An Overview of Creep in Ceramics and Simple Strategies to Reduce its Significance in Equiaxed Investment Casting Moulds.
A summary of creep models and prospective solutions to reduce its impact on the manufacture of equiaxed turbine blades in an investment foundry.
Creep and its related phenomenon superplasticity both occur at high temperatures typically >0.5Tm and more usually > 0.6Tm. Superplasticity is most commonly found in fine grained polycrystalline materials at elevated temperatures. Creep occurs through diffusion assisted grain boundary sliding with little grain elongation and no cavity formation, plastic deformation occurs due to a combination of applied stress and creep. In this case, we have assumed that the shell defect is creep rather than superplastic, both phenomena can have similar outcomes but it is necessary to be sure of the mechanism to provide the right outcome.
Creep is the progressive deformation of materials at a constant load over a period of time. In this case, we are examining a relatively thin ceramic mould with an external temperature of around 1000C, and a hot-face temperature of around 1600C which is loaded with initially by its own mass, the metallostatic pressure of the melt itself, and latterly by contractive forces. Immediately after the casting operation the primary layer of the shell will approach the casting temperature, and over the next few minutes a temperature gradient will be set such that the outside of the shell may well reach in excess of 1250C. Creep will occur during the time that this temperature gradient is being established until such time as the cast metal has solidified fully develops stress integrity, and is hence to self-sustain load.
Creep can occur through lattice mechanisms such as dislocation glide, climb and dissolution; or, it can occur via boundary mechanisms such as Coble and Nabarro-Herring creep which are associated with grain boundary sliding, with, or without the presence of liquid phases, or the nucleation and growth of flaws along grain boundaries. This is the reason that the lack of grain boundaries in Single Crystal materials have inherently high creep resistance. Single crystals such as MgO, CaO and NaCl have highly symmetrical cubic systems which permit easy glide at elevated temperatures and cause low creep resistance , other single crystals with more asymmetric crystallographies have fewer slip planes available to them and thus the creep resistance is higher. Some Oxides of alumina, mullite and YAG have very high creep resistances in certain planes but are also highly anisotropic. Binary Oxides generally exhibit greater creep strength compared to single phase oxides.
It has been stated that having a high melting point will help resist creep, this is certainly not true. The following table (1) gives the creep strength of some single crystal ceramics, it can be noted that there is no correlation between creep resistance and melting point.
In any manufacturing process, for example hot pressing, or indeed casting; Creep may occur by more than one mechanism. This was demonstrated in the form of deformation mechanism maps by Ashby et al. The maps are Stress versus Temperature diagrams which are divided into zones where a particular deformation mechanisms dominate. Therefore, for a given ceramic, under a certain load, at a certain temperature it is possible to predict the creep mechanism.
Most mechanisms of high temperature creep predict a “steady-state” creep as defined by Norton in his seminal work which arrived at the Arrhenius equation known today as Norton’s Creep Equation: here it is:
Where A is constant, d (delta) is the grain size, p is the inverse grain size exponent, Q is the apparent activation energy for creep, R is the gas constant, s (sigma) is the applied stress, n is the stress exponent, and T is the absolute temperature.
The Activation Energy for creep can be determined by measuring the slope e versus 1/T. The dominant creep mechanism can therefore be determined from that value and n. (1)
Nabarro-Herring creep occurs through vacancy diffusion through individual grains, the creep rate is proportional to 1/d2. Whereas Coble creep which occurs mostly at grain boundaries by grain boundary diffusion the creep rate is proportional to 1/d3 and to s. Dislocation Creep usually occurs at high temperatures and stress levels where creep is controlled by dislocation climb and glide.
At high temperatures and stresses it has been proved that creep will occur by grain boundary sliding where adjacent grains become displaced with respect to each other. This mechanism can be accommodated by transgranular plastic flow, vacancy movement, and grain elongation in the tensile direction. The presence of grain elongation typifies this form of creep as Lifshitz sliding. However, more commonly, we see grain boundary sliding without grain elongation as a result of thin glassy intergranular phases, by cavity formation and growth. This is termed Rachinger sliding.(2)
Creep rate is found to increase with porosity. The increase in porosity in Al2O3 from 5 to 50% increases the creep rate by a factor of 50. Similarly the creep rate in MgO increases by a factor of 6 in increasing apparent porosity from 2 – 12%. (1)
At intermediate strain rates and high temperatures, the growth of intergranular flaws can be responsible for creep. Deformation in fine grained single phases of high purity has been proven to take place by grain boundary cavitation and micro-cracking in addition to any diffusional and dislocational flow mechanisms.
