CRITICAL LEARNING FROM MICROGRAVITY SINTERING OF TUNGSTEN ALLOYS: IMPLICATIONS FOR EXTRATERRESTIAL FABRICATION AND REPAIR

Randall M. German, Seong J. Park, John L. Johnson, and Louis G. Campbell

Center for Innovative Sintered Products 147 Research West

Pennsylvania State University University Park, PA 16802-6809 USA

ABSTRACT

Considerable opportunity exists for extraterrestrial repair and construction based on freeform fabrication from powders. Refractory materials are important for propulsion, radiation, and thermal systems. However, little fundamental data exist to predict the sintering behavior and sintered properties in microgravity. Many of the preliminary designs assume reduced gravity sintering will follow the classic trajectories in terms of densification and distortion. However, early microgravity sintering data show very different behavior. Results from several microgravity tungsten heavy alloy sintering experiments are reported here to show the very different behavior, suggesting that pore buoyancy is a critical parameter. Further, gravity may have a beneficial effect on grain-grain contact, and in reduced gravity these contacts are missing. Consequently, extraterrestrial sintering without pore buoyancy will see pores coalesce and grow in a sintering body that has less structural rigidity, resulting in incomplete densification, massive pores, and significant distortion. Critical experiments have been performed to isolate gravity effects on pore buoyancy and grain settling, showing that gravity is very important to densification and distortion. Surprisingly, distortion is larger in microgravity sintering while attaining full density is nearly impossible due to pore agglomeration and the lack of pore buoyancy motion. This means that envisioned space-based repair and fabrication efforts will be much harder than anticipated. Altogether the Penn State team has processed 77 samples in microgravity, with thousands of equivalent samples processed on Earth. The findings show that on Earth densification occurs prior to distortion, but in microgravity, distortion occurs without densification. A conceptual model based on strength evolution, capillarity, and gravity has been formed to explain this behavior.

INTRODUCTION

Future NASA efforts to extend human exploration back to the moon and beyond will require development of techniques and processes that permit fabrication and repair of critical components under reduced gravity conditions, including 10-6 g (in space), 1/6 g (on the surface of the moon) and 3/8 g (on the surface of Mars). This capability is needed to reduce resource requirements and spare parts inventory while

Part 10 296

enhancing mission security. Many of the proposed in-space fabrication techniques involve particulate materials processing, but limited data are available to assess the viability of these proposed techniques under reduced gravity conditions.

The findings from liquid phase sintering experiments conducted aboard the Space Shuttle Columbia during STS-65, STS-83, and STS-94 provide significant knowledge on the role of gravity in processing solid-liquid mixtures. Many of the important findings have been published, and a fundamental contribution has been made in terms of a new strength evolution model for sintering [1]. One of the most important findings from the combined ground-based (1g) and microgravity (μg) studies is that the rules isolated by mankind over many years of Earth-based processing do not carry over into the microgravity environment. New pathways will be needed to obtain densification and satisfactory mechanical properties while minimizing distortion.

In this paper, the effects of gravity on densification, solid-liquid separation, and distortion of liquid phase sintered tungsten heavy alloys are reviewed. Tungsten contents range from 35 to 93 wt.% (35W to 93W) with the matrix phase consisting of Ni and Fe in a 7:3 ratio. Based on the limited available data, a finite element simulation is constructed to predict the effects of various gravitational levels on densification and distortion. The implications on future in-space fabrication processes are discussed.

BACKGROUND

Densification, solid-liquid separation, and distortion are the concerns of this research. Prior work has given little systematic investigation of distortion. This becomes evident in trying to explain the causes of significant distortion when sintering in microgravity.

Densification

Basically, during heating a powder compact gains strength through low-temperature interparticle bonding, usually induced by solid-state surface diffusion, followed by further strength contributions from high-temperature sintering densification. In cases where a liquid phase forms, densification is accelerated because of solid transport in the liquid, capillary forces, and liquid lubrication leading to grain sliding along contacts. As long as there are solid bonds or open pores in the sintering body, then there is sufficient rigidity to avoid distortion.

