Processing of iron aluminides by pressureless sintering through Fe+Al elemental route

(1)

Processing of iron aluminides by pressureless sintering through Fe + Al elemental route

S. Gedevanishvili *, S.C. Deevi

Research Center, Chrysalis Technologies Incorporated,4201Commerce Road, Richmond, VA23234, USA Received 22 January 2001; received in revised form 17 April 2001

Abstract

The sintering behavior of elemental powders of Fe and Al was investigated at various heating rates by dilatometric experiments in the temperature range of 1000 – 1350 °C. It was found that the formation of FeAl was accompanied by expansion due to volume change during the formation of the intermediate phase Fe₂Al₅. The mechanism of FeAl formation and the final density depended on the heating rate. The expansion rate during slow heating (50.5 °C min^{− 1}) was 0.0018 mm mm^{− 1}per min as against 0.34 mm mm^{− 1}per min at high (]1°C min^{− 1}) heating rates. Slower heating rates (50.5°C min^{− 1}) lead to lower expansion during Fe₂Al₅formation and subsequently to a higher density of FeAl up to 94.5% of theoretical. This is the highest density obtained so far (during the last 25 years) by heating elemental Fe and Al powders without pressure assistance to form FeAl or any other intermetallic in the Fe – Al system. The sequence of phases formed during the heating of Fe + Al mixture was identified by X-ray diffraction, optical microscopy and SEM. The results, along with the DSC data were correlated to the expansion and shrinkage behavior of the samples. © 2002 Elsevier Science B.V. All rights reserved.

Keywords:Processing; Iron aluminides; Pressureless sintering

1. Introduction

Intermetallic compounds have been the subject of scientific interest for more than 50 years because of their attractive physical and mechanical properties. In recent years research mostly focused on the use of monolithic intermetallic materials based on Ni₃Al, NiAl, Fe₃Al, FeAl, Ti₃Al and TiAl as replacements for denser structural materials such as steels or superalloys for high temperature service. Among the intermetallics, iron aluminides are attractive for elevated temperature applications due to their low density, low material cost and good high temperature mechanical properties. In addition they exhibit excellent corrosion resistance in oxidizing and sulfidizing atmospheres due to the formation of protective Al₂O₃ scale [1 – 4]. Of the iron aluminides, FeAl, has a B2 structure and exists over a wide range of Al concentrations at room temperature (36 – 50

at.%). Iron aluminides based on FeAl exhibit better oxidation resistance than Fe₃Al alloys and have lower densities compared to the steels and commercial iron based alloys, offering better strength-to weight ratio. In addition, FeAl exhibits higher electrical resistivity, in the range 130 – 170mV cm, comparable to many of the commercial metallic heating elements [5]. These properties allow them to be considered as high temperature structural materials, gas filters, heating elements, and as fasteners. Iron aluminides have been prepared by a variety of methods including melting [6,7], roll com- paction [8 – 10] and mechanical alloying [10 – 12].

The method of FeAl preparation used in the present work is sintering. Sintering may be the only reasonable, cost effective, choice sometimes to form precision, high- performance products operating in demanding applications such as automotive engines, aerospace hardware, manufacturing tools and electronic components. Sinter- ing enables net shape processing, uses limited material and eliminates deformation processing and machining of the components. It also allows better control of the microstructure of the product. Sintering of the powder

* Corresponding author. Fax: + 1-804-2744778.

E-mail address: shalva.gedevanishvili@pmusa.com (S. Gedevan- ishvili).

PII: S 0 9 2 1 - 5 0 9 3 ( 0 1 ) 0 1 4 4 2 - 3

(2)

S. Gede6anish6ili, S.C. Dee6i/Materials Science and Engineering A325 (2002) 163 – 176 164

compacts of a particular shape requires exposure to high temperatures (usually at or above approximately one-half of the absolute melting temperature) to eliminate the porosity and bond the particles. The bonds between the particles grow by one or more of a num- ber of mechanisms that occur at the atomic level. The common mechanisms for metal bonding are solid- state diffusion and liquid state sintering. The classical sintering process includes several stages: contact formation, neck growth, pore rounding, pore closure, and final densification of the product [13]. The sintering process allows densification of powders into net shape products, and attainment of near 100% density is very important for the utilization of high strength precision parts.

As a starting mixture for dense intermetallic pro- duction, many processes use prealloyed powders, which can be obtained by atomization or mechanical alloying. Further densification (sintering) involves the use of complex and costly processes based on hot isostatic pressing or hot extrusion. It is, therefore, important to develop economical processing methods to utilize the important attributes of intermetallics in developing critical aerospace or other high performance products. One of the interesting approaches is based on the use of elemental powders. Several authors investigated the sintering process in the Fe – Al system. Sheasby [14] attempted to sinter an alloy con- taining 6 at.% Al. He observed rapid expansion of the sample up to 15% at a temperature close to the melting point of aluminum due to the exothermic formation of Fe₂Al₅ compound, and outward diffusion of aluminum. Further sintering of the sample led to some consolidation but not a restoration of the origi- nal geometry. Lee and German [15] conducted sintering experiments with Fe + Al mixtures up to 26.7

at.% aluminum. They also observed rapid swelling at 655°C. Swelling was observed to increase with increase in the aluminum content. They noted that swelling was caused by outward diffusion of aluminum into the iron matrix. Rabin and Wright [16]

studied compacts with 28 at.% Al and 50 at.% Al.

