- Email: [email protected]
Phase identification in reactively sintered molybdenum disilicide composites
Materials Science and Engineering A261 (1999) 181 – 187
Phase identification in reactively sintered molybdenum disilicide composites K.K. Chawla a,*, J.J. Petrovic b, Jose Alba Jr. c, R. Hexemer d a
Department of Materials and Mechanical Engineering, Uni6ersity of Alabama at Birmingham, Birmingham, AL 35294, USA b Los Alamos National Laboratory, MST-4, Los Alamos, NM 87801, USA c S&S Energy Products, Houston TX 77530, USA d Ad6anced Refractory Technologies, Inc., Buffalo, NY, USA
Abstract The objective of this research was to identify and quantify the phases formed in reactively sintered MoSi2 and MoSi2 composites made from carbon coated MoSi2 powders. The purpose of adding carbon was to form silicon carbide particles (SiCp) in situ and reduce the inherent presence of SiO2 in MoSi2. The carbon additions were made via polymeric coatings on MoSi2 particles by two processes, Phenolic Resin Based Carbon by Solvent Evaporation (PRBCSE) and Aqueous Dispersion Flocculation (ADF). The sintering temperatures and times ranged from 1600 to 1800°C, and from 1 to 100 h, respectively. The addition of carbon did reduce the presence of SiO2 and there was formation of SiCp. However, the SiCp formation was less than expected, a maximum of 1.4 vol.% SiCp was formed. The Nowotny phase (Mo 5 4 8Si3C 5 0.6) was observed in the sintered samples. The in situ formation of SiCp would increase the toughness of MoSi2 by serving as a reinforcement. The reduction in the amount of SiO2 would reduce the incidence of grain boundary sliding caused by viscous flow of SiO2 at elevated temperatures. © 1999 Elsevier Science S.A. All rights reserved. Keywords: Silicon carbide particles; Grain boundary sliding; Molybdenum disilicide; Composites
1. Introduction Molybdenum disilicide has been predominantly used for furnace heating elements, but recently there has been interest in its use for high temperature structural applications [1,2]. This increased interest stems from its desirable characteristics that include a high melting point, relatively low density, good oxidation resistance, and relatively good thermal conductivity [3,4]. MoSi2 is also electrically conductive which allows it to be electro discharge machined. An important characteristic of MoSi2, common to all silicides, is its tendency to form SiO2 when exposed to air [5]. Although, in the case of a consolidated component, the ability of MoSi2 to form a protective SiO2 surface layer is advantageous, in the powder form the SiO2 forms on the surface of almost all powder silicide particles [6]. The SiO2 on the surface of the powder particles ends up at grain boundaries in the microstruc* Corresponding author. Tel.: +1-1205-975-9725; fax: +1-205934-8485; e-mail: [email protected].
ture after consolidation, causing easy grain boundary sliding at elevated temperatures, which makes for low creep resistance at elevated temperatures. From a processing point of view one can eliminate the presence of the deleterious SiO2 in the microstructure of the consolidated component by two methods. The first would be to simply minimize the exposure of the starting powder to air. This is not economically feasible because it is virtually impossible to avoid air exposure. The second route would be to eliminate the SiO2 by adding a Table 1 Composition of the different powder blends by wt.%a
ARSE 70/30 ARSE 0/100 PRBCSE ADF
Grade A (wt.%)
Grade C (wt.%)
Actual carbon (wt.%)
70 0 0 0
30 100 98.79 97.12
0 0 1.21 2.88
a Grade A and Grade C are pure MoSi2 with powder particle sizes of : 10 and 3 mm, respectively.
0921-5093/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 9 8 ) 0 1 0 6 4 - 8
182
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
Fig. 1. Schematic of the phenolic resin based carbon by solvent evaporation process.
scavenging material coating on the starting powder which would eliminate the SiO2 and form a desirable phase to produce a composite. Carbon would appear to be a prime candidate for such use because it could reduce SiO2 by converting it to silicon carbide particles (SiCp). The SiCp would be an excellent second phase because it is a high temperature refractory material that has been used as a reinforcement and it is thermodynamically stable with MoSi2 [7].
