Combustion synthesis of microstructurally designed green powder compacts

Materials Science and Engineering A 478 (2008) 208–213

Combustion synthesis of microstructurally designed green powder compacts K. Morsi ∗ , N. Wang Department of Mechanical Engineering, San Diego State University, 5500 Campanile Drive, San Diego, CA 92182, USA Received 19 February 2007; received in revised form 27 May 2007; accepted 16 July 2007

Abstract This paper discusses a new elemental powder compact design that promotes considerably improved densities of combustion synthesized nickel aluminide–titanium carbide composites (at 20 vol.% titanium carbide loading) without the application of external pressure. The effect of processing temperature (700 ◦ C, 900 ◦ C and 1164 ◦ C) on the product porosity, microstructure homogeneity and hardness is discussed. Results are also compared to specimens produced using the conventional green compact design and processed at the same investigated temperatures. Both porosity and inhomogeneity are reduced using this new design at a processing temperature of 700 ◦ C, and were also both found to decrease with increase in processing temperature. Moreover, Vickers hardness measurements on the specimens revealed that the new microstructural design resulted in products with significantly superior hardness compared to materials produced through conventional green compact design. Also despite the low levels of porosity of materials produced through conventional green compact design when reacted at 1164 ◦ C, a vastly inferior hardness was still obtained owing to poor TiC distribution, in addition to poor bonding between intermetallic grains. © 2007 Elsevier B.V. All rights reserved. Keywords: Combustion synthesis; Intermetallic composites; Nickel aluminide; Titanium carbide; Ball milling

1. Introduction Intermetallics and intermetallic matrix composites have been formed in the past from powder compacts using low-input energy processes such as combustion synthesis [1,2]. In the combustion synthesis of nickel aluminides under the thermal explosion mode of ignition, the powder compact is uniformly heated to a temperature above the ignition temperature (∼640 ◦ C). Following ignition, the compact reacts throughout its volume converting the nickel (Ni) and aluminum (Al) powder into the desired nickel aluminide intermetallic product depending on initial composition. The reaction is exothermic in nature so the compact is self-heated to very high temperatures that are maintained for short periods of time due to heat losses. In general, materials and composites produced through combustion synthesis contain a significant level of unwanted porosity. One way to overcome this problem is to consolidate the material through the application of pressure using processes like hot pressing, hot isostatic pressing and even pseudo-hot isostatic pressing [1,3–6]. More recently,

∗

Corresponding author. Tel.: +1 619 594 2903; fax: +1 619 594 3599. E-mail address: [email protected] (K. Morsi).

0921-5093/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2007.07.061

one of the authors successfully applied bulk deformation processes (extrusion and forging) during the combustion synthesis process while the specimen is hot to simultaneously shape and consolidated reacted materials [7,8]. All these pressure-assisted processes are geared to resolve the porosity problem encountered in combustion synthesis. It would however be greatly advantageous if we could achieve this without the need for externally applied pressure, especially for intermetallic composites where the porosity problem can be even more pronounced. This could be possible if we consider the basic mechanism of the combustion synthesis process as applied to aluminide intermetallics. One of the pre-requisites for successful combustion synthesis of aluminide intermetallics is for the low-melting point phase (molten aluminum) to efficiently spread throughout the compact, and effectively surround the nickel particles [1]. The problem then arises when ceramic particles are added to the system. The addition of these reinforcements (e.g. titanium carbide (TiC) to the 3Ni + Al powder mixture) has a couple of shortcomings [9]. First, TiC will act as a heat sink thus absorbing part of the heat of reaction and thus the maximum attainable temperature achieved during the reaction, also termed the combustion temperature, will be reduced. Second, these ceramic particles will obstruct the spreading of molten Al. These ultimately lead to

