Ceramics International 46 (2020) 12606–12612

Contents lists available at ScienceDirect

Ceramics International
journal homepage: www.elsevier.com/locate/ceramint

Elimination of grain boundaries and its effect on the properties of silicon T

nitride ceramics
Feng Hua, Tianbin Zhub,c, Zhipeng Xiea,∗, Jian Liua, Zunlan Hua, Di Ana
a
State Key Laboratory of New Ceramics and Fine Processing, School of Materials Science and Engineering, Tsinghua University, Beijing, 100084, China
b
The State Key Laboratory of Refractories and Metallurgy, Wuhan University of Science and Technology, Wuhan, 430081, China
c
National-provincial Joint Engineering Research Center of High Temperature Materials and Lining Technology, Wuhan University of Science and Technology, Wuhan,
430081, China

A R T I C LE I N FO A B S T R A C T

Keywords: To eliminate the excess grain boundary phase in silicon nitride ceramics prepared through hot pressing and a
Silicon nitride ceramics post-sintering heat treatment, carbon was introduced into the sintered ceramics. This was accomplished in two
Thermal conductivity ways: first, a small amount of carbon was used as a sintering additive; however, this creates large pores in the
Grain boundary ceramics, resulting in a decrease in the thermal conductivity and mechanical properties of the samples.
Post-sintering heat treatment
Alternatively, the C/Si3N4 power bed could be used instead of the BN/Si3N4 power bed, which obviously
eliminates the grain boundary phase. Further, the C/Si3N4 power bed possess excellent mechanical properties
while increasing the thermal conductivity of the samples. Therefore, the silicon nitride ceramics with
123 W m−1 K−1 thermal conductivity and 986 MPa bending strength of can be obtained using the C/Si3N4
powder bed during the post-sintering heat treatment.

1. Introduction dislocations) are recognized factors determining the thermal con-

ductivity of silicon nitride ceramics [7–9]. Silicon nitride ceramics
High-power electronic devices have been widely used for a variety comprise silicon nitride grains and glassy phases within grain bound-
of applications, such as high-speed trains, electric/hybrid motor ve- aries, which exhibit only 1–10 W m−1 K−1 thermal conductivity, which
hicles, wind power generation, and industrial robots. To achieve more is much lower than that of the Si3N4 grains [7].
effective electric powder control, increasing voltage, current, and Similar to Si3N4 ceramics, AlN ceramics exhibited high thermal
power density are increasingly adopted in the development of power conductivity, but grain boundaries with a low thermal conductivity
device technologies. However, these will cause significant thermal seriously affect the thermal conductivity of the ceramics. Many studies
stress inside the device and cause brittle fractures in fragile ceramic have shown that sintering under a reducing N2 atmosphere with carbon
substrate materials [1,2]. As such, ceramic substrates with excellent is an effective method for enhancing the thermal conductivity of AlN
mechanical properties and good thermal conductivity are required for ceramics [10–12]. In this sintering process, the AlN ceramics were often
high-power electronic devices. doped with Y2O3, as shown in equation (1), to decrease the grain
Silicon nitride ceramics are considered a superior structural mate- boundary phase [12].
rial owing to their excellent mechanical properties, outstanding che-
Al2O3·2Y2O3 + 3N2 + 9C → 4 YN + 2AlN + 9CO (1)
mical stability, and superior thermal shock resistance [3–5]. Previous
studies have reported that the theoretical thermal conductivity of During these reactions, the grain boundary comprising Al2O3 and
single-crystal β-Si3N4 is estimated to be 200–320 W m−1 K−1 at room Y2O3 migrates to the surface of AlN ceramics and the oxides are nitrided
temperature [6]. Owing to its excellent mechanical and thermal prop- to form YN and AlN.
erties, silicon nitride is considered as a promising high-thermal-con- Similarly, carbon has also been widely used to improve the thermal
ductivity substrate material for power electronic devices. However, the conductivity of silicon nitride ceramics by these methods, for instance,
thermal conductivity of polycrystalline silicon nitride ceramics is by using doped carbon as sintering additives and using a BN/graphite
usually much lower than the theoretical thermal conductivity. Grain powder bed [13,14]. The introduction of carbon was found to promote
boundaries and lattice defects (such as impurity atoms, vacancies, and the elimination of oxygen and increase the N/O atomic ratio of grain

