A Metallographic Investigation Into the Effect of Sintering on an FC-0205 Premix. Thomas F. Murphy, George B. Fillari, Gerard J.

Transcription

1 A Metallographic Investigation Into the Effect of Sintering on an FC-0205 Premix Thomas F. Murphy, George B. Fillari, Gerard J. Golin Hoeganaes Corporation 1001 Taylors Lane Cinnaminson, NJ Abstract The properties of ferrous P/M materials are developed during the sintering process where metallurgical bonds are formed at particle-to-particle contacts and alloying of mixed and bonded additives occurs. Increasing either the sintering temperature or time can produce improvements in the microstructure and, consequently, the ensuing properties. In this paper, an FC-0205 premix, sintered for various times at 1120 C (2050 F), will be used to study the microstructural changes resulting from increases in sintering time. Features of interest include, changes to the surface-to-volume ratios at the particle boundary and pore surface areas, diffusion of the copper in the solid state and as a liquid component, and the homogenization of the microstructrue with increasing sintering times. Stereological techniques, using light optical microscopy, will be employed to examine the diffusion of the added alloying materials and to quantify the improvement in the degree-of-sinter. Additionally, electron microscopy (SEM) will be used to examine Charpy impact fracture surfaces from specimens sintered at the various times. Introduction Iron-copper-carbon materials, especially FC-0205 and FC-0208, have been the workhorse of the North American P/M industry for many years. They probably account for greater powder usage than any other single alloy system. There are many reasons for the success of these compositions. The liquid phase from melting of the copper enhances the sintering of the matrix iron and acceptable properties can be developed with less than optimal conditions. Increasing the sintering times can lead, however, to greater stability and consistency in parts production through the development of a more homogeneous microstructure. In addition, from a property standpoint, the carbon diffuses and forms pearlite and the alloyed copper strengthens the ferrite.

2 In the past, authors have addressed some aspects of the Fe-Cu and Fe-Cu-C systems. Pore formation [1] and dimensional change [2] have been discussed for the Fe-Cu materials, while the dimensional and microstructural characteristics [3] and the effect of the liquid phase have been investigated for the iron-copper materials both with and without carbon additions [4]. Further, the effect of sintering atmosphere on dimensional change [5] was determined for the FC-0208 material. In contrast to these offerings, this paper takes a metallographic approach to describe the changes in microstructure as sintering progresses. Experimental Procedure The material used in this study was a mixture of iron (Ancorsteel 1000) and 2 w/o ACuPowder 8081 copper, 0.5 w/o Asbury 3203H graphite, and 0.75 w/o Lonza Acrawax C lubricant. After sintering, this composition falls within the MPIF specification for an FC-0205 material. The test specimens utilized to record the changes in microstructure were unnotched Charpy impact bars that were compacted at three pressures, 415, 550, and 690 MPa (30, 40, and 50 tsi). Sintering was performed in a pusher furnace with the temperature set at 1120 C (2050 F) under a 90 v/o N 2 /10 v/o H 2 atmosphere. A pusher furnace was used in order to vary sintering time by pushing a tray loaded with the impact bars into the hot zone, starting a timer, and pushing the tray into the cooling zone after the appropriate amount of time had elapsed. Residence times in the hot zone ranged from five to a total of 50 minutes, in five-minute intervals. In addition, one set of bars was sintered for a total of one minute to show the effect of lubricant burn-off without the temperature reaching a point where alloying was initiated. In this furnace, the time required for room temperature samples to reach the sintering temperature varied between 10 and 15 minutes and was dependant on the load being sintered. Impact energy and apparent hardness values were measured for each group of bars prior to microstructural examination. One bar from each sintered set was selected for metallographic testing. Transverse cross-sections parallel to the pressing direction were removed and mounted with the cut surface exposed for examination. Each mount was ground through 600 grit (U.S. Standard) SiC abrasive paper and polished using a vibratory polisher with successively finer Al 2 O 3 particle sizes, beginning with 1 µm, and continuing through 0.05 µm. The prepared mounts were then outgassed in a vacuum chamber to remove the entrapped moisture, vacuum impregnated with warmed epoxy to fill the void spaces and, hopefully, retain any loosely sintered particles. The impregnated mounts were then reground and polished, again using the vibratory polisher with the various sizes of Al 2 O 3 polish. All samples were examined in both the unetched and etched conditions. Unetched specimens were used for general microstructural examination and stereological testing, while etched mounts were used to show the variation in alloying due to diffusion of the added alloying materials. Both specimen conditions were documented using light optical microscopy. The stereological test methods were designed to measure changes in the surface-to-volume ratio (S V ) of the particle boundaries as the sintering time was increased. However, pressed and sintered materials present problems in stereological sampling. The inherent directionality in the microstructure resulting from compaction prevents uniform sampling on a single plane using parallel, straight-line sample probes. To remedy this, vertical sections (cross-sections containing the pressing direction) were prepared and an array of test probes in the form of oriented cycloids was overlaid on the magnified, unetched specimen surfaces. The cycloid shape was shown, first by Baddeley, et. al. [6], and then by others [7,8,9], to sample oriented structures uniformly when coupled with prepared vertical sections. The four cycloid orientations were joined into a sine-like wave pattern to simplify counting. A counter was incremented as the