2. Materials and experimental procedure The four types of MoSi2 powder1 blends used in this work are given in Table 1. Grade A and grade C are pure MoSi2 powders with particle sizes of : 10, and 3 mm, respectively. Carbon coating was put on the Grade C MoSi2 particles by two different processes, Phenolic Resin Based Carbon by Solvent Evaporation (PRBCSE) and Aqueous Dispersion Flocculation (ADF). The actual carbon content in the powders after pyrolysis was measured at ART [8], by using a Leco carbon analyzer. The PRBCSE powder used a phenol formaldehyde resin, Polyophen 2305622, to apply a carbon coating on 1 2
Advanced Refractory Technologies, Inc., Buffalo, NY, USA. Durez Corp., North Yonawanda, NY, USA.
the MoSi2 particles. The Grade C MoSi2 powder, resin, and acetone were ball-milled. The acetone was then evaporated leaving only the resin coating on the MoSi2 particles. The resin coating was then pyrolyzed at 600°C for 1 h in argon to yield a carbon coating on MoSi2 particles. The coated powder was then pressed into a green compact in a steel die and sintered. A schematic of the process is shown in Fig. 1. Due to the nonuniform distribution of SiCp found in some preliminary samples, these powders were ball-milled using WC media to obtain a more uniform distribution of carbon in the MoSi2 powders and SiCp in the consolidated samples. In the ADF process, a high carbon yielding cationic starch flocculant was electrostatically adhered to the surface of the MoSi2 particles. The starch used was Redibond 532033 which is a liquid natural polymer with cationic quaternary amine functionality. The process involved dispersing the MoSi2 powder in an aqueous system by adding an anionic dispersant and ball-milling for 1 h. The starch flocculant and glycerol plasticizer were added to the slurry and ball-milled for another 30 min. The slurry was put in a high shear blender and the pH was increased to activate the flocculant which gelled the slurry. At this point the MoSi2 powder was uni3
National Starch and Chemical Company, Bridgwater, NJ, USA.
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
183
Fig. 2. Schematic of the ADF process.
formly coated with the starch. Water was added to thin the slurry and put into a vacuum forming die to remove the water. The material was dried and passed through a 100 mesh screen. The powder was pressed into a green sample and pyrolyzed at 600°C for 1 h in vacuum to convert the starch into carbon. The sample was then sintered in argon. A schematic of this procedure is shown in Fig. 2. The powders were cold pressed under a pressure of 140 MPa into thin, disk shaped, green samples. These disks were approximately 2.9 cm in diameter (inside diameter of the die), :0.4 cm in height, and weighed :8 g. Before sintering, the samples were embedded in a mixture of 50 wt.% MoSi2/50 wt.% BN or placed on top of SiC disks. The purpose of embedding the material was to isolate the samples from the graphite crucible, minimize volatilization of the carbon, obtain uniform shrinkage, improve microstructural uniformity, and enhance densification. BN powder was used to prevent the MoSi2 bed from densifying during sintering. Samples sintered with the MoSi2/BN powder showed MoB formation at the surface. In order to avoid this, SiC disks were used to isolate the samples from the graphite crucible. Multiple sets of samples were sintered at 1600, 1700 and 1800°C and 1, 10 and 100 h. Samples were heated at 10°C min − 1 and held at 600°C for 1 h in vacuum to pyrolyze the starch in the ADF samples. After this, argon gas was introduced into the chamber
and sintering was done in argon. The starch was not pyrolyzed on the MoSi2 particles before cold-pressing so that it could be used as a binder in the cold-pressed samples to toughen the compacts for handling. The sintered samples were polished to a 600 grit finish using SiC grinding paper to obtain a flat surface and to remove the surface layer for X-ray diffraction (XRD). Siemens D5000 and Philips PW-1720 X-ray diffractometers were used for XRD analysis. After XRD, the samples were mounted in epoxy and polished to a 0.1 mm finish using diamond paste. Optical microscopy (OM) and scanning electron microscopy (SEM) were used to characterize the microstructure. OM (Zeiss and a Unitron Versamet-2) was used to examine the porosity in bright field and grain structure in polarized light. A Hitachi Hi-scan HHS-2R and JEOL high resolution SEM were used for the SEM examination. A JEOL 733 microprobe with energy and wavelength dispersive spectrometry (EDS and WDS) attachments with low atomic number element detection capabilities was used for phase identification and distribution analysis. Software was used for WDS analysis which provided a ZAF (atomic number, absorption, and fluorescence) correction to determine atomic percentages of elements in compounds. Sigma Scan Pro4 was used for the image analysis. 4
Jandel Scientific, San Rafael, CA, USA.