In ceramics with glassy grain boundaries, flow of the intergranular phases may result in growth of microdefects, the growth and coalescence of which will change the elastic modulus of the substrate resulting in non-linear deformation. (4)
In ceramics with high glass contents creep is controlled by the viscous flow of glass. Viscosity, and in turn, creep rate will depend on the composition of the glassy phase and its temperature.
Increasing the creep resistance of creep susceptible ceramics can be improved by creep resistant reinforcements, for example the creep resistance of Alumina has been improved by the addition of 15%SiC whiskers. The improvement is associated with the increased creep resistance of the second phase coupled with the inhibition of grain boundary sliding. Silicon Carbide (SiC) whisker-reinforced ceramic composites were an innovation that came into prominence for potential structural applications because of the significant improvements in the mechanical properties these materials offered as compared to the monolithic materials. The incorporation of SiC whiskers into alumina ceramics resulted in increases in strength, fracture toughness, thermal conductivity, thermal shock resistance and high temperature creep resistance. These discoveries initiated several years of intense study into this class of composites. SiC whiskers used for reinforcement are discontinuous, rod or needle-shaped fibres in the size range of 0.1 to 1 µm in diameter and 5 to 100 µm in length. Because they are nearly single crystals, the whiskers typically have very high tensile strengths (up to 7 GPa) and elastic modulus (up to 550 GPa).
The first commercially available SiC whiskers were introduced in the early 1960s. The use of SiC whiskers to reinforce materials was originally applied to metal matrices, such as aluminum. However, the first application of whisker reinforcement to ceramics did not occur until the 1980’s. Several methods and numerous starting materials can be used to grow SiC whiskers. Much of the early work prior to the mid-1970s employed the vapor-liquid-solid (VLS) mechanism to produce small quantities of whiskers. Later production methods used carbothermic reduction reactions of low-cost silica and carbon precursors, such as rice hulls, to produce large quantities of whiskers at reasonable cost. This allowed SiC whiskers to become economically viable as reinforcing agents in components for large-scale, high-volume applications.
Using the rice-hull technology, Advanced Composite Materials Corporation (ACMC) developed production capability to manufacture large quantities of SiC whiskers in the USA. The major market for the whiskers was to be for reinforcing metal matrix composites as had been done in the past. This led to the first whisker-reinforced alumina composites being fabricated at Oak Ridge National Laboratory by hot-pressing in 1982. The initial results showed such promise that large-scale development programs were started to exploit the materials and several years of research followed.
The mechanical property improvements observed with the incorporation of SiC whiskers into ceramic matrices were unprecedented. For example, the fracture toughness of alumina was increased from ~3.0 MPam exp(1/2) to 8.5 MPam exp(1/2) with the addition of 20 v/o whiskers. This was accompanied by fracture strengths of 700-800 MPa versus <= 400 MPa in unreinforced alumina. Just as importantly, these property improvements were retained to elevated temperatures, unlike some other toughened ceramic systems. Remarkably improved thermal shock and creep resistance were also observed. Subsequent studies have examined the SiC whisker-reinforcement of numerous ceramic matrix systems, including mullite, zirconia, glass, spinel, cordierite, silicon nitride, boron carbide, and combinations of these materials.
Research into the toughening behavior responsible in the composite materials shows that crack-whisker interaction resulting in crack bridging, whisker pullout and crack deflection are the major toughening mechanisms. One of the keys to the behaviour of SiC whisker reinforced composites, is that for this mechanism to operate, debonding along the crack-whisker interface must occur during crack propagation and allow the whiskers to bridge the crack in its wake. A theoretical presentation of this is considered in Bengisu’s Engineering Ceramics (1). Scientifically, the understanding gained in studies of the mechanical behavior of these composites has been used to develop other ceramic systems (e.g., the self-reinforced ceramics) having increased toughness and strength. (9)
Unlike whisker reinforcement second phase particle and macro fibre additions to certain ceramic matrices reportedly decrease creep strength. Known examples occur in Alumina/YAG composites and Alumina/Zirconia composites. It is this method which is also most commonly seen in the Precision Investment Casting Industry. In this application, typically, polydispersed grains sizes, or ceramic or even polymer fibres are added to the slurry which is applied prior to application of coarse grained stucco. The grains and fibres are normally macroscopic and have been found to be typically 0,2 – 2,0 mm in length. Early work on asbestos based fibres were found to be reasonable effective due to the ability of asbestos to continually subdivide to sub micrometre sizes in process. However class actions with respect to prevalence of mesothelioma amongst workers curtailed such developments. In recent years systems based on macro and indivisible fibres such as alumina. Silica and aluminosilcates have been developed. In these systems dislocation and grain boundary pinning does not occur to any degree as a result of fibre effects. Similarly the use of coarser grains has no effect in reducing creep behavior. In spite of this often moulds manufactured in this way are found to creep less. In general this is due to the spurious increase in thickness of such systems due to local reduction in shear forces during the liquid phase application. Effectively, the hot face of the shell mould is supported by the cooler, and, therefore, more creep resistant outer shell. Under true creep conditions cavitation at the fibre or particle matrix interface was observed which may explain the weakening. (1) (11)
It is therefore obvious from the preceeding that Creep is a complex phenomenon and it is necessary to be cogniscent of the ceramic system, heating cycle, degree of nucleation, grain growth and refinement, and thermodynamics being employed in order to best determine the dominating creep mechanism, its method of operation, and hence develop a strategy to reduce known creep.
