Ceramic Fibers and Coatings: Advanced Materials for the Twenty-First Century (1998)

The performance objectives described in the previous chapters limit the material choices of ceramic fibers for CMCs to polycrystalline oxides (e.g., Al₂O₃, mullite), non-oxides (e.g., SiC, Si₃N₄), and amorphous Si-C-N-B-O compositions. Single-crystal monofilaments have certain performance advantages, but their cost is prohibitive. Therefore, they are not discussed in this chapter.

Once fabrication flaws have been eliminated, the strain-to-failure of ceramic fibers is dominated by grain size (d), scaling as ~ d^-1/2. Typically, the grain size must be 0.1µm to 0.5µm (0.004 to 0.02 mils) to achieve the strain-to-failure goal of 0.6 percent for ceramic fibers (Evans, 1997). The creep rate (e') of fine-grained polycrystalline ceramics typically increases as the grain size decreases because of the dominate diffusive deformation mechanisms in these materials. In single-phase ceramics, the creep rate is controlled by the following relationship:

e'~D_b/dⁿ (1)

D_b is the diffusivity, and n is an exponent between 1 and 3, depending on the dominant creep mechanism (e.g., n = 3 for grain boundary diffusion). However, these mechanisms are not sufficiently understood for creep rates to be explicitly predicted from the grain size and diffusivity of the material. The recent experimental findings described below highlight this deficiency and also indicate the opportunity for developing polycrystalline materials that combine high tensile strength with excellent creep resistance.

To facilitate the discussion of potential improvements in fiber properties through microstructural enhancement, some recent creep studies (and the associated mechanistic understanding) are summarized below. Ceramic fibers consisting of mullite and alumina mixtures (e.g., Nextel 720 fibers) have demonstrated creep strengths considerably greater than those observed in single-phase alumina or mullite fibers (Chapter 3). In general, two-phase mixtures exhibit less grain growth than single-phase materials and tend to be more microstructurally stable (French et al., 1990). However, the mechanism by which multiphase ceramic oxide fibers resist creep rupture is not understood. The microstructure of the Nextel 720 fiber consists of needle shaped mullite grains and elongated a-Al₂O₃ grains, suggesting that grain morphology plays a key role in creep retardation. For example, calculations have implied that grain elongation along the fiber axis can enhance creep strength, when diffusion mechanisms dominate creep rate (Figure 5-1) (Sabol, 1994).

Adding SiC nanoparticles to bulk samples of alumina has been shown to enhance the creep strength of alumina (see Figure 5-2), particularly when the SiC particles are preferentially located at the alumina grain boundaries (Ohji et al., 1995). It has been postulated that the SiC particles inhibit grain boundary sliding, setting up a “back stress” that resists creep. It has also been inferred, from sintering studies, that SiC particles decrease the overall grain boundary diffusion rate and that creep inhibition is due to slow diffusion along the alumina/silicon carbide interface (Stearns et al., 1992). However, the fundamental mechanisms whereby SiC particles inhibit creep are not known.

The addition of yttrium and other oversized isovalent cations to alumina has also been shown to enhance creep strength (Figure 5-3) (French et al., 1994). The oversized ions segregate to the alumina grain boundaries (Figure 5-4). One

FIGURE 5-1 Predicted creep rates for alumina fibers as a function of aspect ratio using a two-dimensional model. Source: Sabol, 1994.

FIGURE 5-2 Tensile creep rate of Al₂O₃-SiC nanocomposite containing 5 volume percent of 0.15 µm (0.006 mils) SiC particles and undoped Al₂O₃ of the same grain size. The hatched band represents flexural strain rate data for SiC whisker-reinforced alumina Source: Thompson et al., 1997b.

hypothesis is that the retardation in creep results from inhibition of the grain boundary diffusion kinetics by a simple “steric hindrance ” effect. There is also some evidence in the literature that the presence of dopant ions can significantly increase the proportion of special (i.e., coincident site lattice) grain boundaries in alumina (Lartigue-Korinek et al., 1994). Perhaps the diffusivities along these boundaries are diminished, resulting in grain boundary sliding rates that are also diminished. A comprehensive understanding of the mechanism by which oversized isovalent cations inhibit creep, however, remains elusive.

Although the findings described above regarding creep in fine-grained polycrystals have been limited to oxides, it is anticipated that analogous effects are possible for non-oxide ceramics. For example, certain additives and second-phase particles (e.g., Ti, C, and B ₄C) enhance the creep strength of polycrystalline SiC, while others (e.g., B and Al) degrade it (recall Chapter 3). The mechanisms, however, are not understood.

The development of an amorphous fiber in the Si-N-C-B system has revealed two unexpected possibilities. First, this fiber appears to be stable in the amorphous state up to ~1,600°C (2,912°F), even in air, and retains its tensile strength up to 1,600°C (2,912°F). Second, oxidation of the Si-N-C-B fiber forms silica at rates comparable to the rate that silica is formed upon oxidation of crystalline SiC, and also creates a thin (~50 nm [0.002 mils]) buried interphase of hexagonal BN. It has been hypothesized that this in-situ BN would act as a crack deflection layer in a CMC. If so, the regenerative nature of the deflection interphase could avert composite degradation by stress oxidation.