184
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
Fig. 3. A typical micrograph used to determine the volume fraction of SiCp in an ADF processed sample sintered at 1700°C for 100 h. The average vol.% was 1.4% SiCp.
3. Results and discussion
3.1. General microstructure Fig. 3 shows a typical optical micrograph showing the general view of the distribution of phases, including porosity. When the as-received PRBCSE powder was sintered, it showed the SiC formation around pores (see Fig. 4). This indicated that the carbon was not evenly distributed in the starting powder and that the powder needed to be ball-milled, which was done as described in Section 2.
Fig. 5. Monolithic MoSi2 (ARSE 0/100) sample sintered at 1700°C for 100 h on SiC disks. (a) SEM, and (b) oxygen map of the area in (a) showing high oxygen content.
3.2. Phase identification
Fig. 4. Optical micrograph of as-received, PRBCSE, sintered at 1700°C for 1 h. Notice that the SiC (gray phase) forms at pores.
3.2.1. SiO2 Oxygen maps obtained by energy dispersive X-ray analysis were used to determine if the carbon addition reduced the presence of SiO2 (Fig. 5(b) and Fig. 6). Fig. 5 and Fig. 6(b) show the EDS oxygen maps for the ARSE 0/100 (3 mm particle size MoSi2) and the ADF (3 mm particle size MoSi2 with 2.88 wt.% carbon) sintered on SiC disks at 1700°C for 100 h, respectively. The oxygen maps indicate that indeed the SiO2 content was reduced by the addition of carbon.
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
185
The oxygen maps were used in conjunction with image analysis software to obtain the vol.% of SiO2 in the sintered samples. Fig. 7 indicates that the ADF samples all showed very little SiO2 content compared to the monolithic MoSi2 (ARSE 0/100) indicating that the carbon addition reduced the SiO2 content. Fig. 7 also shows that the PRBCSE samples increased in SiO2 content compared to the ARSE 0/100. This is understandable inasmuch as the ball-milling reduced the particle size and increased the surface area and consequently, the amount of SiO2 in the starting powder. Fig. 7. Vol.% of SiO2 versus sintering time for the samples sintered on SiC disks at 1700°C. Obtained from oxygen EDS maps. Note that the ADF samples showed a lower SiO2 content than the ARSE 0/100. The PRBCSE samples showed a higher SiO2 content, which resulted from the ball-milling.
3.2.2. Porosity Image analysis software was used to obtain volume percentage of porosity. This number was then subtracted from the volume percentage of porosity plus SiO2 obtained using the optical micrographs giving the volume percentage of porosity (Fig. 8). The ADF samples showed higher porosity compared to the monolithic MoSi2 (ARSE 0/100) samples. We attribute this to the porosity in the ADF samples left after the pyrolysis of the starch before sintering. The porosity in the PRBCSE samples exhibited a similar trend as the ARSE 0/100 but was lower than the ARSE 0/100 indicating that the high SiO2 content reduced the porosity found in the sintered samples. 3.2.3. SiC, Mo 5 4.8Si3C 5 0.6 and MoSi2 Molybdenum disilicide was easily detectable by XRD but SiC and Mo 5 4.8Si3C 5 0.6 (the Nowotny phase) were not, because of their small amounts. Thus, a more localized analysis technique was required. The results of
Fig. 6. ADF sample sintered at 1700°C for 100 h on SiC disks. (a) SEM micrograph, and (b) EDS oxygen map of the area shown in (a). Note the low oxygen content.
Fig. 8. Vol.% porosity versus sintering time for the samples sintered on SiC disks at 1700°C. Note that the ADF sample showed more porosity because of voids left after the pyrolysis of the starch. The PRBCSE and ARSE 0/100 samples showed the same trend.
186
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
Fig. 9. (a) SEM micrograph of ADF powder sintered at 1700°C for 10 h. Arrow indicates area analyzed. (b) EDS spectra of the region analyzed. Mo, Si, and C peaks indicate that this region is the Nowotny phase.