Factors affecting creep in an Investment Cast Shell
As far as I can assert there is no creep map for the typical mix of materials used in Investment Casting. This may be a worthy piece of research if funding is available. ( In truth, I actually believe that this work has been done, but it would be fair to say that it is not in the public domain!)
Shells are nearly always always cast above 0.5Tm therefore creep is likely to occur to a greater or lesser degree. Our role therefore is to develop a strategy whereby it effect is minimised or does not occur to a significant degree at the temperatures used.
Norton tells us that: the higher the temperature, the greater the loading, and smaller the grain sizes employed, the more significant creep will be. It is not likely that alloys will be changed, the cast weight reduced, or significantly different particle size distributions be used due to its effect on green and hot strength.
If Nabarro – Herring creep is the dominant mechanism, the procurement of denser more perfectly sintered materials will be beneficial. Some work has been performed on the use of dislocation pinning techniques using Silicon Carbide Whiskers to prevent creep, and with some success. There are a number of specialist refractory companies performing work in this arena.
Experience dictates that Coble creep is likely to be the dominant mechanism. The shell structure being comprised of very small grains, bonded together with a composite of fine flour, amorphous silica and soda provides for a near perfect substrate for Rachinger sliding.
In particular, I note two characteristics of the shell which individually reduce creep resistance, and which in conjunction probably have a significant effect:
The use of chamottes and kaolinitic clays is common throughout the Precision Investment casting industry. However, being a naturally occurring clay it contains many trace elements, and many different phases. As metallurgist are aware the presence of even small amounts of low melting point materials can seriously affect overall performance, especially when there are degrees of solid solubility causing the presence of eutectic or eutectoids. Of interest is the presence of Kaolinite which undergoes a large number of changes in heating before it eventually melts:
Kaolinite AL2O3.2SiO2.2H2O 450 ºC
MetaKaolinite AL2O3.2SiO2 950 ºC
Spinel 2AL2O3.3SiO2. 450 ºC
Mullite 3AL2O3.2SiO2. 1500 ºC + (SiO2)
Whereas these changes are carried out in the calcination process, the material, therefore, being stable up to its calcination temperature of 1550C, densification and dislocation movement is likely to occur in temperature excursions above this temperature. Crucially, it may also contain a certain amount of Mica, commonly encountered as Muscovite, K2O.3Al2O3.6SiO2.2H2O, which has a significantly lower melting point than Kaolinite, and acts as a flux at higher temperature. The flux combines with the amorphous, glassy, free silica phase. (7) Remet can supply Kaolinite of identical thermal expansion that does not contain Mica. Remasil 60 has been studied over a period of years and has proven to be particularly resilient under tensile high temperature conditions (8) and is freely available from Remet outlets.
The Binder itself comprises an amorphous silica which is stabilised in colloid form with the addition of Sodium Hydroxide. In drying, firing and sintering the hydroxide is reduced to Soda NaO. The presence of Sodium, especially in Directionally Solidified or Single Crystal Casting has long been associated with fluxing and softening at high temperature which occurs as a result of it high reduction potential. Remet, in conjunction with Grace Davison has developed a range of deionised binders based on the Ludox HS range of binders that have long been a favourite of Investment Casters, due to their lower sodium content. Ludox HS is first aluminised to create a pH insensitive binder, then deionised to remove all but the last traces of sodium, typically 0.05%, the result Ludox HSA has been used in high temperature applications for a number of years and is currently being qualified by an OEM for primary applications. Where higher green strength is required a water soluble polymer can be added, the resultant material Ludox SK and SK-CV have long been used in the manufacture of Equiaxed, DS and SC components by more than one OEM. Matzek (8) also compares a sodium bearing binder, in the Syton PX30 (Now Ludox PX30), and a sodium free ethyl silicate binder , Silester AR, though actual supporting evidence is unclear due to complexities of formulation and chemistry.
