Solidification of high speed steels
M. Boccalini and H. Goldenstein
High speed steels are ferrous based alloys of the
Fe–C–X multicomponent system where X
represents a group of alloying elements
comprising mainly Cr, W or Mo, V, and Co. These
steels are mainly used for cutting tools, since they
are characterised by their capacity to retain a high
level of hardness while cutting metals at high
speed. The as cast microstructure of high speed
steels consists of dendrites surrounded by a more
or less continuous interdendritic network of
eutectic carbides. These are observed even under
rapid cooling, at rates as high as 106 K s−1. The
main features of the as cast microstructure are the
distribution and morphology of eutectic carbides,
owing to their decisive influence on mechanical
properties, and on the service performance of the
high speed steels, even after hot working. Hence,
Published by Maney Publishing (c) IOM Communications Ltd
the most important improvements on the as cast
microstructure that have been obtained through
progress in alloy design concern the type,
morphology, and volume fraction of the eutectic
carbides, while improvements obtained by
progress in solidification processing are, on the
other hand, mostly related to a significant
reduction in the microstructure scale. The
development of the microstructure in high speed
steels through solidification is reviewed,
emphasising the effects of the alloy chemical
composition and of the cooling rate. The formation
of the eutectic carbides and the techniques used to
control their morphology and distribution, both
under normal and rapid cooling, are discussed.
Microstructures obtained through surface
remelting and surface alloying using laser and
electron beams are described, and the application
of highly alloyed high speed steels to the
manufacture of cast hot rolling mill rolls is
reviewed. IMR/363
© 2001 IoM Communications Ltd and ASM International.
Dr Boccalini is at the Instituto de Pesquisas Tecnológicas –
IPT, São Paulo, Brazil and Professor Goldenstein is in the
Escola Politécnica da Universidade de São Paulo, São
Paulo, Brazil.
Introduction
High speed steels comprise a family of alloys mainly
used for cutting tools. Their name – high speed steel
– is a synthesis of the following two features: (a) the
alloys belong to the Fe–C–X multicomponent system,
where X represents a group of alloying elements in
which Cr, W or Mo, V, and Co are the principal
ones; ( b) the alloys are characterised by their capacity
to retain a high level of hardness even when submitted
to elevated temperatures resulting from cutting metals
at high speed.1,2
A steel developed by Mushet, in England, in the
latter half of the 19th century – Fe–2C–2·5Mn–7W
(wt-%)* – is considered the forerunner of the modern
* Compositions are given in weight-% unless otherwise stated.
92 International Materials Reviews 2001 Vol. 46
high speed steels. That steel presented a great advan-
tage over high carbon steels, which had been largely
applied so far in cutting tools, since it became hard
when air cooled from a temperature from which most
steels required water quenching for hardening.2–4 The
most important development during the following
30 years was the substitution of chromium for
manganese, as a result of work conducted in Western
Europe and the USA.3
The full capabilities of the alloy as cutting material
were however realised only after the work of Taylor
and White on its heat treatment, towards the end of
the 19th century. Hardening the alloy from a temper-
ature close to its solidus temperature, they attained
uncommonly high hardness levels. Roberts3 empha-
sised that, since high speed steel must combine wear
resistance with the essential property of red hardness,
the first milestone of the high speed steels’ history
should not be the date when Mushet’s steel compos-
ition was first proposed, but the date when that
composition was combined with this particular heat
treatment.
The first alloy formally classified as high speed
steel, patented by Crucible Steel Co. in the beginning
of this century, contains 0·7C, 18W, 4Cr, 1V (AISI
T1 in the American designation terminology or S
18–0–1 in the German designation terminology).
During the following 40 years this basic alloy design
was held, the most significant improvements being
the addition of up to 12%Co and the use of higher
carbon and vanadium contents (1·5%C and 4%V,
respectively), the latter being partially substituted for
tungsten.3
During World War II, a general shortage of metallic
raw material induced the development of alloy designs
with a view to leaner alloy contents and cheaper
alloying elements, mainly substituting for tungsten
and vanadium.4 This brought, in the 1950s, a high
popularity to AISI M2 steel (S 6–5–2) 0·9C, 6W, 4Cr,
5Mo, 2V, after some restrictions to its use from
European steelmakers were surmounted, leading to
the overtaking of T1 steel.5 In the USA, on the other
hand, molybdenum rich grades, such as AISI M1
steel (S 1–8–1) 0·8C, 1·5W, 4Cr, 8Mo, 1V, have been
used since the 1930s.4,6 For studying purposes, con-
ventional high speed steel grades have been divided
into three series called tungsten, tungsten–
molybdenum, and molybdenum high speed steels,
according to their W and Mo contents.
In the 1970s, another wave of cost savings based
on the use of niobium as an alloying element was
attempted.4,7–9 The main research line was based in
the idea of partially substituting niobium, a cheaper
and stronger primary carbide forming element, for
vanadium in the grades of the tungsten–molybdenum
and molybdenum series, thus allowing the vanadium
to be utilised at a lower level, predominantly for
secondary precipitation hardening. Although the
research effort on niobium as a tool steel alloying
No. 2 ISSN 0950–6608
 Boccalini and Goldenstein
element was developed world wide, the production of
conventional niobium alloyed high speed steel grades
is currently restricted to South America and the
former Soviet Union countries.10,11
Despite the diversity of compositions resulting from
the alloy design development, the solidified micro-
structure of high speed steels has maintained its
characteristic basic feature, i.e. dendrites surrounded
by a more or less continuous interdendritic network
of eutectic carbides. These microstructures have
evolved over time, however, mainly as a result of
changes in two main parameters:
1. Alloy composition. The most important changes
caused by progress in alloy design concern the type,
morphology, and volume fraction of the eutectic
carbides.
2. Solidification process. When considering the role
of conventional solidification processing (cooling rate
ranging from 10−3 to 102 K s−1) on the solidified
microstructure, the changes in microstructure caused
by changing cooling rate are mainly with regard to
dendrite size and eutectic colony distribution, both
being more homogeneous as a result of the application
Published by Maney Publishing (c) IOM Communications Ltd
of electroslag remelting2,12–14 or refining treatments
through minor additions.15–17 With rapid solidifi-
cation processing (cooling rate ranging from 102 to
107 K s−1) the resultant microstructure is mainly
characterised by its very reduced scale, the dendrite
arm spacing or cell size ranging from 0·01 to 1 mm,
i.e. two to four orders of magnitude smaller than that
of normal solidified microstructures.2 Despite this
remarkable difference in size, the more or less contin-
uous interdendritic (or intercellular) network of eutec-
tic carbides remains as a distinctive characteristic of
the microstructure, even for cooling rates as high as
106 K s−1, since short range segregation still takes
place. For extremely high cooling rates of more than
108 K s−1, however, segregation is suppressed and a
featureless microstructure is observed, probably as a
result of diffusionless solidification.18
The application of rapid solidification techniques,
like powder metallurgy, spray deposition, and laser
remelting/alloying, allows as a consequence pro-
duction of conventional high speed steel grades with
improved end user properties,19 as well as design of
highly alloyed high speed steel grades with carbide
contents up to 50 vol.-%,18,20 thus ensuring superior
wear resistance.21,22
Table 1 gives the nominal chemical composition of
the most commonly used commercial high speed steel
Table 1 Nominal chemical composition of the most commonly used commercial high speed steel grades
Chemical composition, wt-%
Alloy Series Processing† C Cr
T1 Tungsten C 0·7
T15 C; PM 1·5
M1 Molybdenum C 0·8
M7 C 1
M42 C; PM 1·1
M2 Tungsten– C; PM 1·0
M35 molybdenum C; PM 0·8 (1·0)
ASP 60* PM 2·3
CPM Rex 45* PM 1·3
* Proprietary alloys.
† C produced through processing routes involving conventional solidification; PM produced through powder metallurgy routes.
Solidification of high speed steels 93
grades, including both the conventional grades (pro-
duced through processing routes involving conven-
tional solidification) and the P/M grades (produced
through powder metallurgy routes). Figure 1 is a
panoramic sketch showing the evolution of the as
cast microstructure throughout the century of devel-
opment of high speed steels.
Solidification sequence of high speed
steels
Most of the early experimental work on the solidifi-
cation path of high speed steels was based on the
18%W, 4%Cr section of the quaternary Fe–W–Cr–C
diagram published by Murakami and Hata23 in the
1930s and modified by Kuo in the 1950s.24 This
particular section received a lot of attention, as it has
the same W and Cr contents as AISI T1 steel, largely
applied so far as discussed above, Fig. 2.
The following incorrect assumptions were made by
early users of the Murakami–Hata–Kuo temperature–
concentration section (isoplet) for the analysis of the
solidification of high speed steels:
1. The presence of up to 2 wt-%V causes only
small changes in the general shape of the section,
apart from a slight shift of the whole diagram to
higher C contents caused by the stoichiometric con-
sumption of C by the formation of VC carbide,
thermodynamically favoured over the W and Cr
carbide (M C).24 This assumption is based on the
6
hypothesis that the VC carbide precipitates directly
from the liquid, preceding all other solidification
reactions. This was soon reviewed by Hoyle and
Ineson,25 showing that vanadium carbides form at
the end of solidification, through the eutectic reaction
at temperatures lower than that of the formation of
M C carbide. Additionally it was found that there is
6
a substantial solubility of vanadium in the matrix as
well as in M C carbides, i.e. not all V is consumed
6
by the formation of MC carbides.
2. Tungsten and molybdenum have almost ident-
ical behaviour as alloying elements, as can be seen
by the similarity between Fe–W–C and Fe–Mo–C
ternary diagrams.26 On a weight fraction basis this
can be expressed by a ‘tungsten equivalent’ value
given by Mo/W=1:2. Actually, that similarity is very
limited, since, for the composition ranges typical for
high speed steels, the Fe–W–C diagram presents only
M C carbide, while the Fe–Mo–C diagram presents
6
also Mo C.27
2
W Mo V Co
4 18 ... 1 . ..
4 12 ... 5 5
4 1·5 8 1 . ..
4 1·75 8·75 2 . ..
3·75 1·5 9·5 1·15 8
4 6 5 2 . ..
4 6 5 2 5
4 6·5 7 6·5 10·5
4 6·25 5 3 8
International Materials Reviews 2001 Vol. 46 No. 2
 94 Boccalini and Goldenstein Solidification of high speed steels