∗
Corresponding author.
E-mail address: xzp@mail.tsinghua.edu.cn (Z. Xie).

https://doi.org/10.1016/j.ceramint.2020.02.024
Received 7 November 2019; Received in revised form 2 February 2020; Accepted 3 February 2020
Available online 04 February 2020
0272-8842/ © 2020 Published by Elsevier Ltd.
F. Hu, et al. Ceramics International 46 (2020) 12606–12612

boundary, thus stimulating the selective growth of the β-Si3N4 grain. Germany), and the bending strength of the sample was measured by the
Consequently, several grain boundaries were reduced by the reducing three-point-bending method on 1.5 × 2 × 25 mm3 test bars at 0.5 mm/
N2 atmosphere with carbon at high temperatures and the rest of the min loading speed and 20-mm span. According to the continuous
grain boundaries were prone to reside at the triple and multiple points. stiffness measurement, the elastic modulus (E) was measured using a
Thus, the thermal conductivity of silicon nitride ceramics increased nanoindenter (XP, MTSS systems Co., America). The Vickers hardness
significantly. However, the mechanical properties of silicon nitride (H) was analyzed using a Vickers hardness tester (Tukon2500B, Wilson,
ceramics have not been investigated yet. As is well known to all, America), under a 5-kg loading condition for 15 s. The microstructures
bending strength is an important parameter for ceramic substrate ma- of fracture and plasma-etched surfaces were observed using field-
terials. Therefore, methods for maintaining high thermal conductivity emission scanning election microscopy (SEM, Merlin Compact, Zeiss,
and good mechanical properties in ceramic substrate materials should Germany) equipped with EDS. Plasma etching was performed using
be considered in the preparation of silicon nitride ceramics. At present, plasma etching apparatus (L-451D-L, Anelva reactive ion etching
high-thermal-conductivity silicon nitride ceramics are prepared mainly system) using a CF4 gas. Elemental composition analysis was performed
via the sintering of reaction-bonded silicon nitride or gas-pressure via X-ray fluorescence analysis (XRF-1800, SHIMADZU, Japan). The
sintering (GPS) [15–18]. Although high thermal conductivity can be thermal conductivity (κ) of the sample was calculated using the formula
obtained by these sintering methods, the bending strength is often in- κ = ραCp. The thermal diffusivity (α) was measured by the laser-flash
sufficient and has therefore not been reported. The hot-pressing (HP) method (LFA 447, NETZSCH, Germany) with a dimension of
process has been considered an effective sintering process for im- 10 × 10 × 2 mm3. The bulk density (ρ) was measured by the
proving bulk density and mechanical properties [19]. Therefore, in- Archimedes method, and the specific heat (Cp) of the sample was
vestigating the different elimination method of grain boundary and its measured using a physical property measurement system (PPMS-9T,
effect on the properties of silicon nitride ceramics sintered by HP is Quantum Design, USA). The identification of the carbon was performed
important. by Raman spectroscopy (LabRAM HR Evolution, Horriba, France).
In this study, α-Si3N4 powder were used with MgF2–Y2O3 additives
as starting materials for the fabrication of high-thermal-conductivity
silicon nitride ceramics through hot pressing and post-sintering heat 3. Result and discussion
treatment. Two different methods for the introduction of carbon were
used to eliminate the grain boundary of silicon nitride ceramics. One 3.1. Effect of adding carbon as sintering additives on silicon nitride ceramics
method was to use a small amount of carbon as a sintering additive; the
