An Investigation into the Effect of Copper and Graphite Additions
to Sinter-Hardening Steels
Thomas F. Murphy & Michael C. Baran
Hoeganaes Corporation
1001 Taylors Lane
Cinnaminson, NJ 08077
Abstract
Sinter-hardenable powders have been used as replacements for traditional quench-hardened and
tempered materials due to their ability to transform to a martensitic microstructure upon cooling
during a typical sintering operation. In the manufacture of the base sinter-hardenable powders,
alloying additions are made to the melt (prealloyed), with graphite added to the base powder as
the carbon source. However, additions of copper to further improve the hardenability of the mix
are commonplace. The combined admixed additions of graphite and copper to the prealloyed
powder can conceivably lead to increased retained austenite contents. In the present study,
metallographic techniques have been developed to resolve retained austenite in a predominately
martensitic material. Etching and staining techniques, automated image analysis, and scanning
electron microscopy were the metallographic tools used in this study.
Introduction
The addition of sinter-hardenable powders to the arsenal of P/M materials has expanded the range
of applications available for pressed and sintered powder metallurgy parts. Compacts made from
these powders have the capability to transform to martensite upon furnace cooling from the
sintering temperature. Due to this ability to transform to the higher-hardness martensite, the
process of quench-hardening and tempering can be eliminated from the manufacturing process
sequence where surface or through hardening is required.
This capacity to form martensite during the sintering process is developed by an increase in
hardenability, accomplished through both prealloyed and admixed elemental additions. The use
of both alloying techniques results in an increase in the alloy content of the final mix. The base
materials used for sinter-hardening applications are prealloyed. In prealloying, the alloy additions
are made to the molten metal bath prior to atomizing. This creates uniformity in alloy content
and, as the particulates are formed, the chemical composition within each particle is the same. In
most sinter-hardening situations, two additional materials are used as admixed powder additions.
They are carbon, in the form of graphite, and copper. These beneficial elements are almost
always admixed because prealloying them in the melt has deleterious effects on both
compressibility and green strength.
Experimental Procedure
The information presented in this paper is a metallographic analysis of samples taken from two
separate research programs. The tested, commercially available material, was a highly alloyed
prealloyed powder, with additions of graphite and copper as powders. The additions of copper
and graphite provide further enhancement of the hardenability. Due to its high hardenability, this
material has the ability to form martensite upon cooling in the sintering furnace. During the
sintering process, carbon is diffused into the base powder before the copper becomes liquid. The
consequence of this alloying sequence is twofold. The diffusion of copper is inhibited in mixes
with high carbon contents, and as the copper diffuses, high concentrations are often located along
particle and grain boundaries. The high copper concentrations result in a localized increase in
hardenability and a decrease in the Mf temperature, both on a local scale. This generally results
in the formation of high percentages of retained austenite in these boundary regions.
The current authors reported results from the first study in 1999 [1], where the base powder was
mixed with a constant 2 weight percent (w/o) Cu and varying amounts of graphite, from 0.43 to
0.83 w/o. In the second, more recent program, the same base material was mixed with varying
amounts of both copper and graphite. Combinations of 0.8 w/o graphite and 1.5 w/o copper;
0.9 w/o graphite and 2.0 w/o copper; and 1.0 w/o graphite and 2.5 w/o copper were used.
In both programs, the test specimens were tensile bars pressed to densities of 6.80 and
approximately 6.95 g/cm³. Sintering was performed in an Abbott belt furnace fitted with a
VARICOOL cooling system. The atmosphere used in both studies was 90 v/o N2/10 v/o H2. The
sintering temperature in the 1999 program was 1138 °C (2080 °F) while 1120 °C (2050 °F) was
used for the more recent study. After sintering, all tensile bars were tempered at 205 °C (400 °F).
In preparing the specimens using the standard etching techniques with nital and picral solutions,
problems were encountered. The presence of the retained austenite between the martensite
needles was difficult to establish. Insufficient contrast was developed within the highly alloyed
martensite. An alternative etching/staining technique was used to ‘color’ the martensite while
leaving the retained austenite white. In addition, the stain permitted the presence of alloy
variations to be observed through a change in coloration.