An unexpected finding from microgravity sintering is the stability of the pores. The lack of buoyancy forces reduces the ability to eliminate the pores, making densification difficult in microgravity. A plot comparing the densities of 78W and 93W is given in Figure 1. For samples with high W contents and many solid bonds between the particles, high sintered densities can be achieved in microgravity. But for samples that lack a rigid solid skeleton, liquid phase sintering in microgravity does not give any densification.

Although the pores are unable to exit the sample, they are highly mobile and coarsen at a rapid rate. Figure 2 shows a micrograph of a microgravity liquid phase sintered compact and the very large pores captured in the process of coalescence. Note the pores are enormous, since the green compact was formed from normal small carbonyl iron, carbonyl nickel, and reduced tungsten powders, the initial pores were below 1 μm. Now after sintering the pores are up to 200 μm across. This corresponds to a pore coarsening rate 1000-times faster than the grain growth rate.

Part 10 297

0.5

0.6

0.7

0.8

0.9

1.0

0 50 100 150 200 250

Time (min)

Rel

ativ

e si

nter

ed d

ensi

ty78W 1g

78W µg

93W 1g 93W µg

heating18°C/minute

isothermal1500°C

Figure 1. Effect of sintering time and gravitational level on the densification of 78W and 93W.

Figure 2. Coalescence to form large pores in a liquid phase sintered tungsten heavy alloy processed in microgravity.

Solid-Liquid Separation

In the case of W heavy alloys, a 9 g/cm3 density difference exists between the grains and the surrounding liquid (usually an alloy of Ni and Fe or Cu). On Earth, gravity causes the W grains to settle to the base of the sample, resulting in microstructural gradients. Dilute W heavy alloys produce a separated liquid layer above the settled region, making it impossible to obtain W volume fractions below about 0.6 on Earth.

Figure 3 shows the effect of W content on the solid volume fraction at the top and bottom of the settled region of samples sintered on Earth in comparison to samples sintered in microgravity. The gradients between the top and bottom of 1g samples become more pronounced at low solid contents. Similar trends are seen with other microstructural parameters such as contiguity and connectivity. These microgravity

Part 10 298

samples were free from statistically significant microstructural gradients. Microgravity also prevented solid-liquid separation, which was observed in the 78W and 83W 1g samples, enabling homogeneous microstructures with volume fractions near 0.5.

Microgravity enables processing of extremely dilute W heavy alloys that cannot be processed at all on Earth due to grain settling. Micrographs of a 35W alloy after sintering for 600 minutes on Earth and in microgravity are shown in Figure 4. In the ground-based sample all of the grains have settled into a thin layer giving a local solid volume fraction of about 0.6. In the micrgravity samples, the grains are much more dispersed, but still tend to cluster and form chain-like structures due to agglomeration forces. These results confirm an earlier hypothesis regarding grain agglomeration in microgravity [2].

0.5

0.6

0.7

0.8

0.9

1.0

75 80 85 90 95 100W content (wt.%)

W v

olum

e fr

actio

n

µg

1g top

1g bottom

Figure 3. Gravity induces grain settling, producing higher W volume fractions at the bottom than at the top of the settled region. The difference increases with lower W contents. The samples were sintered for 120 minutes at 1500°C. All of the samples were first presintered to full density in 1g.

(a) (b)

Figure 4. Micrographs of 35W sintered for 600 minutes at 1500°C in (a) 1g and (b) μg. Both samples were presintered to full density in 1g.

Part 10 299

While grain settling generally does not occur in microgravity, one 78W sample sintered for 180 minutes in microgravity demonstrated a surface-to-core microstructural gradient. A cross-section of this sample (Figure 5) shows a segregated liquid pool near the center of the sample with the fraction of W grains increasing radially from this pool. The reason for the anomalous behavior of this sample is unknown, but demonstrates some of the unexpected phenomena that can occur in microgravity.