effects of heat treatment at different temperatures were investigated by During the sintering process of silicon nitride ceramics, the sintering
this method. The other method was to use a C/Si3N4 power bed instead additives can obviously affect the properties of the material. The bulk
of BN/Si3N4 during heat treatment at different temperatures. The me- density and thermal conductivity of the SN and SNC samples are clearly
chanical properties, thermal conductivity, and microstructure were shown and compared in Fig. 1. From the graph, the bulk density has
systematically investigated. hardly changed. However, there was an obvious decrease in the thermal
conductivity from 76.7 to 71.6 W m−1 K−1. This decrease indicated
2. Experimental procedure that adding carbon as a sintering additive had an adverse effect on the
thermal conductivity of silicon nitride ceramics. From the XRD pattern
The starting materials were α-Si3N4 (E−10 grade, D50 = 0.33 μm, of the SN and SNC samples, shown in Fig. 2, we can see that the main
oxygen content < 2 wt%, UBE Industries., Ltd., Japan), MgF2 crystalline phase is β-Si3N4 with no other phase present in both sam-
(Sinopharm Chemical Reagent Beijing Co., Ltd., Beijing, China), Y2O3 ples, this indicates that the addition of carbon did not hinder the
(Sinopharm Chemical Reagent Beijing Co., Ltd., Beijing, China). transition from α- Si3N4 to β-Si3N4. The SEM images of the fracture
The compositions of the starting powder mixtures were determined surfaces and polished and plasma-etched surfaces of the SN and SNC
in weight ratios as follows: Si3N4: MgF2: Y2O3 = 94.5 : 3.3: 2.2 and samples are shown in Fig. 3. According to the SEM micrograph, the
Si3N4: MgF2: Y2O3: C = 94.5 : 3.3: 2.2 : 1.18 for the SN and SNC addition of carbon did not result in changes in the microstructure such
samples, respectively. In this work, glucose (C6H12O6, analytically pure, as increase in porosity, this was consistent with the result of bulk
Sinopharm Chemical Reagent Beijing Co., Ltd., Beijing, China) was used density. Fig. 4 shows the Raman spectra of SN and SNC samples, the
as the carbon source. C6H12O6 can be decomposed into highly active SNC sample exhibited characteristic absorption peaks similar to those of
carbon particles under an inert atmosphere at high temperatures and carbon (1349, 1578, 2727 and 2937 cm−1), which confirms the
does not form aggregates. The amount of C6H12O6 added was relatively
small because the purpose of the addition was to promote the elim-
ination of excess grain boundaries. The starting powders were subjected
to planetary milling in ethyl alcohol for 12 h using Si3N4 milling media.
After drying, the dried powders were sieved using 80 mesh grids to
increase the fluidity of powders in a mold. Next, the measured quantity
of the mixed powders was loaded into a 50-mm-diameter graphite die.
Then, they were hot-pressed at 1780°C for 3 h under a 20-MPa pressure
under a flowing N2 atmosphere. Subsequently, the HPS samples were
entirely embedded in the corresponding powder bed in a graphite
crucible: (1) a mixture bed of 50 wt% BN (Purity 97%, Yingkou Liaobin
Fine Chemical Co., Ltd., China) and 50-wt% Si3N4 (E−10 grade,
D50 = 0.33 μm, oxygen content < 2 wt%, UBE industries., Ltd.,
Japan); and (2) a mixture bed of 50-wt% C (Graphite, Purity 99.95%,
Aladdin, China) and 50 wt% Si3N4. The post-sintering heat treatment at
three temperatures (1300°C,1600°C, and 1900°C) was maintained for
3 h under 0.9 MPa N2 pressure. The post-sintering heat treatment
process was conducted in a graphite resistance furnace.
The phase compositions and analysis of the samples were performed
through X-ray diffraction analysis (XRD, D8 Advance, Bruker, Fig. 1. Bulk density and thermal conductivity of the SN and SNC sample.