3 overlaid cycloid wave pattern intersected the individual particle boundaries. The relationship for determining S V is shown in Equation 1: S V = 2P L (1) where S V is the surface-to-volume ratio of the microstructural feature being measured [10,11] and P L is the number of incidents of the cycloid wave crossing the feature boundaries per unit cycloid line length. In this case, the microstructural features of interest were the particle boundaries. Fracture surfaces from samples with selected sintering times were examined in an effort to find trends in the failure characteristics as the sintering time was extended and the diffusion of the alloying elements became more homogeneous. Secondary electron imaging was used to document these features. Results As was described in the Introduction, the present study uses a metallographic approach to describe the response of the Fe-Cu-C material to various sintering times. James [12] has described the three factors governing the properties of P/M materials as the following: Density Porosity (volume fraction, size & shape distributions) Chemical Composition & Alloying Method (distribution of the added alloying materials) Resultant Microstructure All three factors are affected, either directly or indirectly, as the sintering time is varied with development of the final microstructure occurring upon cooling from the sintering temperature. The major effect on the microstructure is from the distribution of the alloying additives during the sintering process, with diffusion increasing with increased sintering time. In many cases, this causes the development of a nonuniform distribution of some of the additives. This leads to variations in local hardenability, and consequently a variety of transformation products are developed as the material is cooled. With the copper containing materials, the copper becomes liquid and diffuses along pore and particle boundaries. With short sintering times the distribution is heterogeneous. The distribution becomes more homogeneous, however, as the sintering time is increased. The process of sintering, from pre-heat to cooling, controls the constituents in the microstructure. Sintering can be separated into three basic areas where each has associated effects on the development of the microstructure. Each section is detailed briefly below: Pre-heat Lubricant is burned out Time at Sintering Temperature Carbon diffuses Copper becomes liquid and diffuses Sintering occurs Other added alloying elements diffuse and homogenize Cooling Transformation products are formed

4 Cross-sectional Analysis (Light Optical Microscopy) The change in microstructure can be seen in Figures 1 through 6 where the samples compacted at 550 MPa resided in the furnace hot zone for increasing periods of time. Figure 1 was taken of the sample sintered for one minute. The boundaries between the particles are evident as are the unmelted copper particles. A faint ferritic grain structure can be seen within the iron particles and no graphite has diffused to form pearlite. 2 v/o nital + 4 w/o picral Figure 1. Photomicrograph of a sample residing for one minute in the hot zone. No sintering has taken place. All particle boundaries are visible. Figure 2 shows the structure after five minutes in the hot zone. The start of carbon diffusion is apparent with formation of the lamellar pearlite; the copper particles, however, remain unmelted. The apparent lack of uniformity in the distribution of carbon is shown also. In Figure 3, there is evidence of a large copper particle that has melted without complete distribution of the liquid copper. The liquid metal has moved to the edge of the surrounding matrix material and the molten copper has been pulled into the surrounding pores by capillary action. The time above the melting point of copper (1083 C, 1981 F) was probably insufficient for the entire volume of the copper particle to distribute completely. It should also be noted, that the area surrounding the melted Cu particle is free of particle boundaries; although some angular features are still apparent along the pore edges. This image is from near the top surface of a bar compacted at 415 MPa and sintered for 5 total minutes. Only a small amount of the cross-section exhibited the initiation of the copper melting.