1 Evolution of as cast microstructure of high speed steels

Published by Maney Publishing (c) IOM Communications Ltd

2 Section of quaternary Fe–W–Cr–C phase

diagram at 17%W, 4%Cr (adapted from Ref. 24)

Despite those incorrect assumptions, experimental

results obtained by directional solidification, and con-
ventional and differential thermal analysis28–31 show,
as a first approximation, good qualitative agreement
with the temperature region relevant to solidification
in the diagram shown in Fig. 2, thus justifying its use
by generations of researchers. Years later, Horn and
Brandis32 obtained the temperature–concentration
section Fe–6W–5Mo–4Cr–2V–C for the study of AISI
M2 steel, reflecting the relative importance achieved
by this particular alloy at that time, Fig. 3.
From 1990 onwards, isoplets of the Fe-rich corner
of the Fe–Cr–W–Mo–V–C diagram have been calcu-
lated by means of the CALPHAD method, using the
ThermoCalc package (program and thermodynamic
data base).34–36 Figure 4 shows the calculated isoplets
Fe–4Cr–18W–1V–C, Fe–4Cr–6W–5Mo–2V–C, and
Fe–4Cr–2W–9Mo–2V–C. These sections refer,
International Materials Reviews 2001 Vol. 46 No. 2
3 Section of Fe–W–Mo–Cr–V–C phase diagram at
6%W, 5%Mo, 4%Cr, 2%V (adapted from Ref. 32)
respectively, to the steels T1, M2, and M7, that typify
the high speed steels series and hereafter are treated
as representatives of these.
Agreement between experimental and calculated
isoplets is quite good, mainly in relation to the phase
stability fields (the tie lines are not in the plane of the
sections and, therefore, the amount and the compos-
ition of the phases must be calculated separately).
Most experimental diagrams do not make distinctions
between specific carbides resulting from eutectic reac-
tions, labelling all carbides under the common letter
 Boccalini and Goldenstein Solidification of high speed steels 95

Precipitation of primary delta ferrite from

(a) liquid (Ld)
In the composition region of the Fe–Cr–W–Mo–V–C
system relevant to high speed steels, carbon and, to
a lesser extent, molybdenum have a significant effect
on the liquidus temperature, which decreases accord-
ing to the increase of the carbon or molybdenum
contents.26,29,37 Table 2 gives liquidus temperatures
for steels T1, M7, and M2 calculated using
ThermoCalc and experimentally measured using ther-
mal analysis techniques (for cooling rates ranging
from 0·08 to 0·10 K s−1).
The partition of alloying elements between delta
ferrite dendrite and the liquid is more intense for C
than for all other elements, even though all liquid/
ferrite partition coefficients are smaller than 1. This
occurs due to the fact that C is the sole austenite
stabiliser among the alloying elements, with a small
(b) solubility in ferrite. Simultaneously, the high diffusion
rate in the ferrite lattice allows the almost complete
homogenisation of ferrite stabiliser alloying elements
during this step of the solidification (diffusion of
Temperature, °C
Published by Maney Publishing (c) IOM Communications Ltd

alloying elements from the borderline to the core of

the dendrites).40–42 Fredriksson40 showed, for a M7
steel quenched from the melt just before the onset of
the peritectic reaction, that the W and Cr concen-
trations are homogeneously distributed throughout
the ferrite dendrite.