K. Morsi, N. Wang / Materials Science and Engineering A 478 (2008) 208–213

an increase in both product porosity and in-homogeneity with an increase in volume fraction of TiC reinforcement. Recently, Yeh et al. [10] also reported an increase in product porosity with increase in TiC content for combustion synthesized NiAl–TiC composites. We here present a new microstructural design of the green compact that basically removes the ceramic reinforcement from the pathway of molten Al to allow more efficient spreading, resulting in high-product densities without the need for an externally applied pressure. This is achieved by ball milling the TiC with Ni (the high-melting point phase) to generate Ni–TiC composite particles, then mixing them with the Al powder (the low-melting point phase). This generates a powder mix that is then compacted and reacted. A recent study by the authors [11] focused largely on the ball milling, powder compactability aspects of processing and had only considered one processing temperature (700 ◦ C). Despite the formation of multiphase microstructures, results have still shown that the density and hardness of materials processed using this new design are vastly superior to those of materials of the same composition but processed through a conventional powder mixing approach. It is important to establish to what extent the low density (which can be improved by processing at higher temperatures) is responsible for the inferior properties of these conventionally processed materials, and whether homogeneity can be improved by processing at higher processing temperatures. This paper discusses the effect of this new microstructural design and the influence of processing temperature on the product porosity, microstructure, homogeneity and Vickers hardness of the final product. Results are also compared to reacted specimens from conventionally formed powder compacts of the same initial composition. 2. Experimental procedures Commercially pure Ni (5–10 ␮m), TiC (2 ␮m) and Al (8–11 ␮m) powders (Atlantic Equipment Engineers, NJ, USA), were used as the starting materials (all particle sizes are manufacturer quoted values). The composition of the powder mixture was chosen to give a final product of Ni3 Al with 20 vol.% TiC assuming complete conversion. The Ni and TiC powders were first gently mixed for 1.5 h using a powder rotator mixer then ball-milled for 10 h (interrupted runs) in a SPEX 8000 mixer under an argon atmosphere. WC–Co balls were used as the milling media. The ball-topowder weight ratio used was 4:1. This process was conducted to embed/disperse the TiC into the nickel powders. The milled powders were then gently rotator mixed for 1.5 h with the remaining Al powder making up the balance. The elemental powders at this stage are referred to as method A. Prior to ball milling, all powders were vacuum degassed at 120 ◦ C for 10 h. This is carried out in order not to trap/embed any moisture or low-boiling point impurities with the TiC or Ni in the new composite particles during ball milling, which could subsequently evaporate and expand during the reaction and cause unwanted porosity. Green specimens were produced by pressing the powders in a rigid die to discs of diameters ∼8 mm, and height ∼3 mm (maintaining a relative density of 70–75%). The specimens were then combustion

209

synthesized in a tube furnace under argon atmosphere by heating at 10 ◦ C/min to temperatures of 700 ◦ C, 900 ◦ C and 1164 ◦ C and dwelling for 15 min followed by cooling. For comparison Ni, Al and TiC were rotator mixed (i.e. using the conventional processing method) for 1.5 h (referred to in the paper as method B) compacted and combustion synthesized under exactly the same conditions (in this case all powders were also vacuum degassed prior to compaction). The green compacts were also pressed to similar green densities to those of method A, this way green density would not be a processing variable in our investigation. Microstructural examination was carried out after the specimens were cut, ground and polished to 1 ␮m finish. Scanning electron microscopy and X-ray diffraction (XRD) were used for microstructural examination and phase analysis, respectively. Product porosity was measured using image analysis. Hardness testing was conducted using a Vickers macrohardness tester using a load of 10 kg. Eq. (1) was used to calculate the Vickers hardness number from the load and diagonal length of the indents produced. The reported hardness results for each specimen are averages of five tests per specimen:   F HV = 1.854 (1) d2 where d = (d1 + d2 )/2, F the load in kgf and d is the average diagonal length in mm. 3. Results and discussion During the ball milling of Ni and TiC powders, Ni (the ductile phase) undergoes deformation, cold working, fracturing and re-welding. TiC particles have also been seen to fracture during the process. These TiC particles become entrapped in between the fractured and re-welded Ni particles during ball milling. A continuation of this process resulted in a homogeneous distribution and dispersion of fractured and refined TiC in Ni particles. These now composite particles can have an increased level of defect concentration which can facilitate high-temperature diffusion, and possibly improved homogeneity following the reaction with Al. As mentioned, for method A, the ball-milled Ni and TiC composite powders were further rotator mixed with Al powder followed by compaction then combustion synthesis. Figs. 1 and 2 show the polished cross-sections of the 3Ni + Al + TiC green compact microstructure of methods A and B. It is clear that TiC is largely well dispersed in the nickel powders for the new design (method A), and would therefore be subsequently removed from the path of the molten Al during the reaction. The TiC also appears to have been fragmented into smaller submicron particles. On the other hand, in the conventional powder mixed green compact (method B); a large fraction of TiC particles are seen to be agglomerated into distinct regions and not well distributed throughout the microstructure prior to reaction. XRD of the precursor powders used for both methods prior to their compaction and combustion synthesis was conducted. The results indicate that all peaks belong to Al, Ni and TiC with no new phases being detected for either methods. Moreover, no significant peak broadening in method A compared to method

210

K. Morsi, N. Wang / Materials Science and Engineering A 478 (2008) 208–213

Fig. 1. SEM of cross-section of compacted specimen (method A) prior to combustion synthesis. Fig. 3. Method A and B specimens after combustion synthesis at 700 ◦ C.

