- Email: [email protected]
A closer look at porous refractories
Technical trends
A closer look at porous refractories
ew studies have been done on the topic of porous refractory metals, iridium in particular. Due to the higher cost of these materials, net shaping is very important to avoid machining waste. However, in low quantity production, hard tooling is not economical, making slurry casting into soft tooling more attractive. More efforts have been made towards high density processing of niobium and molybdenum [1, 2, 3]. Grain growth is a limiting factor in attaining high densities in Nb and Mo based materials, so efforts have been made to pin the grain boundaries in these systems to reach higher density [4]. Rhenium is typically added to Mo to improve ductility and tensile strength by formation of solid solution. At 49%Re addition, used in this study, a solid solution with Re-rich precipitates is observed at room temperature [5]. Without the addition of pore-forming agents, the green pore size of a powder compact is a function of particle size, particle shape, particle size distribution, and packing density. Given the particle size D, and packing density ∂, the mean pore size d can be estimated by the following equation [6]:
F
A research team at Penn State's Centre for Innovative Sintered Products has used slurry casting to produce high-porosity metals with tailored pore sizes…
(equation 1)
Figure 1: Relationship between packing density, particle size, and pore size, predicted by equation 1.
Table I: Powder Characteristics Sample Mo-Re, as received
D10, µm D50, µm D90, µm 5.3 9.8 15.6
Shape Spherical
Mo-Re -32 µm sieved
7.2
13.7
20.3
Spherical
Mo-Re +32 µm sieved
9.2
16.6
28.1
Spherical
Nb, 7 µm, as received
3.2
6.7
12.1
Rounded, irregular
Nb, 7 µm, presintered at 1000 °C
3.3
6.4
10.9
Rounded, irregular
Nb, -45 µm, as received
23.1
39.5
63
Angular
Nb, -32 µm sieved
18.9
31.1
47
Angular
Nb, -45 µm, +32 m sieved
31.2
45
66
Angular
Ir, as received
3.7
8.5
16.8
Agglomerated
Ir, presintered at 1000 °C and milled
3.6
8.4
17.5
Sintered agglomerates
20
MPR September 2004
Figure 1 indicates the predicted pore size from the equation, given a packing density and particle size. This equation assumes that the particles are monosized spheres, so it is only a first approximation of pore size. The equation may also be used to predict the pore size of a presintered structure, since pore shape is only slightly altered in the presintering process. In this case the sintered density replaces the packing density f in the equation. One Ir powder, two Nb powders, and one Mo-49Re powder were used in this study. Each powder was processed to alter its particle size distribution before shaping and sintering. The powders are listed in Table I with the particle size analysis in the as-received condition. It is important to note that the Ir powder and the 7µm Nb powder both contained amounts of agglomerated sub-micron particles, which were interpreted as discreet particles by laser scattering wet particle size analysis. This caused the
0026-0657/04 ©2004 Elsevier Ltd. All rights reserved.
Technical trends particle size measurement to indicate a larger particle size than actually existed. Ultrasonics were used to decrease the severity of agglomeration prior to testing. Powder Pre-Processing Niobium The 7µm Nb powder was first presintered by heating 5°C/min to 300°C, 2°C/min to 800°C for 1 hour and 10°C/min to 1050°C for 1 hour. This cycle served to dehydride and deoxidise the powder, and sinter the finest particles to the larger particles or to themselves, eliminating the agglomerated sub-micron particles. Presintering of the powder did not result in a hard cake. The soft cake was returned to powder condition upon removal from the sintering crucible. No significant changes in particle size were observed, however, the particle size was increased, since the agglomerated fine particles were eliminated. The -45µm (325 mesh) Nb powder was sieved to +/32µm (450 mesh) to create two particle size distributions. The particle size analyses of all three powders are given in Table I. Mo-49Re The Mo-49Re powder was specified as 20µm (-635 mesh) in the as-received condition. Because of the powder's tendency to agglomerate, most of the powder would not go through a 635 (20µm) screen. Two new particle size distributions were obtained by sieving with a 450 (32µm)
