00
Printed in Great Bntain. All rights reserved Copyright ‘, 1984 Pcrgamon Press Ltd
Abrbact-The microstructure of a powder metallurgical high speed steel (ASP 60) has been investigated
by combined analytical electron microscopy and atom-probe field-ion microscopy. The compositions of
MC and M,C primary precipitates were determined. The tempering reaction at 560°C was studied in detail
and it was shown that extremely fine M,C precipitates dominated at optimum hardness (3 h). Some fine
MC precipitates were also present. The compositions of the secondary carbide precipitates were
determined by atom-probe analysis. Overageing at 700°C resulted in the transfo~ation of the fine
secondary carbide dispersion to coarse M,C and M& and a change in the matrix morphology from
martensitic plates to equiaxed recovered ferrite subgrains.
R&mn&Nous avons ttudii la microstructure dune poudre mitallurgique d’un acier rapide (ASP 60) en
combinant la microscopic electronique analytique et la microscopic a emission d’ions. Nous avons
determine les compositions MC et M,C des precipitis primaires. Nous avons hudii en ditaii k revenu
i 560X et nous avons month que des precipitis extr~mement fins de M,C dominent ion de Ia durete
optimale (3 h). Quelques precipitb fins de MC ttaient tgalement presents. Nous avons determine par
analyse avec la sonde atomique les compositions des precipitb de carbures secondaires. Un sur-
vieillissement $ 700°C conduisait P la transformation de la dispersion de fins carbures secondaires en M,C
et Mz3C6 grossiers et a un changement de la morphologie de la matrice de plaquettes martensitiques en
des sous-grains equiaxes de ferrite restauree.
been possible to determine the exact nature of the of the peak hardness condition to a short period of
phases which are responsible for peak secondary over-heating.
hardness. However, depending upon the alloy com- The technique used for the preparation of thin foils
position, they are usually considered to be either MC for transmission electron microscopy has been de-
or M2C [ 1,4-lo]. Sometimes MC (cementite) is also scribed previously [2]. Examination was carried out in
claimed to be present at peak hardness [I, 71. a JEOL 200 CX Temscan instrument (TEM;STEM)
The alloy chosen for the present investigation was fitted with a Link 860 system for energy dispersive
the highly alloyed powder metallurgical steel, ASP X-ray spectroscopy (EDX). Typical detector live-
60. The powder metallurgical technique used in its times of 100-250 set were used for each microanalysis
production enables high concentrations of alloying by EDX. The energy scale of the spectra was cali-
elements to be used while keeping the microstructure brated against a generated zero strobe peak and the
free from large scale heterogeneities (e.g. banding, CuKn peak of a spectrum which was taken up at
segregation, slag inclusions) which are common to regular intervals from a Cu standard specimen. The
cast and wrought high speed steels [ 1I]. As is the case microanalyses were made quantitative by using the
for most other high speed steels, the optimum combi- Link Systems RTS-Z/FLS computer programme
nation of hardness, yield strength and toughness is which compares peak shapes with those from stan-
obtained by repeated tempering at 56OC with inter- dard spectra of pure elemental standards and applies
mediate air-toolings. For ASP 60 a three stage tem- the thin foil approximation [14] with corrections for
pering treatment is necessary to attain complete absorption.
transformation and tempering of the substantial The kxsl sensitivity factors valid for the electron
amounts of austenite, which are retained in the microscope and EDX system used in the investigation
microstructure after hardening. The tempering tem- were checked experimentally [3, 151. For most ele-
perature of 56O’C is slightly over the minimum which ments of interest it was found that the k,,, factors
is necessary to ensure complete transformation of the predicted by theory (161 were satisfactory but it has
retained austenite (121. Tempering at higher tem- been found necessary to use experimentally deter-
peratures or for longer times results in a substantial mined kxs, factors at low X-ray photon energies. In
decrease in secondary hardness. Because of the prac- particular it should be mentioned that the kxs, factors
tical importance of the tempering treatment at 560°C for W, and MoL lines were checked experimentally.
the present investigation has concentrated upon the Although primary compositional information for W
detailed microstructural developments which occur at and MO was obtained from the W, and MO, line
this temperature. The effect of overageing at 7OO’Cof intensities, knowledge of the relative intensity values
an initially optimum microstructure is also consid- W&V, and MoL/Mo, proved extremely useful in
ered. Such a heat treatment simulates a severe over- subsequent quantification.