Formation of austenite through peritectic

reaction (L+dc)
The intense carbon segregation to the liquid during
the crystallisation of delta ferrite along with decreas-
ing temperature leads to the stabilisation of the
(c)
austenite, which precipitates at the d/L interface as
the product of the peritectic reaction. Actually, as
discussed below, in the section ‘Formation of
austenite–peritectic reaction’, the peritectic reaction
is a two-step process controlled by different
mechanisms.
The temperature of the onset of the peritectic
reaction T is strongly affected by the chemical
p
composition26,28,29,40 and by the cooling rate during
solidification:40 it increases with increasing carbon
content and decreases with increasing V, Mo, and W
contents. These effects come from the nature of the
alloying element: carbon enlarges the austenite stabil-

Table 2 Liquidus temperature for steels T1, M2,

and M7
Carbon content, wt-% Carbon, Liquidus,
Steel wt-% Technique* °C Ref.
a Fe–4Cr–18W–1V–C; b Fe–4Cr–6W–5Mo–2V–C; c Fe–4Cr–
2W–9Mo–2V–C T1 0·76 DTA 1441 33
ThermoCalc 1441 35
4 Pseudobinary sections calculated using DTA 1441 38
ThermoCalc (adapted from Ref. 35) M2 0·76 DTA 1440 29
0·81 DTA 1433 33
0·81 ThermoCalc 1434 35
0·88 CTA 1427 39
C, while calculated diagrams show the carbides M C
6 M7 0·93 DTA 1408 33
and MC as stable ones. 0·93 ThermoCalc 1418 35
For carbon levels between 0·8% and 1%, typical 0·93 DTA 1408 38
of the high speed steels, the solidification sequences 0·97 DTA 1410 29
obtained from Figs. 2–4 are similar and can be 1·0 CTA 1400 39
described as follows. * CTA conventional thermal analysis; DTA differential thermal analysis.

International Materials Reviews 2001 Vol. 46 No. 2

 96 Boccalini and Goldenstein Solidification of high speed steels

ity field (in M2 steel C levels larger than 1·4% make

austenite the primary phase to solidify, suppressing
the formation of the d ferrite26), while V, Mo, and W
enlarge the ferrite stability field.26,28 Table 3 gives
1400
values calculated using ThermoCalc, and values
experimentally measured using thermal analysis tech-
niques (cooling rate ranging from 0·08 to 0·10 K s−1),
of the temperature of the onset of the peritectic

Temperature, °C
1350
reaction for steels T1, M7, and M2.

Formation of eutectic carbides through

1300
eutectic reaction (Lc+C)
The residual interdendritic liquid, enriched with
carbon and carbide former elements, decomposes
through different eutectic reactions as it moves down 1250
a eutectic trough, leading to the formation of up to
three carbides: M C, M C, and MC. The resultant
2 6
microstructure contains a continuous network of
interdendritic carbides. 1200
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Table 4 gives values of the temperature of the onset fs
of formation of those eutectics experimentally meas-
ured using thermal analysis techniques (cooling rate 5 Volume fraction of solid phase (fs) as function
Published by Maney Publishing (c) IOM Communications Ltd

of temperature during solidification of T1, M7,

again ranging from 0·08 to 0·10 K s−1). The values of and Nb-containing M2 steels (adapted from
T correspond to instances where eutectic formation Ref. 33)
e
starts with c+M C and/or c+M C, while values of
2 6
T correspond to instances where c+MC eutectic
MC peritectic reaction is usually described, somewhat
appeared before the other eutectics, as discussed paradoxically, as the last step of the solidification
below, in the section ‘Effects of chemical composition process.31,33,38 This reaction occurs almost simul-
and cooling rate (conventional solidification)’. taneously with the eutectic reactions, developing an
austenite plus carbide eutectoid-like product.38
Delta eutectoid transformation
The solid state decomposition of the intradendritic Volume fraction of solid phases
residual delta ferrite not entirely consumed by the Although the solidification sequence described above
is basically the same for T1, M2, and M7 high speed
Table 3 Temperature of onset of peritectic steels, the evolution of the solid/liquid ratio during
reaction L+dc (T ) solidification is somewhat different, expressing the
p balance between austenite and ferrite stabilisers in
Steel Technique* T , °C Ref.
p their compositions. Figure 5 shows the evolution of
T1 ThermoCalc 1324 35 the volume fraction of solid phases as a function of
DTA 1323 33
DTA 1323 38
temperature during the solidification of T1, M7, and
M2 ThermoCalc 1318 35 Nb-alloyed M2 (similar to plain M2) high speed
DTA 1330 33 steels. The volume fractions were measured from
DTA 1310 29 directionally solidified specimens whose solidification
CTA 1350 39 was concluded by quenching.38 One can see how
M7 ThermoCalc 1354 35
DTA 1350 33
different the solidification ranges are, and also the
CTA 1322 39 fractions of the solidification time spent in the regimes
DTA 1294 38 of delta ferrite and austenite related to T1 steel, on
DTA 1290 29 one hand, and to M7 and Nb-alloyed M2 steels on
* CTA conventional thermal analysis; DTA differential thermal analysis. the other.

Table 4 Temperature of formation of c+M C, Dendrite spacing

2
c+M C, and c+MC eutectics The as cast dendritic microstructure is defined in a
6
Steel Technique* T †, °C T †, °C Ref. quantitative basis by the primary and secondary
MC e
T1 DTA 1333 1303 33
dendrite spacings, the latter being more widely used
DTA . .. 1286 38 as a parameter for correlating process variables and
M2 DTA . .. 1242 33 microstructure, since the primary dendrite spacing
DTA . .. 1240 29 intervenes only in the case of directional solidification.
CTA 1270 1255 39
The effects of the liquid phase thermal gradient,
M7 DTA 1234 1225 33
CTA . .. 1233 39 dendrite growth rate, and coarsening/coalescence
DTA . .. 1225 38 phenomena33,43 on the secondary dendrite arm
DTA 1240 1222 29 spacing (simply designated hereafter as the dendrite
* CTA conventional thermal analysis; DTA differential thermal analysis. arm spacing, DAS) have been well characterised
† See text. experimentally.
International Materials Reviews 2001 Vol. 46 No. 2
 Boccalini and Goldenstein Solidification of high speed steels 97

solidification. Table 6 gives the results for five different

Fe-base alloys, including M2 and M7 steels. It can be
Ref. 43
Ref. 39
seen that M2 and M7 steels present a far smaller
Ref. 33 variation of the DAS than other alloys compared,
Ref. 45 despite the larger interval for austenite solidification.
Boccalini45 shows that during the solidification of
DAS, µm