B, for the nickel peaks was observed. Hence nanostructured Ni is not likely to have formed after ball milling. Fig. 3 shows the compacts following combustion synthesis, it is clear that method B had swelled much more than that of method A (the new design). This is primarily due to the increased product porosity and in-homogeneity of method B compared to method A. Fig. 4 is a scanning electron micrograph of the cross-sections of reacted compact of methods A and B for all processing temperatures. As mentioned earlier the melting and spreading of Al is an important part of the process. The Al spreads through capillary forces and surrounds the Ni particles, followed by Ni/Al reaction through diffusion. It is evident for specimens of method B (Fig. 4a–c) that the reaction did not go to completion at the 700 ◦ C processing temperature. Regions are seen with a nickel (un-reacted core, confirmed by energy dispersive spectroscopy (EDS)) with outer concentric rings of intermetallic phases. As the temperature is increased from 700 ◦ C to 1164 ◦ C, the un-reacted nickel core is seen to diminish at 900 ◦ C and

Fig. 2. SEM of cross-section of compacted specimen (method B) prior to combustion synthesis.

almost disappear at 1164 ◦ C. Hence the homogeneity of the microstructure is improved with processing temperature, owing to enhanced diffusion characteristics at the higher temperatures. The porosity is also seen to decrease with increase in processing temperature (less than 5% at 1164 ◦ C), but appear higher than those of specimens from method A reacted at the same temperature (Fig. 4d–f). In Fig. 4a–c, TiC particles can also be seen agglomerated and segregated in regions between the intermetallic regions (resembling the green microstructure). The presence of these groups of particles leads to excess porosity. TiC is a high-melting point material (with a melting temperature in excess of 3000 ◦ C) and the temperatures experienced during the reaction, in addition to the reaction’s short duration will not be enough to sinter these particles together. This will therefore leave inter-particle porosity between these TiC particles. Hence the porosity observed for method B after reaction includes smaller inter-TiC pores. The decrease in porosity with temperature is a result of sintering, and also TiC particle re-arrangement. This is in sharp contrast to reacted specimens of method A, for which the microstructure was almost near fully dense; this is seen in Fig. 4d–f. The microstructure here contains some TiC particles (as in method B) and dark grey regions containing refined TiC particles (milled down during the ball milling process to submicron sizes). Fig. 5a and b shows higher magnification micrographs of these regions for specimens of method A and B reacted at 1164 ◦ C. The same type of features of the refined TiC (for method A) and agglomerated TiC (method B) in the reacted specimens were also seen when the green compacts were sectioned and polished prior to the reaction (Figs. 1 and 2). It is clear that in method B the TiC particle size has not been reduced in size. For method A, although some TiC particles have not been reduced in size, a considerable amount of TiC has indeed been refined as mentioned. The porosity content of reacted method A specimens (<2%) is considerably lower than that of method B at all processing

K. Morsi, N. Wang / Materials Science and Engineering A 478 (2008) 208–213

211

Fig. 4. SEM images of method B (a–c) and method A (d–f) reacted at 700 ◦ C, 900 ◦ C and 1164 ◦ C.

temperatures, showing the effectiveness of the new green compact microstructural design. The TiC is however not distributed evenly in the microstructure of method A (although still much better than that for method B), resembling more a dual matrix, i.e. regions of reinforced material (where the Ni–TiC composite particles have been reacted with Al) surrounded by un-reinforced regions. EDS analysis revealed these surrounding regions to be in-homogeneous in nature but also predominantly Ni3 Al at 700 ◦ C and the in-homogeneity decreased with increase in processing temperature. The median particle size of the ball-milled Ni + TiC composite powders was determined to be ∼23 ␮m which is larger than the Al particle size. This favors a more inter-

connected Al powder network within the green compact which would also help in the densification and reactivity. The lowporosity levels obtained at this TiC content for reacted specimens of method A is unprecedented to the best of the authors’ knowledge for combustion synthesized nickel aluminide composites (without the application of an external pressure). XRD scans on both specimen methods reacted at 700 ◦ C, 900 ◦ C and 1164 ◦ C are shown in Fig. 6. The XRD scans show that TiC is still present following the reaction for both methods at all temperatures. Moreover, at a processing temperature of 700 ◦ C the microstructure is multiphased for both methods signifying the reaction did not go to

212

K. Morsi, N. Wang / Materials Science and Engineering A 478 (2008) 208–213

Fig. 6. X-ray diffraction scans of reacted compacts for methods A and B, respectively.

