BACKGROUND

Since the efficiency of a gas turbine engine scales directly as the difference in temperatures between the input working fluid temperature and the exhaust temperature, the search for higher temperature materials for turbine applications is ongoing. This is particularly the case in sealing applications in the flow path and in secondary cooling circuitry. A common seal in the art is an abradable seal that relies on channels cut in a sealing material by a moving component such as a blade tip to create and maintain a close fit between, for instance, the blade and a shroud in a gas path to minimize leakage to maintain high efficiency and to limit fuel consumption.

Ideal sealing materials are elastomeric in nature that exhibit a high degree of compliance and springback. Common elastomeric materials are organic in nature and are limited to sealing applications in cooler sections of a turbine engine. A metallic, intermetallic, or ceramic material with the requisite compliance in springback would be useful as a seal material in gas turbine engine applications.

SUMMARY

High temperature nanocellular foams for application as a seal material in gas turbine engines are presented. The foams can be described as having pores, interconnecting ligaments, and nodes where three or more ligaments intersect. The ligament cross section thickness is about 5 nanometers to about 200 microns and the distance between nodes is about 15 nanometers to about 1000 microns. Nanocellular foams for seals can be aluminides such as nickel aluminide, silicides, such as molybdenum silicide, carbides, such as silicon carbide, and others.

A method of forming a nanocellular seal material comprises first depositing two or more unreacted precursor powders on a surface to form a coating. Energy is then applied to the coating to cause at least one of the powders to undergo a change in state to form a ligament structure with a cross section thickness of about 5 nanometers to about 200 microns and a distance between nodes of about 15 nanometers to about 1000 microns.

BRIEF DESCRIPTION OF THE DRAWINGS

1 is a schematic representation of a foam. FIG. 2 is a plot illustrating the specific strength and Young's Modulus variability achievable with nickel and silicon carbide nanocellular foams.

FIGS. 3A and 3B are schematics showing how nanocellular foams mitigate internal stress concentrations.

FIG. 4 is a process to produce MoSi ₂ nanocellular foam.

FIG. 5 is a process to produce MoSi ₂ nanocellular foam seal material.

FIGS. 6A and 6B are photomicrographs of nanocellular MoSi ₂ foam before and after thermal processing.

FIGS. 7A-7D are photomicrographs showing different MoSi ₂ nanocellular foam microstructures.

DETAILED DESCRIPTION

Metal, intermetallic, and ceramic foams with pores in the nano or submicron size range are needed to advance cellular material applications in gas turbine engines.

A schematic illustration of a cellular solid variously referred to as a foam or porous material is shown in FIG. 1. Foam 10 comprises pores 12 surrounded by ligaments 14 and nodes 16. Nodes 16 are defined as the junction of three or more ligaments.

The mechanical or structural properties of a foam can be modeled based on standard elasto-plastic analysis of the ligament skeleton. For this purpose, elastic and plastic properties of the ligament material, such as Young's modulus and yield strength, are used as they exist for such materials in the dense or bulk form. Experimental observations show that even this analytical approach is optimistic and generally the performance of most cellular materials fail to exceed the rule of mixture projections. In simple terms, the rule of mixtures suggests the mechanical properties of cellular solids scale with the volume of ligaments. In other words, a cellular solid with 50% open space should have 50% the strength of the dense material. Analytically derived predictions are a bit more complex than that, as the performance depends on the architecture of the ligaments and the uniformity of the architecture all through the cellular solids, but in general the prediction of mechanical properties, M, such as Young's modulus, yield strength, fracture strength or fracture toughness all scale to the ratio of the densities, p, of the cellular solid to the density of the dense or solid mater as:

M _c = M _s ^■ (p _c I p _s Y ^■ f (ligamentarchitecture) Where suffix c refers to cellular solids and the suffix s refers to a dense solid, made of identical material. The exponent n is dependent on the type of mechanical properties and experimental values tend to be lower than the models.

The innovation and core basis of this invention is that if sound material science principles are applied and celluar solids are made with adequate microstructure control, then it is not necessary for the performance of cellular solids to be bound by this "rule of mixture". The innovation rests on two key building blocks: (a) a conceptual understanding that if the cellular solids are built with ligament sizes approaching 100 to 100,000 times the atomic dimensions, mechanical behavior cannot be assumed to be identical to that of a bulk solid, and (b) processing of such a nanocellular solid with ligament sizes approaching the nanometer scale cannot be achieved with routine foaming techniques. Instead, the invention requires an in-situ approach of reactive synthesis involving, either melting, solid state diffusion or chemical reduction of at least one component contributing to the integrity of the nano-scale ligaments. It is critical to note that primary focus of this innovation is the size of ligaments and not just ligament architecture, or the size and volume fraction of pores.