M2 steel, small changes in carbon level cause signifi-

cant changes in parameter A, while changes in
vanadium content do not affect A in the same way,
Fig. 8. This result was attributed to the effects of the
alloying elements on the constitutional supercooling.

dT, K s_1
6 Experimental dendrite arm spacing (DAS)
versus dT /dt plots obtained by different
researchers
Traditionally the effects of the liquid phase thermal
gradient G and dendrite growth rate V (which are
Published by Maney Publishing (c) IOM Communications Ltd
Formation of austenite–peritectic
reaction
Classically, peritectic reactions are a two-step
process:46
1. Nucleation and growth of the peritectic phase
(c-phase, in this case) occurs first, along the interface
between the liquid and the properitectic phase
(d-ferrite), with partial consumption of the latter. This
step is relatively rapid, depending only on alloying
element diffusion in the liquid phase.40

2. Independent growth of the peritectic phase then

independently controlled in directional solidification
experiments) on the DAS are related by an empirical
relationship where both variables are lumped into
one single parameter, namely the cooling rate, as
dT /dt=GV (Ref. 44)
DAS=A(dT /dt)−b . . . . . . . . . (1)
where A and b are material-specific parameters.
Figure 6 shows experimental curves DAS versus
dT /dt for M2 steel obtained by different researchers.
Table 5 gives the A and b parameters used to construct
the curves shown in Fig. 6.
The empirical relationship between DAS and cool-
ing rate incorporates all thermally activated phen-
omena, including coarsening/coalescence. According
to Fischmeister,33 parameter A in equation (1)
expresses the influence of coarsening/coalescence on
DAS, since it correlates predominantly with the solidi-
fication range, which determines the time available
for coarsening. Experimental results show the preva-
lence of coarsening and coalescence phenomena
during austenite growth, after the onset of the peritec-
tic reaction (Fig. 7).
Jernkontoret’s data39 suggest, though, that changes
in DAS during solidification are not directly related
to the solidification range after the peritectic reaction.
This can be tested by comparing measurements of
the DAS on samples quenched from the melt immedi-
ately before the onset of the peritectic reaction with
that found on samples quenched just after the end of
Table 5 A and b parameters of equation (1),
DAS=A (dT /dt )−b
Experimental method A, mm s K−1 b Ref.
Unidirectional growth, 32 −0·28
variable growth rate
Unidirectional growth, 35·7 −0·35
variable growth rate
Unidirectional growth, 43 −0·38<b>−0·34 33
constant growth rate
Unconstrained growth, 42·1 −0·32
variable growth rate
takes place. When the surface of the properitectic
phase is completely covered by the peritectic phase
(no simultaneous contact between the three phases
involved in the reaction), the growth of the peritectic
phase can only be completed through two different
and concomitant processes: dc and Lc, the kin-
etics of which are determined by volume diffusion of
alloying elements through the peritectic phase. This
step is slower than the first and controls the reaction
as a whole.40,46 The properitectic phase, besides that,
may be incompletely consumed depending on the
nature of the phase and on the thermal history during
solidification.40
Steps 1 and 2 are known respectively as the peritec-
tic reaction and the peritectic transformation.40,46 In
order to avoid improper interpretations caused by
the usual terminology, since both the overall process
and its first step are called peritectic reaction, the first
and second steps will hereafter be named, respectively,
peritectic reaction step 1 and peritectic transform-
ation step 2.
Austenite growth through peritectic reaction step 1
and peritectic transformation step 2 is influenced by
thermal parameters and by chemical composition.
Peritectic reaction step 1
Fredriksson,40 using metallographic analysis of M2
samples quenched from the melt at different temper-
atures within the solidification range, showed that
both carbon content and cooling rate during solidifi-
cation exert an influence on the peritectic reaction
step 1: it is favoured by high carbon levels or high
cooling rates in conventional solidification.
The effect of carbon is explained by its austenite
stabilising character. The effect of the cooling rate is
43 indirect: a low cooling rate allows back diffusion of
39
ferrite stabiliser elements during the crystallisation of
the dendrites (Mo, W, and V, mostly), delta ferrite
becoming more stable. Figure 9 shows the micro-
structure of M2 steel samples solidified under different
45 cooling rates and quenched from the melt at temper-
atures just below the beginning of the peritectic
International Materials Reviews 2001 Vol. 46 No. 2
 98 Boccalini and Goldenstein Solidification of high speed steels

A, µm s K_1

(a) (b)
Published by Maney Publishing (c) IOM Communications Ltd

7 Correlation between parameter A and a total solidification range DT and b fraction of solidification
range after nucleation of peritectic austenite DT /DT (adapted from Ref. 33)
aust
A, µm s K_1

Vanadium content, wt-%

$ 0·75%C; & 0·95%C; + 1·15%C
8 Effects of carbon and vanadium contents on
parameter A in M2 steel45

reaction step 1. Slowly cooled samples (Fig. 9a) show

properitectic ferrite regions still in contact with the
liquid (not completely covered by peritectic austenite).
These regions are not seen in the more rapidly cooled
samples (Fig. 9b).
Peritectic transformation step 2
During the peritectic transformation step 2, austenite
growth is accompanied by both the dissolution of the
properitectic ferrite (solid state dc reaction) and
the liquid phase reaction Lc.
dc transformation
Under kinetically favourable conditions (small cool-
ing rates) the dc transformation is accompanied by
the partition of alloying elements between the austen-
ite and ferrite, with a progressive enrichment of the
International Materials Reviews 2001 Vol. 46 No. 2
b
a 0·07 K s−1; b 0·7 K s−1
9 M2 steel samples quenched from melt at
temperatures just below beginning of peritectic
reaction: arrow in a shows properitectic ferrite
region is still in contact with liquid40
ferrite in ferrite stabiliser elements (notably W, V, and
Mo). This partition stabilises ferrite, sometimes to the
point of blocking the completion of the transform-
ation. On the other hand, it promotes the formation
of a liquid phase layer at the d/c interface, which
decomposes into an aggregate of solid phases resem-
 Boccalini and Goldenstein Solidification of high speed steels 99

