CMCs typically consist of a ceramic fiber, a matrix, and one or more fiber/matrix interfacial coatings (fiber coatings). The purpose of the fiber coating is to provide a weak fiber-matrix interface that prevents matrix cracking from penetrating the fibers—thus providing damage tolerance (toughness) to the composite. Fiber coatings also protect fibers from environmental degradation during composite fabrication and use. The fiber coating must be chemically and mechanically stable in high-temperature corrosive environments in order to maintain the fiber-coating-matrix debond characteristics necessary for damage tolerance. Unfortunately, the development of interfacial coatings has not kept pace with the development of ceramic fibers, which has limited the use of CMCs. Current fiber coatings either do not have adequate oxidation resistance or are not stable with fibers and matrices at elevated temperatures. Generally, these limitations have resulted in the degradation of the strength and toughness of CMCs during use.
In this chapter, the current status, major issues, and needs for fiber coatings are discussed. This discussion includes the status of the vendor base needed to commercialize fiber coatings. Although ceramic fibers were discussed in Chapter 4 and matrix materials are beyond the scope of this report, the effectiveness of the fiber/matrix interface is system dependent. Therefore, system approaches, such as the use of matrix additives to inhibit oxidation, will also be discussed in this chapter.
COATINGS FOR NON-OXIDE COMPOSITES
This section centers on fiber coatings for non-oxide composites in which either the fiber or the matrix is a non-oxide ceramic. Although oxide fiber-reinforced composites have been studied, most of the research available in the literature has focused on SiC fiber-reinforced composites. For example, mullite (3Al2O3-2SiO2) fiber-reinforced SiC matrix composites have been fabricated by CVI (chemical vapor infiltration). However, SiC fiber-reinforced SiC matrix (SiC/SiC) composites are superior for the following reasons: (1) mullite fiber-reinforced composites do not improve resistance to oxidation, one of the major factors limiting the use of non-oxide composites; and (2) SiC fibers have mechanical properties superior to those of mullite fibers. This section will be concerned primarily with SiC fiber-reinforced ceramic composites, which offer the best oxidation resistance of any non-oxide fiber at high temperatures (particularly above ~1,100°C [2012°F]).
Both oxide and non-oxide matrices have been used with SiC reinforcements. Examples include alumina matrix composites fabricated by oxidation of an aluminum melt (DIMOX process) (Newkirk et al., 1986), glass-ceramic matrix composites fabricated by hot pressing (Prewo and Brennan, 1980), SiC matrix composites made by CVI (Stinton et al., 1986), SiC or SiC/Si3N4 matrix composites made by polymer pyrolysis, and SiC-Si matrix composites produced by silicon melt infiltration (Luthra et al., 1993).
CMCs must be thermally stable. That is, they must retain a significant fraction of their room-temperature properties, such as strength and toughness, after long-term (desired service life) exposures at operating temperatures. They must also maintain these properties under oxidizing conditions. A composite that is thermally stable, however, may not remain stable when high temperatures are combined with high stresses under oxidizing conditions, a situation that is likely to be encountered in a composites operating environment. A wide variety of fiber coatings have been studied, but tough, thermally stable, non-oxide ceramic composites have only been demonstrated when C or BN fiber coatings were used.
Oxidation of the fiber-coating-matrix interface is one of the major problems that have prevented the widespread use of non-oxide ceramic composites (Luthra, 1997b). This interface can be exposed to oxidizing environments when the ends of coated fibers are exposed to the surrounding atmosphere or when matrix cracks are present, allowing atmospheric oxygen to reach the fiber coatings.
Matrix cracks develop when the composite is exposed to stresses above the matrix cracking strength (proportional limit), which can occur inadvertently even if CMC components are designed to operate below the proportional limit. Once matrix cracks have formed, they remain open even after the operating stress is reduced below the matrix cracking strength. Matrix cracks allow oxygen ingress and, therefore, oxidation of the fiber coating, and, potentially the degradation of the fiber itself. Oxidation of the coating also degrades the debond characteristics of the interface thereby degrading both the strength and toughness of the composite.
Carbon was one of the first fiber coatings that produced tough behavior in ceramic fiber-reinforced CMCs. Carbon coatings can form in-situ by the decomposition of Nicalon fibers, as has been observed in glass-ceramic matrix composites prepared by hot pressing, or they may be applied via CVD (chemical vapor deposition) techniques. Composite toughness appears to increase as the thickness of the coating increases. The load transfer between the fibers and the matrix, however, is reduced as the thickness of the coating is increased. Therefore, the coating thickness is typically on the order of 0.1 to 0.3 µm (0.004 to 0.01 mils). Tough composite behavior is demonstrated in carbon-coated Nicalon fiberreinforced SiC matrix composites prepared by CVI (chemical vapor infiltration), as shown in Figure 6-1 (Headinger et al., 1994).
The oxidation of carbon fiber coatings has been extensively analyzed in the literature. Fillipuzi and Naslain (Fillipuzi et al., 1994; Fillipuzi and Naslain, 1994) were the first to conduct a careful, systematic study of carbon coating oxidation in Nicalon fiber-reinforced composites. Figure 6-2 is a schematic representation of the oxidation phenomenon in carbon-coated SiC fiber-reinforced SiC matrix (SiC/SiC) composites. Carbon oxidation results in the formation of CO and CO2, leaving behind a gap between the fiber and the matrix, which exposes the fiber side and the adjacent matrix to oxidation, producing SiO 2, which tends to close the gap. Oxidation of carbon and the corresponding depth of recession of the carbon coating along the fiber continue by gas phase diffusion of oxygen until the SiO2 formed by the oxidation of the fiber and the matrix seals the gap. Based on the parabolic rate constants of the oxidation of the Nicalon fibers and the SiC matrix, Luthra (Luthra, 1994) calculated the time needed to seal the gap, which varies inversely with the oxidation rate constants of the fiber and the matrix and varies directly with the thickness of the fiber coating.
Figure 6-3 shows the corresponding depths of oxidation of a carbon coating as a function of temperature. Even for a fiber coating only 0.1 µm (0.004 mils) thick, the depth of oxidation of the fiber coating is very high, on the order of millimeters. Thus, the oxidation of fiber coatings is a serious concern when fiber ends are exposed. The depth of oxidation varies inversely with temperature in the range of 900 to 1,200°C (1,652 to 2,192°F). At high temperatures, the gap created by carbon oxidation is sealed fairly quickly, thus limiting the depth of oxidation. Consequently, the ends-on oxidation
problem is most damaging at intermediate temperatures of around 700 to 800°C (1,292 to 1,472°F) when it causes embrittlement of the composite. Embrittlement at intermediate temperatures is frequently called pesting.
Non-oxide composite components will be designed to operate below the proportional limit, but occasional accidental exposures above the proportional limit must be anticipated. When they occur, oxygen can diffuse through the matrix cracks and oxidize carbon fiber coatings. In these cases, the oxidation front will migrate laterally along the length of the fiber, and the depth of oxidation will be similar to the depth observed for the ends-on oxidation described above. Furthermore, sealing by the oxidation of SiC to produce SiO2 does not work effectively at intermediate temperatures of ~ 700 to 800°C (1,292 to 1,472°F). Therefore, an alternative mechanism must be found to seal matrix cracks faster and minimize oxidative degradation.
In the case of carbon fiber-reinforced carbon (C/C) composites, boron has been added to the matrix to help seal the matrix cracks in the intermediate temperature regime. B2O3 glass, formed by the oxidation of boron, acts as the crack sealant. DuPont Lanxide (Gray, 1992) has applied this boron sealing concept to enhanced CVI SiC/SiC composites. It was not clear, however, whether boron additives work fast enough to prevent the degradation of carbon fiber coatings when matrix cracks develop. Further work will be necessary to determine if performance can be improved in non-oxide systems after matrix cracking occurs.