The stain etch used was one proposed by Vilella and Kindle [2] in the late 1950’s and is an
aqueous solution of approximately 21 to 28 w/o sodium bisulphite (NaHSO3). In practice, the
specimens were first ground and polished using standard metallographic techniques. Good
polishing practice should be used with this procedure, because it reveals the presence of fine
scratches. After polishing, the samples were lightly pre-etched using a combination of either
2 volume percent v/o nital and 4 w/o picral or 1 (v/o) nital and 4 w/o picral, depending on the
copper content and the microstructure. Regions containing a higher copper content may etch or
stain at a faster rate using the more highly concentrated combination. Therefore, using the more
dilute solution may slow the pre-etch process sufficiently to provide the operator more flexibility
at this stage of sample preparation. After pre-etching, the samples were immersed in the NaHSO3
solution. A sulfur-rich interference film is deposited on the prepared surface, changing the visual
appearance to a bluish tint. After staining, the specimen is first rinsed under running warm water
and then with alcohol. The coloration of the film appears dark (thick layer) with the lower
alloyed martensite while the retained austenite remains white or nearly white (thin layer) and the
higher alloyed martensite is an intermediate shade. This preparation technique develops contrast
between the various constituents and, if a thick layer is deposited and greater contrast developed,
is applicable for use with an automated image analysis (AIA) system. It was hoped that using this
preparation technique would help locate any areas containing excessive amounts of retained
austenite and the possible causes for degradation of the material properties.
Results and Discussion
Figures 1 and 2 are images at the extremes of the alloy additions from the more recent study.
They illustrate the difficulty in distinguishing the amount and location of the retained austenite
when samples are etched with only the nital and picral combination. Samples are not usable for
testing using an image analysis system because of the lack of contrast and definition of the
various constituents.
2 v/o nital + 4 w/o picral
Figure 1. Section from a 0.8 w/o graphite, 1.5 w/o copper sample. The precise location of the
retained austenite is difficult to determine.
2 v/o nital + 4 w/o picral
Figure 2. Section from a 1.0 w/o graphite, 2.5 w/o copper sample. Retained austenite location
remains difficult to determine.
Using the same samples, but different fields of view, the improvement in contrast is seen in
Figures 3 and 4 when they are stained using the sodium bisulphite solution. The particle interiors
are darkened while the pore and some grain boundary edges show the bright, angular features
typical of retained austenite.
2 v/o nital/4 w/o picral then 25 w/o NaHSO3 in H2O
Figure 3. Sample containing 0.8 w/o graphite and 1.5 w/o copper. Bright features are apparent in
regions along some of the particle/pore regions.
2 v/o nital/4 w/o picral then 25 w/o NaHSO3 in H2O
Figure 4. Sample containing 1.0 w/o graphite and 2.5 w/o copper. As in Figure 3, the bright
regions are retained austenite and higher alloyed martensite.
Figures 3 and 4 show the locations, and relative amounts of higher alloyed martensite and
retained austenite in the two extreme samples. The regions higher in alloy content are along the
pore and particle edges. These are the locations where, after melting, the liquid copper would
flow and alloy. It is apparent from these images that the copper is present at higher
concentrations in the regions where staining is less.
The recent images, Figures 3 and 4, were compared with images from the 1999 study. As can be
seen in Figures 5 and 6, the location of the retained austenite is similar to that observed in the
more recent program.
Figure 5. Sample containing 2 w/o Cu
and 0.74 w/o carbon from the earlier
sinter-hardening study. This is a
monochrome image suitable for analysis
using an automated system.
1 v/o nital and 4 w/o picral then
25 w/o NaHSO3 in H2O
Figure 6 Sample containing 2 w/o Cu
and 0.83 w/o carbon, also from the
earlier study. As with Figure 5, this is a
monochrome image suitable for AIA.
1 v/o nital and 4 w/o picral then
25 w/o NaHSO3 in H2O
For comparison, Figure 7 was taken of a 2 w/o Cu + 0.43 w/o graphite sample. The white,
retained austenite regions decorating the pore edges are not present. Apparently, the change in
carbon content had a large effect on the distribution of the 2 w/o copper, or the 0.43 w/o carbon
content was not sufficient to transform the alloy to martensite. The latter argument does not
appear feasible because the particle interiors are martensitic, as are the pore edges.
Figure 7. Photomicrograph of a 2 w/o Cu
+ 0.43 w/o carbon sample. The retained
austenite at the pore edges is not present.