12607
F. Hu, et al. Ceramics International 46 (2020) 12606–12612

Fig. 2. XRD patterns of the sintered materials; SNC-1900°C represents the Fig. 4. The Raman spectra of SN and SNC samples.
sample of SNC after heat treatment at 1900°C.

exhibits little change within the error range. Combined with the results
presence of carbon in the SNC sample [20,27]. Besides, it can be ob- shown in Fig. 4, the presence of the amorphous carbon may be the main
served that SN and SNC sample have completely different colors, which cause of the decrease of bending strength. As further shown in Fig. 6,
is undoubtedly caused by the addition of carbon. According to the the hardness of the samples was affected by the presence of the
analysis of XRD pattern, the β-Si3N4 is the main crystalline phase, amorphous carbon.
which indicated the residue carbon is present in the SNC sample as
amorphous carbon. The conduction of heat in silicon nitride ceramics is
dominated by phonon transport. Kingery et al. reported that impurities 3.2. Effect of post-sintering heat treatment on the properties of silicon nitride
are one of the important factors that cause lattice scattering. Hence, the ceramics doped with carbon
presence of amorphous carbon leads to a decrease in thermal con-
ductivity of the SNC sample [28]. To facilitate the reaction of the added carbon with the grain
The bending strength of the SN and SNC samples is shown in Fig. 5. boundary glass phase, the post-sintering heat treatment at high tem-
The addition of carbon decreased the bending from 957 to 650 MPa. perature is an optimal method [23]. The SNC sample was entirely
Generally, the bending strength of silicon nitride ceramics depends on embedded in a powder bed comprising 50 wt% BN and 50 wt% Si3N4
microstructures such as porosity and impurity particles [21]. The por- during the post-sintering heat treatment (the corresponding sample was
osity of ceramic materials is inversely proportional to the elastic mod- named SNC(B/S)) at different temperatures (1300°C,1600°C and
ulus; the elastic modulus decreases sharply with increasing porosity 1900°C). In Fig. 8, the bulk density and thermal conductivity of the SNC
[22]. As shown in Fig. 7, the elastic modulus of the SN and SNC samples sample after post-sintering heat treatment at different temperatures are
compared. As shown in Fig. 8, the bulk density of the SNC sample

Fig. 3. SEM pictures on the fracture surfaces of (a) SN and (b) SNC, and the SEM pictures on the polished and plasma-etched surfaces of (c) SN and (d) SNC.

12608
F. Hu, et al. Ceramics International 46 (2020) 12606–12612

Fig. 8. The bulk density and thermal conductivity of sample SNC after heat
Fig. 5. Bending strength of the sintered materials. treatment at different temperatures.

decreased gradually with the increasing temperature of the post-sin-

tering heat treatment. At 1900°C, the bulk density decreased to 2.95 g/
cm3. Simultaneously, the thermal conductivity of the SNC sample in-
creased gradually. However, the growth trend stagnated after 1600°C
because of the low bulk density. XRD patterns of the sample SNC after
the post-sintering heat treatment at 1900°C are presented in Fig. 2 β-
Si3N4 was the only crystalline phase, indicating that no new substance
was produced during the high-temperature post-sintering heat treat-
ment. The bending strength of the SNC sample after the post-sintering
heat treatment at 1900°C was statistically analyzed and plotted in
Fig. 5. The bending strength decreased to 431 ± 118 MPa. The
bending strength scatter plots of the sample showed a wider bending
strength range, which suggested worse material's stability than the one
before post-sintering heat treatment. According to the Griffith Theory of
Brittle Materials, the bending strengths were proportional to the square
root of elastic modulus [24]. This was consistent with the results shown
in Fig. 7 (the value of elastic modulus was 247 GPa).
The SEM micrographs on the fracture surfaces of the SNC sample
after post-sintering heat treatment at 1900°C and EDS analysis of the
whisker in the pore are presented in Fig. 9. Fig. 9(a) shows a low-
Fig. 6. Hardness of the sintered materials. magnification image of the fracture surfaces, and some large pores were
distributed in the sample. These pores were parallel to the compression
surface during the hot pressing; further, the length of the pores can
reach several hundred microns and the width can reach about ten mi-
crons. The high-magnification image of large pores is shown in
Fig. 9(c), and many whiskers with high aspect ratios were stacked in the
large size pores. The EDS pattern (Fig. 9(d)) revealed that the whiskers
contained both Si and N species and that the mole ratio of Si: N was
34.82:65.18. This indicated that silicon nitride was a component of
these whiskers. From the aforementioned analysis, it can be inferred
that carbon and grain boundaries undergo a nitridation reduction re-
action under the post-sintering heat treatment at 1900°C, and the gas
generated by the reaction accumulated inside the sample. Further, be-
cause the temperature was too high, the silicon nitride on the pore walls
was continuously volatilized and deposited, thus generating a large
amount of silicon nitride whiskers. This directly verifies that the addi-
tion of carbon as a sintering aid can affect the excess grain boundary in
silicon nitride ceramics. However, further exploration is required to
solve the problem of larger pores. In Fig. 9(b), high-magnification im-
aged of fracture surfaces were shown. The sample comprised longer
column β-Si3N4 grains, which was in accordance with the XRD patterns.
Besides, some pores could still be observed. The hardness of the sample
SNC after the post-sintering heat treatment was statistically analyzed
Fig. 7. Elastic modulus of the sintered materials. and plotted in Fig. 6. The value of hardness decreased to