Boron Nitride Coatings
Aside from carbon, boron nitride (BN) is the only fiber coating that has enabled graceful failure in non-oxide ceramic composites. It should be noted, however, that not all BN coatings are the same. When deposited at low temperatures, BN can be amorphous or turbostratic. When deposited at high temperatures (e.g., over ~ 1,500°C [2,732°F]), BN typically has an ordered hexagonal crystal structure (the desired structure for providing toughness to CMCs). BN deposited by CVD invariably contains carbon or oxygen impurities. Finally, silicon-doped BN is an intentional variation being evaluated for improved oxidation resistance.
Like the toughness of carbon-coated fiber-reinforced CMCs, the toughness of BN-coated fiber-reinforced CMCs appears to increase with the thickness of the coating. The optimum thickness of the BN coatings might be in the range of 0.3 to 0.5 µm (0.01 to 0.02 mils), thicker than for the carbon coatings. Figure 6-4 shows the tensile test results of Hi-Nicalon SiC fiber-reinforced SiC-Si matrix composites made by melt infiltration. Figure 6-5 shows the tensile test results for Nicalon SiC fiber-reinforced glass-ceramic matrix composites
made by hot pressing. Both have BN fiber coatings. Tough composite behavior is indicated by the nonlinear stress-strain curve (suggesting inelastic deformation) and a strain-to-failure of > 0.6 percent.
The oxidation rate of BN, in dry air or oxygen, is expected to be much lower than that of carbon because the BN oxidation product is a liquid boron oxide rather than a gaseous oxide. The lower rate is reflected in ends-on oxidation studies of composites. The depth of oxidation of the BN coatings in SiC-Si matrix composites has been observed to be very low, on the order of 10 µm (0.4 mils) or less for times up to 100 hours at 700 to 1,200°C (1,292 to 2,192°F) (Brun and Luthra, 1997). It is believed that BN oxidation produces a boron oxide liquid that interacts with silica, the oxidation product of the fiber and the matrix, to form a borosilicate glass. The composition of the glass changes with time due to surface volatilization of the boron oxide. The depth of oxidation is much shallower than the oxidation for carbon coatings. Therefore, BN coatings should perform much better than carbon in dry oxidizing environments.
In wet environments, the oxidation behavior of BN coatings appears to be similar to carbon in oxidizing environments. The oxidation product (B2O3) volatilizes as boron hydroxides before a borosilicate glass forms that can seal the gap created by BN oxidation and volatilization. Figure 6-6 shows the depth of oxidation of the BN coating in SiC-Si matrix composites reinforced with Hi Nicalon fibers with a nominal coating thickness of 0.5 µm (0.02 mils). At 1,200°C (2,192°F) oxidation proceeds for only about one hour before sealing prevents further oxidation; at 900°C (1,652°F) oxidation stops after approximately four hours. At 700°C (1,292°F), however, sealing does not occur and oxidation continues for up to 100 hours. Thus, it appears that embrittlement of these composites will occur more readily at intermediate temperatures of about 700 to 800°C (1,292 to 1,472°F) than at high temperatures of ~1,200°C (2,192°F).
In the presence of matrix cracks, oxidants (oxygen and water vapor) can rapidly diffuse through the cracks and oxidize fiber coatings. The coatings are normally 1 µm thick or less. Therefore, coatings are expected to oxidize very rapidly (in minutes at operating temperatures of more than ~900°C (1,652°F) (Jacobson et al., 1997), and the damage will then extend to the fiber. In dry environments, the damage should be confined because the lateral growth of the oxidation product will be slow, as suggested by ends-on oxidation studies. In wet environments, however, the damage will be more extensive because the BN coating can oxidize rapidly.
Therefore, approaches must be developed to seal the matrix cracks caused by overstressing. Two possible mechanisms by which the matrices could seal cracks are: (1) matrix heating or deformation (e.g., via creep); and (2) matrix additives that (upon oxidation) provide oxidation products that help seal cracks. In principle, the matrix sealing approach can work as long as the sealing occurs before significant damage is done to the composite. Results of stress rupture tests (Sun et al., 1996) on Nicalon fiber-reinforced glass-ceramic
composites, with a layered BN/SiC fiber coating, indicate that sealing does indeed occur. A sample was exposed at a temperature of 1,100 °C (2,012°F) in air at a stress of 138 MPa (20 ksi) (well above the fast fracture proportional limit stress of 88 MPa [13 ksi]). The sample lasted for more than 1.7 years, and the residual strength was comparable to the strength of the unoxidized sample.
Figure 6-7 shows a cross-section of a sample exposed at 1,200°C (2,192°F). The oxidation is confined to a surface depth of about 100 µm (4 mils). Although the microstructure of the sample at 1,100°C (2,012°F) was not shown, the degradation mode was presumably similar. There were no boron additives in the matrix, but the fiber coating was a layered BN/SiC coating. Sealing occurred by either creep deformation of the matrix or by oxidation products of the fiber coatings. There is little or no information in the literature as to whether boron additives will work effectively with BN fiber coatings, particularly in wet environments where BN can volatilize as boron hydroxides.
Crystalline BN and silicon-doped BN have been suggested to improve the oxidation resistance of amorphous BN in water vapor environments (Morscher et al., 1997). Silicon-doped BN should be particularly attractive because it is expected to maintain the debond characteristics of the BN and to improve oxidation resistance significantly over BN, in both dry and wet environments. Although the effect of silicon on oxidation resistance has not been quantitatively evaluated, it is reasonable to expect that the addition of silicon to BN would significantly decrease the oxidation rate of the coating. The oxygen diffusivity through B2O3, the oxidation product of BN, is about 3 orders of magnitude higher than through silica, the oxidation product of SiC and Si3N4. The volatility of silicon as silicon hydroxide is also very much lower than of boron hydroxide. Silicon doping in BN is also expected to reduce the volatilization of the coating in water vapor environments. Thus, silicon-doped BN is potentially an attractive coating that should be explored further.
Alternative Fiber Coatings
Several alternative fiber coatings have been evaluated in non-oxide fiber-reinforced composites, either to improve the oxidation resistance of composites or to improve the stability of carbon and BN coatings. Alternatives include layered coatings (Luthra et al., 1994; Brennan, 1997), such as BN/C/BN, BN/C/Si3N4, SiC/C/SiC, and BN/SiC. Multiple iterations of these layered coatings have also been evaluated. Because each carbon or BN layer in multilayer coatings is not as thick as a single carbon or BN coating, multilayer coatings may be more resistant to oxidative degradation. Multiple layers may also provide multiple debond layers.
Other coatings that have been proposed include carbides, such as Ti3SiC2 (Christian et al., 1991); nitrides, such as AlN; and a variety of oxides. However, no information is available in the literature describing whether or not any of these alternative fiber coating systems provide damage-tolerant, tough ceramic composites in the absence of a carbon or BN layer. Fiber coatings or coating systems that provide improved oxidation resistance (compared to carbon or BN coatings), as well as acceptably tough composites, will have to be developed.
It is difficult to assess the oxidation resistance of all of the coatings that have been proposed in the literature. However, it is worthwhile to discuss the potential capabilities of the best oxidation-resistant coatings. In order to maintain a practical level of load transfer between the matrix and fiber, fiber coatings can be a maximum of ~ 1 µm (0.04 mils) thick for fiber bundles and several microns for monofilaments that are more than 100 µm (4 mils) in diameter. For the present discussion it was assumed that the coating was about 1 µm (0.04 mils) thick.