These findings imply that there is a realistic potential for discovering materials with superior creep strength, through a mechanism-based research approach. Complex oxides and carbides, as well as multiphase materials, could be tailored to create attractive combinations of properties.

Fine-grained (grain size < 1µm [0.04 mils]) oxide poly-crystals creep in accordance with diffusive deformation mechanisms, accompanied by sliding of the grain boundaries. (Note that the stress/temperature domain in which dislocation is the dominant creep mechanism has yet to be rigorously identified). Therefore, if all other parameters remained fixed, the creep rate would simply scale with the grain boundary diffusivity (D_b) and vary inversely with the grain size (d) to the first, second, or third power (n) depending on whether the process is controlled by interface control (or sliding), lattice diffusion, or grain boundary diffusion (see Equation 1).

One implication that has been adequately validated by experiment is that complex oxides with low D_b, particularly mullite, have creep strengths superior to single-phase oxides, such as alumina, yttria, and magnesia. However, the findings that solid solutions and nanoparticles profoundly affect creep rates in bulk polycrystalline oxides was not obvious from

FIGURE 5-3 The steady-state tensile creep rate for undoped alumina and alumina doped with 1,000 ppm Y₂O₃. Note that yttria doping dramatically reduces the creep rate. Source: French et al., 1994.

FIGURE 5-4 High resolution secondary ion mass spectroscopy compositional maps showing dopant segregation in yttrium and lanthanum doped alumina. Source: Chabala et al., 1997.

prior knowledge of the creep mechanism. The current theory is that (1) some solid solutions diminish the diffusivity, and (2) some intergranular nanoparticles inhibit the sliding and rotation of the grain boundaries. The specific mechanisms that dictate the diffusivity and sliding are not clear, but the findings are compelling and suggest trajectories for future research and development.

The best creep results on nanoparticle-reinforced alumina have been achieved with intergranular SiC (at about 5 volume percent). The effect of SiC nanoparticles was unexpected because simple theory indicates that diffusion would occur rapidly around SiC particles located along sliding grain boundaries because of their small size. However, there is no evidence that this occurs. The mechanism(s) responsible for this counterintuitive effect need to be understood so this behavior can be realized in other systems. Despite the promising results indicating excellent creep resistance in Al₂O₃ reinforced by SiC nanoparticles, SiC nanoparticle-reinforced Al₂O₃ does not appear to be appropriate for use as a fiber for reinforcement of CMCs subject to oxidizing conditions. That is, the oxidation characteristics of this material are problematic. Oxidation measurements indicate that a silica outer layer forms on Al₂O₃ reinforced with SiC nanoparticles at a rate comparable to the rate it would form on SiC. However, a much thicker SiO₂ layer forms below the surface of SiC nanoparticle-reinforced Al₂O₃, wherein the SiC particles have a modified chemistry and morphology. Such a thick SiO₂ layer would result in the degradation of a fiber made of this material. The benefit of creep strengthening with nanoparticles could be realized without oxidative degradation, if oxide ceramics were reinforced with oxide ceramic nanoparticles. The committee recommends that this possibility be investigated.

Certain solutes can also have a profound, beneficial effect on creep resistance. For example, it has recently been discovered that the addition of small quantities (less than 1,000 ppm) of rare earth dopants, such as yttrium, lanthanum, or neo-dymium, to fine grained Al₂O₃ (1µm to 2µm [0.04 to 0.08 mils] grain size) can reduce the creep rate by several orders of magnitude. The effect seems to be strongly correlated with the fact that these ions are highly oversized for the alumina lattice and thus segregate very strongly to the grain boundaries. Other isovalent dopant ions, such as chromium and iron, which are closer in size to aluminum and do not segregate strongly to the grain boundaries, do not affect creep strength.

The mechanism for creep reduction is not understood although the relationship between dopant concentration and creep resistance has been confirmed to be a true solid-solution effect. One school of thought is that creep is controlled by grain boundary diffusion and that large segregated ions simply hinder diffusion in the core sites at the boundary. This theory is supported by kinetic measurements of self-diffusion in undoped and yttria-doped alumina and by studies of the oxidation of aluminum alloys (Le Gall et al., 1995). There is

also some evidence to suggest that yttria may promote the formation of special (i.e., high coincidence site lattice) grain boundaries in alumina (Lartigue-Korinek et al., 1994). This theory argues that grain boundary diffusion would be retarded along these boundaries because of their greater structural order, which would lead to slower creep kinetics. The committee recommends that research be continued on the effect of isovalent doping on the creep properties of oxide ceramics.

Much of the work on doping has been performed on bulk materials with grains several times larger than those present in most fibers. An important aspect of future research should be to determine the extent to which these findings translate to fibers with ultrafine grains.