12609
F. Hu, et al. Ceramics International 46 (2020) 12606–12612

Fig. 9. The SEM micrographs (a, b, c) on the fracture surfaces of the sample SNC after the post-sintering heat treatment at 1900°C and EDS analysis of the whisker in
pore (d).

12.72 ± 0.84 GPa. The values of the bending strength scatter plots of
the sample showed a polarization trend, which was in accordance with
the microscopic morphology characteristics shown in Fig. 9.

3.3. Effect of the powder bed composition on the properties of silicon nitride
ceramics during the post-sintering heat treatment

To explore the effect of powder bed composition on the properties of

silicon nitride ceramics during high-temperature post-sintering heat
treatment at different temperatures (1300°C,1600°C, and 1900°C), the
SN sample was embedded in two types of powder beds: (1) a mixture
bed of 50 wt% BN + 50 wt% Si3N4 and the sample using the powder
bed during heat treatment was denoted as SN(B/S); (2) a mixture bed of
50 wt% C + 50 wt% Si3N4, and the sample using the powder bed
during heat treatment was denoted as SN(C/S).
As seen in Figs. 10 and 11, the thermal conductivity of the SN
sample embedded in two different kinds of powder beds improved
greatly with increasing heat treatment temperature. Further, the bulk
Fig. 11. The bulk density and thermal conductivity of sample SN after heat
treatment using C/Si3N4 powder bed at different temperatures.

density first increased and subsequently decreased. According to the

analysis, this was because the grain boundary glass phase was volati-
lized during the long-term heat treatment at high temperatures.
Therefore, the bulk density of the sample was lowered. When the heat
treatment temperature was 1900°C, the thermal conductivity of the
sample SN(B/S) after the post-sintering heat treatment process was
observed to reach 103 W m−1 K−1 and the other SN(C/S) sample after
the post-sintering heat treatment process reached 123 W m−1 K−1. The
thermal conductivity values were increased by 33.9% and 59.9%, re-
spectively, compared with the SN sample. Fig. 12 shows the XRD pat-
terns of the SN(B/S) and SN(C/S) samples, indicating that the post-
sintering heat treatment (1900°C) resulted in the crystallization of the
grain boundary glass phase. Moreover, the C/Si3N4 powder bed pro-
moted the crystallization of the grain boundary glass phase more ob-
viously.
Fig. 13 shows the SEM micrographs of the SN(B/S) and SN(C/S)
Fig. 10. Bulk density and thermal conductivity of SN sample after heat treat- samples on the fracture surfaces and polished and plasma-etched
ment using the BN/Si3N4 powder bed at different temperatures.