electron microprobe analysis involving both light element detection EDS and WDS on an ADF sample sintered at 1700°C for 10 h are shown in Figs. 9 and 10. The light element detection EDS results showed that the brightest phase in the secondary electron micrographs (Fig. 9) contained Mo, Si, and C which indicated that this phase was the Nowotny phase. Similar results were obtained by Costa e Silva et al. [9], for powder compositions in the same region on the Nowotny ternary phase diagram (i.e. MoSi2, SiC, and Mo 5 4.8Si3C 5 0.6) for 1600°C [10]. The light element detection WDS results from this same compound indicated that it was indeed the Nowotny phase with Mo4.7Si3C0.56 stoichiometry, see Table 2. Fig. 8 shows that the amount of the Nowotny phase formed in this sample was less than 1 vol.%. The darkest region in Fig. 10 was thought to be SiC because it only showed Si and C using the light element detection EDS. The light element detection WDS analysis of the same region indicated that the stoichiometry was not perfect but was SiC0.86 (see Table 2). This may be caused by noise coming from the matrix, which would lead to a
Fig. 10. (a) SEM micrograph of ADF powder sintered at 1700°C for 10 h. Arrow indicates area analyzed. (b) EDS spectra of the region analyzed. The Si and C peaks indicate that this region is SiC0.86.
higher amount of Si detected than actually present in the compound. Optical micrographs were used to determine how much SiC formed in the ADF samples sintered at 1700°C for 100 h (Fig. 3). This sample which contained more SiCp than all of the other samples had only 1.4 vol.%. It is likely that silicon monoxide (SiO) formed along with CO and CO2 during the initial stages of sintering. This could account for the small amounts of SiCp and the Nowotny phase in the final product. It
Table 2 Compositional data obtained by WDSa At.% Mo At.% Si At.% C Theoretical composition
33.726 62.692 – MoSi2
– 53.263 45.670 SiC
56.935 36.304 6.761 Mo54.8Si3C50.6
Actual composition
MoSi1.9
SiC0.86
Mo4.7Si3C0.56
a Note the composition for SiC is slightly off stoichiometry possibly caused by signal coming from the matrix.
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
187
4. Conclusions 1. Silica was reduced significantly by the addition of carbon to the starting powders. 2. Silicon carbide and the Nowotny phase were observed. 3. The amount of SiCp formed was very small. The by-products of reactive sintering need to be considered for the formation of CO and CO2.
Acknowledgments
Fig. 11. SEM micrograph of a crack produced using a Vickers indentor. Notice that the SiCp deflects the crack. The arrow indicates the direction of the crack propagation.
would be desirable to analyze the effluent gases during sintering to verify this point.
3.3. Fracture toughness Fig. 11 shows an SEM micrograph of a crack formed by a Vickers indentation (1 kg, 15 s) and its interaction with a variety of particles in the MoSi2. The first unidentified particle stopped the crack and the crack reinitiated on the opposite side. After the crack was reinitiated, it encountered another unidentified particle and the crack deflected around this particle. The crack then penetrated the SiCp and was arrested there. A new crack was then formed in the SiCp that propagated into the MoSi2. Such crack deflection and reinitiation would enhance the toughness of the composite. However, one needs an appreciable amount of SiCp to have some reasonable increase in toughness.
.
This work was supported by Los Alamos National Laboratory under Contract No. 9-X63-0402J-1 Task No. 9 and the US Office of Naval Research under Contract No. N00014-96-1-0846. The starting powders were provided by ART Inc.
References [1] A.K. Vasudevan, J.J. Petrovic, (Eds.), High Temperature Structural Silicides: Proceedings of the First High Temperature Silicides Workshop, Elsevier Science Publishers, Amsterdam, (1992). [2] V. Bizzari, B. Linder, N. Lindskog, Am. Ceram. Soc. Bull. 68 (1989) 1834 – 1841. [3] S. Bose, Mater. Sci. Eng. A155 (1992) 217 – 225. [4] J. Cook, A. Khan, E. Lee, R. Mahapatra, Mater. Sci. Eng. A155 (1992) 183 – 198. [5] J. Cook, R. Mahapatra, E.W. Lee, A. Khan, J. Waldman, Ceram. Eng. Sci. Proc. 9 – 10 (1991) 1656 – 1670. [6] R.K. Wade, J.J. Petrovic, J. Am. Ceram. 75 (1992) 3160–3162. [7] A.K. Bhattacharya, J.J. Petrovic, J. Am. Ceram. Soc. 74 (1991) 2700 – 2703. [8] Advanced Refractory Technologies, Inc., Personal communication. [9] A. Costa e Silva, M.J. Kaufman, Metall. Mater. Trans. A 25 (1994) 5 – 15. [10] H. Nowotny, E. Parthe, R. Kieffer, F. Benesovsky, Monatshefte fur Chemie 85 (1954) 255.