Figure 5. Cross-sectional views of 78W sintered for 180 minutes at 1500°C in microgravity showing surface-to-core microstructural gradients.

Distortion

On Earth, rapid sintering densification leads to a loss of structural rigidity. Substantial weakness occurs when the pores are closed by rapid densification because the solid skeleton is dissolved by newly formed liquid. This condition is evident with high liquid contents or when pores rapidly coalesce. When this semisolid system also rapidly densifies, then the capillary forces associated with open pores are also lost, leaving saturated pores and no secondary source (solid skeleton) strength. Thus, distortion is delayed until densification is practically complete as demonstrated in Figure 6 for 88W.

Surprisingly, gravity helps suppress distortion. The deviatoric stress acting on the solid grain structure induces grain-grain contact, which provides increased structural rigidity. In contrast, without grain settling, sintering in microgravity results in significantly more distortion. This is illustrated in Figure 7 for 83W compacts sintered from identical green compacts in the same furnace for the same time-temperature pathway, but the one on the right was processed in microgravity. The left-most image shows the nearly fully dense W-Ni-Fe compact after pre-sintering at 1400°C and machining into a right circular cylinder. The center compact followed 1g sintering at 1500°C for 2 h, while the right-most image was processed in μg yet everything else was the same.

Part 10 300

0.94

0.95

0.96

0.97

0.98

0.99

1.00

0 20 40 60 80 100Time (min)

Rel

ativ

e si

nter

ed d

ensi

ty

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Dis

tort

ion

para

met

er

Distortion

Density

Figure 6. Distortion parameter and density for liquid phase sintered 88W-8.4Ni-3.6Fe as a function of time above 1400°C. Samples were water quenched during heating at 5°C/min to 1500°C and after different hold times at 1500°C.

Figure 7. 83W compacts (left prior to sintering, center Earth-sintered, right microgravity sintered).

The presence of pores in μg samples further increases distortion unless there are sufficient grain contacts to form a rigid skeleton. For instance, the 93W samples whose densities were plotted in Figure 1 did not distort in either 1g or μg. However, the 78W sample, which lacks a rigid skeleton, gave more distortion and less densification in μg than in 1g. Since there is no buoyancy force on the pore space in μg the compacts distort freely as the pores agglomerate. Also, there is extra liquid since it does not fill the pore space. Most surprising is the evolution toward hollow spheres, where all of the pores have coalesced into one single central pore and the whole body has spheroidized, as shown in Figure 8. Such behavior is contrary to earlier expectations that sintering in space would enable a wider range of compositions to be liquid phase sintered to full density and would lead to greater precision. Instead, we now see that microgravity sintering leads to lower performance, an inability to eliminate pores, and more distortion.

Part 10 301

Figure 8. Cross-section of 78W sintered for 120 minutes in μg. This sample had a green density of 62% and was not presintered to full density in 1g.

MODEL

From the above findings, a model has been developed to be able to predict densification and distortion during liquid phase sintering under reduced gravity conditions. The model is based on constitutive laws of deformation, which are incorporated into a finite element simulation. Continuum models are useful for making macroscopic shape predictions with relatively little computing time. The main features of the model are summarized below.

The deformation rate ij of the porous compact is expressed as [3]

ijsmijij K31

21 ' (1)

where 'ij and m are the deviatoric and hydrostatic (= kk/3) parts of the true stress ij, ij is the Kronecker delta, s is the sintering stress, is the effective shear viscosity, and K is the effective bulk viscosity of the porous powder compact. The formulations for sintering stress and the bulk viscosity are detailed by Kwon et al. [4] and have been adapted for activated and liquid phase sintering.

This model considers the influence of gravity in the force-stress equilibrium equation in the form of a body force fi as follows:

0' ijij f (2)

where fi is defined as follows:

otherwise0direction verticalin thegT

if (3)

Part 10 302

where is the relative density, T is the theoretical density and g is the gravitational acceleration. In the finite element simulation, Equation 2 is solved simultaneously with Equation 1, which uses the ij'j terms.