They synthesized Fe₃Al and FeAl compounds and found swelling to be a major obstacle in the high density synthesis of materials. The highest sintered density achieved by these authors during pressureless sintering was 75% theoretical density (TD) of Fe₃Al. However, Zhuang et al. [17] were able to sinter Fe₃Al to 81% TD. A summary of the experimental details and final densities obtained in the Fe – Al powders by various authors is provided in Table 1. It can be seen that the researchers found it to be very difficult to achieve densities in the range 90 – 99% TD.

Due to the exothermic nature of the Fe + Al powder mixture, heating of the powders is always accompanied by heat generation. The technique which involves initiation of the exothermic reaction between the mixed powders to form intermetallics and other inorganic materials is called self-propagating high- temperature synthesis (SHS) or combustion synthesis.

This subject has been extensively reviewed by Munir and Anselmi-Tamburini [18]. Several authors synthesized iron aluminides by combustion synthesis [16,19 – 24] and found that the main disadvantage of this process was the large porosity in the final products.

To reduce or eliminate the porosity some workers suggest mixing of elemental Fe and Al with prealloyed powders (for example see Ref. [25]); others de- scribe the application of pressure during the combustion or sintering, which makes the process complex [16,20,26 – 29].

Table 1

Details of previous work on pressureless synthesis of FeAl from elemental powder

Composition at.% Al

Authors Heating rate Maximum temperature (°C), Maximum % TD

holding time (min) (°C min⁻¹)

10 –

Itin et al. (1973) [19] – –

Sheasby (1979) [14] 6 \100 900+ –

Lee and German (1985) [15] 526.7 10–200 1300, 60 –

–

25 700

Ranganath et al. (1990) [22] 54

Rabin and Wright (1991) [16] 28–50 10–50 1000 75

25–50 30

Bose et al. (1991) [20] 1200, 20 59.2

28 5–50

Zhuang et al. (1994) [17] 950, 30 81.1

Joslin et al. (1995) [23] 25–54 – 750 –

– 1250, 60 77.2

Yi et al. (1995) [21] 25 Fe+(FeAlx) alloy

5–10 1200, 60 –

Cai and German (1996) [25] 25 Fe₃Al+20%(Fe+Al)

(3)

Fig. 1. SEM photomicrographs of the elemental powders of Fe (a) and Al (b).

a composition of 40 at.% Al (24 wt.% Al) was pressed to obtain rectangular bars with the dimensions 31.75 × 6.35 × 3 mm. No binder was used and no lubricant was added for pressing and shaping the bars. Green densities of the rectangular bars used for subsequent sintering were in the range 77 – 87% of TD of Fe – 40 at.% Al (TD of Fe – Al40 at.% = 6.06 g cm^{− 1}).

The sintering experiments were run in an argon at- mosphere to a maximum temperature of 1350°C. Ex- pansion and shrinkage of the sample during sintering were monitored by employing dilatometry. Phase iden- tification in the green samples and products of sintering was done by X-ray powder diffraction technique. Mi- crostructural evaluation was carried out by optical and scanning electron microscopy. A simultaneous differential scanning calorimeter (DSC) and thermogravimetric analyzer (TGA) was used to determine the nature of the reactions (endothermic or exothermic) and the temperature of onset of the various reactions at different heating rates. The sintered samples were chemically etched by Keller’s reagent for optical microscopic observations.

3. Results and discussion

Fig. 2a shows DSC data of a Fe – 40Al pellet heated at a rate of 5 °C min^{− 1} in the temperature range 350 – 850 °C. Two exhotherms can be seen with the onset of exothermic reactions at 560 and 655 °C. Fig.

2b shows the thermal expansion curve in the same temperature range which illustrates the dramatic expan-

Fig. 2. DSC (a) and dilatometry (b) data obtained by heating a Fe – 40Al mixture at a rate of 5 °C min^{− 1}.

In the present work, pressureless sintering behavior of Fe + Al powders was studied using dilatometry for monitoring the sintering stages of the process, i.e. temperatures at which expansion, shrinkage and phase transformation take place. Pressureless sintering is based on the thermal bonding of the particles into a solid structure without the assistance of the pressure and is widely used in the automotive industry. Dilato- metric data were correlated with the differential scanning calorimetric data, X-ray analysis and microscopic observations.

2. Experimental

The materials used in the present study were an- nealed 99.8% purity Fe powder of − 325 mesh, from Hoeganaes, and 99.5% purity Al powder of − 325 mesh, from Goodfellow. Scanning electron microscope (SEM) images of the elemental powders are given in Fig. 1. A mixture of the two powders corresponding to

(4)

Fig. 3. Selected portion of thermal expansion and sample holder temperature profiles, heating rate 1 °C min^{− 1}.

sion of the sample at 560 °C. This expansion behavior is consistent with the data of Sheasby [13] and matches the temperature of the first exothermic peak in Fig. 2a. No dimensional changes are observed in the thermal expansion curve of Fig. 2b corresponding to the second exothermic peak temperature which appears at approximately the melting point of aluminum. Simi- lar expansion profiles (Fig. 2b) were obtained during heating of the Fe – 40Al pellets at a heating rate of 1, 2 or 10 °C min^{− 1}.