In conventional thinking, for example, a cellular solid or foam with 50% porosity suggests an image of a solid with visible pores and a weak skeletal material that may not stand up to applied stresses or impacts. It is not easily recognized that such a cellular material with 50% porosity can also be made with the pore and ligament size in the range of 0.1-100 microns that are not visible to the naked eye and wherein the structural behavior of ligaments and the pores is not treatable in a conventional mechanical sense. For all practical purposes, such a material may appear like a "dense" solid. For example, even if such a nanocellular material had an open cell structure, the air flow or fluid flow through such small pores may face an unprecedented surface tension resistance. Based on sound and well recognized principles discussed in the following, the potential to achieve unprecedented, high strength-to-density, unusually high strain tolerance, and other unusual combinations of physical and chemical properties exists with such a material when ligament sizes fall below a threshold and surface-to-volume ratio exceeds another threshold.

It is a common experience that thin glass plates can be bent to achieve visible deflection in spite of glass being brittle. Paper thin (~ 25 micron) silicon wafers can be bent double without cracking. Achieving high deflection or "ductility" with a thin brittle material shows that the absolute size of material matters. In brittle materials, it is well recognized that, besides reducing the outer fiber tensile stress in bending of thin materials, fracture strength is also increased with reduced defect density that limits crack initiation, with decreasing volume of the material. Statistically the defect population associated with any material processing method decreases with decreasing volume of the material produced. This is well exploited in manufacturing of high strength whiskers and fibers of ceramic materials such as, AI2O ₃ , SiC, S1 ₃ N4, glass, and metallic glass.

In the same vane, the ductile behavior of metals results from the generation and motion of line defects called dislocations, which are characterized by a Burgers vector of dimension -0.3 nm. The strengthening of metals critically depends on the scale of microstructure that inhibits dislocation motion and it is well known that unless the mean free distance between the inhibiting microstructural features is of the order of Burgers vector, 1 micron or less, no significant strengthening occurs. This forms the backbone of increasing strength of dense metals by grain refining (Hall-Petch hardening) or precipitation or dispersion hardening. On the other extreme, it is well known that when free standing material dimensions are in the neighborhood of 1000 to 10,000 times the Burgers vector, as in whiskers, statistically the dislocation density becomes so low that material starts displaying very high strength, approaching theoretical strength of the order of G/30 where G is the bulk shear modulus.

For most engineering structural analysis, it is not necessary to understand the nature of dislocations. Knowledge of bulk elastic and strength properties and assuming materials to be a continuum body is sufficient. Yet it is well recognized in the classical handbook by Peterson (Stress Concentration Design Factors by R.E. Peterson - 1953) for assessing elastic stress concentration of notches of various shapes that notch sensitivity starts decreasing in steels when the notch radius starts dropping below 0.050 inches or 1,270 microns. This size effect is not anticipated in the framework of classical continuum elastic theory, the solution of which depends only on the relative dimension ratios. This challenges conventional thinking that all pores in a porous material are potential stress concentration sites and therefore potential failure sites, irrespective of size.

A key and well recognized point is that the mechanical behavior of materials such as strength, ductility, and even stress distribution are size dependent. However, it is not generally recognized that the relation only becomes significant below a threshold of 1-100 microns depending on whether the material is ductile or brittle and whether it is a crack initiating defect, or dislocation density. More importantly, below the threshold size, these properties generally are enhanced significantly. Incorporating this favorable deviation from bulk behavior in nanocellular ligaments is the key concept of this patent.

Based on experimental data for metals and ceramic whiskers and a wide range of recent work where testing of free standing nanopillars with low defect content has been possible, it has been shown that strength, σ, follows a power-law relationship with the size, d, as follows:

σ = Αά ^~η

(2) where σ is measured in MPa and d is measured in nm, and A~lxl0 ⁴ -lxl0 ¹⁰ and n~0.5- 1.5.

It is easy to see from the relationship that every 10X decrease in diameter leads to 25X increase in strength. The size range in which this rapid rise in strength over the bulk strength occurs varies from material to material and is sensitive to the process by which the material is made. The available experimental data suggest that the threshold is in the range of 100 microns for brittle material and is of the order of 10 microns for ductile metals.