Because oxide ceramics cannot oxidize, it is commonly believed that oxide coatings represent the best choice in terms of oxidation resistance. Although this is true for the prevention of oxidation of fiber coatings, it is not true for the protection of the underlying non-oxide fibers from oxidation or for the maintenance of weak fiber-matrix interfaces. Using diffusion analysis, Luthra (1997b) demonstrated that non-oxide fibers can start oxidizing at the interface between the fiber and the oxide coating within a matter of seconds, even with an oxidation barrier like SiO2. Once oxidation is initiated at the fiber-coating interface, the fiber bonds to the coating, which may also bond to the matrix, which renders the composite intolerant to damage. The coating composition also changes, which is particularly important if the coating is weak and debonding occurs within the coating (e.g., mica substances), rather than at the fiber-coating interface. Because fiber-coating reactions
can occur at typical composite fabrication temperatures, it is difficult even to fabricate non-oxide fiber-reinforced composites with oxide fiber coatings without degrading the fiber and composite properties.
For non-oxide coatings, silica-forming materials provide the best protection against oxidation at temperatures above ~ 1,100°C (2,012 °F) because silica has the lowest oxygen permeability of any oxide. However, Luthra (1997b) has demonstrated that the oxidation protection provided by a 1 µm (0.04 mils) coating of silica-forming material, such as SiC or Si3N4, is limited, theoretically, to a maximum of ~100 hours at 1,200°C (2,192°F). Therefore, no non-oxide coating can protect SiC fibers for more than 100 hours in the presence of oxygen. Oxidation resistance should be considered in terms of the overall composite system to determine how oxidation degrades composite properties and how the system, including the fiber, coating, and matrix, can be modified to alleviate this problem.
Systems Approach to Inhibiting Oxidative Embrittlement
In the presence of open matrix cracks, no fiber coating can (or will likely ever) protect the fiber from oxidation for more than 100 hours at 1,200°C (2,192°F) in an oxygen containing environment. This does not, however, eliminate the need for
fiber coatings with improved oxidation resistance because theoretical limitations have not yet been reached. Therefore, a two pronged approach should be pursued:
The oxidation resistance of non-oxide fiber coatings must be improved, from the current calculated value of a few minutes (at 1,200°C [2,192 °F]) for 1 µm (0.04 mils) BN coatings to the maximum possible calculated value of ~ 100 hours for 1 µm (0.04 mils) silica-forming coatings.
Techniques should be sought for rapidly sealing matrix cracks once they form. The crack sealing time should be less than the time it takes to oxidize the fiber coating.
A schematic representation of a cracked CMC is shown in Figure 6-8. The crack width is similar to the thickness of the coating, on the order of 1 µm (0.04 mils) or less. The effect of oxidation on one of the crack/coating/matrix regions schematically shown in Figure 6-8 is presented in Figure 6-9. Various processes that occur in the cracked region are shown in Figure 6-9a. Oxidants, such as oxygen and water vapor, diffuse through the open crack to the interior of the composite, oxidizing the sides of the cracks and the fiber coating (indicated by three arrows). Upon oxidation, the volume of the oxidation products is higher than the volume of the oxidizing constituents. For example, the oxidation of SiC to form silica results in a volume expansion of ~100 percent. Therefore, the oxidation products can fill the cracked region and prevent the oxidation of the underlying composite. Figure 6-9b shows schematically the progression of the oxidation region. In this case, the crack is still open, although its width has been reduced, and the coating is not fully oxidized. Figure 6-9d shows a coating that has been breached before the crack was sealed. Oxygen diffusing through the crack can continue to oxidize the fiber by diffusing through the oxidation product of the coating, indicated by a semi-spherical region. The best situation is shown in Figure 6-9c, where the crack has been sealed before the coating was oxidized, thus preventing significant damage to the fiber-coating-matrix interface. This approach might be feasible with multilayer coatings that provide multiple debonding layers.
If cracks develop near the fiber/coating interface, some oxidation of the interface will inevitably occur before the cracks are sealed. If there are multiple debond layers, the debond characteristics of the interface can be maintained even after sealing. However, this has not been demonstrated for any composite system. Questions regarding the survivability of the sealed regions under the cyclic loading conditions that will be encountered in an operating environment have not been addressed. Therefore, in addition to developing matrix crack sealing concepts and fiber coatings with improved lifetimes, the performance of these materials should be evaluated under conditions representative of operating environments.
Coating Processes and Vendors
Chemical Vapor Deposition
CVD is the most common method of depositing fiber coatings for composite systems because it is a conformal process that can deposit fairly uniform coatings on a wide variety of structures. CVD is perhaps the only method that has been used successfully to provide tough composites with acceptable mechanical properties. A wide variety of coatings have been deposited by CVD, including carbon, BN, silicon-doped BN, Si3N4, and SiC.
Carbon is usually deposited by cracking hydrocarbons, such as CH4 and C2H2 at temperatures of 1,000 to 1,800°C (1,832 to 3,272°F). BN is usually deposited using a boron trichloride precursor with ammonia and a nitrogen carrier gas at temperatures of 800° to 1,600 °C (1,472 to 2,912°F). Other precursors, such as borozine and BF3, have also been used. The crystallinity and deposition rate of BN increases with temperature. However, the coatings also become less uniform as temperature is increased. The non-uniformity is expected to be less prevalent when fiber tows are coated and most prevalent when three-dimensional composite lay-ups are coated. The chemistry of the BN coatings depends upon the
precursors, as well as a range of other variables. For example, BN coatings produced in silica reactors are subject to oxygen contamination, but coatings produced in carbon furnaces contain carbon contaminants. Patibandla and Luthra (1992) observed up to 12 atomic percent oxygen in their BN coatings; Sun, Nutt, and Brennan (1994) intentionally used coatings with 40 atomic percent boron, 40 atomic percent nitrogen, and 20 atomic percent carbon. Unfortunately, the effects of BN chemistry and crystallinity on the mechanical properties of composites, both as fabricated and after environmental exposure, are not well characterized or understood.
Silicon nitride is usually deposited using one of the following silicon precursors: silane, dichlorosilane, trichlorosilane, or silicon tetrachloride. Silicon tetrachloride, the most common silicon precursor, is used with an ammonia and nitrogen mixture to obtain silicon nitride at temperatures of 800 to 1,400°C (1,472 to 2,552°F). Silicon carbide is usually produced by cracking methyl trichlorosilane at 800 to 1,400°C (1,472 to 2,552 °F).
This section discusses the domestic coating vendor base. The capabilities of foreign vendors are not available in the open literature and are, therefore, not covered.
The applicability of specific deposition technology to coating a particular ceramic fiber depends on the fiber architecture. For example, CVI is most suitable for depositing multilayer coatings on woven cloth or fibers with a three-dimensional architectures. In the United States, CVI process vendors include—but are not limited to—DuPont Lanxide, BF Goodrich, and Amercom. To produce uniform CVI coatings, deposition has to be carried out at relatively low temperatures of (e.g., 800 to 1,000 °C [1,472 to 1,832°F]). However, significant variations in coating uniformity have been observed from one filament to another within fiber bundles, as well as from the inside ply to the outer ply of a three-dimensional layup. The effect of non-uniformity in the fiber coating on composite behavior has not been well characterized. Non-uniformity is expected to increase if crystal-line BN coatings are used because they require deposition temperatures of 1,400°C (2,552°F) or even higher. Chemical non-uniformity is expected to be another important factor in the deposition doped BN coatings, such as silicon-doped BN.
