Influence of Sintering Conditions and Nanosilicon Carbide Concentration on the Mechanical and Thermal Properties of Si3N4-Based Materials

In the work, silicon nitride ceramics (Si3N4) and silicon nitride reinforced by nano silicon carbide particles (Si3N4-nSiC) in amounts of 1–10 wt.% were investigated. The materials were obtained using two sintering regimes: under conditions of ambient and high isostatic pressure. The influence of the sintering conditions and the concentration of nanosilicon carbide particles on the thermal and mechanical properties was studied. The presence of highly conductive silicon carbide particles caused an increase in thermal conductivity only in the case of the composites containing 1 wt.% of the carbide phase (15.6 W·m−1·K−1) in comparison with silicon nitride ceramics (11.4 W·m−1·K−1) obtained under the same conditions. With the increase in the carbide phase, a decrease in the densification efficiency during sintering was observed, which caused a decrease in thermal and mechanical performance. The sintering performed using a hot isostatic press (HIP) proved to be beneficial in terms of mechanical properties. The one-step high-pressure assisted sintering process in the HIP minimizes the formation of defects at the sample surface.


Introduction
Due to the excellent properties of silicon nitride, it finds applications as high-performance ceramics. The two main branches of silicon nitride applications are wear-resistant materials (e.g., bioceramics, cutting tools, bearings) and high-temperature materials (e.g., crucibles, thermocouple tubes). The combination of high strength, low thermal expansion, and thermal and chemical resistance means that silicon nitride material can be used in combustion chambers for rocket nozzles [1]. The high working temperatures, thermal shock resistance, and resistivity to oxidation can be tailored by careful control of the introduced sintering additives, the porosity, and the presence of a second phase [2][3][4][5].
Silicon nitride particulate composites have been investigated in terms of improvements in fracture toughness and creep behavior [4][5][6][7]. Most works focused on hot-pressed materials. The hot pressing technique guarantees high densification of multiphase materials; however, it can only be applied for the fabrication of elements with limited dimensions and shapes. In the case of the production of bulk elements with complicated shapes, pressureless sintering is usually considered, which leads to increased porosity with increased concentration of the second phase in silicon-nitride-based composites [8]. There are few reports concerning the sintering of silicon nitride materials in a hot isostatic press (HIP). The research proves that the sintering of silicon nitride in a hot isostatic press leads to a decrease in fracture toughness and an increase in mechanical strength [9].
Composite materials in the Si 3 N 4 -SiC system were previously obtained using the HIP technique in a post-sintering process [10]. No references were found on the efficiency of consolidation of the Si 3 N 4 -SiC materials in a HIP in a one-step sintering process.
In the presented paper, the influence of the sintering conditions and the concentration of SiC particles in the Si 3 N 4 matrix on physical, thermal, and mechanical properties was investigated.

Materials and Methods
In this research, silicon nitride granulate Starceram N rtp, grade M provided by HC Starck Ceramics GmbH (Selb, Germany), and silicon carbide nanopowder provided by ABCR GmbH (Karlsruhe, Germany) were used.
The samples were prepared from silicon nitride granulate and granulate with 1, 5, and 10 wt.% addition of silicon carbide nanopowder (nSiC). The silicon nitride granulate contains sintering additives (mainly yttria and alumina) and technological binders and plasticizers, which increase the densification effectivity during forming. The granulates containing nanosilicon carbide were prepared by mixing silicon nitride granulate with an appropriate amount of silicon carbide nanopowder in propanol in a ball mill via planetary milling for 1 h. Silicon nitride balls of 3 mm in diameter were used for the mixing of the powders.
