Grain Boundary Control to Obtain Heat-Resistant Silicon Nitride

Abstract

Silicon nitride ceramics are composed of silicon nitride grains and a grain boundary phase. The grain boundary phase melts or becomes a mass transfer path at high temperatures, even if the temperature is lower than the decomposition temperature of silicon nitride. The high-temperature mechanical properties are strongly affected by the grain boundary phase. The formation of heat-resistant grain boundaries is a key factor in obtaining heat-resistant silicon nitride ceramics. The grain boundary phase is formed by the solidification of the liquid phase during sintering. The liquid phase is formed by the reaction of sintering additives with oxide impurities on the surface of silicon nitride particles. Improvement of the high-temperature properties of silicon nitride ceramics by controlling the grain boundary phase is explained in this chapter.

3.1 Addition of Yb₂O₃ and Lu₂O₃ to Silicon Nitride Ceramics as Sintering Additives

Silicon nitride ceramics are composed of silicon nitride grains and a grain boundary phase, as mentioned in Sect. 1.1. The grain boundary phase softens and/or becomes a mass transfer path at high temperatures. The formation of a heat-resistant grain boundary phase is necessary for to obtain heat-resistant silicon nitride ceramics, which means that control of the addition of sintering additives is important.

Yttrium oxide (Y₂O₃) is added as a sintering additive to obtain heat-resistant silicon nitride ceramics [1]. When Y₂O₃ and SiO₂ are added together, yttrium silicates, Y₂SiO₅ and Y₂Si₂O₇, which have high melting temperatures, are crystallized at multigrain junctions by heat treatment after sintering. Although such crystallization is effective for improving the heat resistance of silicon nitride ceramics, decreased strength and heat resistance have been reported. Probable reasons for the degradation are the existence of an amorphous layer between the crystalline silicate phase and the silicon nitride grains, and residual stress induced by the difference in thermal expansion between the silicates and silicon nitride [2]. These results suggest that the silicates are not effective for improving the heat resistance of silicon nitride ceramics. Various rare-earth silicon oxynitrides exist with the phase relationship RE₂O₃–SiO₂–Si₃N₄ (RE: Sc, Y, and rare-earth elements) (Fig. 3.1). Among the oxynitrides, RE₄Si₂O₇N₂-type and RE₂Si₃O₃N₄-type oxynitrides have high melting temperatures [3] and their chemical composition is closer to that of silicon nitride (Si₃N₄) than silicates because they contain nitrogen (N). They are thus expected to be effective additives to promote the growth of a crystalline grain boundary phase to achieve heat-resistant silicon nitride. The crystallization of Y₂Si₃O₃N₄ was reported by Tsuge et al., but the formation of Y₂Si₃O₃N₄, which lies on the tie line between Y₂O₃ and Si₃N₄, in the phase diagram of Y₂O₃–SiO₂–Si₃N₄ is difficult, because SiO₂ always exists on the surface of Si₃N₄ particles [4]. Y₄Si₂O₇N₂ cannot coexist with silicon nitride under the equilibrium state, as shown in the phase diagram of Y₂O₃–SiO₂–Si₃N₄ (Fig. 3.1a) [5]. Yb₄Si₂O₇N₂ and Lu₄Si₂O₇N₂ can coexist with Si₃N₄ from the phase diagram relationships of Yb₂O₃–SiO₂–Si₃N₄ (Fig. 3.1b) [3] and Lu₂O₃–SiO₂–Si₃N₄ (Fig. 3.1c) [6]. A To investigate the effect of additives on the properties of silicon nitride ceramics, a silicon nitride ceramic including Yb₄Si₂O₇N₂ and silicon nitride ceramics including Lu₄Si₂O₇N₂ were synthesized and their high-temperature mechanical properties were measured.

Fig. 3.1

Phase diagrams among Si₃N₄–SiO₂–Y₂O₃ (a), Si₃N₄–SiO₂–Yb₂O₃ (b), and Si₃N₄–SiO₂–Lu₂O₃ (c)

3.2 Sintering of Silicon Nitride Ceramics

A silicon nitride powder, RE₂O₃ powder (RE:Yb or Lu), and SiO₂ powder were weighed and mixed in n-hexane by planetary milling using a silicon nitride pot and balls. The mixed powders were hot-pressed in a flow of nitrogen gas for Yb addition or at a nitrogen pressure of 1 MPa for Lu addition under a compressive load. Figure 3.2 shows the shrinkage of the Si₃N₄–Yb₄Si₂O₇N₂ powder during hot pressing. Densification curves of the powder, which consists of of 94.1 mol% Si₃N₄, 4.7 mol% Yb₂O_3, and 1.2 mol% SiO_2, are shown in the figure. The powder densified gradually from 1550 °C, rapidly above 1650 °C, and even more rapidly until the middle of the soaking time at 1750 °C [7].