bling that of the solidification product of the inter-
dendritic residual liquid.38,40,47 It was the similarity
between the two structures that actually led to the
hypothesis that a liquid phase layer forms at those
interfaces, i.e. this formation was only indirectly
detected. In addition, Fredriksson,40 using a
semiquantitative approach to the solidification of a
ternary Fe–M–C alloy, M being a strong ferrite
stabiliser element, showed that it is possible for the
peritectic transformation to turn into a metatectic
reaction (dL+c), thus explaining the forma-
tion of liquid inside the solid phase. This liquid
will solidify later through a eutectic reaction
Lc+carbides.38,40
The presence of residual ferrite from incomplete
peritectic transformation in the as solidified micro-
structure (actually ferrite transformed into
c+carbides through the d-eutectoid reaction) is
dependent on the cooling rate and the carbon content;
a low volume fraction of d-eutectoid corresponds to
low carbon content or low cooling rate.40
The influence of the carbon content is related to
the following factors:
Published by Maney Publishing (c) IOM Communications Ltd
(iii) smaller d-ferrite DAS with increasing cooling
rate, leading to higher d/c interface area and
consequently to faster dc transformation.48
Lc transformation
This step of the overall peritectic reaction is analogous
to the formation of austenite from other liquid phases.
The cooling rate affects the morphology of the peritec-
tic austenite resulting from this transformation. For
low cooling rate, transformation occurs with a smooth
c/L interface, the ausenite thus ‘inheriting’ the den-
dritic morphology from the primary ferrite.40 For
high cooling rate, the c/L interfaces become unstable
and dendrite arms start branching from the austenite
layer covering the primary ferrite dendrites. This
branching is due to a large kinetic undercooling of
the liquid promoted by the high cooling rate during
the Lc transformation.40
Formation of carbides – eutectic
reactions
The literature on high speed steels often alludes to a

particular eutectic microconstituent by means of the

1. Ferrite stability phase field in the phase diagram.
carbide formed: M C eutectic, M C eutectic, or MC
An increasing carbon content shrinks the delta ferrite 6 2
field, diminishing its volume fraction precipitated eutectic. Although lacking in precision, this termin-
ology describes the essential difference between the
from the liquid phase. Figure 5 shows that T1 steel
has, in relation to M2 and M7 steels, a higher volume eutectics and thus is also employed in this text. Care
fraction of ferrite at the temperature at which the must be taken, though, to distinguish whether an
author is referring to the eutectic constituent
formation of peritectic austenite starts. Consequently,
the as solidified microstructure of T1 steel presents a (c+carbide) or to isolated carbides, particularly when
discussing volume fractions.
higher d-eutectoid fraction than M2 and M7 steels.
Table 7 gives the compositions of the M C, M C,
2. Kinetics of the properitectic delta ferrite con- 6 2
sumption during the peritectic transformation. The and MC carbides in the as cast microstructure of the
T1, M2, and M7 high speed steels. It can be seen that
kinetics are controlled by the amount of liquid phase
the differences between the compositions of the car-
formed through the metatectic reaction, which favours
bides M C and M C refer to the levels of V, Cr ( both
the diffusion process that controls the peritectic trans- 2 6
being higher in the M C carbide) and, mainly, Fe
formation.47,48 The metatectic reaction occurs only 2
(significantly higher in the M C carbide). The chemi-
above a critical carbon content.40 It can thus, there- 6
fore, be assumed that the higher the C content, the cal composition of MC carbide is different from the
compositions of the M C and M C carbides by virtue
faster the formation of the peritectic austenite and 2 6
of its higher levels of V and low levels of Fe and Cr,
the lower the fraction of ferrite at the end of the
solidification. in addition to its wide range of variability.33
Regarding the effect of the cooling rate during the
solidification ( less residual ferrite with increasing cool- Characteristics of M C, M C, and MC
6 2
ing rate), the following effects are noted: eutectics
(i) a decreasing level of back diffusion of ferrite The main characteristics of the three types of eutectic
stabilisers into ferrite with increasing cooling that can be found in the as cast microstructure of
rate, reducing ferrite stability high speed steels are outlined as follows.
(ii) suppression of the formation of liquid phase
M C eutectic
at d/c interfaces by increasing cooling rate 6
The three-dimensional morphology of this eutectic is
characterised by the presence of a central platelet of
Table 6 Values of DAS from Fe-base alloys
quenched at different temperatures45 Table 7 Chemical compositions of M C, M C, and
6 2
DAS†, mm
MC carbides33
DT *, Variation
aust Chemical composition, at.-%
Alloy Nominal composition, wt-% K P E P–E /P, %
Steel Carbide W Mo V Cr Fe
AISI 1034 Fe–0·34C 55 50 90 80
... Fe–0·3C–1·5Cr–4Ni–0·5Mo 60 60 90 50 T1 M C 35·7 ... 6·6 6·9 50·3
6
... Fe–0·7C–13Cr 168 50 80 60 M2 M C 14·5 20·9 7·6 7·0 50
6
AISI M2 Fe–0·9C–6W–5Mo–2V–4Cr 82 35 40 14 M C 20·8 33·3 27·8 10·6 7·5
2
AISI M7 Fe–1·0C–2W–9Mo–2V–4Cr 89 35 35 0 MC 8·9 11·6 65·8 4·8 8·9
M7 M C 4·9 30·9 6·1 6·2 51·9
* DT solidification range after nucleation of peritectic austenite. 6
aust M C 4·1 50 27·9 11 7·0
† P samples quenched immediately before the onset of the peritectic 2
MC 1·8 14 76·5 4·7 3·0
reaction; E samples quenched just after the end of the solidification.

International Materials Reviews 2001 Vol. 46 No. 2

 100 Boccalini and Goldenstein Solidification of high speed steels

A new terminology was proposed for the classifi-

cation of the morphological types of the M C eutec-
2
tic,56 using the formal classification for all eutectic
systems proposed by Crocker et al.57 This was necess-
ary to describe the characteristics of this eutectic
more precisely and to avoid conflicting uses of the
same expression, as commonly observed in the litera-
ture.56 Figure 11 shows these morphologies and
related classification.
Strictly speaking, there are two types of M C
2
eutectic morphology: irregular and complex regular.
Irregular M C eutectic is characterised by a ragged
2
boundary that does not clearly outline the interface
between the matrix and the eutectic pool, suggesting
a a limited coupling between eutectic phases. In this
eutectic, M C carbide has a platelike morphology
2
and a tendency to assemble as radiating clusters (this
feature is often referred to as ‘fanlike’ or ‘feathery’).
It also shows extensive branching, which is respon-
sible for avoiding overgrowth by austenite and, conse-
quently, for maintaining M C carbide as the leading
2
phase.45 Complex-regular M C eutectic is charac-
2
terised by a smooth boundary that clearly outlines
Published by Maney Publishing (c) IOM Communications Ltd