For Earth-based sintering, friction is considered as follows:

nt (4)

where t is the tangential frictional stress along the substrate and sample interface, n is the normal stress, and μ is the Coulomb frictional coefficient (determined to be 0.2 by fitting to experiment). Equation 4 is applied as a boundary condition in the finite element simulation.

The normal stress induced by the surface tension on the free surface boundary is written as follows:

n (5)

where is the surface energy. The mean curvature of the free surface is defined as:

n (6)

where n is the outward unit normal on the surface. Equation 5 is applied as a free surface boundary condition in the simulation.

This model is used to explore the effects of different gravitational levels on the distortion of 83W during liquid phase sintering. The data from ground-based experiments provide one end-point while the microgravity experiments provide the other. The predicted distortion behavior of 83W in 1g and μg is compared with the experimental results in Figure 10. Since limited data exist for densification in microgravity and essentially indicate that very little occurs, the samples are assumed to be presintered on Earth to full density. A future need is to incorporate grain contacts and pore buoyancy into the simulation.

Figure 11 shows predicted results for sintering on the moon and Mars in comparison to on Earth and in space. Lunar and Martian sintering will give noticeably different distortion behavior than seen in any prior experiments. Different gravitational levels may require different sintering pathways for achieving full density without distortion.

Part 10 303

0

2

4

6

8

10

12

0 1 2 3 4 5 6 7

original shapeexperiment (0.0g)

full densitywithoutdistortion

simulation (0.0g)

simulation (1.0g)

experiment (1.0g)

effect ofW particlemigration

differencedue to friction

Figure 10. Comparison of shape profiles of 83W samples after liquid phase sintering for 120 minutes in either 1g or μg with finite element simulation predictions.

In space (no gravity)

Moon

0

2

4

6

8

10

12

0 1 2 3 4 5 6

original shape(green part)

full densitywithout distortion

Mars

Earth

Figure 11. Predicted shape profiles of 83W right circular cylinders liquid phase sintered for 120 minutes at 1500°C under gravitational levels of interest for space exploration.

Part 10 304

DISCUSSION

So why do liquid phase sintering systems distort after densification? The key is in the microstructure and the low strength the microstructure passes through when the liquid forms. Up to pore closure, a combination of solid skeleton and capillarity compression from open pores provide sufficient rigidity to prevent distortion. When sintered on Earth, gravity keeps the grains in contact and this contributes a small yield strength to the compact (estimated at 0.3 kPa in some cases). Thus, Earth-based processing passes through low strength conditions with a small, finite yield strength up to pore closure. If there is an excess of liquid, then densification is rapid, pores close early, and strength depends only on the degree of solid-solid bonding. Inherently, sintering densification and distortion depend on many of the same parameters. Full density is possible in the rearrangement stage of liquid phase sintering with a high liquid content [5], but with concomitant distortion. With a low liquid content the solid skeleton forms to resist densification, and this solid skeleton helps retain compact shape. Thus, simple comparative experiments show low solid content samples achieved full density and distort, while high solid content samples densified slowly but did not distort.

The low strength transient that allows distortion is traced to the dihedral angle. With a low dihedral angle, the reduction in solid-liquid surface energy on first melt formation leads to liquid penetration of grain boundaries [6]. Consequently, densification by rearrangement takes place rapidly, but solid-solid bonds do not form before pore closure. Thus, the disappearing capillary force removes the only source of component strength. Distortion is the consequence of strength loss. Most interesting, samples sintered to full density still distort on reheating to the liquid phase sintering temperature. This demonstrates that newly formed liquid does attack grain boundaries. On the other hand, high liquid content alloys with a high dihedral angle retain shape up to full density.