Phase identification in reactively sintered molybdenum disilicide composites K.K. Chawla a,*, J.J. Petrovic b, Jose Alba Jr. c, R. Hexemer d a
Department of Materials and Mechanical Engineering, Uni6ersity of Alabama at Birmingham, Birmingham, AL 35294, USA b Los Alamos National Laboratory, MST-4, Los Alamos, NM 87801, USA c S&S Energy Products, Houston TX 77530, USA d Ad6anced Refractory Technologies, Inc., Buffalo, NY, USA
Abstract The objective of this research was to identify and quantify the phases formed in reactively sintered MoSi2 and MoSi2 composites made from carbon coated MoSi2 powders. The purpose of adding carbon was to form silicon carbide particles (SiCp) in situ and reduce the inherent presence of SiO2 in MoSi2. The carbon additions were made via polymeric coatings on MoSi2 particles by two processes, Phenolic Resin Based Carbon by Solvent Evaporation (PRBCSE) and Aqueous Dispersion Flocculation (ADF). The sintering temperatures and times ranged from 1600 to 1800°C, and from 1 to 100 h, respectively. The addition of carbon did reduce the presence of SiO2 and there was formation of SiCp. However, the SiCp formation was less than expected, a maximum of 1.4 vol.% SiCp was formed. The Nowotny phase (Mo 5 4 8Si3C 5 0.6) was observed in the sintered samples. The in situ formation of SiCp would increase the toughness of MoSi2 by serving as a reinforcement. The reduction in the amount of SiO2 would reduce the incidence of grain boundary sliding caused by viscous flow of SiO2 at elevated temperatures. © 1999 Elsevier Science S.A. All rights reserved. Keywords: Silicon carbide particles; Grain boundary sliding; Molybdenum disilicide; Composites
1. Introduction Molybdenum disilicide has been predominantly used for furnace heating elements, but recently there has been interest in its use for high temperature structural applications [1,2]. This increased interest stems from its desirable characteristics that include a high melting point, relatively low density, good oxidation resistance, and relatively good thermal conductivity [3,4]. MoSi2 is also electrically conductive which allows it to be electro discharge machined. An important characteristic of MoSi2, common to all silicides, is its tendency to form SiO2 when exposed to air [5]. Although, in the case of a consolidated component, the ability of MoSi2 to form a protective SiO2 surface layer is advantageous, in the powder form the SiO2 forms on the surface of almost all powder silicide particles [6]. The SiO2 on the surface of the powder particles ends up at grain boundaries in the microstruc* Corresponding author. Tel.: +1-1205-975-9725; fax: +1-205934-8485; e-mail: [email protected].
ture after consolidation, causing easy grain boundary sliding at elevated temperatures, which makes for low creep resistance at elevated temperatures. From a processing point of view one can eliminate the presence of the deleterious SiO2 in the microstructure of the consolidated component by two methods. The first would be to simply minimize the exposure of the starting powder to air. This is not economically feasible because it is virtually impossible to avoid air exposure. The second route would be to eliminate the SiO2 by adding a Table 1 Composition of the different powder blends by wt.%a
ARSE 70/30 ARSE 0/100 PRBCSE ADF
Grade A (wt.%)
Grade C (wt.%)
Actual carbon (wt.%)
70 0 0 0
30 100 98.79 97.12
0 0 1.21 2.88
a Grade A and Grade C are pure MoSi2 with powder particle sizes of : 10 and 3 mm, respectively.
0921-5093/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 9 8 ) 0 1 0 6 4 - 8
182
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
Fig. 1. Schematic of the phenolic resin based carbon by solvent evaporation process.
scavenging material coating on the starting powder which would eliminate the SiO2 and form a desirable phase to produce a composite. Carbon would appear to be a prime candidate for such use because it could reduce SiO2 by converting it to silicon carbide particles (SiCp). The SiCp would be an excellent second phase because it is a high temperature refractory material that has been used as a reinforcement and it is thermodynamically stable with MoSi2 [7].