Vendors are available for coating single plies of woven cloth and individual tows. For example, Synterials currently uses a batch reactor to deposit coatings, and Advanced Ceramics uses a semicontinuous process in which the uncoated woven cloth is carried though the CVD furnace and spooled at the other end. Advanced Ceramics has a 3-spool reactor and a 12-spool reactor for coating fiber tows, which is the most attractive approach if uniform fiber coatings are needed. Unfortunately, to date it has not been possible to weave coated fibers without degrading the fiber coating. Therefore, coated fibers have been successfully used only in processes that do not require a woven cloth or three-dimensional lay-up, such as prepregging.
Although oxide fiber coatings have been successfully deposited using a wide variety of non-CVI processes, such as sol-gel or particulate slurry, only CVD has been successfully used for non-oxide coatings, including carbon and BN. It should be possible, however, to deposit non-oxide coatings by other means. For example, Kim, Cofer, and Economy (1995) obtained a BN matrix using a borazine oligomer.
The costs of fibers and coatings constitute a very large fraction of the high costs of CMCs. The costs of coatings currently are on the order of a few thousand dollars per pound of fiber. Although cost analyses have not been reported as a function of processing conditions, it is expected that uniform coatings will be more expensive to produce than non-uniform coatings. For example, although lowering the coating deposition temperature typically improves uniformity, it also decreases the deposition rate, which is expected to increase costs. Furthermore, it has been demonstrated that moving from multiple layers of cloth (i.e., three-dimensional lay-ups) to a single layer of cloth (i.e., two-dimensional lay-ups) to fiber tows (i.e., one-dimensional lay-ups) improves coating uniformity. Coating several fiber tows prior to weaving them into cloth, however, is expected to be more costly than coating the cloth itself. The development of additional low cost approaches will be necessary to reduce the costs of coatings and, ultimately, of composites.
One approach that appears to be particularly attractive is insitu coatings formed during fiber processing or by subsequent heat treatment. An example of an in-situ coating discussed in Chapter 3 was developed by Bayer of Germany, which has reportedly formed an in-situ BN layer in a Si-B-N-C fiber. It should be possible to form a BN and/or silicon-doped BN coating on SiC fibers doped with boron by appropriate heat treatment in an nitrogen-containing gas environment. These coatings should be more uniform in thickness and much cheaper to produce than coatings produced by CVD. However, it is not clear what the debond characteristics of in-situ formed coatings will be in a finished composite. The behavior of in-situ coated fibers in a composite system should be studied.
OXIDE FIBER COATINGS
The development of ceramic oxide composites has lagged behind the development of non-oxide composites because of the poor creep resistance of oxide fibers (compared to SiC
fibers) and because of the lack of adequate oxide fiber coatings that promote fiber-matrix debonding. Recent advances in creep-resistant oxide fibers (e.g., Nextel 720) and progress on interface control has improved the potential for using oxide composites in industrial and defense applications. However, an effective coating for oxide fibers that provides a weak fiber/matrix interface (and therefore tough composite behavior) remains to be demonstrated.
Oxide fiber coatings are most commonly applied to oxide fibers. Using oxide fiber coatings on silicon-based non-oxide fibers has generally been avoided because these fibers have a tendency to react with the oxide fiber coatings, thus creating strongly bonded interfaces. There are notable exceptions, however (e.g., SiC fibers coated with a SiO 2/ZrO2/SiO2 multi-layered coating).
Carbon-based and BN-based coatings for oxide composite systems have not been investigated as fervently as they have for non-oxide systems. One reason for this is that a primary advantage of using an oxide system is its inherent oxidation resistance, which could be hindered by the presence of an interface that was prone to oxidation. This is especially true for BN, the oxidation product of which (boria) can react with the fiber and the matrix. Using carbon coatings as fugitive interfaces in oxide systems as a means to debond the fiber from the matrix is an example of taking advantage of a coating prone to oxidation. This is discussed in more detail later.
The initial approach to developing fiber coatings for oxide composite systems, beyond the carbon and BN approaches borrowed from non-oxide composites, focused on oxide compounds that do not form compounds with the fibers or the matrices. The focus later shifted to oxide interfacial coatings that mimic the layered crystal structures of carbon and BN, which provide weak interfaces in non-oxide composite systems. Techniques were also developed to provide mechanically weakened interfaces between the fiber and matrix by using porous, or fugitive, coatings. The latest concepts being investigated for oxide systems are focused on creating high energy interfaces between coating compounds and oxide fibers (described further in the following sections).
Porous Coatings and Porous Matrix Approaches
In order to decrease the debond strength of fiber/matrix interfaces, porosity has been incorporated into the fiber coating to provide a weak path along which matrix cracks can be deflected away from the fiber (Davis et al., 1993). Various techniques have been tried for incorporating porosity in a fiber coating, including sol-gel and co-deposited CVD. Porosity within the coating often has to be generated after processing of the composite matrix to prevent the pores from being filled during composite fabrication. To achieve this porosity, the as-deposited coatings often contain carbon, which is removed to create pores after matrix processing has been completed. For example, porous alumina coatings have been obtained by sol-gel processing using boehmite sols in a polyelectrolyte ammonium polymethacrylate (Darvan C) solution (Boakye et al., 1997). After the fibers were coated with this solution, they were fired in an argon atmosphere to yield an oxide-carbon coating (Figure 6-10). No data has been generated for composites that use these fiber coatings, however, nor has a deposition technique that yields high quality coatings over long lengths of fiber been demonstrated.
A major concern about porous coatings is maintaining the stability of the pore morphology after prolonged high-temperature exposure, as there may be a tendency for these coatings to densify. Thus far, porous coatings on fine-diameter fibers have not been subjected to long-term thermal exposures to evaluate their microstructural stability. If coarsening of the porosity proves to be a problem, it may be necessary to use oxides with sluggish diffusion kinetics to prevent sintering of the coating. Possible compounds for porous coatings include mullite, rare earth aluminum garnets, and mixed phases, such as mullite-alumina or alumina-garnet.
CVD has been investigated as a method for creating a porous zirconia coating by co-depositing carbon and zirconium carbide in a ratio that yields a 40 percent porous zirconia coating upon oxidation (Goettler, 1997a). Oxidation of the coating at 1,100°C (2,012°F), however, caused sintering of the zirconia to a dense coating. Some fiber pullout from an alumina matrix was observed when these coatings were used. However, this was probably caused by shrinkage of the fiber
coating (away from the alumina matrix) upon oxidation and sintering of the coating following composite processing.
No Coating/Porous Matrix Approach
The cost of ceramic composites could be lowered and the overall processing simplified if the fiber coating step could be eliminated. This approach is being investigated by General Electric and researchers at the University of California at Santa Barbara (UCSB) for oxide systems composed of Nextel 610 or Nextel 720 fibers in an alumina-mullite matrix. The General Electric material uses a matrix containing alumina particles and a silicon containing organic polymer (Harrison et al., 1994)). The silicon polymer provides green strength to the preform and generates silica upon decomposition, which then reacts with and bonds the alumina matrix particles. The UCSB material uses a bimodal distribution of coarse mullite particles and finer alumina particles to create the majority of the matrix through a pressure filtration technique (Levi et al., 1997). Bonding of the matrix particles is achieved by additional infiltrations with an aluminum hydroxychloride solution. The key to success is retaining sufficient porosity within the matrix to allow for crack branching within matrix regions and crack deflection around the continuous fibers. Total composite porosities typically fall within the 20 to 25 percent range depending on the total fiber loading. This approach is similar to the porous coating approach described above except that the porous interfacial region extends throughout the entire matrix.
The range of potential components that could be fabricated from porous matrix ceramic composites may be limited because of the poor mechanical properties of the porous matrix. Porous matrices generally yield composites with low compressive and interlaminar properties, making it difficult to design structures capable of carrying out-of-plane loads. Furthermore, it is difficult to attach porous matrix composites to neighboring structures. Three-dimensional fiber architectures rather than two-dimensional laminate structures will probably be required for the components made from a porous matrix ceramic composite.