screen, and taking the +450 and -450 portions. The particle size distributions are given in Table I. Iridium The iridium powder as-received was highly agglomerated, much more so than the 7µmNb powder. In order to reduce the agglomeration and increase the particle size, the powder was presintered at 1000°C in 10-4 Torr vacuum tube furnace, resulting in a presintered powder cake. Break-up of the cake was performed by pushing the cake through a 20 mesh sieve with milling rods, then rod milling for 10 minutes. The particle sizes before and after this process are given in Table I. The measured particle size did not change significantly, however, since the asreceived particle size measurement was effectively measuring the agglomerate size. Slurry casting was used to produce the samples, to allow for low green density and shape complexity [7]. In this process, powder is mixed with melted Dussek-Campbell paraffin wax with a melting point of 55°C and poured hot into a rubber mold, avoiding the need for hard tooling. Some settling of the powder occurs before the wax solidifies. The compositions of the slurry for each material and the resulting green densities are given in Table II. All slurry was cast at 110°C into a mold preheated to 80°C and allowed to cool at room temperature. The piece was ground down to the correct height, allowing for removal of the
Figure 2: 7mm Nb sintered at 1450°C
Figure 3: 31mm Nb sintered at 1450°C
Figure 4: 45mm Nb powder sintered at 1500°C
Table II: Sintering Results Material D50 Particle Size, (µm)
Distribution Green Sintering Slope Density, (%) Condition Parameter, Sw
Sintered Density, (%)
Measured Predicted Predicted/ Average Pore Average Pore Measured Diameter, (µm) Diameter, (µm) Ratio
Mo-49Re
9.8
5.5
45
1500 C 4 h
61
3.4
4.1
1.21
Mo-49Re
13.7
5.7
48
1500 C 4 h
56
5.8
7.2
1.24
Mo-49Re
16.6
5.3
44
1500 C 4 h
57
7.9
8.2
1.04
Nb
6.4
4.9
34
1350 C 2 h
43
3.5
5.6
1.63
Nb
6.4
4.9
33
1450 C 2 h
52
2.5
3.8
1.52
Nb
31
6.5
46
1450 C 2 h
52
10.2
19.0
1.87
Nb
45
7.8
44
1450 C 2 h
49
13.2
31.2
2.37
Ir
8.4
3.7
33
1200 C 1 h
35
5.8
10.2
1.77
Ir
8.4
3.7
33
1300 C 1 h
46
3.4
6.4
1.87
metal-powder.net
September 2004 MPR
21
Technical trends
Figure 5: 17mm Mo-Re sintered at 1500°C for 4 hours.
low density material on top due to powder settling. Cylinders with a diameter of 4 mm were cast for use in all green density, sintered density, and porosimetry measurements. Complex shapes produced were 7 cm diameter discs with and without circular surface corrugations. Debinding and Sintering Niobium Debinding and presintering were performed by surrounding the part in 10µm alumina powder and heating at 2°C/min to 120°C for 1 hour, 2°C/min to 250°C for 1 hour, 1°C/min to 450°C for 1 hour, 5°C/min to 1125°C for 1 hour, and 10°C/min to 20°C in high purity argon. The parts were removed from the alumina packing and heated at 5°C/min to 1350°C or 1450°C for 1 hour in 10-5 Torr in an all-metal construction vacuum furnace with tungsten heating elements. Mo-49Re Debinding was performed by surrounding the parts in 10µm alumina powder and heating at 2°C/min to 120°C for 1 hour, 2°C/min to 250°C for 1 hour, 1°C/min to 450°C for 1 hour, 5°C/min to 1200°C for 1 hour, and 10°C/min to 20°C in high purity argon. The parts were then presintered in the alumina packing at 1500°C for 4 hours in 10-5 Torr.
Figure 6: 14mm Mo-Re sintered at 1500°C for 4 hours.
milled powder was sintered in a vertical dilatometer at 5°C/min to 1400°C for 1 hour in high purity argon. Sample Evaluation Green density and sintered density were measured by Archemedes technique, while sintered pore size was evaluated by mercury porosimetry. Metallography was performed on selected samples. Table II summarises the sintered samples and their properties. Pore size was predicted based on equation 1, substituting the sintered density for the packing density, and using the average particle size. Since the samples are all in the presintered state, this is a reasonable approximation. During the presintering stage, the pore shape does not change significantly, since surface diffusion is dominant [8]. There is a small increase in density from the green to sintered condition, and this is a combination of constriction of the compact as wax is removed in debinding, and sintering shrinkage. Metallography indicates that no real particle morphology changes have occurred; only sintering necks have been formed. In general, the predicted pore size is larger than the measured pore size. Only the Mo-Re powders are spherical. It is observed that the best agreement with the equation is observed with the Mo-Re
Iridium Debinding and presintering were performed by surrounding the parts in 10µm alumina powder and heating at 2°C/min to 120°C for 1 hour, 2°C/min to 250°C for 1 hour, 2°C/min to 450°C for 1 hour, 5°C/min to 1100°C for 1 hour, and 10°C/min to 20°C in high purity argon. Parts were removed from the alumina before heating at 5°C/min to 1200°C or 1300°C for one hour in 10-5 Torr. Tapped
22
MPR September 2004
Figure 8: Ir sintered at 1300°C
Figure 7: 10mm Mo-Re sintered at 1500°C for 4 hours.