5 2 v/o nital + 4 w/o picral Figure 2. Five-minute sintered sample showing diffusion of the graphite leading to the formation of pearlite, but the unmelted copper particles are still visible. Unetched Figure 3. Melted Cu particle before the completion of copper distribution through the local porosity. The liquid copper has moved into the surrounding region through capillary action. The image is from the upper surface of a 415 MPa compacted bar sintered for 5 total minutes.

6 As the copper becomes liquid, the molten material travels along the particle boundaries and diffuses at both pore and grain boundaries. This can be seen in Figure 4, where a sample sintered for ten minutes exhibits partial diffusion of the copper and small areas of free copper along the prior particle and grain boundaries. The brown staining indicates regions with a higher copper content in solution with the iron. This will become less evident as sintering proceeds and the copper is more evenly distributed. 2 v/o nital + 4 w/o picral Figure 4. A ten-minute sintered sample showing diffusion of the copper along particle and grain boundaries. The brownish staining indicates regions where copper has a higher concentration than the grain interiors. One consequence of liquid phase sintering is the presence of the spaces vacated by the particles after melting. This is shown in Figure 5, in a sample sintered for a total of 30 minutes. The total sintering time equates to a time at temperature of approximately 15 minutes. The large pores are prior sites of the larger copper particles, and it is apparent that, at this sintering time, the carbon is more uniformly distributed.

7 2 v/o nital + 4 w/o picral Figure 5. Low magnification photomicrograph showing the distribution of the carbon and the pores produced from the melting of the large copper particles in a sample sintered for a total time of 30 minutes. 2 v/o nital + 4 w/o picral Figure 6. Also taken of the 30 minute sintered sample. The carbon and copper distributions appear more uniform. No obvious particle boundaries remain.

8 In Figure 6, the microstructure has become more uniform with no obvious free copper or particle boundaries. Increasing the sintering time will, however, improve the distribution of the alloying materials and give a more stable material. Stereological Testing The samples used in the previous discussion were part of the larger population used to evaluate the evolution of the microstructure through stereological testing; although for the latter, the samples were in the unetched condition. The evaluation procedure required IUR (isotropic, uniform, and random) sampling, which presented difficulties with pressed and sintered P/M materials. During compaction, the deformation of the particles, caused by the high compaction pressures, elongates the particles and pores in a plane 90 to the pressing direction. This directionality precludes the use of parallel, straight-line probes as the sampling mechanism. An alternative sampling method was found that used of an array of cycloid-shaped probes. The oriented cycloids have been shown to sample directional structures randomly when combined with sections containing the orientation. In the present study, the cross-sections were taken in an orientation that included the pressing direction. Consequently, it was permissible to use the oriented cycloid array as the sampling means. Image analysis, using automation both with and without operator interaction, was used to determine the S V of the particle boundaries and the pore surfaces. Figure 7 shows an example of an area sampled using the cycloid array. Figure 7. Example of the cycloid array (the red sine-like waves) overlaid on a sample sintered for a total of five minutes. Using the automated image analysis system, the pores have been colored green in this example. Red arrows point to the approximate locations of coincidence of the particle boundaries and the cycloids. A P L count of 12 would be made for this field. In practice, as the array is overlaid on the field, the locations of cycloid and particle boundary coincidence are counted and totaled in a multi-field analysis. These counts are then entered into