Combining this relationship with the "rule of mixture" relationship stated in equation (1), it is easy to show that strength prediction for cellular solids based on "rule of mixture" can be easily surpassed in nanocellular material if the ligament size is brought below the threshold at which material properties such as strength are size dependent.

Thus specifically

(J _c = a _s ^■ Ad ^~ " - (p _c I p _s Y ^■ f (ligamentarchitecture)

(3)

Based on a similar analysis specifically applied to square, triangular and honeycomb cellular architecture and based on some published data on SiC and Ni whiskers, the strength-to-density ratio of hypothetical nanocellular materials with ligament thickness in the range of 500-1000 nm was modeled and compared with many dense solid systems as shown in Figure 2. The strength-to-density data are plotted against Young's modulus since it is specifically meaningful for the nanocellular materials for high temperature structural applications. Indicators for the fields shown in FIG. 2 are as shown:

It is apparent from the figure that the specific strengths of nanocellular nickel foams with 500-1000 nm ligaments are superior to both conventional nickel foams and dense nickel. In addition, the modulus of both nickel and SiC nanocellular foams is a function of cell structure and can be intentionally varied.

In practice, dense high temperature structural materials are always subjected to thermal gradients, ΔΤ. Thermal gradients impose a strain on the material as a result of thermal expansion. The imposed strain then imposes stress, σ, on the material proportional to Young's modulus, E, in the direction of the thermal gradient and strain as expressed by the following relationship:

σ = EaAT

(4) where a is the coefficient of thermal expansion (CTE).

The imposed stress varies as the thermal gradient varies with heating, holding and cooling cycles, along with other structural loading, which then causes material to fail in fatigue. This kind of failure is referred to as thermal mechanical fatigue. From Equation (4) it is clear that, for dense solids, a lower Young's modulus, E, and a, will lower the imposed stress, σ. For cellular solids, while overall Young's modulus is reduced with reduced density, the structure can become even more strain tolerant than indicated by Equation (4). As shown in Figures 3 A and 3B, while dense material subjected to hot spot 20 experiences compressive stresses in the center of the hot spot, a cellular material subjected to the same hot spot locally absorbs the stress by distorting individual ligaments elastically. The situation is much more complex but it is easy to see that in cellular solids, with finer cell sizes and stronger ligaments, the material can tolerate sharper thermal gradients with improved thermal fatigue resistance since constraining effects that cause high stresses in dense materials are absent. Both these characteristics are inherent in the nanocellular material of the invention as discussed before. This opens up design space for smaller features, such as cooling holes in turbine blades; as such features will be hundreds of time coarser than the cell size in nanocellular materials.

It is well known that porosity in high temperature refractory bricks improves their thermal shock resistance. However, these observations are not fully exploited in high temperature structural materials. This invention achieves that goal with nanocellular solids.

In summary, the benefit of nanocellular structures results from decoupling the mechanical properties of the ligament from its bulk behavior by exploiting the strong size effect that exists below certain threshold ligament sizes. Smaller ligament size for a given porosity level leads to smaller pore size since these factors are geometrically coupled. The overall decrease in the cell size and ligament size in nanocellular material not only leads to high strength-to-density but also manifests higher strain tolerance in thermal environments with improved thermal fatigue resistance. In addition, increasing strength with decreasing ligament size in conjunction with improved strain tolerance due to lesser local constraints is enabling for the application of brittle high temperature materials such as intermetallic silicides and aluminides for gas turbine engine applications.