1 v/o nital and 4 w/o picral then
25 w/o NaHSO3 in H2O
To verify the effect of the carbon on the diffusion of the copper, chemical analysis was performed
on the samples with 2 w/o copper, with both 0.43 w/o and 0.74 w/o carbon. An SEM equipped
with an Energy Dispersive X-Ray Analysis system was used for the analysis. Conditions for the
analysis were: accelerating voltage of 15 keV, magnification of 2 kx, in spot mode for a dwell
time of 120 sec. livetime at 25-30% deadtime. The sampled region was along a line starting at an
interparticle sinter neck and stepped towards the middle of one of the particles at the neck. The
step size was 5 µm and the analysis was continued until either the center of the particle was
reached or until the copper content was near zero. Copper concentrations were determined using
standardless quantitative software, which is less accurate than analysis using certified standards;
but the concentrations of the other alloyed elements in the material were within acceptable limits.
Figure 8 is a graph showing the approximate copper concentrations with respect to distance from
the sintered neck. The solid line represents the high carbon, 0.74 w/o carbon, while the dashed
line the 0.43 w/o carbon. It is apparent that the copper was more concentrated at the surface of
the high carbon material, dropping from near 4 w/o to < 1 w/o within a distance of approximately
5 µm. In contrast, the lower-carbon material contained a lower copper content at the surface, near
3.5 w/o, with a measurable content, 0.5 w/o, at a depth of approximately 45 µm.
4.5
0.74 w/o Carbon
Approximate Copper Content (w/o)
4
3.5
3
2.5
2
1.5
1
0.43 w/o Carbon
0.5
0
0
5
10
15
20
25
30
35
40
45
50
Distance from Interparticle Sinter Neck (µm)
Figure 8. Graph showing approximate copper concentrations from an interparticle sinter neck.
Additional testing was performed on the four, 2 w/o copper content materials to determine the
amount of retained austenite present as the carbon content was increased. The analysis was
performed on the sodium bisulphite stained samples using an automated image analysis system.
Samples containing 0.43, 0.63, 0.74, and 0.83 w/o sintered carbon were analyzed using
monochrome images at a pixel resolution of 0.11 µm/pixel. A multi-field analysis totaling
0.6 mm² was examined on each sample. The average amount of porosity was subtracted from the
total area measured, resulting in an adjustment of the retained austenite values to account for only
the metallic portion of the microstructure. Figure 9 shows the relationship of the retained
austenite to the carbon content of the tested tensile bars. It should be noted that the increased
carbon levels is of value as the section size of the parts is increased, and the cooling rate is
changed.
0.9
Carbon Content (w/o)
0.8
0.7
0.6
0.5
0.4
0.3
0
2
4
6
8
10
12
14
Retained Austenite (v/o)
Figure 9. The relationship between the measured volume percent retained austenite and the
carbon content of the sinter-hardened specimens.
Conclusions
Metallographic determination of the retained austenite content in sinter-hardened materials has
proven difficult due to the lack of visual contrast between the martensite needles and the small,
angular white patches of retained austenite within the needles. The typically used etchants of
nital and picral reveal the martensitic structure. Although the orientation effects and lack of
definition of the individual martensite needles can be misleading when making estimates of
retained austenite contents. The use of stain etchants, in particular the aqueous sodium bisulphite
solution, alters the appearance of the martensite and makes quantitative assessment of the retained
austenite possible.
The evidence from these two studies appears to show that the carbon content of the sinterhardened compact has a strong effect on the diffusion of the added copper, and, that at higher
carbon contents, the diffusion of the copper is inhibited by the diffused carbon. This effect was
shown by German [3] and Kuroki, et. al. [4] where the amount of carbon diffused in γ-Fe reduces
the diffusion of Cu at pore surfaces and grain boundaries. This situation causes a higher local
hardenability at the particle/pore surfaces, and promotes the formation of retained austenite in the
high copper regions. However, in commercial practice, this level of retained austenite did not
interfere with part performance. More detailed investigations are in progress to understand the
material behavior and properties more thoroughly.
References
1. M.C. Baran and T.F. Murphy, Metallographic Testing to Determine the Influence of Carbon
and Copper on the Retained Austenite Content on a Sinter-Hardening Material, P/M Science
& Technology Briefs, Vol. 1, No. 3, 1999, pp. 22-26.
2. J.R. Vilella and W.F. Kindle, Sodium Bisulphite as an Etchant for Steel, Metals Progress,
Dec. 1959, pp. 99-100.
3. R.M. German, Liquid Phase Sintering, Plenum Press, New York, NY, 1985, P. 93.
4. H. Kuroki, G. Han, and K. Shinozaki, Solution-Reprecipitation Mechanism in Fe-Cu-C
During Liquid Phase Sintering, International Journal of Powder Metallurgy, March 1999, pp.
57-62.