It is the liquid penetration of grain boundaries that causes compact strength loss. A high solid solubility in the liquid correlates with a low dihedral angle that is a precursor to liquid penetration of grain boundaries. The fractional atomic solid solubility in the liquid can be approximately linked to the dihedral angle by the following empirical expression [6]:

2tan

214.011.0Ak (7)

where kA is the fractional atomic solid solubility in the liquid, and is the dihedral angle in radians. High solid solubility in the liquid kA indicts a low dihedral angle according to Equation 7. The atomic solid solubility in liquid can also be linked to the dihedral angle by the following empirical equation:

Ak63875 (8)

where is in radians and kA is the fractional atomic solid solubility change in newly formed liquid as compared with the solid solubility in the additive. Equation 8 indicates that if the atomic solid solubility in the liquid is much larger than the atomic solid solubility in the additive, then the systems have low dihedral angles. In turn, low dihedral angles imply easier liquid penetration of the grain boundaries and often are associated with distortion.

Solid volume fraction and dihedral angle are the dominant factors controlling densification and distortion. Our research has determined microstructural links to distortion [7, 8]. From the microgravity experiments, a critical contiguity of 0.38 is needed to prevent distortion of W-Ni-Fe heavy alloys [9]. This critical contiguity is also likely necessary to enable densification in microgravity. Shape retention can be maintained at lower contiguities for systems with higher dihedral angles. For example, the critical contiguity for shape retention of W-Cu is 0.22 [10]. W-Cu is a strongly bonded system and has critical microstructural parameters for both

Part 10 305

shape retention and rearrangement. W-Ni-Fe is less strongly bonding due to the solubility of W in the matrix, which can produce liquid films at the grain boundaries. Prevention of distortion requires more solid-solid contacts, giving the higher critical contiguity. Table 2 compares the critical microstructural parameters of W-Cu with those of W-Ni-Fe. These parameters are also likely required for densification to occur in microgravity.

Table III. Critical microstructural parameters to prevent distortion for W-Ni-Fe and W-Cu. W-Ni-Fe W-Cu

Critical contiguity 0.38 0.22

Critical connectivity 2.7 1.0

Critical coordination number 10.2 1.5

DISTORTION CONTROL STRATEGIES

Various strategies emerge to improve distortion control in liquid phase sintering. Since the compact strength evolution determines densification and distortion onset, the design of compositions for densification without distortion requires manipulation of the microstructure and its evolution to sustain enough compact strength to avoid distortion. As compact strength has contributions from both sinter bonds and capillary forces, in principle, processing strategies that preserve sinter bonds or capillary forces will resist distortion. Liquid phase sintering systems that densify slowly resist distortion; also they do not show a propensity for secondary rearrangement. Most practical systems that resist distortion inherently have high solid volume fractions and high dihedral angles. So one option for distortion control is to use high solid volume fraction and high dihedral angle systems. Systems with low solid solubility in the liquid phase inherently have a high dihedral angle and resist distortion. Compositions with presaturated liquid forming agents (for example, prealloy matrix) will inhibit dissolution of sinter bonds into newly formed liquid and resist distortion. A slow heating rate at the point of liquid formation also allows solid bonds to reform before capillary forces are lost with pore closure, thereby resisting distortion. On Earth, densification and distortion are sequential events; the first focus is on densification which occurs prior to distortion. Only when open pores are eliminated and no solid skeleton has formed will gross distortion be observed. If a solid skeleton has formed prior to pore closure the compact will resist distortion. Not so in microgravity. One of the surprising observations in microgravity is the migration of pores to the compact center with a sealed, dense outer compact shell. The absence of gravity induced solid grain contacts reduces the skeletal strength and the compact forms a thick shell of dense material.

Solid-liquid mixtures in which solid bonds do not form between the particles will be more porous and more distorted when processed in space than when processed on Earth. This has significant implications on proposed extraterrestrial fabrication and repair efforts, many of which are based on freeform fabrication processes using powder-binder mixtures. These results show the microgravity fabricated products will not be dense, will be weaker and not as hard and planned, will be imprecise, and might be permeable. Thus, they cannot be qualified for structural or living quarter uses without new fabrication strategies. So, freeform fabrication will not necessarily be useful in space.