The granulates were uniaxially pressed in molds with zirconia liner into cylindrical samples with a diameter of about 16 mm and height of 2.2 mm for the flexural strength test and 3.8 mm for other tests. The samples were subsequently isostatically densified in a cold isostatic press at a pressure of 200 MPa. The green bodies, placed in a graphite crucible, were then sintered in two regimes: (1) pressureless sintering in nitrogen flow in a graphite furnace at a temperature of 1800 • C; (2) hot isostatic pressing (HIP) with a sintering temperature of 1800 • C and pressure of 200 MPa of nitrogen. In both furnaces, nitrogen of 5.0 purity was used.
The density and open porosity of the sintered bodies were measured by the Archimedes method. The measurements were conducted in water. The results were averaged, and the standard deviation was calculated based on the measurements of at least ten samples. The relative density is the ratio between the apparent and theoretical density. The theoretical densities of each composition were calculated according to the rule of mixtures.
The particle size was characterized by the dynamic light scattering technique (dls). Particle size distribution analysis was conducted by means of the zeta potential and a Zetasizer Nano ZS particle size analyzer (Malvern Instruments Ltd., Worcestershire, UK) for diluted suspensions, which underwent ultrasonication (VibraCell VCX130, Sonics & Materials, Inc., Newtown, CT, USA) prior to measurement. The analysis results are presented in terms of the Z average (Z ave , also referred to as the cumulant mean or harmonic intensity averaged particle diameter) and polydispersity index (PdI); these data were derived from the intensity of the overall mean particle diameter value and the overall average polydispersity, respectively.
The specific surface area was measured according to the Brunauer-Emmett-Teller (BET) theory using a gas adsorption analyzer (Gemini VII, Micromeritics Instrument Corp., Norcross, GA, USA). Based on the results of the BET, the equivalent spherical particle diameter (d BET ) was calculated.
Imaging and energy-dispersive spectroscopy (EDS) analyses were performed by means of scanning electron microscopy (SEM; Nova NanoSEM 200, FEI Company, Hillsboro, OR, USA). For the EDS, three areas were analyzed, from which 3-5 points were selected for measurement.
The X-ray diffraction (XRD) was performed in a Bragg-Brentano system using a Bruker-AXS D8 DAVINCI diffractometer equipped with a copper X-ray source. The XRD analysis was performed on the unpolished surfaces. Diffractograms were recorded in the angular range of 5 to 120 • 2θ (Cu Kα, with radiation wavelength 0.154 nm), the measurement step was 0.01 • , and the time was 2 sec per step. The optical system of the diffractometer consisted of a 0.3 • divergence slit, a 1.5 • anti-scattering slit, two 2.5 • Soller slits, a Ni filter, and a LynxEye strip detector with a field of view of 2.94 • . Identification of the phases was performed by comparing the recorded diffractograms with patterns found in the COD database using the DIFFRACplus EVA-SEARCH program. Quantitative X-ray analysis was performed based on published crystal structures (COD) using the Rietveld method in Topas v5.0 software [11,12].
The hardness of the materials obtained was determined by the Vickers method using a LECO LV800 hardness tester. The measurement was performed according to the standards PN-EN ISO 14705:2021-06 and PN-EN ISO 6507-1:2018-05. Vickers pyramid impressions were made using a force of 1 kgf on the surface of the polished samples. The rate of movement of the pyramid was 0.2 mm/min. After reaching the set force, the load was maintained for 15 s.