The exothermic nature of the reaction can also be inferred from the dilatometric data in Fig. 3. which shows the temperature – expansion profile in an experiment performed with a heating rate of 1 °C min^{− 1}. In order to measure the temperature close to the sample, a thermocouple was placed inside the specimen holder just under the sample without actually touching it. The data indicate that at the maximum expansion, the temperature of the sample was higher than the furnace temperature. The dramatic increase in the sample expansion can be attributed to the exothermic reaction occurring between Fe and Al powders. The rapid decrease at the tip of the profile is probably associated with the shrinkage during fast cooling of the sample back to the furnace temperature after the completion of the exothermic reaction. It is also important to note that the expansion rate was very high, 0.34 mm mm^{− 1} per min.

X-ray analysis of the products after the first exothermic reaction reveals that it is primarily due to the formation of the aluminum rich Fe₂Al₅. Along with the intermetallic Fe₂Al₅, the product contains unreacted Fe and minor amount of FeAl. The second exotherm is primarily due to the formation of FeAl by the reaction

of Fe₂Al₅with Fe. However the amount of FeAl is very small even after the second reaction; the formation of a narrow ring of FeAl is observed at the interface of Fe₂Al₅ and Fe, which will be referred to later.

Fig. 4 shows the temperature – expansion/shrinkage profile in a sintering experiment carried out at 5 °C min^{− 1}with a sintering time of 150 min at 1300 °C. It may be noted that a significant shrinkage was observed after reaching a temperature of 1200 °C and a continu- ous shrinkage thereafter up to a temperature of 1300

°C. The ‘sintering limit’ was reached after the first 40 min of sintering at 1300 °C. Additional holding of the sample at 1300 °C for another 1.8 h did not lead to any further shrinkage as evidenced by the horizontal profile in Fig. 4. Cooling of the sample from 1300 °C led to thermal shrinkage. The highest densities obtained during the experiments at heating rates from 1 to 10 °C min^{− 1} were in the range 86.5 – 89.8% TD and the maximum linear expansion observed at 560 °C was

18%.

Fig. 5 describes the thermal expansion behavior of the pellet sintered at a heating rate of 0.5°C min^{− 1}. Interestingly, the behavior is completely different from that depicted in Fig. 4 which is general of the heating rates in the range 1 – 10 °C min^{− 1}. The expansion is nearly linear from room temperature to 540 °C (region A). This can be attributed to the natural expansion of the Fe – 40Al mixture, which matches with the expansion profile of cast Fe – 40Al included here for comparison. The sample dramatically expands from 540 °C onwards as shown by the near vertical line in region B and the linear expansion reaches 3%. On further heating, the expansion rate slows down, the linear expansion value reaching about 5%, followed by another

(5)

dramatic expansion to about 6%. This region B of intensive expansion covers the approximate temperature range 540 – 655 °C. At about 655 °C which is very close to the melting temperature of aluminum as well as the eutectic temperature, the expansion rate significantly slows down (region C). In region D, that starts at about 900 °C, natural expansion of the synthesized material occurs. At about 1200 °C the linear expansion curve starts to drop sharply which is indicative of the beginning of the intensive sintering process (region E).

In order to investigate the mechanism of simultaneous reaction synthesis, experiments were performed

under the same conditions as the experiment shown in Fig. 5 except that the experiments were interrupted at 500, 600, 700, and 1000 °C. The products obtained after each interruption were characterized by X-ray diffraction, optical microscopy and SEM. The X-ray diffractograms of Fig. 6 permit an evaluation of the synthesis/sintering steps. The sample after heating to 500 °C contained elemental Fe and Al, just as in the case of the initial green sample (Fig. 6a). The sample obtained after interruption at 600 °C indicates the formation of the Fe₂Al₅phase (Fig. 6b), while interruption at 700 °C shows the onset of FeAl (Fig. 6c) and no

Fig. 4. Thermal expansion and set temperature profile for a Fe – 40Al sample heated at a rate of 5 °C min^{− 1}.

Fig. 5. Thermal expansion profile of a Fe – 40Al sample heated at a rate of 0.5 °C min^{− 1}.

(6)

Fig. 6. X-ray diffractograms of the interrupted samples at various temperatures: green sample at (a) RT, (b) 600 °C, (c) 700 °C, (d) 1000 °C.

Heating rate 0.5 °C min^{− 1}.

evidence for any remaining free Al. It may also be noted that at 700 °C FeAl coexists with Fe₂Al₅and free iron. X-ray analysis of the sample obtained at 1000 °C shows a single phase FeAl without any detectable presence of Fe₂Al₅ phase. Furthermore, the samples quenched from 500, 600, and 700 °C were magnetic, as opposed to the sample quenched from 1000 °C, which may be attributed to the presence of free iron in the products formed in the former case and its near total absence in the latter.

To determine the extent of sintering and the residual porosity, the samples were also observed in optical microscopy in the polished and unetched condition.