powders. The worst agreement is observed with the 31µm and 45µm Nb powders that are very angular in shape. The particle size distribution width is given by the slope parameter [9]:
(eq. 2)
This parameter increases as particle size distribution becomes narrower. Comparison between this value and the predicted/measured ratio of pore size reveals that as particle size distributions become narrower in Nb and Mo-Re powders, the prediction of pore size becomes less accurate, and more oversized. This is the opposite trend than expected, since the equation is derived for monosized spheres. As particle size distributions become wider, the smaller particles can fit into the spaces between the larger particles, resulting in higher packing densities and smaller pores [10,11]. Based on this, the wider particle size distributions should result in pore size smaller than predicted from using the D50 particle size in equation 1. Figures 2-4 display the sintered micrographs of the Nb powders. The 7µm Nb is observed to be polygonal in particle shape, with good sintering necks. The 31µm and 45µm Nb powders are very angular in shape, with smaller sintering necks than the 7µm Nb. Figures 5-7 show the sintered micrographs of the three Mo-Re powders. As indicated in Table I, there is not a large variation in particle size between the three samples, and the powders are all spherical. Figure 8 is a micrograph of the Ir powder sintered at 1300°C. Figure 9 is the dilatometry plot obtained from
metal-powder.net
Technical trends tapped, milled Ir powder that had not been presintered, heated at 5°C/min to 1400°C for 1 hour in high purity argon. It is observed that sintering begins at low temperatures (~1000°C) relative to the melting point (2443°C). This is due to the small particle size within the agglomerates. Similar behavior is observed in dilatometry of 7µm Nb powder that had been dehydrided at 800°C, but not presintered, shown in Figure 10. The larger Nb powders and Mo-Re powders used in this study did not produce significant shrinkage in dilatometry due to their large particle size. Figure 11 is a dilatometry plot of the -635 mesh asreceived Mo-49Re powder, showing very little sintering response, even at 1550°C. The findings indicate that Equation 1 may be used to predict the presintered pore size if the particles are spherical or near spherical. The sintered density is substituted for the packing density. In the case of the three Mo-Re powders, which are all spherical, very good agreement is reached between the predicted and measured values. The prediction does become less accurate as the particle size distribution width decreases, which is not expected, however; such small deviations from the prediction could easily be caused by other measurement errors. For example, the mercury porosimetry measurement may be more accurate at larger pore sizes. In the case of the Ir, where agglomerates were sintered, a bimodal pore size resulted because the pores within the agglomerates were smaller than those between the agglomerates. Since the measured particle size reflects the size of the agglomerates, the porosity within the agglomerates is not accounted for in the prediction of pore size. This would explain why the predicted pore size is larger than the measured pore size by a factor of 1.77 at 1200°C and 1.87 at 1300°C. It is observed that as the Nb particle size increases, the particles become less spherical and more angular. It was also observed, as in the Mo-Re powders, that the pore size predictions were increasingly oversized as the particle size distribution width decreased. The micrographs show that the angular particles have flat edges that often lie against each other, which creates wedge shaped pores, which may measure smaller than they actually
metal-powder.net
Figure 9: Dilatometry plot of milled, tapped Ir powder heated at 5°C/min to 1400°C for 1 hour. The green density was 9.5 g/cm3 and the sintered density measured 15.7 g/cm3.
Figure 10: Dilatometry plot of dehydrided, tapped 7mm Nb powder heated at 5°C/min to 1400°C for 1 hour. The green density was 3 g/cm3 and the sintered density measured 3.8 g/cm3.
are in mercury porisimetry, from ink well effects. Correspondingly, the predicted pore size becomes increasingly oversize as the particles become more angular. It is considered that the shape factor is dominant over particle size distribution factors in the case of Nb powders. Levinskii, et al. [12, 13] studied sintering of niobium to produce porous materials using a variety of particle sizes and sintering temperatures. In their work, samples were compacted from 1µm, 2µm, and 8µm powder and samples were measured for pore size via mercury porosimetry. This study indicated that
in the initial stages of sintering, the pore size increased as sintering time or temperature increased because the smallest pores were eliminated, leaving behind
The team THE CISP team investigating porous refractory materials included roject engineer Neal Myers, research technologist Tim Mueller and Randall German, Brush Chair Professor in Materials.