The indentation fracture resistance was calculated based on Palmquist cracks (0.25 < l/a < 1.5) propagating from the apex of the Vickers indent using Equation (1) [13][14][15]: where a is half of the diagonal of the indentation induced by the Vickers indenter (a = 0.5 d), H V is the Vickers hardness, E is Young's modulus, and c is the length of the Palmquist crack. The length of the crack was measured in a straight line from the apex of the indent to the tip of the crack. In order to determine the flexural strength, the ball-on-ring testing method was used, which involves measuring the maximum failure force of cylindrical specimens laid flat on a ring-shaped stand and loaded axially. The tests were performed using a Tinius Olsen H10-KS testing machine. The flexural strength index using the biaxial loading method was determined from the Kirstein and Wooley equation [16,17]: where P is the failure load [N], t is the thickness of the disk, a is the radius of the supporting ring (12 mm), b is the radius of uniform load (b ≈ t/3), R is the radius of the disk, and ν is Poisson's ratio for Si 3 N 4 (v[Si 3 N 4 ] = 0.255 was used for calculations). The choice of the ball-on-ring technique over the 3-and 4-point flexural strength was motivated by several reasons. First of all, in this test, the friction between the tested sample and the jig is minimized. Secondly, the stresses usually present at the edges of rectangular samples have a great influence on the 3-point or 4-point flexural strength test results. In the case of the ball-on-ring technique, this phenomenon is avoided [18]. Last but not least, test samples in the cylindrical shape are easy to prepare, and their geometry introduces fewer defects related to shrinkage stresses into the ceramic body. The flexural tests were conducted on "as received" samples.
The value of the Weibull modulus was estimated for both the Si 3 N 4 and Si 3 N 4 -nSiC specimens. The Weibull modulus was determined from the results of a flexural strength test by a graphical method using Equation (3): where F is the probability of brittle decohesion of the specimen, m is the Weibull modulus, σ is the flexural strength, and C is the constant resulting from the double logarithmic operation of the basic expression relating the probability of failure of a shape to its strength: where σ u is the parameter of asymmetry of the distribution of strength values σ, and σ 0 is a normalizing constant [19,20]. The thermal diffusivity measurement was carried out by the LaserFlash (LFA) technique using an LFA 427 analyzer (Netzsch GmbH). In this method, the lower surface of the sample placed in a holder is submitted to a short energy pulse produced by a neodymium laser. This results in a temperature change of the second, upper surface of the plane parallel surface. The temperature changes measured by an infrared detector, type InSb, are then registered as a function of time. The thermal diffusivity was calculated from Equation (5) given by Parker et al. [21]: where α is the thermal diffusivity, L is the sample thickness, and t 1/2 is the half-time (time value at half the temperature signal height). The thermal conductivity was calculated based on Equation (6) [22]: where λ is the thermal conductivity, d is the density of the measured sample, and c p is the specific heat of the material.

Results
The starting materials were characterized in terms of their particle size (Table 1). Additionally, the silicon carbide nanopowder was studied by the SEM technique ( Figure 1). where σu is the parameter of asymmetry of the distribution of strength values σ, and σ0 is a normalizing constant [19,20]. The thermal diffusivity measurement was carried out by the LaserFlash (LFA) technique using an LFA 427 analyzer (Netzsch GmbH). In this method, the lower surface of the sample placed in a holder is submitted to a short energy pulse produced by a neodymium laser. This results in a temperature change of the second, upper surface of the plane parallel surface. The temperature changes measured by an infrared detector, type InSb, are then registered as a function of time. The thermal diffusivity was calculated from Equation (5) given by Parker et al. [21]: where α is the thermal diffusivity, L is the sample thickness, and t1/2 is the half-time (time value at half the temperature signal height).
The thermal conductivity was calculated based on Equation (6) [22]: where λ is the thermal conductivity, d is the density of the measured sample, and cp is the specific heat of the material.

Results
The starting materials were characterized in terms of their particle size (Table 1). Additionally, the silicon carbide nanopowder was studied by the SEM technique ( Figure  1). Table 1. Properties of the powders used in the research (Zave-cumulant mean, PdI-polydispersity index, SBET-specific surface area, dBET-BET equivalent spherical particle diameter).   According to the specific surface measurements results, the Si 3 N 4 powder, which was the basis of the used granulate, was of submicrometric size (d BET = 135 nm); however, in the dls measurements, the presence of agglomerates was detected (Z ave = 1.1 µm).

Zave, nm
The silicon carbide nanopowder also showed some agglomeration in the dls measurement, as the cumulant mean was higher than the BET equivalent particle diameter (Z ave = 397 nm, d BET = 66 nm). In the SEM image, it was observed that most of the grains were of nanometric size (<100 nm), and there were few grains of submicrometric size visible (100-400 nm; Figure 1).