2. Materials and experimental procedure The four types of MoSi2 powder1 blends used in this work are given in Table 1. Grade A and grade C are pure MoSi2 powders with particle sizes of : 10, and 3 mm, respectively. Carbon coating was put on the Grade C MoSi2 particles by two different processes, Phenolic Resin Based Carbon by Solvent Evaporation (PRBCSE) and Aqueous Dispersion Flocculation (ADF). The actual carbon content in the powders after pyrolysis was measured at ART [8], by using a Leco carbon analyzer. The PRBCSE powder used a phenol formaldehyde resin, Polyophen 2305622, to apply a carbon coating on 1 2
Advanced Refractory Technologies, Inc., Buffalo, NY, USA. Durez Corp., North Yonawanda, NY, USA.
the MoSi2 particles. The Grade C MoSi2 powder, resin, and acetone were ball-milled. The acetone was then evaporated leaving only the resin coating on the MoSi2 particles. The resin coating was then pyrolyzed at 600°C for 1 h in argon to yield a carbon coating on MoSi2 particles. The coated powder was then pressed into a green compact in a steel die and sintered. A schematic of the process is shown in Fig. 1. Due to the nonuniform distribution of SiCp found in some preliminary samples, these powders were ball-milled using WC media to obtain a more uniform distribution of carbon in the MoSi2 powders and SiCp in the consolidated samples. In the ADF process, a high carbon yielding cationic starch flocculant was electrostatically adhered to the surface of the MoSi2 particles. The starch used was Redibond 532033 which is a liquid natural polymer with cationic quaternary amine functionality. The process involved dispersing the MoSi2 powder in an aqueous system by adding an anionic dispersant and ball-milling for 1 h. The starch flocculant and glycerol plasticizer were added to the slurry and ball-milled for another 30 min. The slurry was put in a high shear blender and the pH was increased to activate the flocculant which gelled the slurry. At this point the MoSi2 powder was uni3
National Starch and Chemical Company, Bridgwater, NJ, USA.
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
183
Fig. 2. Schematic of the ADF process.
formly coated with the starch. Water was added to thin the slurry and put into a vacuum forming die to remove the water. The material was dried and passed through a 100 mesh screen. The powder was pressed into a green sample and pyrolyzed at 600°C for 1 h in vacuum to convert the starch into carbon. The sample was then sintered in argon. A schematic of this procedure is shown in Fig. 2. The powders were cold pressed under a pressure of 140 MPa into thin, disk shaped, green samples. These disks were approximately 2.9 cm in diameter (inside diameter of the die), :0.4 cm in height, and weighed :8 g. Before sintering, the samples were embedded in a mixture of 50 wt.% MoSi2/50 wt.% BN or placed on top of SiC disks. The purpose of embedding the material was to isolate the samples from the graphite crucible, minimize volatilization of the carbon, obtain uniform shrinkage, improve microstructural uniformity, and enhance densification. BN powder was used to prevent the MoSi2 bed from densifying during sintering. Samples sintered with the MoSi2/BN powder showed MoB formation at the surface. In order to avoid this, SiC disks were used to isolate the samples from the graphite crucible. Multiple sets of samples were sintered at 1600, 1700 and 1800°C and 1, 10 and 100 h. Samples were heated at 10°C min − 1 and held at 600°C for 1 h in vacuum to pyrolyze the starch in the ADF samples. After this, argon gas was introduced into the chamber
and sintering was done in argon. The starch was not pyrolyzed on the MoSi2 particles before cold-pressing so that it could be used as a binder in the cold-pressed samples to toughen the compacts for handling. The sintered samples were polished to a 600 grit finish using SiC grinding paper to obtain a flat surface and to remove the surface layer for X-ray diffraction (XRD). Siemens D5000 and Philips PW-1720 X-ray diffractometers were used for XRD analysis. After XRD, the samples were mounted in epoxy and polished to a 0.1 mm finish using diamond paste. Optical microscopy (OM) and scanning electron microscopy (SEM) were used to characterize the microstructure. OM (Zeiss and a Unitron Versamet-2) was used to examine the porosity in bright field and grain structure in polarized light. A Hitachi Hi-scan HHS-2R and JEOL high resolution SEM were used for the SEM examination. A JEOL 733 microprobe with energy and wavelength dispersive spectrometry (EDS and WDS) attachments with low atomic number element detection capabilities was used for phase identification and distribution analysis. Software was used for WDS analysis which provided a ZAF (atomic number, absorption, and fluorescence) correction to determine atomic percentages of elements in compounds. Sigma Scan Pro4 was used for the image analysis. 4
Jandel Scientific, San Rafael, CA, USA.
184
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
Fig. 3. A typical micrograph used to determine the volume fraction of SiCp in an ADF processed sample sintered at 1700°C for 100 h. The average vol.% was 1.4% SiCp.