The green pellets consisted of an Al and Fe powder mixture with a density of 5.3 g cm^{− 3}corresponding to 87% TD. The sample interrupted at 500 °C (Fig. 7a) shows clearly the presence of two phases: the dark Fe phase and the light Al phase. This confirms that no significant reaction occurs from room temperature to 500 °C except natural expansion (Fig. 5). By volume, the Fe – 40Al mixture should consist approximately equal amounts of the two elements (Fe – 52 vol.% and Al – 48 vol.%). The sample interrupted at 600 °C contains three phases (Fig. 7b) Fe, Al, and Fe₂Al₅. Using energy dispersive spectroscopy (EDS), the various phases were identified as follows: lightest, Al; light gray, Fe; and dark gray phase surrounded by iron, Fe₂Al₅. It can also be seen from Fig. 7b that the amount of aluminum is less than iron, which agrees with the X-ray diffraction and EDS data (formation of aluminum rich phase). According to the X-ray diffraction results (Fig.

6), three phases should be present in the sample quenched from 700 °C namely, Fe₂Al₅, FeAl, and Fe

while only two phases could be detected by optical microscopy, (the dark regions are pores) (Fig. 7c). High resolution SEM images (Fig. 8) shows three phases, which differ in contrast. EDS spectra obtained from these regions indicate that between Fe and aluminum rich Fe₂Al₅phase there is another layer with Fe concentration much higher than expected in Fe₂Al₅. This layer is believed to be consisting of FeAl or a mixture of FeAl with Fe or Fe₂Al₅. The peaks of Au and Pd present in the spectra of Fig. 8 are due to the Au – Pd coating given to the SEM specimen. It can be con- cluded that at this stage (700 °C) all the aluminum has reacted to form iron aluminide and left behind large pores. At 1000 °C the pellets consist of only FeAl as per the X-ray analysis, Fig. 6. This is consistent with the single phase material (with pores) observed in the microstructure of Fig. 7d. Further development of the microstructure is associated with an increase in the linear thermal expansion up to 8.5% due to natural expansion of the material (Fig. 5, region D) followed by the beginning of sintering at about 1120 °C (Fig. 5, region E). After exposure of the material to 1350 °C, the density of the compact reaches 94.5% TD. Fig. 9 shows unetched (a) and etched (b) microstructures of the sample heated at 0.5 °C min^{− 1}and sintered at 1350

°C for 3 h. The average grain size is 25mm and isolated pores with a near spherical shape can be seen.

Changes occurring during the synthesis/sintering of FeAl were also monitored by measuring the density at different stages of processing by the water displacement method. Results are presented in Fig. 10. Also indicated in Fig. 10 are the densities of the different phases found in the product. It is clear that in the initial stages of

(7)

processing the density of the sample decreased with temperature up to the point (1000 °C) where complete conversion to the desired FeAl phase occurred.

At this temperature, however, the density is only 4.46 g cm^{− 3} (73.6% TD of Fe – 40Al). With further

increase in the temperature, the density starts increasing, reaching a value of 5.73 g cm^{− 3} (94.55%

TD). These results are consistent with the data on expansion/shrinkage obtained by dilatometry (Fig. 5).

Fig. 7. Microstructures of the interrupted samples at various temperatures: (a) 500 °C, (b) 600 °C, (c) 700 °C, (d) 1000 °C. Heating rate 0.5 °C min^{− 1}.

Fig. 8. Microstructure and EDX spectra of a sample interrupted at 700 °C.

(8)

Fig. 9. Unetched (a) and etched (b) microstructures of Fe – 40Al sample sintered at 1350 °C for 3 h.

Fig. 10. Dependence of density of interrupted sample as a function of temperature.

As mentioned earlier, the expansion profiles of Fe – 40Al samples obtained at heating rates of 1, 2, 5 and 10

°C min^{− 1}are different from that obtained at a heating rate of 0.5 °C min^{− 1}. In addition, the expansion rate during the first stage reaction (540 – 590 °C) and the final sintered density depend on the rate of heating. Fig.

11 shows the variation of the sintered density and the linear expansion rate with the heating rate. It is evident that the sintered density decreases with increasing heating rate. The highest density of 94.5% TD ever achieved in this intermetallic alloy (see Table 1 for comparison to the previously reported values) was obtained at the heating rate of 0.5 °C min^{− 1}. The linear expansion rate shows an increase with the heating rate. The linear expansion rate of the sample at 540 °C in the case of 0.5 °C min^{− 1} is 0.0018 mm mm^{− 1} per min, which is about 180 times lower than the lowest expansion rate value obtained during heating rates ]1 °C min^{− 1}. It is

possible to identify a ‘threshold’ value for the heating rate somewhere between 0.5 and 1 °C min^{− 1}that yields the highest density coupled with lowest expansion rate.

To the authors’ knowledge, there have been no system- atic studies reported in the literature which identified the desirable heating rate for yielding such high density sintered products. The density obtained in this work (94.5% TD) is also the highest reported so far for pressureless sintered iron aluminides.

The effect of the density of the green compact or the starting density on the sintering process has also been studied in some detail. Several experiments have been carried out with a starting green density of 77%TD.