A potential problem with this type of composite system is the possibility of reaction or sintering between matrix particles and the fibers at points of contact. Exposure tests of UCSB Nextel 610 composite at 1,200°C (2,192°F) for 100 hours shows reasonable mechanical property retention (Figure 6-11). Similar tests have yet to be performed on Nextel 720 composites. Therefore, the data on long-term exposure is insufficient to address the issue of matrix-fiber bonding (Levi et al., 1997). Because degradation is likely to be kinetically driven, residual fiber strength will be dictated by time at temperature.
Many industrial applications (e.g., industrial gas turbines or heat exchangers) require that composites operate nearly continuously for many years at high temperatures. Over time, the fiber properties, and hence the composite properties, may deteriorate as matrix particles react or sinter with fibers. Another concern is the potential densification of the matrix after long-term exposure at operating temperatures, causing a change in the pore-size distribution of the matrix and probably changes in the mechanical properties of the composite. Coarsening of the fine porosity within the silica bond phase of the General Electric Gen IV system reportedly limits the maximum operating temperature for the material (Hay, 1991). This characteristic of the Gen IV system was the motivation for the development of the more stable porous
matrix of the UCSB composite system, which has an alumina bond phase.
The stability of the interface and matrix morphology may not be as critical for defense applications, where the service time of the ceramic component, especially at peak temperatures, is relatively short. Uncoated fibers in a porous matrix, however, may subject the fibers to corrosive species in the operating environment that could degrade fiber strength. These composite systems must be tested in their service environments to determine if the absence of a fiber coating accelerates the degradation of fiber strength.
Fugitive and Segregant Weakened Interfaces
The concept of fugitive interface coatings is based on processing a composite with a carbon coated fiber (molybdenum coatings have also been proposed) (Sambasivan et al., 1993). After composite densification, this coating can be oxidized away to create a void along the fiber-matrix interface (Mah et al., 1991; Keller et al., 1993). In essence, the fibermatrix interface is debonded from the start rather than relying on the stress intensity of a propagating crack front to initiate fiber-matrix debonding.
A major technical issue with this approach is how to ensure load transfer to the fibers if the interface is totally debonded. Test data, based on the rule of mixtures, shows that, even with the carbon coating removed, some load transfer appears to occur between the matrix and fiber (Keller et al., 1993). The degree of load transfer may depend on the thickness of the fugitive layer (which is present during composite processing) relative to the roughness of the fiber surface and the variation in diameter along the fiber's length.
If the fugitive zone (between the fiber and matrix) is not too thick in comparison to these fiber characteristics, some contact may be made between the matrix and the fiber, allowing for load transfer. Fiber-matrix contact, however, raises the possibility that fiber strength could be degraded by reaction or sintering between the fiber and the matrix. Good strength and modulus retention have been demonstrated for a Nextel 720 fiber-reinforced, calcium aluminosilicate glassceramic matrix composite (with carbon coating removed) following a 500 hour, 1,000°C (1,832°F) unstressed heat treatment (Figure 6-12) (Keller et al., 1997).
A second issue with this concept is that bare fiber surfaces are exposed to the operating environment, which could lead to the degradation of fiber strength. Composites that utilize the fugitive interface coating concept must be tested under realistic operating conditions. This is a fairly cost effective coating approach because of the ease with which carbon coatings can be deposited by both sol-gel and CVD techniques.
Segregant Weakened Interfaces
Segregant weakened interfaces is an intriguing debonding concept based on the development of a nonporous, weak fiber/matrix interfacial boundary. It has been proposed that the fiber-matrix interface in dense oxide ceramic composites may be weakened by using segregants (Sambasivan et al., 1993). The basis for this approach was the observation that adding calcia to alumina reduces the fracture toughness of alumina (Jupp et al., 1980). Model composites were investigated in which a single crystal of YAG (yttria-alumina-garnet) was hot pressed in a matrix of either strontium oxide (SrO) or calcia (CaO) doped alumina (Sambasivan et al., 1993). Strontium and calcium were found to exist preferen-
tially at the YAG-alumina interface. Using Vickers indentations, the segregant enhanced interfaces were shown to be weaker than an undoped YAG-alumina interface. These results raise the possibility that this low cost approach could be used to produce high-temperature ceramic composites. So far, only preliminary research has been performed on this coating concept, and only on model composite systems. Quantitative tests will have to be done to determine if interfacial weakening is sufficient to promote crack deflection in CMCs. Furthermore, preliminary results were for a single crystal-polycrystalline matrix interface. Selective segregation may be much more difficult for a polycrystalline-polycrystalline fiber-matrix interface.
Dense Oxide Fiber Coatings
A prerequisite for an effective fiber coating is chemical stability with the fiber and matrix phases of a composite. If reactions occur, the fiber-coating-matrix interface would become strongly bonded, and the composite would exhibit brittle fracture behavior. Based on currently available phase diagrams, oxide compounds have been identified that do not show compound formation with commercially available oxide fibers. Particular attention has been paid to fibers that show the highest temperature capability (alumina fibers, such as Fiber FP, PRD-166, and Nextel 610). Oxide compounds identified as stable with alumina included tin oxide, zirconia, titania (at temperatures below ~1,150°C [2,102°F]), zirconia titanate (ZrTiO4, also at temperatures below ~1,150°C [2,102°F]), and possibly zirconia tin titanate (Zr(Sn,Ti)O4).
Tests on tin oxide fiber coatings in model composite systems indicated some crack deflection at the coating-fiber interface (Siadati et al., 1991; Venkatesh and Chawla, 1992). However, tensile tests of tin oxide coated alumina fiber-reinforced alumina matrix composites demonstrated a decrease in the extent of fiber pullout as the density of the matrix phase was increased. This led to increasingly brittle fracture behavior in these composites (Goettler, 1993). Tin oxide also has thermal stability problems at elevated temperatures (Norkitis and Hellmann, 1991). For example, in the presence of air at temperatures above 1300°C (2,372°F), tin oxide (solid) decomposes into SnO (gas) and O2 (gas). This decomposition occurs at even lower temperatures when the partial pressure of oxygen in the test environment is reduced.
Zirconia coatings require intercoating porosity to allow for crack deflection at the alumina-zirconia interface in zirconia coated alumina fiber-reinforced alumina matrix composites (Davis et al., 1991). Testing of titanate coatings is incomplete, but in general these nonreactive isotropic oxides appear to provide a reaction barrier between the fibers and the matrix. Because they are dense coatings, however, their interfacial debond strengths are too high to allow for crack deflection around alumina fibers. Therefore, they do not enable graceful failure in ceramic composites.
Many layered oxides have been identified as potential fiber coatings for oxide ceramics as analogs to the layered carbon and BN fiber coatings used for non-oxide composites. One of the first layered oxides to be investigated was a class of sheet silicates known as fluoromicas. A potassium-tetrasilic mica (KMg2.5(Si4O10)F2) and a potassium fluorphlogopite (KMg3(A1Si3O10)F2) have been examined (Beall et al., 1990; Chyung and Dawes, 1993). These synthetic micas exhibit easy delamination along the crystal planes containing the fluorine and interlayer alkali cations. The presence of these weak crystal planes along the fiber-matrix interface encourages crack deflection. Because of the complex chemistry of these multicomponent compounds, however, they are chemically incompatible with the currently available fibers and matrices, which causes the composite to become unstable and ineffective after prolonged exposure to elevated temperatures (Cooper and Hall, 1993).