Such a combination of granulometric sizes of the powders should provide good densification in a green state, as the green density of the silicon nitride-silicon carbide mixture of powders was higher than the density of silicon nitride green samples. The green relative density for silicon nitride samples was 60.5%, and the value measured for the Si 3 N 4 -nSiC samples was in the range of 61.3-62.3% (Table 2). Table 2. Selected physical and thermal properties of the materials were obtained by sintering under ambient pressure at a temperature of 1800 • C and in the hot isostatic press at a temperature of 1800 • C.

Material
Sintering Conditions The powders were investigated in terms of their phase composition (Figures 2 and 3). According to the specific surface measurements results, the Si3N4 powder, which was the basis of the used granulate, was of submicrometric size (dBET = 135 nm); however, in the dls measurements, the presence of agglomerates was detected (Zave = 1,1 μm).

Relative Green Density
The silicon carbide nanopowder also showed some agglomeration in the dls measurement, as the cumulant mean was higher than the BET equivalent particle diameter (Zave = 397 nm, dBET = 66 nm). In the SEM image, it was observed that most of the grains were of nanometric size (<100 nm), and there were few grains of submicrometric size visible (100-400 nm; Figure 1).
Such a combination of granulometric sizes of the powders should provide good densification in a green state, as the green density of the silicon nitride-silicon carbide mixture of powders was higher than the density of silicon nitride green samples. The green relative density for silicon nitride samples was 60.5%, and the value measured for the Si3N4-nSiC samples was in the range of 61.3-62.3% ( Table 2).
The powders were investigated in terms of their phase composition (Figures 2 and  3).     The silicon nitride granulate consisted mainly of the α-Si 3 N 4 phase (78.7 wt.%). The β-Si 3 N 4 phase was also detected, and its concentration was 12.4 wt.%. The oxide phases Y 2 O 3 and α-Al 2 O 3 were the sintering agents, and they were present in an amount of about 4.4 wt.% each.
The silicon carbide nanopowder consisted mainly of the β-SiC phase (ca. 70%), which is a cubic crystallographic form of silicon carbide. The rest of the material was the hexagonal α-SiC.
The samples sintered at a temperature of 1800 The silicon nitride granulate consisted mainly of the α-Si3N4 phase (78.7 wt.%). The β-Si3N4 phase was also detected, and its concentration was 12.4 wt.%. The oxide phases Y2O3 and α-Al2O3 were the sintering agents, and they were present in an amount of about 4.4 wt.% each.
The silicon carbide nanopowder consisted mainly of the β-SiC phase (ca. 70%), which is a cubic crystallographic form of silicon carbide. The rest of the material was the hexagonal α-SiC.
The samples sintered at a temperature of 1800 °C were investigated in terms of their phase composition. The results of the XRD analysis of the Si3N4 sample and the sample containing 5 wt.% of nanosilicon carbide, together with the characteristic reflections of the phases present in the samples, are presented in Figures 4 and 5.    In the samples sintered at a temperature of 1800 °C, silicon nitride occurred predominantly in the form of β-Si3N4. XRD analysis performed on a polished sample led to the detection of only β-Si3N4. The α-Si3N4 present in the granulate is more reactive and reacts with the oxides, forming SiAlONs, which then form an amorphous phase [7,23]. At the unpolished surface of the sample, SiAlON occurs in a hexagonal phase. Some residual amounts of oxides were detected in the surface layer; their quantity was estimated to be beneath 1%.
The silicon carbide transformed into the hexagonal α-SiC crystalline form. The β-SiC phase is metastable and transforms into α-SiC [24].