the interface between the matrix and the eutectic

pool; it shows a regular distribution of M C eutectic
2
carbide over small areas, forming cells with macro-
facets.56 It was shown that complex-regular M C
2
eutectic grows as a spiral eutectic,58 though perfectly
developed spirals are rarely observed in the micro-
b structure of as cast high speed steel. Instead of this,
a optical: etching KOH+K Fe(CN) ; b SEM degenerate or interrupted spirals characterise the
3 6
10 Typical morphology of M C eutectic in as cast complex-regular M C eutectic, mainly due to the low
6 2
microstructure of M2 steels45 volume fraction of eutectic, as well as to the highly
branched dendritic structure in high speed steel.45
Irregular M C eutectic is chiefly promoted by low
M C carbide, from which arises secondary platelets 2
6 cooling rate or high vanadium content and the
of M C separated from each other by austenite. These
6 complex-regular M C eutectic by high cooling rate
secondary platelets are usually thicker at the end, 2
or low vanadium content.29,31,45,52 The effect of higher
interrupting the continuity of the austenite and form-
vanadium content is to decrease the volume fraction
ing a ‘wall’ of carbide around the eutectic colony.33,49 of M C carbide in M C eutectic,45 thus increasing
Optical microscopy reveals that the cross-sections of 2 2
the difficulty of keeping M C carbide as the leading
the platelets are lamella-like (a ‘fishbone’ morph- 2
phase in the eutectic, and an irregular structure then
ology). Figure 10 shows the typical morphology of arises. On the other hand, as the cooling rate
the M C eutectic as seen in optical and scanning
6 increases, the difference between the growth rates
electron microscopes. The morphology of the M C
6 of M C carbide (faceted phase) and austenite
eutectic is not influenced by chemical composition 2
(non-faceted phase) also increases and the eutectic
nor cooling rate (considering values as high as
morphology becomes strictly controlled by the
106 K s−1), except that at faster cooling rates the characteristics of the M C carbide growth ( hexagonal
distance between platelets is decreased.33,49 2
hopper crystals45,58), ensuring the formation of a
In the chemical formula for the M C carbide, the
6 complex-regular structure.
letter M corresponds to the elements W, Mo, V, Cr, A third type of morphology associated with faster
and, mainly, Fe. The elements W and Mo are dis- cooling rates and characterised by the presence
solved in the same proportion as they are present in
of rod-shaped M C carbide was identified by
the composition of the steel, which is not the case for 2
Fredriksson and Nica31 and Lichtenegger et al.59
V, whose low level of dissolution is a characteristic Fredriksson and Nica verified that this morphology
of the M C carbide.33 The M C carbide has a complex
6 6 predominated in the regions of the samples whose
fcc crystalline structure and its hardness is around directional solidification was concluded by quenching.
1500 HV.50
It is believed that this morphological type just corres-
M C eutectic ponds to a ‘broken complex-regular’ morphology that
2
Contrary to what occurs with M C eutectic, the comes from frequent interruptions of the spiral growth
6
morphology of M C eutectic is substantially influ- of the M C carbide caused by localised overgrowth
2 2
enced by various factors, namely, the chemical com- by austenite, leading to its rod-shaped appearance.45
position,29,31,49,51 the cooling rate,29,31,51,52 and minor The M C carbide dissolves all the principal
2
additions of certain elements such as aluminium and elements of the high speed steels. Moreover, the
nitrogen.53–55 composition of M C carbide presents great variability
2
International Materials Reviews 2001 Vol. 46 No. 2
 Boccalini and Goldenstein Solidification of high speed steels
11 Morphological types of M C eutectic45
2
Published by Maney Publishing (c) IOM Communications Ltd
and is influenced by the cooling rate.33 M C carbide
2
has a hexagonal crystalline structure and its hardness
is around 2000 HV.50
The M C carbide, being metastable, decomposes
2
into MC and M C carbides when heated to between
6
900 and 1150°C, and it has been shown that the
kinetics of this breakdown and the distribution of the
carbide products is influenced by the morphology of
M C.52,60 In the microstructure of the end products
2
of M2 and M7 steel, therefore, M C carbide is not
2
found, since the manufacturing routes of these prod-
ucts make use of heat and/or thermomechanical treat-
ments within that temperature range.
MC eutectic
Within the range of chemical composition established
by the specifications of M2 and M7 steels, MC carbide
precipitates out during the eutectic decomposition of
the interdendritic liquid. However, the addition of
alloying elements that are strong MC carbide formers,
such as Nb, the use of high V or C contents, and the
presence of inoculants bring about a shift in the
precipitation temperature to levels above that of
the primary ferrite, causing the formation of MC
carbides directly from the bulk liquid phase (‘primary
MC carbides’) and sometimes allowing for a complete
suppression of the formation of the MC
eutectic.4,33,52,61,62
Three types of MC eutectic can be distinguished:
divorced (MC carbide occurs as isolated massive
crystals), irregular (poorly coupled eutectic in which
the MC carbide has a petallike morphology), and
complex-regular (coupled eutectic in which MC car-
bide has a branched petallike morphology). In the
literature, the latter was improperly classified as reg-
ular31 or lamellar.31,49 Figure 12 shows these morpho-
logies and related classification.45
In the divorced MC eutectic, it is likely that the
dendrites of peritectic austenite play the role of hetero-
geneous nuclei for the formation of the eutectic
austenite, making it impossible to distinguish them
metallographically. In irregular MC eutectic, it seems
101
Boccalini45 Other authors
Feathery26,49
Fanlike29,31,51
Lamellar49,51,52
Platelike26,31,39
Lamellar31,49,55
Rodlike29,51
that austenite and carbide begin to grow with a
certain level of cooperation though the carbide is
somewhat ahead, and at some point the austenite is
able to overgrow the carbide and close the halo
around it, possibly due to its greater volume fraction
in the eutectic. The formation of the complex-regular
MC eutectic can be understood in the same way, but
without the formation of the austenite halo (with the
increase in the volume fraction of the MC carbide in
the eutectic, the austenite is no longer able to over-
grow it).45
The difference between the peritectic temperature
and the temperature of eutectic formation determines
the prevailing MC eutectic type: the larger this differ-
ence, the greater the tendency for the formation of
the divorced eutectic;45 consequently, the higher the
cooling rate, the higher the tendency to the formation
of less coupled MC eutectic.45
In the conventional high speed steels, the MC
carbide dissolves mainly vanadium. In Nb-alloyed
high speed steels, vanadium and the weaker carbide
formers are partially replaced by niobium.33 The
composition of primary MC carbides shows variation
from the centre outward, owing to their growth from
a melt whose composition is continuously changing
during solidification.33
MC carbide has a fcc structure and its hardness is
around 3000 HV.50
Effects of chemical composition and cooling
rate (conventional solidification)
The total eutectic volume fraction, the volume fraction
of each different eutectic, and the sequence by which
they precipitate depend on the chemical composition,
the cooling rate, and the presence of residual
elements.4,27,29,31,51–53,63,64
Generally, the total eutectic volume fraction
increases with increasing C content,29,31 mainly
increasing W/Mo ratio,64 increasing V content,31,64
and decreasing cooling rate.29,33,39 Table 8 gives the
volume fraction of the M C and M C eutectics and
6 2
International Materials Reviews 2001 Vol. 46 No. 2
 102 Boccalini and Goldenstein Solidification of high speed steels
Published by Maney Publishing (c) IOM Communications Ltd