12610
F. Hu, et al. Ceramics International 46 (2020) 12606–12612

Table 1
Elemental compositions of the SN, SN(B/S) and SN(C/S) samples.
Sample Si (wt%) N (wt%) Y (wt%) O (wt%) Mg (wt%) F (wt%)

SN 61.77 31.08 3.63 1.64 0.772 0.803

SN(B/S) 64.30 30.88 2.23 1.79 0.263 0
SN(C/S) 65.91 31.35 0.884 1.43 0.139 0

other elements (Y, Mg, and O) introduced by the sintering aid were
lower than those of the SN(B/S) sample.
This indicated that in the SN(C/S) sample, under a reducing N2
atmosphere with carbon at 1900°C, the following reactions occur:

3xSiO2· yY2O3·3zMgO + C + (y + z)N2 →

2yYN + zMg3N2 + (6x + 3y + 3z)CO(g) + x Si3N4 (2)

During reaction (1), the liquid phase produced by the grain

boundary glass phase migrated to the surfaces of the sample and the
grain boundary was nitrided to form a surface layer comprising Mg3N2,
Fig. 12. The XRD patterns of the samples SN(B/S) and SN(C/S).
YN, and Si3N4, which can be knocked off after heat treatment [26].
In terms of mechanical properties, when the heat treatment tem-
surfaces. After heat treatment, the aspect ratio of the grains increased perature was 1900°C, the bending strength of the SN(B/S) and SN(C/S)
after a long period of dissolution and precipitation mechanism. By samples reached 988 and 986 MPa (as shown in Fig. 5), respectively. No
comparing the polished and plasma-etched surfaces of the samples after significant change in the bending strength was observed compared with
heat treatment using two different powder beds, the SN(C/S) sample the SN sample. The SN(C/S) sample had a higher hardness value than
was found to have a lower grain boundary glass phase. It is well known SN(B/S) sample (as shown in Fig. 6). The hardness of ceramic materials
that the thermal conductivity of the grain boundary glass phase was was related to the compositions of the material, and the less grain
only 1–10 W m−1 K−1, which is much lower than that of the grains boundary glass phase was bound to higher hardness. In addition, the
(200–300 W m−1 K−1) [25]. Therefore, the silicon nitride ceramic with elastic modulus of SN(B/S) and SN(C/S) samples was greatly improved
lower grain boundary glass phase content will have higher thermal compared with the SN sample (as shown in Fig. 7).
conductivity. Table 1 lists the elemental compositions of the SN, SN(B/
S), and SN(C/S) samples. It can be seen from the table that the Y and Mg
contents in the SN sample were reduced after the heat treatment at 4. Conclusions
1900°C. According to the analysis, the grain boundary glass phase was
indicated to have volatilized under high temperatures, causing the de- In this study, the silicon nitride ceramics were prepared using α-
crease in the contents of Y and Mg in the SN sample. By comparing the Si3N4 powder and Y2O3–MgF2 additives through hot pressing and post-
content of each element of the SN(B/S) and SN(C/S) samples, the sintering heat treatment. The use of carbon as a sintering aid resulted in
contents of nitrogen and silicon in the SN(C/S) samples were found to a decrease in the overall performance of the silicon nitride ceramics,
be higher than those in the SN(B/S) sample; further, the contents of the such as their thermal conductivity and mechanical properties.
However, by changing the composition of the powder bed during the

Fig. 13. SEM pictures of the fracture surfaces of (a) SN(B/S) and (b) SNC(C/S), and the SEM pictures on the polished and plasma-etched surfaces of (c) SN(B/S) and
(d) SNC(C/S).