Less chemically complex layered oxides, which have the ß-alumina (Me1+Al11O17) and magnetoplumbite (Me2+Al12O19) structures, have also been investigated. These materials are phase compatible with alumina making them promising candidates for alumina fiber-reinforced composites (Morgan and Marshall, 1992; 1993). Although these layered oxides are not as readily cleaved at the basal planes as fluormicas, their fracture energies appear to be sufficiently low to promote crack deflection along the fiber-matrix interface (Phillips and Griffin, 1981; Beevers and Ross, 1937). The structures of both ß-alumina and magnetoplumbite consist of layered spinel blocks ((Al11O16)+) with alkali, alkaline-earth, and rare-earth cations in c-axis planes between the spinel layers (Beevers and Ross, 1937; Utsonomiya et al., 1988). The similar structures of these two classes of compounds are shown in Figure 6-13.
Investigations into ß-alumina fiber coatings have been discontinued because of problems with alkali loss during heat treatment, especially with sodium and potassium ß-aluminas. However, it has been reported that the use of larger alkali cations, such as rubidium, which has a reduced mobility within the ß-alumina structure because of its larger ionic radius, may minimize vaporization (Sambasivan et al., 1996). The magnetoplumbite structures containing alkaline-earth or rare-earth cations do not exhibit this vaporization problem and are, therefore, more promising candidates for fiber coatings. Compatibility studies between Nextel 720 fiber and RbAl11O17 show a degradation of fiber strength for coated fibers after a one hour heat treatment at 1,100°C (2,012°F) (Goettler, 1996). RbAl11O17 appears to be stable with Nextel 610
after similar heat treatment; however, the lower temperature capability of Nextel 610 precludes its use in many high-temperature structural applications.
The most detailed investigations of a layered oxide have focused on the magnetoplumbite known as hibonite, CaAl12O19 (Cinibulk and Hay, 1996). No composites have yet been fabricated from fibers with hibonite coatings. Studies of model composite systems consisting of hibonite coated, single-crystal fibers (both Al2O3 and YAG fibers) in hot pressed matrices of alumina and YAG have shown encouraging crack deflection results. Figure 6-14 is a transmission electron micrograph of a cross-section of a hibonite-coated sapphire fiber that shows the desired crack deflection along the basal plane (Cinibulk and Hay, 1996). There are, however, several technical obstacles to incorporating this coating into actual composite systems. First, high-temperature processing of the coating, typically higher than 1,300°C (2,372°F), is required to generate the magnetoplumbite structure with the layered planes parallel to the fiber axis (Cinibulk, 1997b). This temperature is higher than the temperature at which commercially available polycrystalline oxide fibers show strength degradation. Solution-based precursors in which calcium-aluminum inorganic polymers are created or in which nucleating agents are added to induce lower temperature formation of the magnetoplumbite structures may have to be developed.
Another problem with hibonite is the tendency for calcium to diffuse from the coating into the matrix during hot pressing (Cinibulk, 1994). The use of alternate basal plane cations, which would have less of a tendency to diffuse into the matrix and fibers, may solve this problem.
Hibonite has been chosen for initial investigation because it can be converted to the magnetoplumbite structure at a lower temperature than other layered oxides with basal plane cations other than calcium. Novel solution precursor chemistry that would enable lower temperature conversion to the magnetoplumbite structure could expand the compositional ranges to include cations less susceptible to migration within the matrix and fiber. Magnetoplumbites that contain alkaline-earth and rare-earth elements are likely to be chemically incompatible with fibers that contain mullite, such as Nextel 720 (as was observed with the ß-aluminas). In order to utilize magnetoplumbite fiber coatings, a higher-temperature fiber that does not contain Si, such as YAG or creep-resistant doped alumina, will probably be required to ensure interfacial stability and allow for the high processing temperatures currently required to generate the magnetoplumbite structure.
The latest class of layered oxides to be proposed as fiber coatings are layered perovskites, including potassium calcium niobate (KCa 2Nb3O10) and barium neodymium titanate (BaNd2Ti3O10) (Petuskey et al., 1996). Potassium calcium niobate has a melting temperature of only 1,460°C (2,660°F), whereas barium neodymium titanate has a melting temperature of greater than 1,800°C (3,272°F). Only preliminary investigations have been made thus far to assess these compounds as coatings in oxide composites. The main concern with these complex chemical compounds is their stability with candidate fibers and matrices. Both compounds appear to be stable with alumina, at least up to moderate temperatures of 1,250°C (2,282°F) (Petuskey et al., 1997), but their stability with fibers of more complex chemistry, such as Nextel 720 fiber (which contains mullite) and YAG fibers, is suspect. The potassium in potassium calcium niobate shows some volatility and thin films of the niobate heated to 800°C (1,472°F) exhibit a significant loss of potassium. However, there are indications that a sodium calcium niobate may exhibit less volatility.
One favorable aspect of these layered compounds is their ability to form a layered structure from sol-gel precursors at temperatures as low as 600°C (1,112°F). This ability is critical, especially with complex chemistries, because a lower formation temperature can help avoid reactions between the fibers and intermediate coating phases. Additional work will be required to determine if these classes of compounds are stable with potential and commercially available fiber reinforcements.
Weakly Bonded, Nonlayered Oxides
The segregant approach described earlier has the potential for creating a weak fiber-matrix interface without fiber coatings. An alternate approach is to use a fiber coating that exhibits a high energy, therefore weak, interface with the fiber or the matrix, or both. The high energy interface created between the coating and the fiber will have a tendency to debond readily in the presence of the stress intensity created by a propagating crack. Compounds that exhibit weak bonding to alumina include the rare-earth phosphates of the general formula Me3+PO4 (Morgan and Marshall, 1996), including the monazite family of minerals, which is composed of the larger rare-earth elements of the lanthanide series (La, Ce, Pr, Nd, Pm, Sm, Eu, Gd, and Tb) that are sometimes substituted with divalent and tetravalent cations, such as calcium or strontium. Also included are the xenotime family of minerals comprised of scandium, yttrium, and the smaller rare-earth
elements of the lanthanide series (Dy, Ho, Er, Tm, Yb and Lu). In general, the rare-earth phosphates have a high melting point. For example, monazite (LaPO4) melts at 2,072 ± 20°C (3,761± 68°F).
Model composite systems of monazite-coated fibers in hot pressed billets of alumina have shown very promising interfacial debonding induced by propagating cracks generated by Vickers indentations ( Figure 6-15) (Morgan and Marshall, 1995). Micromechanical analysis, based on theoretical models incorporating interfacial energies and elastic mismatches (He and Hutchinson, 1989), predict debonding at the monazitealumina fiber interface (Morgan and Marshall, 1995). The ease of debonding observed in the model composite systems has yet to be fully demonstrated in an actual dense composite system.
A significant obstacle to implementing these coatings has been depositing fiber coatings with a stoichiometric lanthanum-to-phosphorous ratio. It appears that nonstoichiometric monazite can significantly degrade the strength of the fibers and may affect the ability of the interface to debond (He and Hutchinson, 1989). Monazite appears to be stable with Nextel 720 fibers when the stoichiometry is met (Goettler, 1997b), which correlates with geological information showing that monazite is stable with aluminosilicates (Morgan and Marshall, 1995).
Additional compounds of the general ABO4 formula have been proposed as weakly bonded, nonlayered oxide fiber coatings for alumina-based fibers. These include the broad family of tungstates (Me2+WO4), molybdates (Me2+MoO4), tantalates (Re3+TaO4), and niobates (Re3+NbO4) (Goettler et al., 1997a). Alkaline earth tungstates have melting temperatures ranging from 1,358°C (2,476°F) for MgWO4 to 1,580°C (2,876°F) for CaWO4 (scheelite). Molybdates have lower melting temperatures than the corresponding tungstates. Data is incomplete on the melting points of many tantalates and niobates although they generally have higher melting temperatures than tungstates and molybdates. Tantalates have melting temperatures near 1,900°C (3,452°F), whereas niobates melt at approximately 1,600°C (2,912°F).