The synthesis/sintering patterns are very similar in both the low density (77% TD) and the high density (87%TD) green samples for all the heating rates (0.5, 1, 2, 5, 10 °C min^{− 1}). As in the case of higher green density sample (87%TD), the threshold cooling rate is

(9)

located between 0.5 and 1 °C min^{− 1} heating rates in the lower green density sample. The only difference is the slightly higher temperature at which expansion starts and the higher linear expansion during the exothermic reactions in the case of lower density compacts. Furthermore, sintering of the lower (77% TD) green density samples resulted in a lower sintered density.

It is of importance to understand the rate controlling mechanism operative during sintering. This is attempted in the present work via an estimation of the activation energy for the sintering process. Sintering experiments were carried out at different temperatures for a fixed interval of 1 h and the shrinkage rate was

recorded. Fig. 12 presents the variation of shrinkage rate with sintering temperature in the form of an Arrhe- nius plot. The resultant activation energy, Q, for the sintering process is of the order of 320 kJ mol^{− 1}. There are no studies in the literature on the activation energy data for sintering of iron aluminides.

However, sintering is regarded as a diffusion con- trolled phenomenon. The experimental Q is higher than the activation energy of 265 kJ mol^{− 1}for the diffusion of iron in FeAl [30]. However, the Q value is closer to that for diffusion of aluminum in FeAl [31]. The esti- mated Q value is also in agreement with the deforma- tion activation energies reported in the literature [32 – 34] and the high temperature deformation is again

Fig. 11. Influence of the heating rate on the sintered density and linear expansion rate during the first exothermic reaction.

Fig. 12. Plot for estimation of activation energy for sintering of Fe – 40Al.

(10)

Fig. 13. Schematic diagram of FeAl formation during heating of Fe + Al powder mixtures from room temperature to 1000 °C at slow (0.5 °C min^{− 1}) and fast (5 °C min^{− 1}) heating rates.

believed to be diffusion controlled. The present sintering activation energy can, therefore, be interpreted as corresponding to a diffusion mechanism, with the densification process being controlled probably by the diffusion of aluminum atoms.

It is evident from the experimental results presented above that the heating rate influences the reaction mechanism of the FeAl compound formation in the Fe – Al40 at.% system. Some authors [15,16] noted earlier that increasing the heating rate actually increased the final density of the product in metal – Al system.

This effect was attributed to the extent of liquid phase formation during the combustion reaction. According to these authors, a higher heating rate decreases the amount of pre-combustion phases formed due to slow solid state interdiffusion. This results in a larger amount of liquid during combustion reactions which was believed to lead to a decrease in the porosity through rearrangement of sintering, [35,36]. However, there is a controversy regarding the validity of this mechanism. Savitski et al. [37,38] showed that swelling increases with the amount of liquid forming additive

for swelling systems. It must be mentioned that previous research on reaction synthesis of iron aluminides (6 – 54 at% Al content) was conducted by using heating rates in the range 5 – 200 °C min^{− 1}. The processing of the compacts was always associated with swelling and further pressureless sintering did not lead to densification of the intermetallic for any concentration of aluminum. By contrast, the present work employed heating rates in the range 0.5 – 10 °C min^{− 1} and the results have shown that lower heating rates (51 °C min^{− 1}) are conducive to achieving the desirable attributes of lower expansion and lower porosity. The low heating rates are thus far more beneficial in causing final densification and in enabling near net shape forming of FeAl products.

We now turn our attention to the specific processes that are likely to occur in the sintering process and how they are affected by the cooling rate. Fig. 13 shows schematically the proposed reaction sequence at 0.5 and at 5 °C min^{− 1}heating rates. In the case of slow heating (0.5 °C min^{− 1}), first reaction starts at 540 °C with the formation of the aluminum rich Fe₂Al₅ phase.

(11)

Formation of this compound is also predicted from its heat of formation, which is − 34.3 kcal mol^{− 1} [39].

The heat generated by this reaction is not sufficient to induce melting of aluminum, as also inferred by the microstructure presented in Fig. 7b. Microscopic exam- ination and X-ray analysis showed evidence of a Fe₂Al₅ layer formation around the Fe particles by solid state reaction between Fe of Al. This is considered to be the most important reaction (contributing to swelling) in this process of synthesizing iron aluminides. Therefore, several authors paid special attention to this reaction.

Sheasby [14], observed Fe₂Al₅ compound formation after exothermic reaction at 640 °C during the heating of the initial mixture (Fe – 6 at.% Al) at a rate of ]100

°C min^{− 1}but he could not avoid swelling. Li et al. [40], with a view to avoid the swelling of the sample due to the Fe₂Al₅formation, used Fe₂Al₅ obtained by casting.

After atomizing the ingot into powder they mixed it with Fe and Cr and sintered. The authors claim to have achieved a density of greater than 95%, by using Fe₂Al₅ powders with Fe and Cr powders, although they did not report any values. The second reaction in the sintering phenomenon occurs near the melting point of aluminum, at 655 °C. During this process aluminum completely melts and reacts with iron and Fe₂Al₅. The result of the interfacial reaction between Fe and Fe₂Al₅ is the formation of the target phase FeAl. After the formation of a small amount of FeAl phase, the domi- nant process is solid state diffusion. It is believed that formation of the desired FeAl phase is completed close to 900 – 920 °C. Both exothermic reactions and the diffusion processes were accompanied by expansion of the sample up to 8.5% (linear). However, expansion of the compact continues up to 1120 °C due to the natural expansion of the FeAl compound, Fig. 5 (region D).