12 Morphological types of MC eutectic45

the volume fraction of the MC carbide for T1, M2,

and M7 steels solidified at different cooling rates.33
Figure 13 shows the total eutectic volume fraction in
M2 steel as a function of the C content and solidifi-
cation rate.29 The values of total eutectic volume
fraction in Fig. 13 were indirectly obtained through 0.05 cm min_1
the volume fraction of liquid that remained just before
the start of the formation of poorly or fully coupled
Volume fraction of coupled eutectic, %

eutectic in samples whose solidification was concluded

by water quenching, i.e. divorced MC eutectic was
not considered. This measurement method explains 0.5 cm min_1
the apparently increasing volume fraction with
decreasing carbon content from 0·90% to 0·75%: with
lower carbon content the peritectic reaction ceases
earlier, thus leaving a higher volume fraction of liquid
remaining.29
5 cm min_1
The volume fraction of each one of the three
eutectic types is determined, within this framework,

Table 8 Volume fractions of M C and M C

6 2
eutectics and MC carbide (Ref. 33)
Volume fraction, %
Cooling rate,
Steel K s−1 c+M C c+M C c+MC
6 2
T1 0·09 21·1 ... 0·1
0·41 18·3 ... 0·1
1·40 17·5 ... 0·1
M2 0·10 Trace 9·0 0·1
0·43 Trace 8·5 0·3 Carbon content, wt-%
1·48 Trace 8·0 0·5
M7 0·10 nd* 14·0 0·5 13 Volume fraction of coupled eutectic as function
0·45 nd* 11·3 0·5 of carbon content and solidification rate
1·63 nd* 9·9 0·7 (adapted from Ref. 29)
* Not detected.

International Materials Reviews 2001 Vol. 46 No. 2

 Boccalini and Goldenstein Solidification of high speed steels
14 Correlation between effects of W, Mo, and V
contents, for constant W equivalent, on
formation of eutectics in high speed steels
(adapted from Ref. 64)
by competition between them. The following relation-
Published by Maney Publishing (c) IOM Communications Ltd
ships arise:
1. M C eutectic is favoured with respect to M C
2 6
eutectic by increasing the Mo/W ratio,33,52,64 increas-
ing the V content31,45,64 or the C content,29,31,45 and
by increasing the cooling rate.31,45,52 An interdepen-
dence of these variables must be considered in
determining quantitative limits of their influence on
eutectic formation, as is shown in Fig. 14 for the V
content and Mo/W ratio.64
Different theories have been proposed to explain
the prevalence of the M C eutectic with increasing
2 2
cooling rate. Fredriksson and Brising29 proposed that
increasing cooling rate causes increasing segregation
of vanadium in the residual liquid from the peritectic
reaction, thus favouring the formation of the M C
2
eutectic because M C carbide is richer in vanadium
2
than M C carbide. Later on, Fredriksson and Nica31
6
revised this explanation, since they observed that even
with very low V content in M2 steel the precipitation
of M C eutectic was promoted by increasing cooling
2
rate, and suggested that the precipitation of meta-
stable M C eutectic should be kinetically favoured.
2
Golczewski and Fischmeister34 also presented
different explanations over time. Their first proposal
was also based on increasing segregation with increas-
ing cooling rate; they calculated the composition of
the residual liquid in the non-equilibrium solidifi-
cation of M2 steel and obtained C, Mo, and V
contents as high as 5, 7, and 5 wt-%, respectively.
Later on, they proposed that the competition between
M C and M C eutectics is related to the driving force
6 2
for nucleation, the eutectic with the largest driving
force prevailing over the other.36 Based on this, it was
shown that the larger the undercooling of the residual
liquid, as a consequence for instance of high cooling
rate, the higher the driving force for the nucleation
of the M C eutectic, Fig. 15. It is important to note
2
that this is rather in accordance with the kinetic effect
previously suggested by Fredriksson.
Figure 15 helps us to make a reasonable hypothesis
to explain the effect of V in favouring the M C
2
eutectic. It is known that M C carbide incorporates
2
103
Free energy (G)
(a)
(b)
a T ; b T <T ; DG M2C Gibbs free energy variation for formation
1 1 2
of M C eutectic (Ref. 36)
2
15 Driving force for formation of M C and M C
6 2
eutectic carbides (schematic): driving force for
formation of M C eutectic carbide increases
with increasing undercooling (adapted from
Ref. 36)
V in the ratio of its overall content in the alloy33 and
that its thermodynamic stability increases with
increasing V content.52 Then, increasing overall V
content will shift down the free energy curve of the
M C carbide in Fig. 15, thus increasing the driving
2
force for its nucleation at a given undercooling. In
other words, the larger the V content, the smaller the
minimum undercooling needed to allow the formation
of M C in place of M C and the lower the minimum
2 6
cooling rate needed for obtaining the same effect.
Additionally, for an undercooling large enough to
form both M C and M C eutectics, M C should
2 6 2
grow much faster than M C, since the former has a
6
structure full of defects and faults (Fig. 16).
2. MC eutectic is favoured with respect to M C or
6
M C eutectic by increasing the V content.31,33,64
2
Increasing the overall V content makes the residual
liquid from the peritectic reaction richer in V, increas-
ing the driving force for MC eutectic formation, as
well as increasing both the temperature of its forma-
tion and its volume fraction; consequently, the frac-
tions of the M C or M C eutectics decrease.
6 2
Generally, the volume fraction of MC carbide
increases with increasing cooling rate (Table 8);33
however, the opposite was observed in M2 steel with
low V content ( less than 1·4 wt-%).31,65
3. Besides the effects of the nominal chemical alloy
composition, the competition between eutectics is
International Materials Reviews 2001 Vol. 46 No. 2
 104 Boccalini and Goldenstein Solidification of high speed steels

(a)