September 2004 MPR
23
Technical trends
Nb and Ir, when sintering temperature was increased, pore size decreased. Dilatometry of the Ir and 7µm Nb powders demonstrates that sintering shrinkage begins at relatively low temperatures, and that the shrinkages for the cast samples was lower than that of the dilatometry powders. This is because the cast samples used powder
that had been pre-processed, which included a presintering step of 1000°C for the Ir and 1050°C for the 7µm Nb. These temperatures are high enough to cause the smallest particles to sinter, so that when the cake is broken back down into powder and made into pellets, the effective particle size has been increased, and high temperature sintering produces less shrinkage than predicted by dilatometry of as-received powder. This allows for lower density and larger pore size to be attained than if the powders were not presintered. The difference in particle size from preprocessing was not observed in laser scattering because in the as-received state, many of the smallest particles were agglomerated, and measured as larger particles. This investigation shows that porous Nb, Mo-49Re, and Ir may be produced by slurry casting into soft tooling and vacuum sintering. The pore size and density can be controlled by adjusting the particle size, packing density, and sintering temperature. In the case of spherical or near spherical powders, the pore size can be accurately predicted by a particle packing equation.
5. R L Mannheim and J L Garin, Strain Hardening of Rhenium and two Typical Molybdenum-Rhenium Alloys Manufactured by Powder Sintering, Zeitschrift fur Metallkunde, 6. R M German, Particle Packing Characteristics, Metal Powder Industries Federation, pp. 298-300, 1989. 7. R M German, S. V. Atre, J. A. Thomas, Large, Low-Production Quantity Components via PolymerAssisted Shaping and Sintering Technologies, PM Science and Technology Briefs, 2002, Vol. 4, No.1, pp. 9-13. 8. R M German, Sintering Theory and Practice, John Wiley & Sons, pp. 12-13, 1996. 9. R M German and A. Bose, Injection Molding of Metals and Ceramics, Metal Powder Industries Federation, p. 62, 1997. 10. R M German, Prediction of Sintered Density for Bimodal Powder Mixtures,
Metallurgical Transactions A, 1992, Vol. 23A, pp. 1455-1465. 11. R M German, Sintering Densification for Powder Mixtures of Varying Distribution Widths, Acta Metallurgica et Materialia, 1992, Vol. 40, No. 9, pp. 2085-2089. 12. Y V Levinskii, A B Zaitsev, Y M Polyakov, Y. B. Patrikeev, Interrelationship Between the Properties of Powders, Pressing and Sintering Conditions, and the Structure of Porous Niobium II. Features of Sintering of Niobium Powders, Poroskovaya Metallurgiya, 1992, No. 5, X pp. 29-33. 13. Y V Levinskii, A B Zaitsev, Y M Polyakov, Y B Patrikeev, Interrelationship Between the Properties of Powders, Pressing and Sintering Conditions, and the Structure of Porous Niobium III. The Porous Structure of Material of Sintered Niobium Powders, Poroskovaya Metallurgiya, 1990, No. 7, pp. 23-28.
Figure 11: Dilatometry of tapped -635 Mo-Re powder, heated at 5°C/min to 1550°C for 1 hour.
the larger pores. The pore size measurements from sintered 8µm powder correspond well with the results from 7µm powder in the present work. In their work, a pore size of 4.4µm was observed in 8µm powder, sintered at 1400°C for one hour. The effect of increasing pore size with increased sintering was not confirmed by the present work. In both
References 1. H R Z Sandim and A F Padilha, Sinterability of Commercial-Purity Niobium, Key Engineering Materials, Vols. 189-191, pp 296-301. 2. H S Huang and K S Hwang, Deoxidation of Molybdenum during Vacuum Sintering, Metallurgical and Materials Transactions A, 2002, Vol 33A, pp. 657 - 664. 3. T Patrician, V Sylvester and R Daga, High-Temperature PM - Molybdenum Alloys, Physical Metallurgy and Technology of Molybdenum and its Alloys, Proceedings of AMAX Materials Research Center Symposium, eds K Miska, M Semchyshen and E Whelan, 1986, pp. 1-11 4. Y Hiraoka, S Yoshimura, Effects of Complex Additions of Re or Ti with C on the Strength and Ductility of Recrystallized Molybdenum, Proceedings of the 13th International Plansee Seminar, 1993, Vol. 1, pp. 574-584.