Above 1120 °C shrinkage of the sample takes place, Fig. 5 (region E).

The reaction sequence at the high heating rate of 5 °C min^{− 1} (in general at rates higher than 1 °C min^{− 1}) is however different from what was postulated above for the slow heating rate (0.5 °C min^{− 1}). First, exothermic interaction between Fe and Al occurs at 560 – 565 °C according to the reaction Fe + Al = Fe2Al₅+ Fe + FeAl(minor). The maximum combustion temperature

was measured as 670 °C, which is higher than the melting point of aluminum. It must be emphasized that aluminum was not present in the mixture after reaction.

The microstructure was similar to that obtained at 700

°C for 0.5 °C min^{− 1}heating, but with larger pores. The ignition temperature of the second exothermic reaction was dictated by the eutectic temperature of the Fe – Al system (655 °C). Microscopic observations revealed rings of FeAl around the iron particles on the Fe and Fe₂Al₅interface. It is important to note that no expansion of the sample was observed during the second exothermic reaction as can be seen from Fig. 2. Subse- quently, the process of FeAl formation is similar to the case of the slow heating rate (0.5 °C min^{− 1}). The formation of FeAl was complete around 900 °C. The sequence of steps involved in the formation of the final sintered product is thus strongly dependent upon heating rate and the experimental results on phase identifi- cation, microstructure observation, expansion measurement and density determination are all in agreement with the proposed reaction sequence (Fig.

13) as a function of heating rate.

As mentioned earlier, swelling of the compacts is a major obstacle in the Fe – Al system for producing dense intermetallic materials for commercial applications. As can be seen from Table 2, the density of Fe₂Al₅ compound is much lower than the density of Fe + Al(40 at.%) mixture and Fe – 40Al. This means that if Fe₂Al₅ is an intermediate product in FeAl synthesis, the compact is expected to undergo some volume change during the process. Fig. 14 shows the calculated volume percent expansion/shrinkage as a function of weight percent of aluminum utilized in the reaction, the calculation being based on the diffusion of aluminum to form only Fe₂Al₅. The plots are calculated (a) with the assumption that pores previously occupied by aluminum remain unfilled, and (b) the pores previously occupied by aluminum are completely filled with Fe and Fe₂Al₅. In the case of (b) if all (100%) aluminum is utilized to form Fe₂Al₅ and all pores left by aluminum are filled with the iron and/or Fe2Al₅, the volume of the sample should undergo 1.8% volume shrinkage. In the worst scenario (a), the volume expansion can reach up to 45%. The experimental results show that in case of

Table 2

Data on physical and thermodynamic properties

Density Melting point (°) Heat of formation,DH

Aluminum content at.% Crystal structure

(kcal mol⁻¹)

(wt.%) (g cm⁻³)

7.86 1538 – BCC

– Fe

2.7 660 – FCC

Al –

−12.0 BCC

6.06 40 (24)

FeAl 1370

71.4 (54.7)

Fe₂Al₅ 3.96 1171 −34.3 Orthorhombic

Fe+Al 40 (24) 5.39^a – – –

aCalculated from the individual densities.

(12)

Fig. 14. Calculated dimensional change of a sample as a function of aluminum consumed in the reaction to form Fe₂Al₅.

fast heating rates (]1 °C min^{− 1}) the situation is very close to scenario (a) where the linear expansion reaches 11 – 18%. There are probably some other reasons for expansion besides the simple volume change due to the pore refilling factor. For instance, it is possible that the melt penetrates along the interparticle junctions and causes particle separation. The experiments with low heating rates (50.5 °C min^{− 1}) match with neither curve (a) nor curve (b) in Fig. 14. The results indicate that there is only a partial filling of the pores because of the outward diffusion of aluminum. It is necessary to note that the initial porosity of the green compact was taken into consideration in performing the above calculations.

At high heating rates, the first exothermic reaction between Fe and Al leads to excess heat, and the temperature of the compact exceeds the melting point of aluminum (660 °C). The liquid phase quickly covers the free iron surface owing to capillary action. Conse- quently, large pores are left behind at the sites previously occupied by aluminum (Fig. 7c). Because of the volume difference between Fe₂Al₅and Fe, iron particles covered by freshly formed Fe₂Al₅push each other apart resulting in an overall volume expansion of the compact. It follows that in this case because of the reaction of aluminum the refilling of the pores previously occupied by aluminum is minimal. The second reaction between Fe and Fe₂Al₅ does not influence significantly the dimensions of the compact since there is only a minor amount of FeAl formed at the interface of Fe and Fe₂Al₅. Further slow solid state diffusion process continues without significant changes in dimension up to 1120 °C.