3. Results and discussion
3.1. General microstructure Fig. 3 shows a typical optical micrograph showing the general view of the distribution of phases, including porosity. When the as-received PRBCSE powder was sintered, it showed the SiC formation around pores (see Fig. 4). This indicated that the carbon was not evenly distributed in the starting powder and that the powder needed to be ball-milled, which was done as described in Section 2.
Fig. 5. Monolithic MoSi2 (ARSE 0/100) sample sintered at 1700°C for 100 h on SiC disks. (a) SEM, and (b) oxygen map of the area in (a) showing high oxygen content.
3.2. Phase identification
Fig. 4. Optical micrograph of as-received, PRBCSE, sintered at 1700°C for 1 h. Notice that the SiC (gray phase) forms at pores.
3.2.1. SiO2 Oxygen maps obtained by energy dispersive X-ray analysis were used to determine if the carbon addition reduced the presence of SiO2 (Fig. 5(b) and Fig. 6). Fig. 5 and Fig. 6(b) show the EDS oxygen maps for the ARSE 0/100 (3 mm particle size MoSi2) and the ADF (3 mm particle size MoSi2 with 2.88 wt.% carbon) sintered on SiC disks at 1700°C for 100 h, respectively. The oxygen maps indicate that indeed the SiO2 content was reduced by the addition of carbon.
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
185
The oxygen maps were used in conjunction with image analysis software to obtain the vol.% of SiO2 in the sintered samples. Fig. 7 indicates that the ADF samples all showed very little SiO2 content compared to the monolithic MoSi2 (ARSE 0/100) indicating that the carbon addition reduced the SiO2 content. Fig. 7 also shows that the PRBCSE samples increased in SiO2 content compared to the ARSE 0/100. This is understandable inasmuch as the ball-milling reduced the particle size and increased the surface area and consequently, the amount of SiO2 in the starting powder. Fig. 7. Vol.% of SiO2 versus sintering time for the samples sintered on SiC disks at 1700°C. Obtained from oxygen EDS maps. Note that the ADF samples showed a lower SiO2 content than the ARSE 0/100. The PRBCSE samples showed a higher SiO2 content, which resulted from the ball-milling.
3.2.2. Porosity Image analysis software was used to obtain volume percentage of porosity. This number was then subtracted from the volume percentage of porosity plus SiO2 obtained using the optical micrographs giving the volume percentage of porosity (Fig. 8). The ADF samples showed higher porosity compared to the monolithic MoSi2 (ARSE 0/100) samples. We attribute this to the porosity in the ADF samples left after the pyrolysis of the starch before sintering. The porosity in the PRBCSE samples exhibited a similar trend as the ARSE 0/100 but was lower than the ARSE 0/100 indicating that the high SiO2 content reduced the porosity found in the sintered samples. 3.2.3. SiC, Mo 5 4.8Si3C 5 0.6 and MoSi2 Molybdenum disilicide was easily detectable by XRD but SiC and Mo 5 4.8Si3C 5 0.6 (the Nowotny phase) were not, because of their small amounts. Thus, a more localized analysis technique was required. The results of
Fig. 6. ADF sample sintered at 1700°C for 100 h on SiC disks. (a) SEM micrograph, and (b) EDS oxygen map of the area shown in (a). Note the low oxygen content.
Fig. 8. Vol.% porosity versus sintering time for the samples sintered on SiC disks at 1700°C. Note that the ADF sample showed more porosity because of voids left after the pyrolysis of the starch. The PRBCSE and ARSE 0/100 samples showed the same trend.
186
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
Fig. 9. (a) SEM micrograph of ADF powder sintered at 1700°C for 10 h. Arrow indicates area analyzed. (b) EDS spectra of the region analyzed. Mo, Si, and C peaks indicate that this region is the Nowotny phase.