9 the expression previously shown as Equation 1, along with the total line length of the cycloid array. The resulting calculation produces the surface-to-volume ratio of the particle boundaries within the total material volume. Figure 8 is a graph showing the reduction of the S V in samples processed from 1 to 10 total minutes. 40 FC Particle Boundary SV (mm²/mm³) Compaction Pressure (MPa) Total Sintering Time (min) Figure 8. Graph showing the reduction in S V of the particle boundaries as the sintering time was increased for the three compaction pressures. The one-minute sintered impact bars compacted at 690 MPa suffered considerable damage during testing and were not fit for S V measurement. From the graph, it appears that the higher compaction pressures result in an increase in the amount of particle boundaries, regardless of the sintering time. This is due to the higher densities and the corresponding increase in the amount of particle contact as the compacting pressure is raised. The differences become less, possibly insignificant, as the sintering time reaches ten total minutes. The measurable difference in sintering at this point is significant because, at ten minutes total sintering time, the samples may not have reached the actual set-point sintering temperature. Consequently, the sintering may have taken place at temperatures between the melting point of copper (1083 C) and 1120 C. This helps to explain the relatively forgiving sintering response of the Fe-Cu-C materials. Fractography Fracture surfaces from tested impact bars were examined in an SEM to find possible trends in the failure mechanism as the sintering time was extended. In this case, the group of bars compacted at 690 MPa was examined. Secondary electron imaging (SEI) was used as the signal source for this portion of the sintering evaluation. Other than simple removal of dust from the surfaces, the samples required no additional preparation. The surface of a one-minute sintered sample is shown in Figure 9. No apparent sintering has taken place. The particle surfaces appear as failed green materials. 11

10 Figure 9. Fracture surface of a one-minute sintered bar. No sintering has taken place and separations between adjacent particles are clearly seen. At the five-minute time, sintering has apparently started as evidenced by the small ductile regions at the sintered particle necks (Figure 10, a and b). The location of the image on the right is outlined in white within the image on the left. a Figure 10, a and b. Five-minute sintered sample showing the start of sintering. At ten minutes, the effects of increased sintering are clearly visible. Figure 11, a and b, show the presence of, not only an increase in the amount of ductility, but also the start of transgranular cleavage. Additionally, the progress of sintering, as indicated by the amount of material fracture, has increased considerably. b

11 a Figure 11, a and b. Surface from a ten-minute sintered sample showing a combination of ductile dimples and brittle transgranular fracture. At 30 minutes total furnace time, the greater amount of metallurgical bonding that has occurred results in more evidence of fractured surfaces. Ductile rupture, transgranular cleavage, and fracture through pearlite are all seen in significant proportions in Figure 12. b TC P D a b Figure 12, a and b. An area showing the fracture from a 30-minute sintered sample. The b image contains transgranular cleavage, labeled TC, pearlite, labeled P, and ductile rupture, labeled D. The samples from the 50-minute total sintering time were also examined. An increase in the amount of transgranular cleavage fracture was apparent as seen in Figure 13, a and b. Significant amounts of ductile rupture and fractured pearlite also are present.

12 a b Figure 13, a and b. A fracture surface from one of the 50-minute sintered samples. An increase in the amount of brittle fracture, in the form of transgranular cleavage is apparent. Discussion & Conclusions Sintering of the FC-0205 material appears to begin well before the typical sintering temperature of 1120 C (2050 F) is reached. Evidence for this is presented in Figures 4, 10, and 11; where the sample time in the hot zone was below the minute pre-heat time. Diffusion of the carbon and copper is illustrated in Figure 4, while ductile and brittle fractures are evident in both Figures 10 and 11. In addition, the loss of particle identity through the loss of particle boundaries occurs during heating to the sintering temperature as may be observed in Figure 8. The fracture surface examinations appeared to show an increase in the amount of brittle, transgranular fracture as the sintering time was increased to a total of 50 minutes. This may be due to homogenization of the copper distribution and consequent hardening of the ferrite. The impact energy and apparent hardness measurements from these sintered materials were compared with the MPIF Standard 35 typical values for FC The results are shown in Figures 14 and 15. The apparent hardness results in Figure 14 indicate the diffusion of carbon and at least partial copper diffusion, at 10 to 20 minutes total time, produces values approaching and in excess of the standard values. In reality, these samples may barely have reached sintering temperature, or, were at temperature for less than 10 minutes. The vertical gray lines on both figures indicate the probable time to sintering temperature. Impact results appeared to exhibit a similar trend. At a density of 6.6 g/cm³, the values either nearly met or exceeded the MPIF Std. 35 typical values for a density of 6.7 g/cm³ at all times of 10 minutes or longer in the hot zone. The 6.8 g/cm³ samples exceeded the 6.7 g/cm³ MPIF Std. 35 typical values at 10 minutes and longer sintering times. In addition, the 6.8 g/cm³ samples nearly met the value of the 7.1 g/cm³ MPIF Std. 35 typical value at 20 minutes and surpassed these values at 25 minutes and longer total sintering time. The 7.0 g/cm³ samples exceeded the 7.1 g/cm³ MPIF Std. 35 typical values at all times of 20 minutes and longer. As this material reaches the sintering temperature, there appears to be a region of inconsistency in the mechanical properties as exhibited by the impact results at the 10 and 15 minute total sintering time. However, with an increase in the time-at-temperature, the impact energy appears to stabilize.