There were differences in the phase composition both on the surface and in the sample body of the Si3N4 and Si3N4-nSiC materials. At the surface of Si3N4 ceramics, the amounts of the α and β phase of silicon nitride and SiAlON were 7.7%, 88.8%, and 3.1%, respectively. In the Si3N4-nSiC material, the mass composition of the mentioned phases was 7.0%, 80.2%, and 12.2%, respectively. Additionally, in the Si3N4-nSiC material, the measured phase content of SiC was smaller than the amount of introduced material and equaled about 1%. The XRD of the Si3N4-nSiC material performed on the polished sample revealed that the body consisted mainly of β-Si3N4 with minor amounts of α-Si3N4 and hexagonal SiAlON phase. From the analysis, it can be concluded that the silicon carbide reacts with the oxides and is a competitive phase for the reaction with the oxides in relation to α-Si3N4. In the samples sintered at a temperature of 1800 • C, silicon nitride occurred predominantly in the form of β-Si 3 N 4 . XRD analysis performed on a polished sample led to the detection of only β-Si 3 N 4 . The α-Si 3 N 4 present in the granulate is more reactive and reacts with the oxides, forming SiAlONs, which then form an amorphous phase [7,23]. At the unpolished surface of the sample, SiAlON occurs in a hexagonal phase. Some residual amounts of oxides were detected in the surface layer; their quantity was estimated to be beneath 1%.
The silicon carbide transformed into the hexagonal α-SiC crystalline form. The β-SiC phase is metastable and transforms into α-SiC [24]. There were differences in the phase composition both on the surface and in the sample body of the Si 3 N 4 and Si 3 N 4 -nSiC materials. At the surface of Si 3 N 4 ceramics, the amounts of the α and β phase of silicon nitride and SiAlON were 7.7%, 88.8%, and 3.1%, respectively. In the Si 3 N 4 -nSiC material, the mass composition of the mentioned phases was 7.0%, 80.2%, and 12.2%, respectively. Additionally, in the Si 3 N 4 -nSiC material, the measured phase content of SiC was smaller than the amount of introduced material and equaled about 1%. The XRD of the Si 3 N 4 -nSiC material performed on the polished sample revealed that the body consisted mainly of β-Si 3 N 4 with minor amounts of α-Si 3 N 4 and hexagonal SiAlON phase. From the analysis, it can be concluded that the silicon carbide reacts with the oxides and is a competitive phase for the reaction with the oxides in relation to α-Si 3 N 4 .
Sintered Si 3 N 4 ceramics and Si 3 N 4 -nSiC material samples were tested in terms of their density, open porosity, and thermal diffusivity. The results are presented in Table 2.
The theoretical density of silicon nitride and silicon nitride containing 1 wt.% of silicon carbide nanopowder sintered pressureless at a temperature of 1800 • C exceeds 98%. The Si 3 N 4 -nSiC materials containing 5 and 10 wt.% of nSiC sintered in the same regime showed Samples sintered via HIP have lower density. The Si 3 N 4 ceramics and 1 wt.% Si 3 N 4 -nSiC materials were densified in the HIP process to theoretical densities of 93.9 and 94.7%, respectively. The materials containing higher concentrations of nSiC underwent less effective densification, and their theoretical densities were 86.8 and 80.7%, respectively, for bodies containing 5 and 10 wt.% of the second phase. The open porosity of Si 3 N 4 ceramics and Si 3 N 4 -1 wt.% nSiC bodies was 0.3%. In samples containing 5 and 10 wt.% of nSiC, the open porosity values were 3.7 and 17.5%, respectively.
The thermal conductivity of the bodies sintered in atmospheric pressure was higher than the values calculated for the Si 3 N 4 and Si 3 N 4 -nSiC materials sintered in the HIP process. The highest thermal conductivity was calculated for the sample sintered in a pressureless regime with 1 wt.% of nSiC. All the HIP samples had similar thermal conductivity in the range of 10-13 W·m −1 ·K −1 .
