Traditional manufacturing processes using powder metallurgy initially produce a near net shape part which is only 50–70% dense. These ‘green’ parts then undergo further processing to achieve full density and the desired mechanical properties. The densification is done either by lightly sintering and infiltrating with a lower melting temperature alloy or by a high temperature sintering alone. In the first method, the part's dimensional change is typically only ˜1% making it suitable for fairly large (˜0.5 m on a side) parts, but the resulting material composition will be a heterogeneous mixture of the powder material and the lower melting temperature infiltrant. Sintering the powder to full density will result in a homogeneous final material, but a part will undergo ˜15% linear shrinkage if it starts out at 60% density. For this reason, full density sintering is typically only used for smaller (<5 cm on a side) parts.
In many critical applications (structural, aerospace, military), a material of homogeneous composition is preferable because of certification issues, corrosion issues, machinability, or temperature limitations that might be imposed by the lower melting point infiltrant. Further, designers of metal components are not accustomed to working with composites of heterogeneous composition, and so this creates a psychological barrier.
The ability to create very large parts with homogeneous composition via powder metallurgy builds on all of the benefits of PM processing. The key here is in using an infiltration step to densify the green part without any significant dimensional change, but resulting in a final material with homogeneous composition. This allows fabrication of homogeneous net shape parts in a wide variety of sizes using solid freeform fabrication, metal injection molding, or other PM processes. There also exists the potential of matching an existing commercial material system.
The general concept is to use an infiltrant to fill a powder skeleton, that is similar to the base powder, but contains a melting point depressant. The infiltrant will quickly fill the powder skeleton, then as the melting point depressant diffuses into the base powder, the liquid will undergo isothermal solidification and the material will eventually homogenize. This process will allow more accurate control of dimensions in large parts with uniform or homogeneous microstructure.
To further explain this concept,
The success of such an infiltration requires the time scale of the infiltration to be much faster than the diffusion of the melting point depressant and the subsequent homogenization. There are various techniques that can enhance this tradeoff, but the selection of a material system has the greatest impact on the infiltration and diffusion rate.
These and other features, aspects, and advantages of the present invention will become better understood with regard to the following description and accompanying drawings, where:
Transient liquid phase (TLP) brazing is commonly used to repair cracks and bond materials together. This traditional process involves the mechanism of a melting point depressant diffusing into a base material and undergoing isothermal solidification. Narrow gaps are necessary for the nickel brazing alloys to fill the capillary channel and solidify in a reasonable amount of time. The solidification time is limited by the diffusion of the melting point depressant into the base metal. Gaps wider than ˜50 μm would result in excessively long solidification times.
Wide gap brazing has been developed to allow brazing of gaps in excess of 100 μm. Powder similar to the base material is used to fill the gap prior to the addition of the brazing alloy. This allows the liquid brazing alloy to fill large gaps and solidify faster.
There are two studies from the early 1980's involving the infiltration of a powder skeleton with the aim of creating a useful steel part. Banerjee attempted to use cast iron to infiltrate a skeleton of pure iron. The cast iron would freeze off within a few millimeters of contacting the skeleton due to the high diffusivity of carbon. Thorsen was successful in infiltrating a sintered steel skeleton with a Fe—C—P alloy, but an interconnected network of phosphides resulted in a very brittle final part.
Infiltration of a pure nickel powder skeleton with any of the commercial nickel brazing alloys. The melting point depressants in the preexisting nickel brazes are phosphorous, boron, and silicon. The alloys also typically contain other elements that provide additional strength such as chromium, iron, molybdenum.
Infiltration of pure nickel powder skeleton with a binary alloy of nickel and silicon.
Infiltration of a nickel chromium skeleton with a nickel chromium silicon alloy.
Infiltration of a high melting temperature inconel alloy powder skeleton with a similar alloy containing a melting point depressant such as boron, phosphorous, silicon, tin or a combination thereof.
Infiltration of a pure aluminum or aluminum alloy skeleton with a similar alloy containing silicon or lithium as a melting point depressant.
Infiltration of a pure copper or copper alloy skeleton with a copper silver, copper titanium or other alloy with melting point depressed.
Several techniques have been developed and used to overcome difficulties with diffusion occurring during the infiltration of a skeleton.