24
MPR September 2004
metal-powder.net
A closer look at porous refractories
ew studies have been done on the topic of porous refractory metals, iridium in particular. Due to the higher cost of these materials, net shaping is very important to avoid machining waste. However, in low quantity production, hard tooling is not economical, making slurry casting into soft tooling more attractive. More efforts have been made towards high density processing of niobium and molybdenum [1, 2, 3]. Grain growth is a limiting factor in attaining high densities in Nb and Mo based materials, so efforts have been made to pin the grain boundaries in these systems to reach higher density [4]. Rhenium is typically added to Mo to improve ductility and tensile strength by formation of solid solution. At 49%Re addition, used in this study, a solid solution with Re-rich precipitates is observed at room temperature [5]. Without the addition of pore-forming agents, the green pore size of a powder compact is a function of particle size, particle shape, particle size distribution, and packing density. Given the particle size D, and packing density ∂, the mean pore size d can be estimated by the following equation [6]:
F
A research team at Penn State's Centre for Innovative Sintered Products has used slurry casting to produce high-porosity metals with tailored pore sizes…
(equation 1)
Figure 1: Relationship between packing density, particle size, and pore size, predicted by equation 1.
Table I: Powder Characteristics Sample Mo-Re, as received
D10, µm D50, µm D90, µm 5.3 9.8 15.6
Shape Spherical
Mo-Re -32 µm sieved
7.2
13.7
20.3
Spherical
Mo-Re +32 µm sieved
9.2
16.6
28.1
Spherical
Nb, 7 µm, as received
3.2
6.7
12.1
Rounded, irregular
Nb, 7 µm, presintered at 1000 °C
3.3
6.4
10.9
Rounded, irregular
Nb, -45 µm, as received
23.1
39.5
63
Angular
Nb, -32 µm sieved
18.9
31.1
47
Angular
Nb, -45 µm, +32 m sieved
31.2
45
66
Angular
Ir, as received
3.7
8.5
16.8
Agglomerated
Ir, presintered at 1000 °C and milled
3.6
8.4
17.5
Sintered agglomerates
20
MPR September 2004
Figure 1 indicates the predicted pore size from the equation, given a packing density and particle size. This equation assumes that the particles are monosized spheres, so it is only a first approximation of pore size. The equation may also be used to predict the pore size of a presintered structure, since pore shape is only slightly altered in the presintering process. In this case the sintered density replaces the packing density f in the equation. One Ir powder, two Nb powders, and one Mo-49Re powder were used in this study. Each powder was processed to alter its particle size distribution before shaping and sintering. The powders are listed in Table I with the particle size analysis in the as-received condition. It is important to note that the Ir powder and the 7µm Nb powder both contained amounts of agglomerated sub-micron particles, which were interpreted as discreet particles by laser scattering wet particle size analysis. This caused the
0026-0657/04 ©2004 Elsevier Ltd. All rights reserved.
Technical trends particle size measurement to indicate a larger particle size than actually existed. Ultrasonics were used to decrease the severity of agglomeration prior to testing. Powder Pre-Processing Niobium The 7µm Nb powder was first presintered by heating 5°C/min to 300°C, 2°C/min to 800°C for 1 hour and 10°C/min to 1050°C for 1 hour. This cycle served to dehydride and deoxidise the powder, and sinter the finest particles to the larger particles or to themselves, eliminating the agglomerated sub-micron particles. Presintering of the powder did not result in a hard cake. The soft cake was returned to powder condition upon removal from the sintering crucible. No significant changes in particle size were observed, however, the particle size was increased, since the agglomerated fine particles were eliminated. The -45µm (325 mesh) Nb powder was sieved to +/32µm (450 mesh) to create two particle size distributions. The particle size analyses of all three powders are given in Table I. Mo-49Re The Mo-49Re powder was specified as 20µm (-635 mesh) in the as-received condition. Because of the powder's tendency to agglomerate, most of the powder would not go through a 635 (20µm) screen. Two new particle size distributions were obtained by sieving with a 450 (32µm)