heating which could possibly occur during service of Correction of the EDX results for absorption of
alloys of this type. Some of the results which are X-rays in the specimen required determination of the
presented here have previously appeared in pre- specimen thickness at the position where analysis was
liminary reports [2, 131. carried out. Because of the high amount of strain
contrast within the martensitic matrix this was gener-
2. EXPERIMENTAL ally impossible to establish by normal experimental
procedures. Instead, the thickness of the matrix and
The composition of the alloy under investigation is primary M,C and MC carbides were determined by
shown in Table 1. Specimens for subsequent micro- finding which estimated thickness gave the same
structural examination were austenitized in a salt compositional results from the RTS-2IFLS computer
bath at 1180°C for 3 min and hardened by quenching programme when the MO line intensities were deter-
into a molten salt bath at 55OC and air cooling to mined using either the MO, or MO, lines. The general
room temperature. The step cooling treatment elimi- accuracy of this method has been confirmed pre-
nates quench cracking. The standard tempering treat- viously (171, and in the present case was checked by
ment of three times 1 h at 56OC, which eliminates studies of spade shaped field ion microscope speci-
retained austenite and provides optimum hardness, mens which could be tilted 90’ in the microscope for
was investigated together with shorter and longer visual determination of local thickness. In situations
times at the same temperature. Other heat treatments where the local W composition was relatively high
were carried out at 7OO’C in order to investigate the (e.g. -30 wt% in M6C) it was found that the peak
tempering reaction at a temperature higher than that stripping routine in the computer programme did not
for peak hardness and also to investigate the response always satisfactorily separate the Sik peak from the
Whl peak. (These two lines are only separated by Prior austenite grain boundaries were often decorated
36eV in the EDX spectrum.) In these cases the two by MC carbides.
peaks were stripped manually, knowing from pre- The composition of the martensitic matrix (includ-
vious calibration work the true relative intensities of ing retained austenite). which provides the relevant
W, and W, lines. For all of the carbide precipitates starting composition for the tempering reaction, was
analyzed. the particles penetrated both top and bot- determined by both STEMI;EDX and atom-probe
tom surfaces of the thin foils and they were field-ion microscopy. Results from these analyses are
sufficiently large (- 1 pm) to avoid errors due to
beam spreading.
For all elements of interest, the experimental errors
in individual EDX analyses due to counting statistics
and the errors in she sensitivity factor k,,, were, as
has been shown previously 1181.in the order of 10:;
relative at the 990/, confidence level. Consideration of
other possible sources of error such as surface films
on the specimen and errors in estimation of specimen
thickness suggested that the most satisfactory expres-
sion of confidence in the results was to state the
standard deviation for a series of measurements made
at different positions in the same phase. In general a
total of 10 such measurements gave a satisfactory low
standard deviation. The validity of the results were
further cross checked with analyses made by atom-
probe field ion microscopy where the most important
source of error arises from counting statistics.
The method of specimen preparation for atom-
probe field ion microscopy has been described pre-
viously [2] as has the particular instrument which was
used in this investigation [ 191.All field ion specimens
were examined by transmission electron microscopy
prior to inserting in the atom-probe field ion micro-
scope in order to provide a cross check with the thin
foil electron microscopy work and also to determine
the suitability of the specimens for field ion micros-
COPY.
3. RESULTS
Table 2 STEM EDX and atom-probe anal- mens. Analyses of these precipitates by both atom-
yses (at.“,) of martensitic matrix prior to probe and STEM/EDX techniques confirmed the
tempermg
general validity of the thin foil results shown in
STEM atom-probe Table 3 [2,3]. The carbon contents of the carbides
Si - 1.2_+0.? could aLo be determined by atom-probe analysis and
v 1.1 IO.2 0.8 f 0.2
Cr 4.6 2 0.2 4.2 * 0.4
the results from these few measurements are included
Fe 76.2 + 0.8 75.9 2 0.P in Table 3 in order to give an indication of the
Mn 114~0.2 stoichiometry of the primary carbides. It can be noted
co II.8 = Il.2 I I .4 f0.5
Md 1.7 r0.4 ?.I r0.3 that the MC carbides were rich in V but also con-
w 0.8 IO.1 1.010.2 tained a significant quantity of MO while the M,C
C 3.4 * 0.3
-. carbides contained large amounts of. in order of
Total 96.6”,, 100.0”” importance, Fe, MO and W. As expected, the MC
‘Inciudes Mn (peak overlap). carbides were found to be understoichiometrjc. No
The STEM EDX results have been adjusted significant differences were noted in the compositions
to sum to 96 6”,, in order to account for
C detected b) atom-probe but not by of the primary carbides between the as-quenched and
STEM EDX. The indicated error hmtts different tempered conditions.