Gate Mechanism to Separate Molten Infiltrant from Skeleton
Physical separation of the liquid infiltrant from the skeleton prevents premature interaction or diffusion before the infiltration begins. If the infiltrant is already in physical contact with the skeleton prior to melting, the liquid will begin to wick into the part as soon as it becomes molten. In this case, the melting of the infiltrant or other transient thermal processes will control the infiltration rate. Controlling the introduction of the liquid can be done via a gate that can be actuated at a controlled point in time, once the liquid infiltrant has reached the desired steady state temperature. Several such gating mechanisms have been used in practice.
A simple method is to suspend the skeleton prior to infiltration and dip it into a bath of the molten infiltrant. If the part is too delicate to hang under its own weight, then a special mechanism should be used to allow a gated infiltration with the part resting in a crucible. It can be difficult to create a fluid seal that will hold at the infiltration temperature, but using a crucible material that is not wet by the infiltrant makes a seal possible. Two simple mechanisms have been used successfully so far. The first is a vertical alumina plate used to separate the two halves of a rectangular crucible. The shape of the plate must match the cross-sectional profile of the crucible, so a bisque fired alumina was cut and filed to maintain less than 1 mm gap when fitted to the crucible. This gap was sufficient to hold a 2 cm deep pool, but a deeper pool would require closer tolerances or filling of any gaps with a coarse alumina powder. A more elegant solution is to use an alumina tube with a cleanly cut end to sit vertically with the end flush with the bottom of the crucible. The infiltrant is placed inside the tube and will contain the melt until the tube is lifted from above.
Several other methods can be used for gating the infiltration. A custom crucible could be fabricated with a hole at the bottom. This hole could be plugged with a simple rod to prevent infiltrant flow until the rod is removed. Another method is to tip a container of infiltrant allowing the liquid to flow out of the tundish. Further, the vessel used to contain the infiltrant could be flexible. A woven cloth of alumina fibers has been used to contain liquid metal. A cloth bag could be used to contain the melt and then opened up to allow the liquid to flow out.
The actuation of any type of gate requires a linear or rotary motion actuator passing through the gas-tight shell of the furnace. In the case of nickel parts fired in a forming gas atmosphere, the feedthrough can be a rod sliding through a slightly oversized hole in the shell. If the internal pressure in the furnace is maintained to several inches of water, the leak will not allow air into the furnace to contaminate the atmosphere. In applications where atmosphere purity is more critical, several linear and rotary motion feedthroughs are available commercially for high vacuum applications.
If the liquid infiltrant has a composition that is greater than its equilibrium liquidus composition at a given temperature, it will have the capacity to absorb additional material from the skeleton and dissolve the part. This can happen very quickly because of the high diffusivity in liquids. It can be a significant problem especially when a large melt pool is used.
If the infiltrant composition is known exactly, the process temperature could be selected to exactly match the liquidus temperature for that composition, but this requires very accurate process control. A more robust method for ensuring that the liquid is saturated, is to put it in contact with solid and allow it to reach its equilibrium composition for whatever process temperature it is at. The liquid must be in contact with the solid for a long enough time to reach equilibrium. This time will depend on the surface area of liquid solid interaction and mass transfer in the liquid, determined by diffusion and convection.
For example, to presaturate the nickel silicon infiltrant, excess nickel powder is added to the crucible of infiltrant. The large surface area of the powder enables equilibration in a reasonable amount of time. The amount of excess nickel added is important. Too little would result in it completely dissolving and the liquid still not reaching its equilibrium liquidus composition. Too much would result in solidification of the infiltrant pool. The appropriate amount is determined by considering the extreme cases for a window of processing temperatures.
There are several main considerations that are fundamental to successfully creating homogeneous parts via infiltration of a powder skeleton. Problems arise associated with premature freeze off of the infiltrant, erosion of the skeleton, and part distortion. This section addresses each of these issues by identifying probable causes and possible solutions.
Preventing Premature Freeze-Off of the Infiltrant Before it Fills the Skeleton
As was mentioned earlier, the time for the liquid to fill the skeleton must be significantly shorter than the time it takes for diffusion of the melting point depressant and the resulting isothermal solidification. If the alloying element diffuses too quickly, it will freeze off before the part has filled. Utilizing a gating mechanism during the infiltration as mentioned under details of execution is critical to minimizing the infiltration time. The other factors that control the infiltration rate are based on fluid mechanics.
