Butchereit and K. G. Nickel, J. Am. Ceram. Soc., 2001, 84, 2184.
33. B. Lu and Y. Zhang, Ceram. Int., 2015, l41, 1023.
34. K. G. Nickel, E. Butchereit and C. Schumacher, J. Electrochem. Soc., 2000, 99, 365.
35. E. L. Bourhis and G. Patriarche, Phys. Status Solidi C, 2009, 6, 1846.
36. J. W. Fergus and W. L. Worrell, J. Am. Ceram. Soc., 1995, 78, 1961.
37. J. R. G. Ervin, J. Am. Ceram. Soc., 1958, 41, 347.
38. D. S. Fox, J. Am. Ceram. Soc., 1998, 81, 945.
39. C. G. Cofer and J. Economy, Carbon, 1995, 33, 389.

---