In summary we can hypothesise that the Creep mechanism is based on the Coble model, occurs most likely at the grain boundaries, and is made more likely to occur due the presence of Mica and Sodium in the finely divided matrix existing at grain boundaries.
Reduction of Shell Creep – possible actions
Conduct Research and Development to assert a deformation mechanism map for current shell composite.
There is no risk to this project, however the expense to gain this information may difficult to justify, and harder to justify it proves what we already guess. However, in terms of the cost of scrap caused as a result of shell failure, the payback on the project may be much quicker than anticipated.
Remove Ultra-Fines from flour fraction of shell.
There exists the risk of loss of strength and reduction in working viscosity of slurry though this is easily catered for in a working environment.
Remove Chamottes and Replace with Remasil 60.
Risks exist due to the differences in thermal expansion and particle size distributions. Close control of the development programme should ensure that possible negative effects in shell build as a result of viscosity differences are countered. The use of coarser back-up grains shall also reduce the volume fraction of fine particles interstitial to the grains which will also incrementally decrease the predisposition to creep, and provide potential for coat reduction. The discrete nature of the application, and the limited change in Method of Manufacture would mean that this option could be quickly evaluated at low cost.
Develop a low sodium binder alternative.
The use of deionised binders across a spectrum of OEM and first tier suppliers over the past twenty years indicates that there is no intrinsic risk with the material and those significant advantages such as extended life and robust high temperature properties must be present. The materials tend to be a higher cost than conventional alkali stabilised binders, so the cost benefit ratio must be carefully measured. The cost of revalidation is going to be considerably higher, and the nature of the change means that the trial would need to be more complex. If this is to be evaluated then I would advise that it is trialled in addition to kaolinitic clay substitution.
Development and subsequent use of Advanced Ceramic Technology in the form of Whisker stabilisation, and “theoretical density” materials.
Highly densified materials are available, though the high cost of temperature, time and pressure, needed to make these grades probably outweigh any advantage in using them. Whisker stabilisation (Whisker Pinning) has seen some development work in advanced ceramics, however, due to the very small size of whisker necessitated by the need to pin dislocations issues have arisen due to the potential for inhalation of finely divided ceramic fibres.
Keeping the bulk of the shell cooler
Norton advises that Creep has an activation energy associated with a temperature and that creep rate has an Arrhenius function with temperature. Whereas it is probably either impossible or at least highly unlikely that the cast temperature or the shell temperature can be reduced, it may be viable to increase the time that the shell takes to get hot.
A variety of materials have been developed either with a deliberately low density (so called Popsicles); or indeed hollow spheres. These shapes may have some advantages in reducing the rate of heat transfer through to the outside of the shell. It must be noted of course that they will also reduce the bulk strength of the shell.
To achieve greatest effect in smallest amount of time without excessive cost:
Evaluate substitution of mica bearing kaolinitic clays, or chamottes with Remasil 60 of slightly coarser grain.
Evaluate Sodium free binders.
SAFETY PRECAUTIONS: Read MSDS before using any product listed.
NOTE: Unless specifically identified as a specification value, the above chemical, physical and particle size distribution values are typical properties. They are not specification values. Contact your nearest REMET Sales Office regarding product specifications.
Contact your REMET Territory Manager and visit www.remet.com if you have any questions or require additional information.
This article is the authors opinion only. Information and/or recommendations based on research and technical data believed to be reliable. Offered free of charge for use by persons with technical skills, at their own discretion and risk, without guarantee of accuracy. REMET makes no warranties, express or implied, and assumes no liability as to the use of its products or of any information pertaining thereto. Nothing herein is intended as a recommendation to infringe any patent.
(1) Engineering Ceramics: Bengisu, M.
(2) Metals Ceramics and Polymers: Wyatt and Dew-Hughes
(3) Ashby et al: Deformation maps.
(4) Brown R.J. “Grain size measurements in fused ceramics”; Thesis, 1988.
(5) www.intota.com/expert-consultant.asp?bioID=614520&perID=712373. Accessed 0730 19/4/09
(6) Handbook of material selection; Kutz,M
(7) Brown R.J. :Refractories : An Overview. REMET presentation to J&J Medical, 2002.
(8) Remasil 60: A refractory for casting parts under high thermal stress; Matzek, C.H. Proceedings of 7th World Conference on Investment Casting.
(9) http://www.ms.ornl.gov/researchgroups/process/cpg/cpg_homepage.html Accessed 0655:21/4/09
(10) Bengisu, M : Engineering Ceramics
(11) Smart, R : Investment Casting Handbook< Back to insights