Fig. 3.2

Densification of Si₃N₄ ceramic containing 4.7 mol% Yb₂O₃ and 1.2 mol% SiO₂ during hot pressing [7]

Figure 3.3 shows the shrinkage of Si₃N₄–Lu₄Si₂O₇N₂ ceramics during hot pressing. Densification starts at about 1570 °C, regardless of the additive content. The ceramics containing 1.2 and 4.8 mol% Lu₂O₃ (Lu12 and Lu48, respectively) were almost completely densified, although the ceramic with 0.6 mol% Lu₂O₃ (Lu06) was not densified after heating to 1950 °C for 1 h.

Fig. 3.3

Densification of silicon nitride ceramics with 0.6, 1.2, and 4.8 mol% Lu₂O₃ during hot pressing

3.3 High-Temperature Strength of Silicon Nitride Ceramics

The dependence of the bending strength of silicon nitride ceramics with Yb₄Si₂O₇N₂ on the temperature is shown in Fig. 3.4 along with the results of previous investigations [2, 8, 9]. The bending strength was 977 MPa at room temperature (RT) and decreased slightly with increasing temperature up to 1350 °C. The strength at 1350 °C of 728 MPa is about 75 % of that at RT. The strength was 484 MPa at 1500 °C. Smith and Quackenbush reported the strengths of hot-pressed silicon nitride ceramics containing Y₂O₃; the strengths of silicon nitride including 6 mass% of Y₂O₃ and 13 mass% of Y₂O₃ at 1500 °C were 380 and 500 MPa, respectively [10], comparable with our data. The load-displacement curves obtained in the bending strength test at 1200, 1350, and 1500 °C are shown in Fig. 3.5. The load-displacement curve is almost straight, indicating no yielding before fracture, which suggests that the fracture was brittle at each temperature. A report on the effect of oxides added to gas-pressure-sintered silicon nitride showed that adding Nd₂O₃ or La₂O₃ together with Y₂O₃ was more effective in increasing the high-temperature strength than adding MgO or Al₂O₃ together with Y₂O₃. The load-displacement curves for the materials including Nd₂O₃ and Y₂O₃ at 1200 and 1350 °C indicated brittle fracture, and the degradation of strength and yielding were clearly observed at 1500 °C.

Fig. 3.4

High-temperature bending strength of silicon nitride with Yb₄Si₂O₇N₂. The bending strengths of silicon nitride with ytterbium silicates in previous reports are also plotted [2, 8, 9]

Fig. 3.5

Load-displacement curves of silicon nitride with Yb₄Si₂O₇N₂ obtained in bending strength test [9]

The strength of the Si₃N₄ ceramic with 1.2 mol% Lu₂O₃ added was about 1 GPa at RT, which decreased gradually up to 1200 °C, then decreased sharply up to 1600 °C. The strength of the Si₃N₄ ceramic with 4.8 mol% Lu₂O₃ added Si₃N₄ ceramics changed in a similar manner but the strength at 1350 °C was the same as that at 1500 °C. The strength at 1600 °C was 400 MPa for the ceramic with 1.2 mol% Lu₂O₃ added and 450 MPa for the ceramic with 4.8 mol% Lu₂O₃ added (Fig. 3.6). This result suggests that the Si₃N₄ ceramic including 4.8 mol% Lu₂O₃ is more heat resistant than the Si₃N₄ ceramic including 1.2 mol% Lu₂O₃ [11].