screen, and taking the +450 and -450 portions. The particle size distributions are given in Table I. Iridium The iridium powder as-received was highly agglomerated, much more so than the 7µmNb powder. In order to reduce the agglomeration and increase the particle size, the powder was presintered at 1000°C in 10-4 Torr vacuum tube furnace, resulting in a presintered powder cake. Break-up of the cake was performed by pushing the cake through a 20 mesh sieve with milling rods, then rod milling for 10 minutes. The particle sizes before and after this process are given in Table I. The measured particle size did not change significantly, however, since the asreceived particle size measurement was effectively measuring the agglomerate size. Slurry casting was used to produce the samples, to allow for low green density and shape complexity [7]. In this process, powder is mixed with melted Dussek-Campbell paraffin wax with a melting point of 55°C and poured hot into a rubber mold, avoiding the need for hard tooling. Some settling of the powder occurs before the wax solidifies. The compositions of the slurry for each material and the resulting green densities are given in Table II. All slurry was cast at 110°C into a mold preheated to 80°C and allowed to cool at room temperature. The piece was ground down to the correct height, allowing for removal of the
Figure 2: 7mm Nb sintered at 1450°C
Figure 3: 31mm Nb sintered at 1450°C
Figure 4: 45mm Nb powder sintered at 1500°C
Table II: Sintering Results Material D50 Particle Size, (µm)
Distribution Green Sintering Slope Density, (%) Condition Parameter, Sw
Sintered Density, (%)
Measured Predicted Predicted/ Average Pore Average Pore Measured Diameter, (µm) Diameter, (µm) Ratio
Mo-49Re
9.8
5.5
45
1500 C 4 h
61
3.4
4.1
1.21
Mo-49Re
13.7
5.7
48
1500 C 4 h
56
5.8
7.2
1.24
Mo-49Re
16.6
5.3
44
1500 C 4 h
57
7.9
8.2
1.04
Nb
6.4
4.9
34
1350 C 2 h
43
3.5
5.6
1.63
Nb
6.4
4.9
33
1450 C 2 h
52
2.5
3.8
1.52
Nb
31
6.5
46
1450 C 2 h
52
10.2
19.0
1.87
Nb
45
7.8
44
1450 C 2 h
49
13.2
31.2
2.37
Ir
8.4
3.7
33
1200 C 1 h
35
5.8
10.2
1.77
Ir
8.4
3.7
33
1300 C 1 h
46
3.4
6.4
1.87
metal-powder.net
September 2004 MPR
21
Technical trends
Figure 5: 17mm Mo-Re sintered at 1500°C for 4 hours.
low density material on top due to powder settling. Cylinders with a diameter of 4 mm were cast for use in all green density, sintered density, and porosimetry measurements. Complex shapes produced were 7 cm diameter discs with and without circular surface corrugations. Debinding and Sintering Niobium Debinding and presintering were performed by surrounding the part in 10µm alumina powder and heating at 2°C/min to 120°C for 1 hour, 2°C/min to 250°C for 1 hour, 1°C/min to 450°C for 1 hour, 5°C/min to 1125°C for 1 hour, and 10°C/min to 20°C in high purity argon. The parts were removed from the alumina packing and heated at 5°C/min to 1350°C or 1450°C for 1 hour in 10-5 Torr in an all-metal construction vacuum furnace with tungsten heating elements. Mo-49Re Debinding was performed by surrounding the parts in 10µm alumina powder and heating at 2°C/min to 120°C for 1 hour, 2°C/min to 250°C for 1 hour, 1°C/min to 450°C for 1 hour, 5°C/min to 1200°C for 1 hour, and 10°C/min to 20°C in high purity argon. The parts were then presintered in the alumina packing at 1500°C for 4 hours in 10-5 Torr.
Figure 6: 14mm Mo-Re sintered at 1500°C for 4 hours.
milled powder was sintered in a vertical dilatometer at 5°C/min to 1400°C for 1 hour in high purity argon. Sample Evaluation Green density and sintered density were measured by Archemedes technique, while sintered pore size was evaluated by mercury porosimetry. Metallography was performed on selected samples. Table II summarises the sintered samples and their properties. Pore size was predicted based on equation 1, substituting the sintered density for the packing density, and using the average particle size. Since the samples are all in the presintered state, this is a reasonable approximation. During the presintering stage, the pore shape does not change significantly, since surface diffusion is dominant [8]. There is a small increase in density from the green to sintered condition, and this is a combination of constriction of the compact as wax is removed in debinding, and sintering shrinkage. Metallography indicates that no real particle morphology changes have occurred; only sintering necks have been formed. In general, the predicted pore size is larger than the measured pore size. Only the Mo-Re powders are spherical. It is observed that the best agreement with the equation is observed with the Mo-Re
Iridium Debinding and presintering were performed by surrounding the parts in 10µm alumina powder and heating at 2°C/min to 120°C for 1 hour, 2°C/min to 250°C for 1 hour, 2°C/min to 450°C for 1 hour, 5°C/min to 1100°C for 1 hour, and 10°C/min to 20°C in high purity argon. Parts were removed from the alumina before heating at 5°C/min to 1200°C or 1300°C for one hour in 10-5 Torr. Tapped
22
MPR September 2004
Figure 8: Ir sintered at 1300°C
Figure 7: 10mm Mo-Re sintered at 1500°C for 4 hours.