12611
F. Hu, et al. Ceramics International 46 (2020) 12606–12612

post-sintering heat treatment, these properties were improved. earth oxide additives on the properties of silicon nitride ceramic substrates, Ceram.
Compared with the traditional BN/Si3N4 powder bed, the C/Si3N4 Int. 45 (9) (2019) 12436–12442.
[10] F. Ueno, A. Horiguchi, Grain boundary phase elimination and microstructure of
powder bed improved the thermal conductivity and did not have a aluminum nitride, Proceedings of 1st European Ceramics Society Conference,
significant impact on the bulk density and bending strength of the Elesevier Applied Science, London, New York, 1989, pp. 383–387.
samples. When using the C/Si3N4 bed in the post-sintering heat treat- [11] T. Yagi, K. Shinozaki, M. Kato, Y. Sawada, N. Mizutani, Migration of grain boundary
phases of AlN ceramics in joined sample of sintered and hot-pressed body, J. Ceram.
ment, the grain boundary glass phase reacted with N2 and C, thus Soc. Jpn. 98 (1134) (1990) 198–203.
contributing to the increased thermal conductivity. As a result, the si- [12] K. Watari, M. Kawamoto, K. Ishizaki, Sintering chemical reactions to increase
licon nitride ceramics with a higher thermal conductivity of thermal conductivity of aluminium nitride, J. Mater. Sci. 26 (17) (1991)
4727–4732.
123 W m−1 K−1 and high bending strength of 986 MPa were obtained [13] Y. Li, H.N. Kim, H. Wu, Enhanced thermal conductivity in Si3N4 ceramic by addition
using the C/Si3N4 powder bed during the post-sintering heat treatment. of a small amount of carbon, J. Eur. Ceram. Soc. 39 (2–3) (2019) 157–164.
[14] Yinsheng Li, Ha-Neul Kim, Haibo Wu, Mi-Ju Kim, Jae-Woong Ko, Jin-Myung Kim,
Young-Jo Park, Zhengren Huang, Hai-Doo Kima, Improved thermal conductivity of
Declaration of competing interest
sintered reaction-bonded silicon nitride using a BN/graphite powder bed, J. Eur.
Ceram. Soc. 37 (15) (2017) 4483–4490.
The authors declare the following financial interests/personal re- [15] You Zhou, Hideki Hyuga, Dai Kusano, Yu‐ichi Yoshizawa, Kiyoshi Hirao, A tough
lationships which may be considered as potential competing interests: silicon nitride ceramic with high thermal conductivity, Adv. Mater. 23 (39) (2011)
4563–4567.
[16] You Zhou, Hideki Hyuga, Dai Kusano, Chika Matsunaga, Kiyoshi Hirao, Effects of
Acknowledgements yttria and magnesia on densification and thermal conductivity of sintered re-
action‐bonded silicon nitrides, J. Am. Ceram. Soc. 102 (4) (2019) 1579–1588.
[17] Yinsheng Li, Ha‐Neul Kim, Haibo Wu, Mi‐Ju Kim, Jae‐Woong Ko, Young‐Jo Park,
This work was financially supported by the National Key Research Zhengren Huang, Hai‐Doo Kim, Enhanced thermal conductivity in Si3N4 ceramic
and Development Program of China (Grant no. 2017YFB0310400), the with the addition of Y2Si4N6C, J. Am. Ceram. Soc. 101 (9) (2018) 4128–4136.
National Natural Science Foundation of China (Grant no. 51802230, [18] Young-Jo Park, Mi-Jung Park, Jin-Myung Kim, Jae-Wook Lee, Jae-Woong Ko, Hai-
Doo Kim, Sintered reaction-bonded silicon nitrides with high thermal conductivity:
51672147 and 51427802). the effect of the starting Si powder and Si3N4 diluents, J. Eur. Ceram. Soc. 34 (5)
(2014) 1105–1113.
Appendix A. Supplementary data [19] C.V. Rocha, C.A. Costa, Fracture toughness of Si3N4 processed by gas pressure
sintering and hot pressing, Mater. Res. 9 (2) (2006) 143–146.