The mechanical properties are presented in Table 3. Table 3. Mechanical properties of the materials obtained by pressureless sintering at a temperature of 1800 • C and in the hot isostatic press at a temperature of 1800 • C.  Samples sintered under high isostatic pressure had generally lower hardness, with 16.2 GPa for silicon nitride and 17.1 GPa for Si 3 N 4 -1 wt.% nSiC material. With increasing nSiC content to 5 and 10 wt.%, the hardness dropped to 11.8 and 6.7 GPa.

Material
The indentation fracture resistance of HIP samples was higher than that of samples sintered without external pressure, in the ranges of 4.2-4.4 MPa·m 1/2 and 3.5-3.7 MPa·m 1/2 , respectively. Due to the high porosity of samples containing 10 wt.% silicon carbide sintered in the HIP process, the measurement of the length of the cracks was impossible.
In Figure 6, micrographs of the silicon nitride ceramics and silicon nitride containing 5 wt.% silicon carbide sintered in nitrogen at atmospheric pressure and in HIP conditions are presented.
Materials 2023, 16, x FOR PEER REVIEW 10 of 14 Figure 6. Micrographs of the Si3N4 material sintered pressureless at a temperature of 1800 °C (a-c) and in the hot isostatic press at a temperature of 1800 °C (d-f); Si3N4-nSiC material containing 5 wt.% of silicon carbide sintered pressureless at a temperature of 1800 °C (g-i) and in the hot isostatic press at a temperature of 1800 °C (j-l). In the first column (a,d,g,j), the polished surfaces are presented; in the second (b,e,h,k), the fracture surfaces of the samples are presented; and in the third column, the fracture surfaces at the edge of the samples are presented (c,f,i,l).
According to the EDS analysis (not shown), the dark grey areas in the images of Si3N4 ceramics (Figure 6a-f) are the silicon nitride grains (indicated with solid arrows). The lighter areas are the bonding phase consisting of Y-Si-Al-O-N and Si-Al-O-N systems (indicated with a dashed arrow). These are the Y-α-SiAlON and Y2Si3O3N4 crystalline phases detected in the XRD analysis and amorphous phases, which can also form under these conditions [7]. In the fracture surfaces, longitudinal β-Si3N4 particles are visible in both the materials sintered under pressureless conditions (Figure 6a-c) and those subjected to the HIP process (Figure 6d-f). The samples obtained by pressure-assisted sintering in the HIP are characterized by smaller grains and higher porosity, which was confirmed by the density measurements (Table 2).
In the cross sections of the Si3N4-nSiC materials, the contrast between the Si3N4 matrix Figure 6. Micrographs of the Si 3 N 4 material sintered pressureless at a temperature of 1800 • C (a-c) and in the hot isostatic press at a temperature of 1800 • C (d-f); Si 3 N 4 -nSiC material containing 5 wt.% of silicon carbide sintered pressureless at a temperature of 1800 • C (g-i) and in the hot isostatic press at a temperature of 1800 • C (j-l). In the first column (a,d,g,j), the polished surfaces are presented; in the second (b,e,h,k), the fracture surfaces of the samples are presented; and in the third column, the fracture surfaces at the edge of the samples are presented (c,f,i,l).
According to the EDS analysis (not shown), the dark grey areas in the images of Si 3 N 4 ceramics (Figure 6a-f) are the silicon nitride grains (indicated with solid arrows). The lighter areas are the bonding phase consisting of Y-Si-Al-O-N and Si-Al-O-N systems (indicated with a dashed arrow). These are the Y-α-SiAlON and Y 2 Si 3 O 3 N 4 crystalline phases detected in the XRD analysis and amorphous phases, which can also form under these conditions [7]. In the fracture surfaces, longitudinal β-Si 3 N 4 particles are visible in both the materials sintered under pressureless conditions (Figure 6a-c) and those subjected to the HIP process (Figure 6d-f). The samples obtained by pressure-assisted sintering in the HIP are characterized by smaller grains and higher porosity, which was confirmed by the density measurements (Table 2).