13 Standard 35 Typical Values 7.1 g/cm g/cm 3 Hardness (HRB) Impact Bar Sintered Density (g/cm³) Total Sintering Time (min) Figure 14. Rockwell B apparent hardness values for the three density levels and the variety of sintering times Impact Energy (joules) Standard 35 Typical Values 7.1 g/cm g/cm Total Sintering Time (min) Impact Bar Sintered Density (g/cm³) Figure 15. Unnotched Charpy impact test results for each set of sintered materials. The microstructure of this material undergoes an evolution as the time in the hot zone is increased. At one minute in the hot zone, the maximum temperature reached is probably high enough to burn-out the lubricant but not to start diffusion of the graphite or copper. Extending the time in the hot zone to 5 minutes, the graphite starts to diffuse and the sintering at particle contacts commences. Nevertheless, in most cases, the copper remains unmelted, although the

14 temperature seen by the samples at this time is probably very close to the melting point of copper. At 10 minutes total time in the hot zone, the graphite has diffused, the copper has melted, and the majority of particle boundaries have been sintered. Apparently, the copper has diffused into the matrix and strengthened some of the ferrite, resulting in the beginning of brittle transgranular failure and fracture through pearlite colonies. Sintering for longer times homogenizes the chemical composition and improves the sintered particle bonds. As sintering continues, the fracture surfaces exhibit progressively greater amounts of ductile rupture and brittle transgranular failure. References 1. Danninger, H. Pore Formation During Sintering of Fe-Cu and Its Effects on Mechanical Properties, Powder Metallurgy International, Vol. 19, No. 1, 1987, pp Kaysser, W.A., Huppmann, W.J., Analysis of Dimensional Changes During Sintering of Fe-Cu, Powder Metallurgy, 1980, Vol. 23, No. 2, pp Lawcock, R.L., Davies, T.J., Effect of Carbon on Dimensional and Microstructural Characteristics of Fe-Cu Compacts During Sintering, Powder Metallurgy, Vol. 33, No. 2, 1990, pp Jamil, S.J., Chadwick, G.A., Investigation and Analysis of Liquid Phase Sintering of Fe- Cu and Fe-Cu-C Compacts, Powder Metallurgy, Vol. 28, No. 2, 1985, pp Berlinger, M.,m Wasiczko, B., Nayar, H., Kiefer, R., Chen, Y.T., A Study of Dimensional Changes in Fe-Cu-C Parts in Different Furnace Zones and Different Sintering Atmospheres, Advances in Powder Metallurgy & Particulate Materials-1992, Compiled by J.M.Capus and R.M. German, Metal Powder Industries Federation, Princeton, NJ, 1992, Part 3, pp Baddeley, A.J., Gundersen, H.J.G., Cruz-Orive, L.M., Estimation of Surface Area from Vertical Sections, Journal of Microscopy, Vol. 142 Pt. 3, June 1986, pp Gundersen, H.J.G., et.al., Some New, Simple and Efficient Stereological Methods and Their Use in Pathological Research and Diagnosis, APMIS, Vol. 96, 1988, pp Karlsson, L.M., Cruz-Orive, L.M., Estimation of Mean Particle Size from Single Sections, Journal of Microscopy, Vol. 186 Pt. 2, May 1997, pp Russ, J.C., DeHoff, R.T., Practical Stereology, 2 nd Edition, 1999, Plenum Press, New York, NY, pp Underwood, E.E., Surface Area and Length in Volume, Quantitative Microscopy, Edited by R.T. DeHoff and F.N. Rhines, McGraw-Hill Book Co., New York, NY, 1968, pp Underwood, E.E., Quantitative Stereology, Addison-Wesley Publishing Co., Reading, MA, 1970, pp James, W.B., High Performance Ferrous PM Materials for Automotive Applications, Metal Powder Report, Vol. 46, No. 9, 1991, pp