electron microprobe analysis involving both light element detection EDS and WDS on an ADF sample sintered at 1700°C for 10 h are shown in Figs. 9 and 10. The light element detection EDS results showed that the brightest phase in the secondary electron micrographs (Fig. 9) contained Mo, Si, and C which indicated that this phase was the Nowotny phase. Similar results were obtained by Costa e Silva et al. [9], for powder compositions in the same region on the Nowotny ternary phase diagram (i.e. MoSi2, SiC, and Mo 5 4.8Si3C 5 0.6) for 1600°C [10]. The light element detection WDS results from this same compound indicated that it was indeed the Nowotny phase with Mo4.7Si3C0.56 stoichiometry, see Table 2. Fig. 8 shows that the amount of the Nowotny phase formed in this sample was less than 1 vol.%. The darkest region in Fig. 10 was thought to be SiC because it only showed Si and C using the light element detection EDS. The light element detection WDS analysis of the same region indicated that the stoichiometry was not perfect but was SiC0.86 (see Table 2). This may be caused by noise coming from the matrix, which would lead to a
Fig. 10. (a) SEM micrograph of ADF powder sintered at 1700°C for 10 h. Arrow indicates area analyzed. (b) EDS spectra of the region analyzed. The Si and C peaks indicate that this region is SiC0.86.
higher amount of Si detected than actually present in the compound. Optical micrographs were used to determine how much SiC formed in the ADF samples sintered at 1700°C for 100 h (Fig. 3). This sample which contained more SiCp than all of the other samples had only 1.4 vol.%. It is likely that silicon monoxide (SiO) formed along with CO and CO2 during the initial stages of sintering. This could account for the small amounts of SiCp and the Nowotny phase in the final product. It
Table 2 Compositional data obtained by WDSa At.% Mo At.% Si At.% C Theoretical composition
33.726 62.692 – MoSi2
– 53.263 45.670 SiC
56.935 36.304 6.761 Mo54.8Si3C50.6
Actual composition
MoSi1.9
SiC0.86
Mo4.7Si3C0.56
a Note the composition for SiC is slightly off stoichiometry possibly caused by signal coming from the matrix.
K.K. Chawla et al. / Materials Science and Engineering A261 (1999) 181–187
187
4. Conclusions 1. Silica was reduced significantly by the addition of carbon to the starting powders. 2. Silicon carbide and the Nowotny phase were observed. 3. The amount of SiCp formed was very small. The by-products of reactive sintering need to be considered for the formation of CO and CO2.
Acknowledgments
Fig. 11. SEM micrograph of a crack produced using a Vickers indentor. Notice that the SiCp deflects the crack. The arrow indicates the direction of the crack propagation.
would be desirable to analyze the effluent gases during sintering to verify this point.
3.3. Fracture toughness Fig. 11 shows an SEM micrograph of a crack formed by a Vickers indentation (1 kg, 15 s) and its interaction with a variety of particles in the MoSi2. The first unidentified particle stopped the crack and the crack reinitiated on the opposite side. After the crack was reinitiated, it encountered another unidentified particle and the crack deflected around this particle. The crack then penetrated the SiCp and was arrested there. A new crack was then formed in the SiCp that propagated into the MoSi2. Such crack deflection and reinitiation would enhance the toughness of the composite. However, one needs an appreciable amount of SiCp to have some reasonable increase in toughness.
.
This work was supported by Los Alamos National Laboratory under Contract No. 9-X63-0402J-1 Task No. 9 and the US Office of Naval Research under Contract No. N00014-96-1-0846. The starting powders were provided by ART Inc.
References [1] A.K. Vasudevan, J.J. Petrovic, (Eds.), High Temperature Structural Silicides: Proceedings of the First High Temperature Silicides Workshop, Elsevier Science Publishers, Amsterdam, (1992). [2] V. Bizzari, B. Linder, N. Lindskog, Am. Ceram. Soc. Bull. 68 (1989) 1834 – 1841. [3] S. Bose, Mater. Sci. Eng. A155 (1992) 217 – 225. [4] J. Cook, A. Khan, E. Lee, R. Mahapatra, Mater. Sci. Eng. A155 (1992) 183 – 198. [5] J. Cook, R. Mahapatra, E.W. Lee, A. Khan, J. Waldman, Ceram. Eng. Sci. Proc. 9 – 10 (1991) 1656 – 1670. [6] R.K. Wade, J.J. Petrovic, J. Am. Ceram. 75 (1992) 3160–3162. [7] A.K. Bhattacharya, J.J. Petrovic, J. Am. Ceram. Soc. 74 (1991) 2700 – 2703. [8] Advanced Refractory Technologies, Inc., Personal communication. [9] A. Costa e Silva, M.J. Kaufman, Metall. Mater. Trans. A 25 (1994) 5 – 15. [10] H. Nowotny, E. Parthe, R. Kieffer, F. Benesovsky, Monatshefte fur Chemie 85 (1954) 255.