Fig. 5. SEM image of un-etched specimens of (a) method A specimen processed at 1164 ◦ C (different size TiC are seen), and (b) method B specimen processed at 1164 ◦ C.

completion (i.e. did not form 80 vol.% Ni3 Al and 20 vol.% TiC). Instead, a number of phases have been formed including Al3 Ni, Al3 Ni2 , Ni3 Al, NiAl in addition to un-reacted Ni and Al for method B and solid solutions of Ni/Al for method A. What is evident is that a larger amount of Ni3 Al is present in method A (the new microstructure design), i.e. the reaction was more complete, compared to that for method B. It can also be seen that as the processing temperature is increased the microstructures becomes less in-homogeneous, where the Ni3 Al yield is increased significantly for both method. The microstructure for the specimens reacted at 1164 ◦ C, although predominantly Ni3 Al is still not fully converted. High-temperature heat treatments or optimizing the ball milling parameters to control the Ni–TiC composite particle defect structure may be needed to fully homogenize these microstructures (the subject of future work). The ball milling process can introduce defects (generally termed mechanical activation) within nickel which could facilitate faster diffusion rates at the processing temperature, and therefore yielding more

homogeneous microstructures for method A, especially at the lower processing temperatures compared to method B. As seen from Fig. 7, the Vickers hardness number for the reacted specimens with the new microstructural design is superior to the conventional design for all processing temperatures investigated. Although phase content may play a role, we believe that the drastic difference between the hardness values of these two designs is not only due to the difference in porosity levels between the two, but also largely due to the ineffectiveness of TiC at reinforcing the material in the conventional design (method

Fig. 7. Vickers hardness number for specimens obtained using methods A and B reacted at 700 ◦ C, 900 ◦ C and 1164 ◦ C.

K. Morsi, N. Wang / Materials Science and Engineering A 478 (2008) 208–213

B), where they are seen to form agglomerates with inter-particle porosity. Moreover the TiC in the new design has been refined and better distributed through the ball milling process making them more effective at reinforcing and strengthening the material. For reacted specimens of method B, the slight increase in hardness at 1164 ◦ C is largely due to the reduced porosity content at this temperature. However, despite the low level of porosity (<5%) at the highest processing temperature for method B, the hardness values are still very low. We believe that in addition to the ineffectiveness of the agglomerated and segregated TiC in the conventional design at reinforcing the material, the presence of the TiC particles at certain locations in specimens of method B prevented proper bonding between the intermetallic grains. This is clearly seen in Fig. 5b.

213

5. Despite the low levels of porosity found in the 1164 ◦ C combustion synthesized specimens of method B, the hardness values were still vastly inferior to those of method A. This being a result of poor TiC distribution (and therefore their ineffectiveness at reinforcing the material), in addition to poor bonding between intermetallic regions within the specimens brought about by the presence of these TiC particles. The present work shows the potential of this new microstructural design for processing aluminide intermetallic composites over conventional means of powder processing. Future work will address the effect of heat treatments, and the influence of ball milling parameters on the composite powder defect structure and its resulting influence on product homogeneity.

4. Conclusions Acknowledgements The following conclusions can be drawn from the present study: 1. The new green compact microstructural design (method A) involving the embedment of TiC into the nickel particles and removing them to a large extent from the path of the molten aluminum during combustion synthesis (at all investigated processing temperatures) resulted in highly dense products, compared to reacted specimens from method B (conventionally produced). 2. The conventional green compact microstructural design involving the simple mixing of TiC, Al and Ni powders resulted in porous combustion synthesized products when reacted at 700 ◦ C. The porosity however decreased with increase in processing temperature to less than 5% at 1164 ◦ C. 3. Ni3 Al intermetallic yield was found to increase with processing temperature for both methods. 4. The Vickers hardness of the reacted material using the new green compact design is found to be vastly superior to that of method B at all investigated processing temperatures, due to the lower porosity and better distribution of refined TiC enabled by the new design.

The authors wish to thank the San Diego State University Research Foundation for their support. The authors also wish to thank Dr. Steve Barlow, Mr. Greg Morris and Mr. Mike Lester for their help with electron microscopy and the machining of the ball milling vial, and general technical assistance throughout the project. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

K. Morsi, Mater. Sci. Eng. A 299 (2001) 1–15. J.J. Moore, H.J. Feng, Prog. Mater. Sci. 39 (1995) 243–273. H. Doty, R. Abbaschian, Mater. Sci. Eng. A 195 (1995) 101–111. D. Alman, N.S. Stoloff, Int. J. Powder Metall. 27 (1) (1991) 29–41. D. Padmavardhani, A. Gomez, R. Abbaschian, Intermetallics 6 (4) (1998) 229–241. S.K. Bhaumik, C. Divakar, L. Rangaraj, A.K. Singh, Mater. Sci. Eng. A 257 (2) (1998) 341–348. K. Morsi, S.O. Moussa, J. Wall, J. Mater. Sci. 40 (2005) 1027–1030. K. Morsi, J. Rodriguez, J. Mater. Sci. 39 (2004) 4849–4854. K. Morsi, H. McShane, M. McLean, Metall. Mater. Trans. A 31 (2000) 1663–1670. C.L. Yeh, S.H. Su, H.Y. Chang, J. Alloys Compd. 398 (1/2) (2005) 85– 93. K. Morsi, N. Wang, J. Alloys Compd., in press.