The capillary force that draws the liquid into the skeleton is controlled by the surface tension of the liquid infiltrant. This force acts at the solid-liquid interface, which can be controlled by the powder size. Smaller powder will have a larger driving force proportional to 1/r. However, the smaller pore size will cause a larger restriction to the flow due to viscous drag. For flow through a cylindrical tube, the viscous drag is proportional to 1/r2. This means that infiltration should occur faster in a skeleton made from larger powder.
There will be a limit to the maximum powder size imposed by the necessary capillary rise height. If the pore space in the skeleton is modeled as a cylindrical channel of radius r, the driving force would be equal to 2πr·γst·cos(θ), where θ is the wetting angle. As the liquid rises, it must supports its own weight, equal to πr2·ρgh. The pore size radius must be small enough to yield a capillary rise greater than the height of a part. Using the value of surface tension for pure nickel at 1500° C. (1.7 N/m), assuming perfect wetting, and a part height of 0.5 meters, the pore radius must be less than 80 μm.
We have been able to measure some typical infiltration rates of the Ni-10Si infiltrant filling a 50–150 μm nickel skeleton. This was done through hanging the skeleton by a wire through the roof of the furnace and measuring the force increase on the wire. By isolating the surface tension and buoyancy forces, we were able to relate the force to the increasing mass of the part due to picking up the liquid. The liquid filled an 8 cm tall skeleton in approximately one minute. Other liquid metals have viscosity and surface tension that are similar so this rate should not change drastically with material system.
Now we move to discussion of the diffusion rate that will control the isothermal solidification of the infiltrant and the eventual homogenization of the skeleton. Since the liquid fills a small skeleton in approximately one minute, the isothermal solidification would ideally take place over an hour or two. The diffusion rate will be controlled primarily by the material system chosen. This was a reason for using Si as a melting point depressant in Ni rather than B or P, which diffuse faster. Diffusivity can have a strong dependence on temperature, since it is an activated process that follows an Arhennius dependence. Controlling infiltration temperature allows for some control of the diffusivity for a given material system. Reduced temperature should allow more time for the liquid to fill the skeleton before freezing.
In experimental tests, we have observed much faster mass transport than would correspond to the experimental values of diffusivity. It is possible that there is a reaction occurring at the solid liquid interface. For some material systems, the formation of a particular phase of intermetallic at the interface could accelerate the mass transport. The motion of the solid liquid interface could also be leaving behind solid that is high in composition of the melting point depressant.
Selection of a material system is critical to controlling the time scale of the isothermal solidification. In particular, the diffusivity of the melting point depressant will have the greatest effect on the freezing. Using a slower melting point depressant, such as tin, could drastically increase the amount of time the skeleton has to fill with infiltrant before freezing begins to occur.
Coating the powder skeleton (or just the raw powder) with some type of finite time diffusion barrier would keep the melting point depressant from leaving the infiltrant until the liquid has filled the part. Such a diffusion barrier could be another metal that has a lower diffusivity of solute. The thickness of the barrier could be selected so that it would only last for the duration of the infiltration. It would then allow the solute to diffuse through, allowing isothermal solidification and eventual homogenization.
As the liquid infiltrant enters the skeleton, it has a tendency to leave an erosion path. This occurs to some extent in most powder metal infiltrations, but it usually is limited to the initial 1 cm at the base of a part. In those cases, the part to be infiltrated can be placed on top of a sacrificial stilt where the erosion occurs. In the nickel silicon system, the erosion tends to propagate for several centimeters into the part. The appearance is similar to a riverbed and one example is shown in
Since the infiltrant was presaturated with nickel, it is surprising that more nickel is dissolved (erosion of the skeleton) as the liquid fills the part. Even if previously saturated, the infiltrant would have the capacity to absorb additional nickel if it increased in temperature. An exothermic reaction at the solid liquid interface could be generating heat and causing the erosion. The free energy of the solid at the homogenized composition is substantially lower than that of the initial heterogeneous system. Limiting the speed of that reaction could allow dissipation of the heat and minimize the erosion. This could be done by slowing down the flow of the infiltrant using some type of flow restriction.
Temperature control within the furnace could change the diffusion rate and the solubility of the infiltrant. A temperature variation with time as the part fills could compensate for heat generation within the part. Alternatively, a temperature gradient could be set up within the part. To gain insight into the formation of the erosion paths, visual inspection of the part during the infiltration would show when the erosion develops and how it grows.