powders. The worst agreement is observed with the 31µm and 45µm Nb powders that are very angular in shape. The particle size distribution width is given by the slope parameter [9]:
(eq. 2)
This parameter increases as particle size distribution becomes narrower. Comparison between this value and the predicted/measured ratio of pore size reveals that as particle size distributions become narrower in Nb and Mo-Re powders, the prediction of pore size becomes less accurate, and more oversized. This is the opposite trend than expected, since the equation is derived for monosized spheres. As particle size distributions become wider, the smaller particles can fit into the spaces between the larger particles, resulting in higher packing densities and smaller pores [10,11]. Based on this, the wider particle size distributions should result in pore size smaller than predicted from using the D50 particle size in equation 1. Figures 2-4 display the sintered micrographs of the Nb powders. The 7µm Nb is observed to be polygonal in particle shape, with good sintering necks. The 31µm and 45µm Nb powders are very angular in shape, with smaller sintering necks than the 7µm Nb. Figures 5-7 show the sintered micrographs of the three Mo-Re powders. As indicated in Table I, there is not a large variation in particle size between the three samples, and the powders are all spherical. Figure 8 is a micrograph of the Ir powder sintered at 1300°C. Figure 9 is the dilatometry plot obtained from
metal-powder.net
Technical trends tapped, milled Ir powder that had not been presintered, heated at 5°C/min to 1400°C for 1 hour in high purity argon. It is observed that sintering begins at low temperatures (~1000°C) relative to the melting point (2443°C). This is due to the small particle size within the agglomerates. Similar behavior is observed in dilatometry of 7µm Nb powder that had been dehydrided at 800°C, but not presintered, shown in Figure 10. The larger Nb powders and Mo-Re powders used in this study did not produce significant shrinkage in dilatometry due to their large particle size. Figure 11 is a dilatometry plot of the -635 mesh asreceived Mo-49Re powder, showing very little sintering response, even at 1550°C. The findings indicate that Equation 1 may be used to predict the presintered pore size if the particles are spherical or near spherical. The sintered density is substituted for the packing density. In the case of the three Mo-Re powders, which are all spherical, very good agreement is reached between the predicted and measured values. The prediction does become less accurate as the particle size distribution width decreases, which is not expected, however; such small deviations from the prediction could easily be caused by other measurement errors. For example, the mercury porosimetry measurement may be more accurate at larger pore sizes. In the case of the Ir, where agglomerates were sintered, a bimodal pore size resulted because the pores within the agglomerates were smaller than those between the agglomerates. Since the measured particle size reflects the size of the agglomerates, the porosity within the agglomerates is not accounted for in the prediction of pore size. This would explain why the predicted pore size is larger than the measured pore size by a factor of 1.77 at 1200°C and 1.87 at 1300°C. It is observed that as the Nb particle size increases, the particles become less spherical and more angular. It was also observed, as in the Mo-Re powders, that the pore size predictions were increasingly oversized as the particle size distribution width decreased. The micrographs show that the angular particles have flat edges that often lie against each other, which creates wedge shaped pores, which may measure smaller than they actually
metal-powder.net
Figure 9: Dilatometry plot of milled, tapped Ir powder heated at 5°C/min to 1400°C for 1 hour. The green density was 9.5 g/cm3 and the sintered density measured 15.7 g/cm3.
Figure 10: Dilatometry plot of dehydrided, tapped 7mm Nb powder heated at 5°C/min to 1400°C for 1 hour. The green density was 3 g/cm3 and the sintered density measured 3.8 g/cm3.
are in mercury porisimetry, from ink well effects. Correspondingly, the predicted pore size becomes increasingly oversize as the particles become more angular. It is considered that the shape factor is dominant over particle size distribution factors in the case of Nb powders. Levinskii, et al. [12, 13] studied sintering of niobium to produce porous materials using a variety of particle sizes and sintering temperatures. In their work, samples were compacted from 1µm, 2µm, and 8µm powder and samples were measured for pore size via mercury porosimetry. This study indicated that
in the initial stages of sintering, the pore size increased as sintering time or temperature increased because the smallest pores were eliminated, leaving behind
The team THE CISP team investigating porous refractory materials included roject engineer Neal Myers, research technologist Tim Mueller and Randall German, Brush Chair Professor in Materials.