In general, when material deformation is dominated by diffusion, fibers may be strengthened by texturing and elongating grains along the fiber axis. Methods for developing these microstructures should be investigated in the future.

The creep mechanisms in fine-grained polycrystalline SiC should be the same as the mechanisms that operate in fine-grained oxides. The major difference is the inherently low grain boundary diffusivity of SiC compared to most oxides. Consequently, the baseline creep properties of SiC, before either solid-solution or particle strengthening, are superior to those of oxide ceramics. The only information available regarding solid-solution or particle strengthening of SiC ceramics, however, concerns an apparent solid-solution weakening effect caused by boron-doping, and an apparent particle strengthening caused by excess carbon and, possibly, titanium diboride. It seems that the potential of these mechanisms for creep strengthening has not been explored at all. The committee recommends this area as a focus of future research.

As opportunities for enhanced creep strength are explored, it is essential that the chemistries be retained within a domain that ensures that oxidation rates are comparable to, or better than, the rates for nominally pure SiC. Otherwise oxidative degradation of the composite will reemerge as the performance limiting mechanism. The rupture strength of polycrystalline SiC fibers at elevated temperatures has not been evaluated comprehensively and should be further investigated. A few systematic studies on the effect of heat treatment (aging) on rupture strength have been conducted, but this area is worthy of future research.

The development of amorphous fibers composed of Si, C, B, and N with exceptional creep resistance and thermal stability (against crystallization) has been reported by several domestic and foreign researchers independently (Baldus, 1997). These reports suggest that the creep resistance of amorphous Si, C, B, and N (Si-C-B-N) fibers are comparable to stoichiometric, crystalline SiC. The committee recommends that research be undertaken to verify these results and to develop a basic understanding of the Si-C-B-N material. The object of this research is to find ways to produce these fibers using less expensive precursors. A successful outcome would be the development of an affordable fiber with the same performance characteristics as crystalline SiC. However, confirming the utility of this fiber for CMCs would require that the following issues be rigorously investigated:

• susceptibility to crystallization at long exposure to high temperatures, including the influence of oxidation, temperature, and stress (the expectation is that the onset of crystallization would cause rapid degradation)

• stability of the surface oxide and of the in-situ BN layer at intermediate temperatures (for many non-oxides, intermediate temperatures are the most critical for property retention)

• creep viscosity

The properties of ceramic fibers are dominated by fabrication flaws and fiber microstructure. These two factors are not mutually exclusive and have been discussed in this report in several different contexts. Recommendations pertinent to fiber processing to reduce fabrication flaws and develop tailored microstructures were discussed in Chapter 4. Here the committee recommends directions for research that are likely to determine the microstructures required for attaining the properties that would facilitate the use of ceramic fibers in advanced high-temperature CMCs.

Oxide fibers (which are inherently resistant to oxidation) with improved creep resistance will, with the concurrent development of suitable fiber-matrix interfaces, enable CMCs to be used in higher-temperature applications in oxidative environments (e.g., gas turbine engine exhaust nozzles and heat exchangers for externally-fired combined-cycle power systems). Therefore, the committee recommends that the following research be supported:

• determining the mechanism by which certain solutes lower the creep rate of bulk polycrystalline oxides (e.g., yttria in alumina) and applying them to oxide fibers and determine if analogous effects exist in non-oxide systems

• establishing the mechanism(s) by which SiC nanoparticle dispersions inhibit creep in oxide ceramics and developing oxide nanoparticle systems analogous to the alumina (SiC) nanocomposite system

• exploring and exploiting multiphase microstructures for enhancing high-temperature microstructural

stability (e.g., resistance to grain growth) and improving creep resistance in both oxide and non-oxide systems

Non-oxide fibers, which are inherently more creep resistant than oxide fibers, could provide the higher-temperature performance needed for near-term CMC applications (e.g., furnace hardware, such as pipe hangers). But a systems approach to preventing oxidative degradation (pesting) of nonoxide CMCs will have to be developed. With further improvements in the creep resistance of non-oxide fibers, they could be used in longer-term applications (e.g., gas turbine engine combustor liners). Therefore, the committee recommends that research in the following areas be supported:

• verification and study of the mechanisms of creep enhancement, stability against crystallization, and oxidation resistance in amorphous Si-C-B-N fibers and the exploration of other families of amorphous non-oxide fiber systems

• the development of new materials conducive to the formation of in-situ coatings either during fiber processing or in service (e.g., the BN layer formed by oxidation in amorphous Si-C-B-N fibers)

• clarification of the effect of solid-solution and second-phase additives (especially C and TiB₂) on the creep behavior and grain growth of polycrystalline SiC fibers

• study of the effects of texture on the creep strength of fibers and the development of processing methods to induce the optimum orientation

• systematic studies of the effects of high-temperature heat treatment (aging) on the rupture strength of polycrystalline SiC fibers