New computer models are being generated to handle the effects of gravity on key variables, such as degree of grain contact and pore buoyancy. However, we have little knowledge on pore space changes in microgravity, so predictions of final porosity, shape, and properties are hindered by the limited number of space-based experiments. Current efforts are focused on ground-based experiments to lay in place a better foundation for eventual space experiments.

Part 10 306

CONCLUSIONS

Solid volume fraction and dihedral angle are dominant factors with respect to densification and distortion during liquid phase sintering. Distortion decreases with increasing solid volume fraction and increasing dihedral angle. Green pores affect distortion during rapid heating, especially when nearly full density is achieved or capillary forces are lost (latter happens in microgravity sintering without densification as pores cluster away from the surface). On Earth, distortion is inhibited until pores close and the compact is nearly fully densified. Hence, distortion follows densification on Earth. From this insight, new processing strategies emerge to improve distortion control in liquid phase sintering. The implications are that freeform materials, process, and products cannot be qualified for space fabrication and repair based on ground-based results. In microgravity the materials will be porous, the components will be less precise, the properties will be degraded, and overall performance will be far short of expectations. Microgravity results in a lower strength compact since there is no grain compression to form a solid skeleton. As pores cluster and coalesce, they become closed but not eliminated. As a consequence, microgravity compacts are weaker with more distortion and less densification.

ACKNOWLEDGEMENTS

Funding was provided by NASA under the program “Gravitational Effects on Distortion in Sintering,” Mike Purvey as Project Manager and Witold Palosz as Project Scientist, both are with the Marshall Space Flight Center in Huntsville, Alabama. Several students and staff helped with these studies, and thanks to all of the P/M Lab at CISP for supporting these very difficult experiments.

REFERENCES

1. German, R.M. "Manipulation of Strength During Sintering as a Basis for Obtaining Rapid Densification without Distortion," Mater. Trans., Vol. 42, 2001, pp. 1400-1410.

2. German, R.M. "Grain Agglomeration in Solid-Liquid Mixtures under Microgravity Conditions," Metall. and Mater. Trans. A, Vol. 26B, 1995, pp. 649-651.

3. McMeeking, R., and Kuhn, L.T. "A Diffusional Creep Law for Powder Compacts," Acta Metall. Mater., Vol. 40, 1992, pp. 961-969.

4. Kwon, Y.-S., Wu, Y., Suri, P., and German, R.M. "Simulation of the Sintering Densification and Shrinkage Behavior of Powder Injection Molded 17-4 PH Stainless Steel," Metall. Mater. Trans. A,Vol. 35A, 2004, pp. 257-263.

5. Kingery, W.D. "Densification during Liquid Phase Sintering in the Presence of a Liquid Phase. I. Theory," J. Appl. Phys., Vol. 30, 1959, pp. 301-306.

6. German, R.M. "Strength Loss and Distortion in Liquid Phase Sintering," Sintering Science and Technology, compiled by R.M. German, G.L. Messing, and R.G. Cornwall, State College, PA, 2000, pp. 259-264.

7. Johnson, J.L., and German, R.M. "Microstructural Effects on Distortion and Solid-Liquid Segregation during Liquid Phase Sintering under Microgravity Conditions," Metall. and Mater. Trans. A, Vol. 29B, 1998, pp. 857-866.

8. Wu, Y., German, R.M., Marx, B., Bollina, R., and Bell, M. "Characteristics of Densification and Distortion of Ni-Cu Liquid Phase Sintered Tungsten Heavy Alloy," Materials Science and Engineering A, Vol. 344, 2003, pp. 158-167.

9. Liu, J., Upadhyaya, A., and German, R.M. "Application of Percolation Theory in Predicting Shape Distortion during Liquid Phase Sintering," Metall. Mater. Trans. A, Vol. 30A, 1999, pp. 2209-2220.

10. Johnson, J.L., Brezovsky, J.J., and German, R.M. "Effect of Liquid Content on Distortion and Rearrangement Densification in Liquid Phase Sintered W-Cu," Metall. Mater. Trans. A, Vol. 36A, 2005, pp. 1557-1565.

Part 10 307