In the cross sections of the Si 3 N 4 -nSiC materials, the contrast between the Si 3 N 4 matrix and the SiC phase is low due to the small difference in atomic mass of carbon and nitrogen. The SiC phase in the form of submicrometric grains is visible in the SEM image as light-grey areas. There was no delamination between nSiC and the matrix observed. According to the SEM observations and the EDS analysis, the SiC phase was uniformly distributed in the matrix. Both Si 3 N 4 and SiC grains were surrounded uniformly by the bonding phase constituted of the Y-Si-Al-O-N and Si-Al-O-N systems. Similarly to Si 3 N 4 ceramics, the pressureless sintered Si 3 N 4 -nSiC materials were less porous, and the grains of the matrix underwent higher grain growth.
There were significant differences in the microstructure of the surfaces of the samples depending on the sintering regime. The pressureless sintered Si 3 N 4 ceramics (Figure 6c) and Si 3 N 4 -nSiC materials (Figure 6i) had defects at the surface. At the surface, there was some porosity visible, reaching a depth of 60-80 µm (indicated in Figure 6c,i). A greater amount of oxide phase (light areas) was observed in this region as well.
The samples obtained by the HIP process had no defects visible on the sample surface. The fracture surface images taken of the sample surface (Figure 6f,l) did not vary significantly from the images of the middle of the sample (Figure 6e,k).

Discussion
The physical, mechanical, and thermal properties of the Si 3 N 4 ceramics and Si 3 N 4 -nSiC materials sintered under nitrogen at atmospheric pressure and under nitrogen at high pressure in the hot isostatic pressing process are presented in the form of radial graphs in  [25][26][27]. Thus, the measured open porosity must result from defects present at the surface of the sample ( Figure 6). These defects could not be avoided by sintering in the Si 3 N 4 powder bed. The furnace chamber was vacuumed and flushed with nitrogen before the sintering process; nonetheless, at the surface of the powder, some water and air still could have been adsorbed, which reacted with the silicon nitride matrix of the samples. The Si 3 N 4 powder, in which the samples were buried for the sintering process, itself might have been the source of oxidation of the sample surfaces.
These defects are responsible for low flexural strength. The porosity at the surface might be the origin of the failure of the samples. In the case of materials sintered via HIP, which had no major defects at the surface, the flexural strength was higher, even by over 100%, in the case of Si 3 N 4 -1 wt% nSiC materials in comparison to the pressureless sintered samples. The relation between the surface roughness and the mechanical behavior of silicon nitride was also confirmed by other researchers [28]. Additionally, the increased amount of oxide phases at the surface may also have caused the decrease in flexural strength. The observed fracture surfaces ( Figure 6) are typical for this intergranular character. This means that the intergranular SiAlON phase is weaker than the Si 3 N 4 grains.
The beneficial effect of the presence of silicon carbide on mechanical strength was not confirmed. The materials containing 1 wt.% of nanosilicon carbide were characterized by comparable flexural strength in comparison to the silicon nitride matrix, with values of 487 and 485 MPa, respectively. With increasing silicon carbide content, the flexural strength decreased to 416 and 250 MPa for Si 3 N 4 -5 wt% nSiC and Si 3 N 4 -10 wt% nSiC, respectively. The porosity of the Si 3 N 4 -nSiC materials is one of the factors responsible for this phenomenon. The second is the increase in the amorphous SiAlON phase in the Si 3 N 4 -nSiC materials, which may determine the mechanical strength.
The samples obtained by the HIP process had no defects visible on the sample surface. The fracture surface images taken of the sample surface (Figure 6f,l) did not vary significantly from the images of the middle of the sample (Figure 6e,k).