The elimination of local constraints in nanocellular materials is also enabling for non-cubic materials with anisotropic CTE such as TisSi ₃ . Dense, non-cubic materials are sometimes self destructive in polycrystalline form as each grain imposes stress at a grain boundary owing to anisotropic thermal contraction. In this context it is critical to bring the ligament size below a threshold at which bulk behavior is not emulated. Processing such compounds as nanocellular structures may help suppress such behavior. The projected improvement in the strength-to-density data for nanocellular materials shown in Figure 2 relies on the anticipated power law increase in strength with decreasing ligament size expressed in equation 2. A complete scientific understanding of this behavior is still a subject of intense research but a few basic concepts are indisputable. As alluded to in the preceding section, in ductile metals deformation occurs by dislocation motion and the strength improvement occurs due to a decrease in dislocation density. In this context it is also known that a large surface-to- volume ratio also causes easier annihilation of dislocations and contributes to this behavior. In brittle materials, where strength is primarily limited by premature crack initiation and propagation, smaller section size with decreased defect density benefits. It is understood that the creation of dislocation-free metal or defect-free brittle material is process dependent. In general whiskers or extremely fine single crystal fibers made from vapor or liquid phase show the best improvement in strength. It is also known that, in this case, if a broken whisker is continued to be tested, it may manifest even higher strength as locations with higher dislocation or defect density are eliminated. All commercially available high strength fibers used for composite materials are typically grown from vapor or liquid phase. Current research also shows that the same material carved out of small dimensions behaves differently depending on the processing history. For example, nanosize pillars machined from commercial grade molybdenum do not display as high a strength as a molybdenum fiber grown by eutectic phase separation in a Ni-Al-Mo eutectic system. The key point is, if one were to exploit the high strength of nanosize ligaments in nanocellular materials, adequate processing of ligaments is very critical.

Based on the preceding discussion, ligaments in a nanocellular material are likely to have the projected higher strength if they are processed in situ either via reactive synthesis between two or more elements or compounds, or via vapor phase, eutectic separation, solid state precipitation, or direct reduction to a metal from its most stable oxide, hydride, sulfide, nitrate, carbonate, or other compound state. In conventional foam manufacturing, even if the ligaments are produced from liquid state, the ligament size is generally too large since some fumigating compound is used to create pores. With some model intermetallic systems based on silicides and aluminides and ceramic systems based on SiC, we have identified a powder metallurgical approach coupled with reactive synthesis for making nanocellular foams with good integrity of the ligaments and with varying compliance for application as seal materials. In accordance with this invention, a nanocellular foam is defined as a foam wherein the ligament cross section thickness is from about 5 nm to about 200 nm and the distance between nodes is about 30 to about 1000 nm. In another embodiment, the ligament cross section thickness is from about 5 nm to about 10 μιη and the ligament length is at least three times the cross section thickness.

The nanocellular foam seal material of the invention may be a metal, intermetallic compound, ceramic or mixtures thereof, depending on the operating environment. In an embodiment, the foam may be a closed or open cell structure and may have a porosity from about 5% to about 95%.

One example of a nanocellular MoSi ₂ foam was produced by the steps outlined in Fig. 4. In this processing approach elemental Mo, and Si powders were mixed in the correct stochiometric proportion of 33.3 atom % of Mo + 66.6 atom % of Si, to make MoSi ₂ (Step 30). The blended powder mixture was cold compacted at sufficiently low temperatures as not to react the elemental species (Step 32). A photomicrograph of the as-compacted SiMo powder is shown in Fig. 5 A. Larger Si particles 40 are surrounded by finer Mo particles 42. Subsequently the cold compact was heat treated at a temperature slightly below the 1410°C melting point of Si. Local reaction of Mo and Si ensued, resulting in Si melting and being wicked into the surrounding ligament structure formed by Mo powder. This led to in situ formation of a cellular structure with porosity created at every place formerly occupied by Si powder. The process is greatly helped by the exothermic reaction between Mo and Si leading to the formation of high melting and more stable MoSi ₂ . A photomicrograph of the structure after foaming is shown in Fig. 5B. MoSi ₂ ligaments 14 and nodes 16 are shown encasing pores 12.

Different cellular architectures can be achieved in nanocellular foams produced by the powder metallurgical process of the invention by varying the input powder shape. In addition, any and all known powder metallurgical techniques for mixing, aligning, and otherwise altering the powder distribution can be applied. Powders with bimodal, trimodal, or higher order size distributions can be used. It is a requirement of the invention that at least one component undergoes a transformation as a result of energy applied to the powder compact. The transformation can be in the form of melting, selective or transient melting, welding, evaporating, chemically reacting, solid state diffusion, or a combination thereof to form a ligament of appropriate size, architecture, and defect density. Two or more powders may also react as a result of the applied energy to form a ligament.

In an embodiment, coated powders, fumigating compound powders, fibers, whiskers, and other forms known in the art can be used to form the powder compact. In another embodiment, powders may be a high temperature elastomer that may react to form a nanocellular foam with enhanced compliance.

Four different cellular architectures of MoSi ₂ nanocellular foam are shown in FIGS. 6A-6D. In FIG. 6 A, the foam was produced using equivalent sized Mo and Si powder. In FIG. 6B, the foam was produced using spherodized Si powder to produce spherical pores 12. In FIG. 6C, bimodal spherical Si powder was used to produce spherical pores 12 with a bimodal size distribution. In FIG. 6D, the foam was produced using Mo powder in the form of platelets. In an embodiment, if hollow Si powder were used, a higher volume fraction of porosity will result.