Fig. 3.6

High-temperature strength of silicon nitride ceramics with 1.2 mol% Lu₂O₃ (12Lu) and 4.8 mol% Lu₂O₃ (48Lu) [11]

Figure 3.7 shows the load-displacement relationships obtained from a bending test carried out on Si₃N₄ ceramics with different amounts of added Lu₂O₃ at 1600 °C. The load-displacement curve of the ceramic with 1.2 mol% Lu₂O₃ is almost straight, but that of the ceramic with 4.8 mol% Lu₂O₃ is convex, indicating that it deformed plastically before fracture. These materials can be used up to the stress at which plastic deformation starts, which is much lower than the apparent strength. Figures 3.6 and 3.7 suggest that the ceramic with 1.2 mol% Lu₂O₃ is more heat resistant than that with 4.8 mol% Lu₂O₃, although the apparent strength of the ceramic with 4.8 mol% Lu₂O₃ is higher than that of the ceramic with 1.2 mol% Lu₂O₃.

Fig. 3.7

Load-displacement curves of silicon nitride ceramics sintered with Lu₂O₃ obtained in bending test at 1600 °C. Lu06, Lu12, Lu24, and Lu48 indicate the load-displacement curves for silicon nitride with 0.6, 1.2, 2.4, and 4.8 mol% Lu₂O₃, respectively

3.4 High-Temperature Creep of Silicon Nitride Ceramics

Tensile creep tests were conducted on a silicon nitride ceramic containing Yb₂O₃ and on silicon nitride ceramics containing Lu₂O₃. The silicon nitride ceramic with Yb₂O₃ was fabricated from a powder mixture of 97.6 mol% α-Si₃N₄ and 2.4 mol% Yb₄Si₂O₇N₂ (denoted as Yb) by sintering at 1750 °C for 1 h under compression of 20 MPa in nitrogen gas at a pressure of 0.925 MPa. The silicon nitride ceramics with Lu₂O₃ consisted of 95.2 mol% Si₃N₄ and 4.8 mol% Lu₂O₃ (denoted as Lu48) and 98.8 mol% Si₃N₄ with 1.2 mol% Lu₂O₃ (designated as Lu12). Both ceramics were sintered at 1950 °C for 1 h undercompression of 20 MPa in nitrogen gas at a pressure of 1 MPa. Tensile creep tests were then conducted at 1400 and 1500 °C in air. Figure 3.8 shows the results of the tensile creep test at 1500 °C for the silicon nitride ceramic with 4.8 mol% Lu₂O₃ (Lu48). The curves are typical for silicon nitride ceramics, i.e., transient creep was followed by steady-state creep but the tertiary stage was not observed. The creep rate increases with increasing applied stress [12].

Fig. 3.8

Typical creep curves of silicon nitride with 4.8 mol% Lu₂O₃ (Lu48) at 1500 °C [12]

According to Norton’s law, the creep rate in the steady state, $\dot{\varepsilon }$, is expressed as a power function of stress σ as follows:

$$\dot{\varepsilon } = \mathop D\nolimits_{0} \exp \left( { - \frac{Q}{RT}} \right)\mathop \sigma \nolimits^{n}$$

(3.1)

where n is the stress exponent, R is the gas constant, T is the absolute temperature, and Q is the activation energy. The relationship between the minimum strain rate and the applied stress for Lu48, together with the data for silicon nitride sintered with 2.4 mol% Yb₄Si₂O₇N₂ (Yb), is shown in Fig. 3.9. The minimum strain rate of Lu48 at 1500 °C increases monotonically and that at 1400 °C increases slightly up to 250 MPa then rapidly above 250 MPa. The value of n is calculated assuming that the creep rate in the steady state is the minimum strain rate. The minimum strain rate of the fractured samples is used because the minimum strain rate obtained from interrupted test data may not be that of the steady state. The value of n for Lu48 is 6.8 at 1400 °C and 7.7 at 1500 °C. The value for Yb is 3.1 at 1300 °C, calculated from interrupted test data, and 1.7 at 1400 °C. Comparing Lu48 and Yb, Lu48 is more sensitive to stress than Yb [13].

Fig. 3.9

Time to failure and strain rate for Lu48 and Yb at 1400 and 1500 °C [12]

Figure 3.10 shows the effect of the applied stress on the time to failure of Lu48 at 1400–1550 °C, together with the data for Yb and SN281, which denotes Lu₂O₃-containing silicon nitride produced by Kyocera [14]. The time to failure decreases with increasing applied stress. Comparing the data obtained at 1400 °C, the plots of Lu48 are much higher than those of Yb, meaning that Lu48 is more creep-resistant than Yb. The Lu48 data obtained at 1400 °C almost overlaps that of SN281 data obtained at 1450 °C. Also, the Lu48 data obtained at 1500 °C is below the SN281 data obtained at 1500 and 1550 °C.