[20] P. Rutkowski, L. Stobierski, G. Górny, Thermal stability and conductivity of hot-
Supplementary data to this article can be found online at https:// pressed Si3N4–graphene composites, J. Therm. Anal. Calorim. 116 (1) (2014)
doi.org/10.1016/j.ceramint.2020.02.024. 321–328.
[21] Yao Han, Shuang Li, Tianbin Zhu, Weiwei Wu, An Di, Feng Hu, Fengrui Zhai,
Zhipeng Xie, Enhanced toughness and reliability of Si3N4-SiCw composites under
References oscillatory pressure sintering, Ceram. Int. 44 (11) (2018) 12169–12173.
[22] Paul F. Becher, Chun-Hway Hsueh, Peter Angelini, Terry N. Tiegs, Toughening
[1] H. Okumura, Present status and future prospect of widegap semiconductor high- behavior in whisker‐reinforced ceramic matrix composites, J. Am. Ceram. Soc. 71
power devices, Jpn. J. Appl. Phys. 45 (10R) (2006) 7565. (12) (1988) 1050–1061.
[2] H. Lu, C. Bailey, C. Yin, Design for reliability of power electronics modules, [23] Thanakorn Wasanapiarnpong, Shigetaka Wada, Masamitsu Imai, Toyohiko Yano,
Microelectron. Reliab. 49 (9–11) (2009) 1250–1255. Effect of post-sintering heat-treatment on thermal and mechanical properties of
[3] P.F. Becher, Microstructural design of toughened ceramics, J. Am. Ceram. Soc. 74 Si3N4 ceramics sintered with different additives, J. Eur. Ceram. Soc. 26 (15) (2006)
(2) (1991) 255–269. 3467–3475.
[4] Y. Inagaki, N. Kondo, T. Ohji, High performance porous silicon nitrides, J. Eur. [24] A.A. Griffith, VI, The phenomena of rupture and flow in solids, Philos. Trans. R. Soc.
Ceram. Soc. 22 (14–15) (2002) 2489–2494. Lond. - Ser. A Contain. Pap. a Math. or Phys. Character 221 (582–593) (1921)
[5] Jihong She, Jian‐Feng Yang, Daniel Doni Jayaseelan, Naoki Kondo, Tatsuki Ohji, 163–198.
Shuzo Kanzaki, Yoshiaki Inagaki, Thermal shock behavior of isotropic and aniso- [25] Naoto Hirosaki, Shigenobu Ogata, Cenk Kocer, Hiroshi Kitagawa,
tropic porous silicon nitride, J. Am. Ceram. Soc. 86 (4) (2003) 738–740. Yasuhiro Nakamura, Molecular dynamics calculation of the ideal thermal con-
[6] J.S. Haggerty, A. Lightfoot, Opportunities for enhancing the thermal conductivities ductivity of single-crystal α-and β− Si3N4, Phys. Rev. B 65 (13) (2002) 134110.
of SiC and Si3N4 ceramics through improved processing, Ceram. Eng. Sci. Proc. 16 [26] Akihiro Horiguchi, Fumio Ueno, Mitsuo Kasori, Yoshiko Sato, Masaru Hayashi,
(1995) 475–487. Hiroshi Endo, Kazuo Shinozaki, Akihiko Tsuge, AlN sintered body having high
[7] Mikito Kitayama, Kiyoshi Hirao, Motohiro Toriyama, Shuzo Kanzaki, Thermal thermal conductivity and a method of fabricating the same, U.S. Patent 4 (847)
conductivity of ß‐Si3N4: I, effects of various microstructural factors, J. Am. Ceram. (1989) 221 7-11.
Soc. 82 (11) (1999) 3105–3112. [27] H.H. Lu, C.Y. Chen, Investigation of nano-silicon nitride ceramics containing an
[8] Mikito Kitayama, Kiyoshi Hirao, Akira Tsuge, Koji Watari, Motohiro Toriyama, yttria sintering additive and the carbon thermal reduction reaction, Ceram. Int. 42
Shuzo Kanzaki, Thermal conductivity of β‐Si3N4: II, effect of lattice oxygen, J. Am. (10) (2016) 12452–12459.
Ceram. Soc. 83 (8) (2000) 1985–1992. [28] W.D. Kingery, The thermal conductivity of ceramic dielectrics, Prog. Ceram. Sci. 2
[9] Wei Liu, Wenxin Tong, Xinxin Lu, Shanghua Wu, Effects of different types of rare (5) (1962) 181.

12612