In another embodiment, the mechanical integrity of the ligaments can be improved by adding Ni as a sintering aid for Mo. It is suggested that a low melting Ni- Mo eutectic allowed the Mo ligaments with increased density to form during the heat treat of step 34 in FIG. 4. FIGS. 7A and 7B are photomicrographs showing the microstructure of MoSi ₂ nanocellular foam with (FIG. 7A) and without (FIG. 7B) small Ni additions. The improved mechanical integrity of ligaments 14 is evident from the apparent higher density of the ligaments in FIG. 7A.

In one embodiment, a nanocellular foam seal coating comprising MoSi ₂ is deposited on a substrate for a gas turbine application by the steps outlined in FIG. 8. Mo and Si powders are first sourced for the deposition (Step 40). Proper size distributions of the powders are then created by classification and other methods known in the art (Step 42). A large controllable variation in nanocell foam seal microstructure can be obtained by proper selection of powder size distribution including bimodal, trimodal, and higher order pore size distributions. The next step is to form an unreacted composite layer of Mo and Si particles on a turbine component substrate (Step 44). In one embodiment, the powders can be deposited by cold spray. In cold spray deposition, the particles are accelerated to velocities between 500 to 1000 meters per second. Upon impact, the deformable particles experience plastic deformation and bond to the substrate. In the case of Mo and Si deposition, the deformable Mo particles will encase Si particles to form an unreacted composite foam nanocellular foam precursor. A MoSi ₂ nanocomposite nanocellular foam is formed by heating the composite seal structure to slightly below 1410°C to form the foam as described earlier. (Step 46)

Other methods of depositing unreacted Mo and Si composite nanocellular foam precursor structures include the separate deposition of Mo and Si by thermal spray, plasma spray, high velocity oxy fuel (HVOF), cathodic arc deposition, and others known in the art.

In order to form an unreacted nanocellular foam precursor layer, each component is preferably deposited from a separate source.

The process outlined in FIG. 9 describes in general the formation of nanocellular foam sealant layers of the invention for application in the gas path and other cooling circuitry in turbine engines. In the first step, unreacted nanocellular foam precursor particles are deposited on a substrate to form a coating (Step 50). The particles can be metals, intermetallic compounds, ceramics, or mixtures thereof that will at least undergo a change in state when energized to form a nanocellular foam layer with ligaments (Step 52). Candidates include, but are not limited to, silicides, aluminide intermetallics, ternary intermetallics, carbides, oxides, silicates, nitrides, ternary or multicomponent compounds, metallic glasses, superalloys, MAX phases and mixtures thereof. Examples are Ni and Al to form nickel aluminide, Ti and Al to form titanium aluminide, titanium and silicon to form titanium silicide (TisSi ₃ ), Mo and Si to form molybdenum disilicide, Si and C to form silicon carbide, and other compounds. MAX phases such as Ti3SiC2 are high temperature machinable ternary ceramics and are described by Barsoum eta 1. in American Scientist, 89, 334 (2001). It is advantageous to design a nanocellular foam sealant layer with a ligament architecture with a high length to diameter ratio that exhibits extremely high compliance and spring back to minimize gap width in the seal.

In another embodiment, the nanocellular foam seal material may be a nickel based, cobalt based, iron based superalloy or mixtures thereof.

Coated polymer templates such as PMMA microspheres as mentioned above can be included to increase porosity of the sealant coating. The templates can be coated with any proven nanocellular foam precursor to produce foam architecture with varying and controllable density. Again it is important that the precursor components are deposited from separate sources to form a composite of unreacted components before energizing the coating. Energizing the coating to form a nanocellular foam can be by applying any form of thermal energy, electromagnetic energy, electric energy, or other energy sources.

In another embodiment the deposited powder technique can be easily combined with a conventional foaming technique. Coated fumigating material powder can be incorporated in the process. As long as there is some sub-step that involves a change of state such as, for instance, melting and wicking, diffusion of one or more of the components, alloying or others, a defect free ligament structure is reasonably guaranteed.

While the invention has been described with reference to an exemplary embodiment(s), it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment(s) disclosed, but that the invention will include all embodiments falling within the scope of the appended claims.