Detailed investigations have been performed for scheelite and ErTaO 4 fiber coatings on Nextel 610 fibers in duplex matrices of alumina-scheelite and alumina-ErTaO4, respectively. Scheelite appears to be chemically compatible, as well as show easy debonding, with Nextel 610 fibers (Figure 6-16) (Goettler et al., 1997a). Unidirectional composites have shown strengths of more than 340 MPa (50 ksi) and extremely fibrous pullout at composite porosity levels of approximately 20 percent at room temperature (Goettler et al., 1997a). The scheelite-alumina debonding mechanism must still be verified, however, at the high temperatures representative of projected operating environments. Preliminary investigations suggest that, like sheelite, ErTaO4 is stable with alumina although substantial debonding of this coating from Nextel 610 fiber was not observed (Goettler et al., 1997a). Because tantalates have a crystal structure similar to monazite, the debonding characteristics are also expected to be similar. Additional investigations must be done, however, to demonstrate this. The grain size of the ErTaO4 fiber coatings were about an order of magnitude smaller than the grain size of CaWO4 coatings, indicating that the grain size of the ABO4 coatings may have a strong affect on the interfacial energy and, hence, the ability of the interface to debond readily.
Initial mechanical testing and transmission electron microscope studies of scheelite-coated Nextel 720 fiber-reinforced oxide composites suggest compatibility between scheelite and Nextel 720 fibers (Goettler et al., 1997b). Furthermore, scheelite is found in nature with quartz deposits, providing some geological evidence that scheelite should be stable with silica-containing fibers, such as Nextel 720. However, detailed testing will be required to verify the applicability of scheelite as a fiber coating for Nextel 720. The fact that only some ABO4 compounds (all of which fall into closely related crystal groups) show weak bonding to alumina and aluminasilicate-based fibers underscores the need for basic studies of oxide-oxide interfaces on the atomic scale. Determining the reason for the weak bonding of these interfaces may lead to other candidate interfaces that could extend the range of oxide composite systems that can survive the environments to which these composites may be subjected.
Coating Processes and Vendors
Immiscible Liquid Coating Technique
The most promising oxide fiber coatings appear to be multicomponent oxides. Liquid-based techniques are generally the most viable for producing these complex oxide coatings. Using CVD, it is often very difficult to maintain stoichiometry in the coatings, although some success has been achieved with LaAl11O18 (Brown, 1995) and monazite (Chayka, 1997). Using CVD precursors are often expensive, and using a precursor in a continuous coating process is often inefficient. Solution-based precursors allow accurate stoichiometric control, and, with inorganic polymer chemistry, it is possible to create precursor solutions that convert to the desired phases at low temperatures upon decomposition. For instance, monazite has been reported to form at as low a temperature as 100°C (212°F) from an aqueous solution (Morgan, 1996; Goettler and Sambasivan, 1996).
The major drawback of solution-based coating techniques is the difficulty in obtaining continuous, uniform, bridge-free coatings. CVD coatings are typically superior, morphologically, to coatings obtained by sol-gel techniques. A new technique that minimizes the bridging problem exhibited by solution-derived coatings involves using immiscible liquids to displace excess coating solution that is normally retained
between fibers within a tow bundle after passing through a coating precursor (Figure 6-17a and Figure 6-17b) (Hay, 1991). The basic idea behind this technique is to take advantage of surface energy relationships to keep the individual fibers, which are covered by the coating precursor, separated from one another by a layer of hydrocarbon, such as pentadecane, until the coating precursor gels. Figure 6-18 shows CaWO4 fiber coatings obtained by making multiple passes through an immiscible layer formed by the coating precursor and pentadecane (Goettler et al., 1997b). This technique can yield fairly uniform, thick coatings. Because the nature of the precursor determines the quality of the coatings, the most desirable precursors will form thin films upon drying and exhibit minimal gas evolution upon decomposition. Excessive gas evolution tends to dislodge coatings from the fibers and leads to fiber bridging.
Only a few organizations are known to be using the sol-gel immiscible liquid coating technique to produce moderate quantities of fiber; these include the U.S. Air Force Wright Laboratory and McDermott Technology Incorporated (formerly Babcock and Wilcox), Lynchburg Research Center. An advantage of this technique is the relatively low capital cost required to establish a coating capability. The critical element of this technique is the chemistry of the liquid coating precursor.
The heterocoagulation technique takes advantage of electrostatic attraction, which can be generated between fibers and slurry or colloidal particles in liquid media of high dielectric constant, such as water (Malghan et al., 1990; Cinibulk, 1997a). Surfactants are often required to shift the isoelectric point of either the fiber or the coating particle in order to create electrostatic attraction at a given pH. Figure 6-19 shows LaPO4 fiber coatings obtained by dipping Nextel 720 fibers in Betz 1260 flocculant, washing them, and dipping them in a milled monazite slurry (repeated six times) (Goettler, 1997a).
The attractions of this technique are twofold. First, it utilizes an electrostatic driving force to help ensure the uniform coating of fibers. Second, the coatings are deposited as the desired phase, eliminating the chance of reaction between the fibers and intermediate phases often encountered during sol-gel decompositions. A less attractive characteristic of the technique is that it often requires multiple passes to achieve sufficient coating thickness. A continuous coating process utilizing this technique is complicated by the need to perform multiple passes, combined with the need to wash away excess sol after the electrostatic attraction, often (depending on the fiber/coating system) reabsorbing a surfactant, and then washing away the excess surfactant before reexposing the fiber to the sol. Complete washing of the surfactants is often necessary to prevent contamination of the sol, which could change the zeta potential of the sol and therefore the electrostatic relationship between the sol bath and the fibers.
Research into maximizing the electrostatic potential between common oxide fibers and candidate coating particles using surfactant technology in order to obtain sufficiently thick coatings in a single pass of a fiber tow through a colloidal sol would be useful. The technique would still be limited by having to use fine particle slurries or colloidal sols of the desired coating composition. Formation of these slurries or sols could be difficult for some of the complex oxides being considered as fiber coatings.
Currently, no organizations are known to be using the heterocoagulation technique to coat large quantities of fiber tow. Organizations that have evaluated the technique for composite processing include the U. S. Air Force Wright Laboratory and McDermott Technology, Incorporated, Lynchburg Research Center. Neither group is active in this area currently. This technique is attractive, however, because it would require relatively low capital costs to establish a significant coating capability.
RECOMMENDATIONS AND FUTURE DIRECTIONS
Non-Oxide Fiber Coatings
CMCs reinforced with SiC fibers have used a wide variety of fiber-matrix interface coatings, but acceptable toughness (debond characteristics) has only been demonstrated with coatings that contain either a carbon or a BN layer. Oxidation of both carbon and BN has been a major factor limiting the widespread use of non-oxide CMCs. Carbon coatings can readily oxidize in all oxidizing environments, and BN coatings are particularly vulnerable in environments that contain water vapor. However, not all BN coatings are alike, and the effects of BN chemistry and crystallinity on composite properties have yet to be systematically evaluated.
Composites are generally designed to operate below the proportional limit or matrix cracking strength. However, matrix cracking will inevitably occur by accidental exposures to
stresses above the matrix cracking strength. Therefore, a non-oxide, silica-forming fiber coating—capable of preventing oxidation of the fiber-matrix interface—with adequate debond characteristics is highly desirable. Unfortunately no fiber coating can prevent the oxidation of this interface for more than 100 hours at the desired operating temperatures (~ 1,200°C [2,192°F]). Long-term applications such as industrial gas turbines and commercial aircraft engines, however, have life requirements on the order of several thousand hours. Because occasional matrix cracking probably cannot be prevented, oxidation of fiber coatings is a serious concern—limiting the widespread use of non-oxide composites. Therefore the committee recommends the following approach to inhibiting oxidative embrittlement of non-oxide CMCs:
The oxidation resistance of non-oxide fiber coatings should be improved from the current calculated value of a few minutes (at 1,200°C [2,192 °F]) for 1 µm (0.04 mils) BN coatings to the theoretical limit of ~ 100 hours for 1 µm silica-forming coatings.