During the slow heating rate the first reaction starts at a temperature (540 °C) lower than in the case of high heating rate, and the reaction temperature does not

exceed the melting point of aluminum. This reaction leads to the formation of Fe₂Al₅ phase around iron particles, followed by prolonged solid state diffusion of Al into the iron. It is believed that, during this process Fe₂Al₅ phase already occupied some volume left by diffused aluminum. At 660 °C, melting of aluminum triggers the second exothermic reaction resulting in a complete disappearance of aluminum but with less dramatic expansion of the pellet followed by slow diffusion process towards complete FeAl formation.

The experimental results also indicate that one of the important parameters in pore-free synthesis of FeAl is the ratio between the expanded volume V_e(induced by particles pushing each other) and the shrinkage volume V_s(created by pore refilling). Based on the experimental observations and volume calculations it can be deduced that for the present experimental conditions in Fe – 40 at.% Al system:

V_e/Vs (slow heating)BVe/Vs (fast heating)

where, as referred to earlier, slow heating is 50.5 °C min^{− 1}and fast heating ]1 °C min^{− 1}. It is impossible to avoid the formation of an intermediate phase during the synthesis of FeAl. Therefore, the major task during processing is to keep the ratio V_e/V_s very low, which might be possible by changing the particle size of the initial constituents or by introducing prealloyed FeAl as an additive.

After the synthesis of a single phase FeAl material, the next step is the densification of the compact involv- ing diffusion, which is driven by a reduction in the surface area. The experimental results show that final densification of the samples depends on the expansion, which occurs prior to the complete formation of FeAl.

In the case of high heating rates the linear expansion of the compact is 15 – 18%, whereas the low heating rate

(13)

leads to 8.5% expansion. Samples with a linear expansion of 18% can be sintered only up to 87% TD, whereas samples with 8.5% expansion can be consoli- dated to 94.5% TD. In both cases densification starts at 1120 °C, and the process intensifies beyond 1200 °C.

Results show that temperatures in the range of 1200 – 1350 °C are needed in order to achieve 95% density under pressureless sintering. Experiments carried out at various sintering temperatures in this range confirmed that sintering limit can be achieved faster at higher (1350 °C) temperatures. The same extent of densification can be obtained at lower temperatures (1200 °C), but with longer times.

One of the big challenges in Fe – Al sintering is to reduce the size of pores left by the diffusion of aluminum. Microstructure of the FeAl sample in Fig. 9 shows isolated and almost spherical pores at the grain boundaries as well as in the interior of the grains. Few of them are interconnected. According to the theory of sintering [13], early in the sintering process pores are attached to the grain boundaries. At that time the rate of grain growth is small. With an increase in temperature the rate of grain boundary migration increases, and after isolation of the pores from each other, mov- ing grain boundaries break away from the pores leaving the pores trapped in the interior of the grain. Generally pores in the grain interior shrink at a much slower rate than the pores on the grain boundary, in view of the fact that lattice diffusion is slower than grain boundary diffusion. It is, therefore, important to avoid the formation of pores inside the grains, which can be achieved by carefully controlling the densification process. Cur- rently our experiments are directed towards solving this problem and obtaining near 100% dense FeAl compound from the elemental powders.

4. Summary and conclusion

This paper presents results on the application of a reaction based sintering process to enable fabrication of FeAl components. By employing dilatometry the sintering behavior has been investigated in the temperature range 1000 – 1350 °C. It is demonstrated that synthesis and sintering of the FeAl intermetallic compound are possible by heating the elemental Fe + Al mixture. The sintering behavior is observed to depend strongly on the heating rate; the synthesis/sintering occurs in differ- ent modes at low heating rates of 50.5 °C min^{− 1}and at high heating rates of ]1 °C min^{− 1}. In both cases exothermic reaction occurs between Fe and Al during the heating process to form Fe₂Al₅ and there upon to form FeAl due to the reaction between iron and Fe₂Al₅. The formation of Fe₂Al₅at high heating rates is accom- panied by a large thermal expansion of up to 18% as against 8.5% at a low heating rate. Samples sintered at

low heating rates are denser than the samples sintered at high heating rates. A maximum density of 94.5% TD was reached in the present work, which is perhaps higher than any reported value hitherto. In general, the green density does not change the mode of sintering although higher green density results in higher final sintered density.

Acknowledgements

The authors thank Professor D.H. Sastry for many valuable discussions and reviewing this manuscript. The authors also thank F. Logan for her assistance in the experimental set up.

References

[1] K. Natesan, Mater. Sci. Eng. A258 (1998) 126.

[2] N.S. Stoloff, Mater. Sci. Eng. A258 (1998) 1.

[3] I. Baker, E.P. George, in: S.C. Deevi, V.F. Sikka, P.J. Maziasz, R.W. Cahn (Eds.), Nickel and Iron Aluminides — Processing, Properties, and Applications, ASM International, Materials Park, OH, 1997, p. 145.

[4] C.T. Liu, Mater. Sci. Eng. A258 (1998) 84.

[5] S.C. Deevi, V.K. Sikka, Intermetallics 4 (1996) 357.

[6] S.C. Deevi, V.K. Sikka, C.T. Liu, Prog. Mater. Sci. 42 (1997) 177.