Discussion
The physical, mechanical, and thermal properties of the Si3N4 ceramics and Si3N4-nSiC materials sintered under nitrogen at atmospheric pressure and under nitrogen at high pressure in the hot isostatic pressing process are presented in the form of radial graphs in Figure 7. Note that the open porosity of the samples is presented in the inverted scale.  A concentration of nanosilicon carbide above 5 wt.% hindered the densification of the Si 3 N 4 -nSiC material bodies. In the case of the pressureless sintering regime, the density of the materials dropped to about 95%. However, in the case of samples sintered via HIP, the relation between the concentration of the second phase and the density and porosity was more pronounced. The HIP materials containing 10 wt.% of nSiC were poorly densified (d rel = 80.7%) and were porous (P o = 17.5%). The mechanism responsible for the poor densification of the materials is the pinning effect of the SiC phase located at the graifn boundaries hindering densification by grain growth [29,30], as the grains observed in the materials were smaller.
The presence of silicon carbide in the silicon nitride matrix had a positive influence on the increase in thermal conductivity. The improvement was best visible in the material containing 1% silicon carbide nanopowder sintered pressureless, which showed thermal conductivity about 37% higher in comparison to the silicon nitride matrix. With increasing silicon carbide content in the silicon nitride bodies, the thermal conductivity gradually decreased, and the 10 wt.% nSiC materials showed thermal conductivity similar to that of the Si 3 N 4 body. This effect is attributed to the increasing porosity of the bodies with higher nSiC content. The presence of pores compensates for the effect of the presence of highly conductive nSiC in the matrix.
The presence of silicon carbide in the silicon nitride matrix also influenced the hardness of the material. Despite the decreasing relative density, the hardness was about 19 GPa for the pressureless sintered samples. Only for the material containing 10 wt.% of nSiC was a decrease in hardness observed. In the case of the samples obtained by pressure-assisted sintering, the measured hardness decreased with increasing nSiC content, which is related to the worse densification of the materials.
The presence of the second phase in the silicon nitride matrix did not have much influence on the indentation fracture resistance, equaling 3.5-3.7 MPa·m 1/2 for the pressureless sintered materials and 4.2-4.4 MPa·m 1/2 for the HIP materials. It was observed that the cracks propagated at the grain boundaries and were usually terminated at the elongated β-Si 3 N 4 grains positioned perpendicularly to the cracks. Interaction of the cracks with nSiC grains was not observed. The presence of silicon carbide may even cause a decrease in the indentation fracture resistance of the silicon nitride matrix [7,31]. A beneficial influence of the second phase was detected only in the case of the materials containing coarse silicon carbide particles and whiskers [29,31]. However, the toughening effect was observed only up to a definite amount of the additive, which is attributed to the increased porosity of the materials. A similar effect was observed for the materials sintered in HIP: silicon nitride and silicon nitride materials containing 1 and 5 wt.% of nSiC had K IFR = 4.2-4.4 MPa·m 1/2 , while materials containing 10 wt.% showed porosity, which disabled the measurement of the crack lengths.
The presence of the silicon carbide had a minor effect on the indentation fracture resistance. On the other hand, a significant increase in indentation fracture resistance was observed for materials sintered under HIP conditions. This suggests that this parameter is controlled by the quality of the silicon nitride matrix. According to the literature, the indentation fracture resistance in silicon nitride materials is controlled by the formation of β-Si 3 N 4 and the intergranular phases [29,32,33]. In both Si 3 N 4 ceramics and Si 3 N 4 -nSiC, the main phase of the matrix was β-Si 3 N 4 , and the oxide binding phases were the same. The presence of the silicon carbide grains did not affect the formation of the matrix material.

Conclusions
Pressure-assisted sintering of silicon nitride hindered the formation of defects at the surface of the material and improved the mechanical performance of the as-prepared materials.
The addition of highly conductive silicon carbide particles increased thermal conductivity. However, only small amounts of the second phase brought a desirable effect, as with increasing nSiC content, the effectiveness of densification during the sintering process decreased.
Enhanced thermal conductivity indicated that the silicon carbide particles were well bonded with the silicon nitride matrix, with no cavitations at the phase boundaries, which was also confirmed by the microstructural observations.

Conflicts of Interest:
The authors declare no conflict of interest.