Techniques should be developed for sealing matrix cracks rapidly.
Clearly, the crack sealing time should be less than the time it takes to oxidize the fiber coatings, which are typically on the order of 1 µm (0.04 mils) thick. Coatings that could potentially improve oxidation lives are silicon-doped BN, AlN, multilayer coatings of BN, and silicon-doped BN with SiC and Si3N4. Boron-additives might be useful for sealing matrix cracks in matrices that contain SiC. A critical question that remains to be addressed is whether the sealed cracks will hold under the cyclic stresses encountered during operation, even if the stress is below the incipient cracking strength of the unoxidized matrix. Therefore, the committee recommends that materials be evaluated under conditions representative of their operating environments.
CVD is the most common method of producing fiber coatings for non-oxide composites. This method is limited in some respects because coating non-uniformity increases with the complexity of the substrate architecture (i.e., from tows to weaves to three-dimensional layups). Because the costs of fiber coatings constitute, perhaps, the largest fraction of the cost of composites today, approaches that lower costs and produce more uniform coatings are needed. Coatings formed from liquid precursors represent one possible approach. Another is in-situ coatings, which have been demonstrated with the Bayer fiber. Therefore, the committee recommends that fiber coatings derived from liquid precursors and in-situ coating approaches be investigated further.
Oxide Fiber Coatings
The development of coatings for ceramic oxide fibers has lagged behind the development of coatings for non-oxide fibers partly because of the insufficient creep resistance of
oxide fibers at elevated temperatures. Recent improvements in creep-resistant oxide fibers, however, have improved interface control in oxide systems. The majority of oxide composite studies, however, have been conducted with porous matrix composites (with no coatings). All oxide coating concepts discussed in the literature have been demonstrated only with model, rather than actual, composite systems. Therefore, although several coating approaches promise of providing damage tolerant oxide composites, they will require further study to prove or disprove their viability. These coating concepts are discussed below.
Porous Coatings and Porous Matrix Approaches
Research will be needed to identify porous coating compositions that are microstructurally stable enough to prevent densification of the coating during composite use. Multiphase coatings may be required to suppress the tendency for porous coatings to become dense, which would make chemical compatibility between the fiber, coating, and matrix more difficult to maintain.
More research is needed to support long-term exposure under anticipated operating conditions for systems without an interfacial coating that rely on porous matrices for graceful failure. For example, mechanical testing and interface characterization, especially transmission electron microscopy, of these porous matrix composites will be required to assess the stability of fiber-porous matrix interfaces and the degree of possible fiber degradation. For applications that require longer operating times at elevated temperatures, alternatives to the UCSB alumina-mullite matrix that are more chemically stable with the fiber reinforcement and more resistant to densification may have to be developed.
Fugitive and Segregant Weakened Interfaces
Both fugitive and segregant-weakened interface concepts represent potentially lower cost options because fugitive carbon interfaces can easily be deposited by CVD or solgel techniques, and the use of segregants eliminates the need for fiber coatings. Therefore, these concepts merit further investigation.
Fugitive interfaces will require substantial long-term testing to confirm that fiber strength does not degrade by contact with the matrix. Most long-term testing of fugitive interfaces has involved only unstressed oxidation heat treatments. Heat treatments under load, when the matrix and fibers are more likely to be in contact, and subsequent mechanical testing have not been done. Fugitive interfaces could also benefit from research into alternative matrices that are stable with commercially available fibers for extended exposures to high temperatures.
Segregant-weakened interfaces are an attractive concept based on the possibility of eliminating the coating phase from the composite fabrication process. This concept has only been demonstrated, however, on model composite systems. A major technical issue is whether the matrix or fiber, or both, would have to be doped to control the fiber-matrix interfacial energy. If polycrystalline fibers require doping, then the effects of dopants on creep performance must be considered. Therefore, additional basic research of this interface concept, utilizing available polycrystalline fibers and candidate oxide matrices, is recommended.
Dense Oxide Fiber Coatings
Many layered oxide coatings show promise as crack deflecting interfaces for oxide-oxide composites. Although the ß-aluminas have been shown to be unstable with Nextel 720 fibers, they are still candidate fiber coatings (especially the less volatile RbAl11O17) for ceramic composites if other creep-resistant oxide fibers become available. Research into an improved creep-resistant Nextel 610 fiber that incorporates additions of rare-earth oxides or garnet-based fibers, may make the ß-aluminas more viable as fiber coatings.
The same recommendations hold true for magnetoplumbitebased coatings. Additional research is needed into sol-gel or solution-based inorganic polymers that could convert to the ß-alumina and magnetoplumbite structures at lower temperatures. The development of ceramic oxide fibers with improved high-temperature creep resistance that are stable with these layered oxides would reduce the need for a low-temperature coating process. It would be preferable, however, to create the desired coating phase at as low a temperature as possible to prevent intermediate phases from reacting with the fibers.
Additional investigation of the layered perovskites (KCa2Nb3O10 and BaNd2Ti3O10) should be limited because of the low probability that these compounds will be stable with available creep-resistant fibers. Layered perovskites may be stable with a creep-enhanced polycrystalline alumina fiber although it has been reported that these coatings are stable with alumina only at moderate temperatures (~1,250°C [2,282°F]). The simpler chemistry of layered ß-aluminas and magnetoplumbites are more likely to be used as coatings for available oxide fibers.
Basic research of weakly bonded, nonlayered oxides will be required to understand the mechanism that controls debonding of some ABO4 compounds from alumina and aluminamullite fibers. Preliminary data on the strength retention of monazite-coated fibers underscore the need for testing the potentially detrimental effect of cation diffusion (solid solubility) on the strength of the polycrystalline fibers. Perhaps certain “A” cations for the ABO4 coating will have to be selected that have limited fiber grain boundary solubility. Substantial, long-term exposure and mechanical testing
programs will be required to prove the viability of ABO4 compounds as fiber coatings.
Coating Processes and Vendors
Only two institutions are known to be using the sol-gel immiscible liquid coating technique (Wright Laboratories and McDermott Technology, Incorporated, Lynchburg Research Center). The immiscible liquid coating process is rather slow. For example, McDermott Technology, Incorporated, deposits carbon coatings onto Nextel 610 fiber at a rate of approximately 152 cm (60 in.) of fiber tow per minute. If coating speeds could be increased, integrating the coating step into the fiber manufacturing process would become much more feasible, thus potentially lowering costs. Therefore, programs focused on investigating and improving the quality of these coatings as a function of coating speed are recommended.
No organizations are known to be using the heterocoagulation technique for coating large quantities of fiber tow, despite its potential for depositing uniform coatings without (often problematic) intermediate phases. This is most likely because the heterocoagulation technique often requires multiple passes to build up sufficient coating thickness. Heterocoagulation, however, is attractive because it requires a relatively low capital investment. Therefore, the committee recommends that research be done on surfactants that could yield higher charges on the coating particles and/or fibers. It may be possible to absorb soluble inorganic polymer precursors onto the fiber surfaces by heterocoagulation, which would eliminate the need to form a fine powder of the desired coating composition. The inorganic polymer-derived coatings would probably be very thin, however, and would therefore work best for coating concepts that rely on debonding at the interface rather than within the interphase.