[7] A. Bahadur, B.R. Kumar, O.N. Mohanty, J. Mater. Sci. 30 (1995) 3690.

[8] M.R. Hajaligol, S.C. Deevi, V.K. Sikka, C.R. Scorey, Mater.

Sci. Eng. A258 (1998) 249.

[9] S.C. Deevi, M.R. Hajaligol, V.K. Sikka, J. McKernon, C.R.

Scorey, Proc. Mater. Res. Soc. Symp. 552 (1999) KK4.6.1.

[10] S.C. Deevi, Intermetallics 8 (2000) 679.

[11] D.A. Eelman, J.R. Dahn, G.R. MacKay, R.A. Dunlap, J. Alloys Compd. 266 (1998) 234.

[12] D. Oleszak, P.H. Shingu, Mater. Sci. Eng. A181/A182 (1994) 1217.

[13] R.M. German, Sintering Theory and Practice, Wiley, New York, 1996.

[14] J.S. Sheasby, Int. J. Powder Metall. Technol. 15 (1979) 4.

[15] D.J. Lee, R.M. German, Int. J. Powder Metall. Technol. 21 (1985) 1.

[16] B.H. Rabin, R.N. Wright, Metall. Trans. A22 (1991) 277.

[17] L.Z. Zhuang, L. Buekenhout, J. Duszczyk, Scr. Mater. 30 (1994) 909.

[18] Z.A. Munir, V. Anselmi-Tamburini, Mater. Sci. Rep. 3 (1989) 277.

[19] V.I. Itin, Y.S. Nayborodenko, V.P. Ushakov, Phys. Sinter. 5-2/2 (1973) 359.

[20] A. Bose, R.A. Page, W. Misiolek, R.M. German, Adv. Powder Metall. 6 (9 – 12) (1991) 131.

[21] D.-Q. Yi, C.-H. Li, J.-H. Wang, R. Warren, I. Olefjord, Mater.

Sci. Technol. 11 (1995) 650.

[22] S. Ranganath, T.L. Prakash, J. Subrahmanyam, J. Mater. Lett.

10 (4 – 5) (1990) 215.

[23] D.L. Joslin, D.S. Easton, C.T. Liu, S.S. Babu, S.A. David, Intermetallics 3 (1995) 467.

[24] S. Gedevanishvili, D. Agrawal, R. Roy, J. Mater. Sci. lett. 18 (1999) 665.

(14)

[25] L. Cai, R.M. German, Adv. Powder Metall. Part. Mater. 2 (1996) 139.

[26] Y.-L. Jeng, R.W. Hayes, J. Wolfenstine, E.J. Lavernia, Int. J.

Powder Metall. 31 (1995) 155.

[27] S.C. Deevi, V.K. Sikka, Proc. Mater. Res. Soc. Symp. 364 (1995) 917.

[28] S.C. Deevi, V.K. Sikka, in: S.C. Deevi, V.K. Sikka, P.J. Massi- asz, R.W. Cahn (Eds.), Proceedings of the International Sympo- sium on Nickel and Iron Aluminides — Processing, Properties and Applications, ASM International, Materials Park, OH, 1997, p. 283.

[29] S.C. Deevi, V.K. Sikka, In: Proc. Int. Conf. On Heat-Resistant Materials, ASM International, Materials Park, OH, 1995, p. 275.

[30] M. Eggersmann, B. Sepiol, G. Vogl, H. Mehrer, Defect Diffus.

Forum 339 (1997) 143.

[31] E. Kentzinger, M.C. Cadeville, V. Pierron-Bohnes, W. Petry, B.

Hennion, J. Phys. Condens. Matter 8 (1996) 5535.

[32] D.H. Sastry, Y.V.R.K. Prasad, S.C. Deevi., Mater. Sci. Eng.

A299 (2001) 157.

[33] U. Reimann, G. Sauthoff, Intermetallics 7 (1999) 437.

[34] D. Lin, in: S.C. Deevi, V.K. Sikka, P.J. Massiasz, R.W. Cahn (Eds.), Proceedings of the International Symposium on Nickel and Iron Aluminides — Processing, Properties and Applica- tions, ASM International, Materials Park, OH, 1997, p. 187.

[35] Z.A. Munir, Ceram. Bull. 67 (1988) 342.

[36] K.A. Philpot, Z.A. Munir, J.B. Holt, J. Mater. Sci. 22 (1987) 159.

[37] A.P. Savitskii, L.S. Martsunova, M.A. Emelyanova, Sov. Pow- der Metall. Met. Ceram. 20 (1981) 4.

[38] A.P. Savitskii, N.N. Burtsev, L.S. Martsunova, Sov. Powder Metall. Met. Ceram. 21 (1982) 760.

[39] S.C. Deevi, V.K. Sikka, C.J. Swindeman, R.D. Seals, J. Mater.

Sci. 32 (1997) 3315.

[40] J. Li, A. Crysanthou, P. Tsakiropoulos, in: T. Haulik, V. Bailey, R. Burton (Eds.), Proc. 5th Jpn. Int. SAMPE Symp., Tokyo, Society for the Advancement of Material and Process Engineer- ing, October 1997, SAMPE, Covina, CA, USA, p. 265.