September 2004 MPR
23
Technical trends
Nb and Ir, when sintering temperature was increased, pore size decreased. Dilatometry of the Ir and 7µm Nb powders demonstrates that sintering shrinkage begins at relatively low temperatures, and that the shrinkages for the cast samples was lower than that of the dilatometry powders. This is because the cast samples used powder
that had been pre-processed, which included a presintering step of 1000°C for the Ir and 1050°C for the 7µm Nb. These temperatures are high enough to cause the smallest particles to sinter, so that when the cake is broken back down into powder and made into pellets, the effective particle size has been increased, and high temperature sintering produces less shrinkage than predicted by dilatometry of as-received powder. This allows for lower density and larger pore size to be attained than if the powders were not presintered. The difference in particle size from preprocessing was not observed in laser scattering because in the as-received state, many of the smallest particles were agglomerated, and measured as larger particles. This investigation shows that porous Nb, Mo-49Re, and Ir may be produced by slurry casting into soft tooling and vacuum sintering. The pore size and density can be controlled by adjusting the particle size, packing density, and sintering temperature. In the case of spherical or near spherical powders, the pore size can be accurately predicted by a particle packing equation.
5. R L Mannheim and J L Garin, Strain Hardening of Rhenium and two Typical Molybdenum-Rhenium Alloys Manufactured by Powder Sintering, Zeitschrift fur Metallkunde, 6. R M German, Particle Packing Characteristics, Metal Powder Industries Federation, pp. 298-300, 1989. 7. R M German, S. V. Atre, J. A. Thomas, Large, Low-Production Quantity Components via PolymerAssisted Shaping and Sintering Technologies, PM Science and Technology Briefs, 2002, Vol. 4, No.1, pp. 9-13. 8. R M German, Sintering Theory and Practice, John Wiley & Sons, pp. 12-13, 1996. 9. R M German and A. Bose, Injection Molding of Metals and Ceramics, Metal Powder Industries Federation, p. 62, 1997. 10. R M German, Prediction of Sintered Density for Bimodal Powder Mixtures,
Metallurgical Transactions A, 1992, Vol. 23A, pp. 1455-1465. 11. R M German, Sintering Densification for Powder Mixtures of Varying Distribution Widths, Acta Metallurgica et Materialia, 1992, Vol. 40, No. 9, pp. 2085-2089. 12. Y V Levinskii, A B Zaitsev, Y M Polyakov, Y. B. Patrikeev, Interrelationship Between the Properties of Powders, Pressing and Sintering Conditions, and the Structure of Porous Niobium II. Features of Sintering of Niobium Powders, Poroskovaya Metallurgiya, 1992, No. 5, X pp. 29-33. 13. Y V Levinskii, A B Zaitsev, Y M Polyakov, Y B Patrikeev, Interrelationship Between the Properties of Powders, Pressing and Sintering Conditions, and the Structure of Porous Niobium III. The Porous Structure of Material of Sintered Niobium Powders, Poroskovaya Metallurgiya, 1990, No. 7, pp. 23-28.
Figure 11: Dilatometry of tapped -635 Mo-Re powder, heated at 5°C/min to 1550°C for 1 hour.
the larger pores. The pore size measurements from sintered 8µm powder correspond well with the results from 7µm powder in the present work. In their work, a pore size of 4.4µm was observed in 8µm powder, sintered at 1400°C for one hour. The effect of increasing pore size with increased sintering was not confirmed by the present work. In both
References 1. H R Z Sandim and A F Padilha, Sinterability of Commercial-Purity Niobium, Key Engineering Materials, Vols. 189-191, pp 296-301. 2. H S Huang and K S Hwang, Deoxidation of Molybdenum during Vacuum Sintering, Metallurgical and Materials Transactions A, 2002, Vol 33A, pp. 657 - 664. 3. T Patrician, V Sylvester and R Daga, High-Temperature PM - Molybdenum Alloys, Physical Metallurgy and Technology of Molybdenum and its Alloys, Proceedings of AMAX Materials Research Center Symposium, eds K Miska, M Semchyshen and E Whelan, 1986, pp. 1-11 4. Y Hiraoka, S Yoshimura, Effects of Complex Additions of Re or Ti with C on the Strength and Ductility of Recrystallized Molybdenum, Proceedings of the 13th International Plansee Seminar, 1993, Vol. 1, pp. 574-584.
24
MPR September 2004
metal-powder.net