for STEM EDX are standard deviations
of the spread between individual mea-
surements. The atom-probe results are
from IO.OXi atoms analyzed from 2 3.3. Tempering at 560 C
spectmens. Here the indicated error
limits are based upon the standard devi- The standard three-stage tempering treatment of
atlon of the \tatlstlcal counting error. three times 1 h at 560 C with intermediate air cool-
ings, which gives optimum hardness for this alloy.
presented in Table 2 where it can be seen that good resulted in the transformation of all of the retained
agreement was obtained between the two techniques. austenite to plate martensite. The original mor-
It can be noted that. because of the existence of the phology of the highly twinned plate martensite was
primary carbides MC and M&. the concentration of retained and as shown in Fig. 5 a very fine, but
the strong carbide-forming elements (V. MO and W) dense, dispersion of secondary carbides precipitated
was considerably less in the martensite than given by both on the martensite plate boundaries and within
the bulk composition of the steel (Table 1). The high the plates. In some regions of the microstructure.
carbon content of the martensite (3.4 at.::), as deter- precipitation had apparently occurred on the bound-
mined by atom-probe. indicates a high super- aries of the internal twins within the martensite plates
saturation with respect to carbide phases prior to the [Fig. 5(b)] while in other regions precipitation was
tempering reaction. Approximately 1.2at.?; Si was independent of such features [Fig. 5(a)]. The very fine
detected in the martensite by atom-probe but the Si precipitate sizes combined with the highly strained
content was too low for statistically significant mea- nature of the martensite and the ferromagnetism of
surements by STEM,‘EDX. There exists a possibility the material made imaging of the secondary precip-
that the atom-probe measurements give a slight over- itates in the electron microscope exceedingly difficuft
estimate of the Si content because of selective field and, moreover, chemical analysis of these precipitates
evaporation between the applied high voltage pulses by STEM/EDX was impossible because of the large
[22]. This possible overestimate of the Si content is amount of matrix material which was present in any
also present in the atom-probe analyses of the tem- volume sampled by this method. Satisfactory extrac-
pered matrix in Table 6. Another atomic species, tion replicas which would enable analysis of the fine
which was dificult to anafyse accurately in the atom-
probe is Mn. This is due to peak overlaps with Fe.
Energy i keVf
5 10 15
Energy 1keV)
Fig. 4. Characteristic BDX spectra obtained from primary carbides (a) MC; (b) M,C.
secondary carbides by STEM/EDX could not he and were generally highly streaked. As shown in
obtained. Reflections from the precipitated phase in Fig. 6 selected area diffraction suggested that the
selected area diffraction patterns had a low intensity secondary precipitates were either MC obeying the
Baker-Witting [23] crystallographic orientation re-
Table 3. Average STEM/EDX analyses of s&riruticurul elements
(at.%) in primary carbides lationship with the matrix or M2C being close to the
MC Pits&-Schrader [24] o~entat~on relationship.
M.5C
The high num~r density and very fine particle
si - 4.4 It 2.9
V 59.3 f 2.1 4.7 + 0.3 sizes of the secondary precipitates was confirmed by
Cr 7.6 f 1.8 5.6 i 0.6 field-ion microscopy. The most frequently observed
Fc 4.2 * 0.8 40.7 + 0.9
co 0.7 * 0.5 7.5 * 0.4 precipitate morphology was fine elongated platelets
MO 19.4 * 1.6 23.1 iO.5 of approximately 2 nm thickness [see Fig. 7(a)] but in
W 8.8 * 0.8 14.0 + 0.4 one particular area larger platelets (up to 20nm in
Stoichiomctry MCo,s+oos M 1*+01C
(from atom-probe anatysw) diameter) of the type shown in Fig. 7(b) were ob-
Indicated error limits are based upon standard deviations of the
served. Results of atom-probe analyses of these two
spread bctwccn individual measurements. types of precipitate are given in Table 4. The fine
1462 STILLER er al.: PRECIPITATION IN POWDER METALLURGICAL STEEL
(b)
Fig. 7. Field-ion microscopy of secondary carbide precip-
itates (3 x I h. 560 C). (a) Fine platelets of M,C. (b) Larger
plate shaped MC.