Fig. 3.10

Time to failure and applied stress for Lu48, Yb, and SN281 at 1400–1550 °C [12]

The time to failure t _f may be described as a power function of the stress σ:

$$t_{f} = B\sigma^{ - N}$$

(3.2)

where N is a parameter describing the stress-rupture behavior. The value of N for Lu48 is 10.7 at 1400 °C and 11.1 at 1500 °C. The value for Yb was not calculated because of the large scattering of the data [15]. The value of N for SN281 is 8.58 at 1450 °C, which was obtained using the data in Ref. 14. Lu48 fractured at 1400 °C under a stress of 200 MPa after 1000.5 h, indicating that Lu48 can sustain creep for over 1000 h at 1400 °C under a stress of 137 MPa. Lu12 contains the smallest amount of additive among the ceramics investigated. The creep resistance of Lu12 was expected to be higher than that of Lu48. Lu12 sustained creep for 1678.5 h at 1500 °C under a stress of 137 MPa [14]. This result shows that Lu12 sustains creep for 1000 h at 1500 °C under a stress of 137 MPa, which is the goal of a project of National Institute for Materials Science (NIMS) for developing heat-resistant silicon nitride ceramics. The minimum strain rate could not be calculated because the displacement was below the minimum measurable value, particularly below 1500 °C.

The relationship between the applied stress and the time to failure is shown for Lu12 in Fig. 3.11. The time to failure tends to increase with decreasing applied stress. Plots are typically distributed from 0.08 to 228 h with a change in applied stress from 160 to 180 MPa at 1400 °C. At 1500 °C, the time to failure is distributed from 0.1 to 690 h with a change in applied stress from 145 to 165 MPa. The high sensitivity of the time to failure to the applied stress and the inability to detect deformation during creep tests suggest that the fracture of Lu12 is dominated by subcritical crack growth, not creep rupture, particularly at 1400 °C. The stress-rupture parameter N is 41 at 1400 °C and 73 at 1500 °C, larger than the values for Lu48 (10.7 at 1400 °C and 11.1 at 1500 °C). The minimum strain rate of SN281 has been reported to be on the order of 10⁻¹⁰–10⁻¹¹ s⁻¹ at 1400 °C and 10⁻⁹ s⁻¹ at 1450 °C and above, much smaller than that of Lu48. No cavitation was detected in the creep specimens by transmission electron microscopy (TEM). A non-cavitation creep mechanism has previously been proposed [14]. Lu12 is not deformed by creep and is fractured by subcritical crack growth at 1400 °C. The characteristic features of Lu12 are its high sensitivity of the time to failure to the applied stress and its low fracture strength during subcritical crack growth. These features are not favorable as a structural material.

Fig. 3.11

Time to failure and applied stress for Lu12 at 1400 and 1500 °C [11]

3.5 Conclusion

The use of grain boundary control to improve the heat resistance of silicon nitride ceramics was explained. The high-temperature properties of silicon nitride ceramics sintered with RE₂O₃ (RE: Yb and Lu), which were selected with the aim of achieving grain boundary control, were measured. The heat resistance of silicon nitride with Lu₂O₃ was greater than that of silicon nitride with Yb₂O₃ and SiO₂. The silicon nitride ceramic including a small amount of Lu₂O₃ (Lu12) had a lower strength at 1500 and 1600 °C than the silicon nitride ceramic with a large amount of Lu₂O₃ (Lu48). However, the creep resistance of Lu12 was higher than that of Lu48. These results suggest two important points. One is that a small amount of the grain boundary phase is more favorable for increasing heat resistance. Lu12 has lower strength at 1500 and 1600 °C but its fracture was still brittle, its creep rate at 1400 and 1500 °C was too small to be detected, and its fracture during the tensile test might have been caused by, subcritical crack growth rather than creep. The oxidation resistance of Lu12 was higher than that of Lu48 [13]. When we consider densification, a certain amount of sintering additive is needed for complete densification. High-temperature properties, creep and oxidation resistance, are incompatible with sinterability from the viewpoint of sintering additives. The other point is that high-temperature strength is not always an indicator of heat resistance; although Lu12 had higher creep resistance than Lu48, Lu48 had higher high-temperature strength than Lu12. We must measure the creep to obtain information on heat resistance, considered here to be stability at a high temperature for a long time.