Since the processing is done at temperatures close to the melting point of the skeleton, the mechanical strength is very low. Part distortion was first encountered when suspending odd shaped parts above the melt. A mild manifestation of this can be seen in the serpentine part in
In this homogenizing infiltration technique, the capillary body being filled is a powder skeleton, rather than a crack or narrow channel as is the case in known techniques for crack filling or brazing. This powder skeleton has been created as a net shape or near net shape part through a powder metallurgy process such as solid freeform fabrication or metal injection molding. Part size often dictates that the filling distance for the infiltrant is much greater than in traditional brazing applications. The corresponding bulk flow of infiltrant, especially through the entrance region, is quite large and can lead to erosion at the base of the part. Finally, the isothermal solidification and homogenization in a powder skeleton is different from in a narrow channel, with walls of semi-infinite thickness. The final composition of the part will be determined by the equilibrium composition of infiltrant and initial powder and their volume fractions.
Several techniques have been developed to overcome the challenges of a homogenizing infiltration. Gating the infiltration controls the time the liquid and solid are in contact with each other and prevent premature freezing. Several gating mechanisms are described and some have been successfully used in practice. Presaturation of the infiltrant is necessary to prevent excessive dissolution of the skeleton. Supporting the part in a bed of loose ceramic powder can prevent slumping of delicate parts, since the base material can soften at the infiltration temperature. A large skeleton should be filled with infiltrant prior to its isothermal solidification. Choice of materials, powder size and infiltration temperature can maximize the filling distance according to the relationships described. A coating can be applied to the powder to act as a diffusion barrier and slow down the solidification. The erosion of the skeleton could be caused by an exothermic reaction during the infiltration. Imposing a flow restriction would allow time for the generated heat to be dissipated and prevent the dissolution of the skeleton.
H. Zhuang, J. Chen and E. Lugscheider, “Wide gap brazing of stainless steel with nickel-base brazing alloys” Welding in the World. Vol. 24, No. 9/10, pp. 200–208 (1986)
S. Banerjee, R. Oberacker and C. G. Goetzel “Experimental Study of Capillary Force Induced Infiltration of Compacted Iron Powders with Cast Iron”. Modern Developments in Powder Metallurgy. v 16. Metal Powder Industries Federation: Princeton, N.J. pp. 209–244 (1984)
K. A. Thorsen, S. Hansen and O. Kjaergaard, “Infiltration of Sintered Steel with a Near-Eutectic Fe—C—P Alloy,” Powder Metallurgy International, Vol 15, No. 2, pp. 91–93 (1983)
This claims priority to U.S. Provisional application No. 60/206,066, filed on May 22, 2000, the full disclosure of which is fully incorporated by reference herein.
The United States Government has certain rights in this invention pursuant to the Office of Naval Research Award # N0014-99-1-1090, Research in Manufacturing and Affordability, awarded on Sep. 30, 1999.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/US01/16427 | 5/21/2001 | WO | 00 | 5/19/2003 |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO01/90427 | 11/29/2001 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
3652261 | Taubenblat | Mar 1972 | A |
4286987 | Matthews | Sep 1981 | A |
4327156 | Dillon et al. | Apr 1982 | A |
4455354 | Dillon et al. | Jun 1984 | A |
4478638 | Smith, Jr. et al. | Oct 1984 | A |
4705203 | McComas et al. | Nov 1987 | A |
4710273 | Okamoto | Dec 1987 | A |
4964908 | Greetham | Oct 1990 | A |
4971755 | Kawano et al. | Nov 1990 | A |
5236032 | Nukami et al. | Aug 1993 | A |
5509555 | Chiang et al. | Apr 1996 | A |
5745834 | Bampton et al. | Apr 1998 | A |
5791397 | Suzuoki et al. | Aug 1998 | A |
5848349 | Newkirk et al. | Dec 1998 | A |
6719948 | Lorenz et al. | Apr 2004 | B1 |
Number | Date | Country |
---|---|---|
613 041 | Nov 1948 | GB |
WO 9118122 | Nov 1991 | WO |
Number | Date | Country | |
---|---|---|---|
20040009086 A1 | Jan 2004 | US |
Number | Date | Country | |
---|---|---|---|
60206066 | May 2000 | US |