Alloy content, wt-%

a (b)
Published by Maney Publishing (c) IOM Communications Ltd

b (1) start of solidification; (2) start of peritectic reaction; (3) start of
16 Details from carbides a M C and b M C
2 6
showing high incidence of faults and defects
in former; SEM45
affected by the presence of minor alloying elements.
The formation of M C eutectic is favoured by sili-
6 M C eutectic precipitation does not have a vanadium
con,31,53 niobium,33 rare earth metals,45 and yttrium,52
at the expense of M C eutectic, while the opposite
2 MC eutectic.65
happens for additions of nitrogen,45,53 calcium,53 and
aluminium.53
The order in which the precipitation of M C, M C,
2 6
and MC eutectics occurs is a function of the segre-
gation profiles resulting from the ferrite and austenite
formation and even of the eutectic decomposition
reactions themselves. Those profiles are also influ-
enced by the overall chemical composition and cool-
ing rate.
Figure 17 shows the influence of the V content on
the segregation profiles during the solidification of
M2 steel at 0·09 K s−1 and on the sequence of the
precipitation of the eutectics.31 It can be seen that the
higher the vanadium content, the higher the temper-
ature of the MC eutectic formation. In the solidifi-
cation of M2 steel with 1·44%V at higher cooling
rates, however, no MC and M C eutectics were
6
observed and only M C eutectic was formed.31 For
2
low overall V content, the formation of the MC
eutectic depends on the previous formation of the
M C eutectic. Once M C eutectic is formed (low
6 6
cooling rate), the residual interdendritic liquid is
enriched with vanadium, since M C has a low
6
vanadium content, leading to the formation of the
MC eutectic. If, on the other hand, the M C eutectic
6
is absent (as a result of a high cooling rate), the liquid
International Materials Reviews 2001 Vol. 46 No. 2
Distance, mm
a 0·89C–1·44V; b 0·85C–3·15V
M C eutectic; (4) start of MC eutectic; (5) start of M C eutectic;
6 2
(6) end of eutectic reactions
17 Evolution of chemical composition of
interdendritic liquid during directional
solidification of M2 steel (adapted from Ref. 31)
remaining from the peritectic reaction or from the
2
content high enough to induce the precipitation of
Effect of processing route
In terms of the as cast structure, the processing routes
of high speed steels can be divided into two groups:
normal cooling (solidification is not sufficiently rapid
to suppress reactions predicted by the phase diagram),
and rapid cooling (solidification may be sufficiently
fast to suppress reactions predicted by the phase
diagram).
Normal cooling – ingot casting
The as cast microstructure of high speed steel gener-
ally consists of dendrites of austenite decomposition
products (normally martensite with some retained
austenite) and a network of eutectic colonies with
carbides. These colonies are distributed hetero-
geneously in the interdendritic regions, tending to
form eutectic ‘pools’, and are thicker in the centre of
the ingot.66 Subsequent hot working breaks up this
network, diminishing the size and improving the
distribution of carbides; nevertheless, even after
reductions greater than 90%, this distribution does
not become entirely uniform and the carbides are
arranged in stringers,2,6 the thickness of which is a
 Boccalini and Goldenstein Solidification of high speed steels 105

function of the size and the distribution of the eutectic

colonies.2
For products obtained by conventional processing
routes ( hot working of cast or remelted ingots), the
main target concerning the as cast microstructure
refers to well distributed small-size eutectic car-
bides,16,43,67 since the behaviour of the material during
subsequent plastic deformation or machining, as well
as the properties of the final product, are influenced
by these characteristics.67–69
The techniques for controlling the cast micro-
structure act on two basic microstructural
characteristics: a
(i) the dendritic structure (including secondary
interdendritic spacing, columnar or equiaxed
morphology and size of the macroscopic grains)
(ii) the morphology of the eutectic carbides.
Dendritic structure
The methods commonly used for controlling the
dendritic structure of high speed steels are based on
the influence of thermal parameters (cooling rate and
thermal gradient) and of the addition of refiners on
Published by Maney Publishing (c) IOM Communications Ltd

this structure. There is nothing in the literature con-

cerning manipulation of the basic chemical compos-
ition (probably because it is largely fixed according
to other objectives, the performance of the end-use
product in particular), while mechanical methods
during solidification (vibration) have not achieved
significant results.2,13
Concerning the influence of thermal parameters,
the following can be noted:
1. The greater the cooling rate, the lesser the pri-
mary14,33 and secondary interdendritic spacings and
the lesser the volume fraction of the eutectic colon-
ies.29,33,39 The combination of these results naturally
leads to a more homogeneous distribution of the
interdendritic eutectic carbides. However, in conven-
tional ingot production the positive effects of the
cooling rate are limited to the regions lying nearer
the surface, owing to the large dimensions of the
ingots. Although of limited effectiveness, there are
some methods aimed at increasing the cooling rate
such as low pouring temperature and ingot moulds
with thicker walls.67,70
2. For a given cooling rate, the greater the thermal
gradient in the liquid phase, the greater the fraction
of columnar grains in the macrostructure.43 This effect
is verified in ingots produced by remelting consum-
able electrodes, particularly electroslag remelting (the
macrostructure of the ingots is made up entirely of
columnar grains), whose main characteristic is an
elevated thermal gradient in the liquid phase through-
out the solidification process.2,12–14 The greater micro-
structural homogeneity is characterised by the
uniformity of the size of eutectic cells throughout the
cross-sections of the remelted ingot,70 which results
from the unidirectional progression of the solidifi-
cation front during remelting,5,12,71 Fig. 18.
High speed steels produced by routes utilising ESR
provide better end-use performance than those pro-
duced from conventionally cast ingots.70 The great
advantage of ESR steels is related to the absence of
regions exhibiting greater concentration of eutectic
colonies (as is typical in conventional ingots) and,
b
a conventionally cast ingot; b ESR ingot
18 Microstructure of central region of M2 ingots71
consequently, of thicker stringers in the micro-
structure after plastic deformation.6,70 In fact, the
interdendritic network of eutectic colonies still exists
in ingots produced by ESR process, and even pos-
sesses a strongly directional phase distribution as a
result of a totally columnar macrostructure, in spite
of attempted improvements by the addition of refiners.
The effect of refiners is mostly to decrease the
macroscopic grain size (dendrites of primary ferrite),
notably through the formation of a higher fraction of
equiaxed grains.2,15,17,33,50,54,72–79 The decrease in the
size of the dendrites of primary ferrite increases the
interdendritic solid/liquid interface and, therefore, for
a given volume fraction of this liquid, the colonies
resulting from its eutectic decomposition are more
homogeneously distributed in the microstructure.
There are two mechanisms by which the refiners work:
1. The formation of compounds that play the role
of nuclei for the crystallisation of primary ferrite,
allowing for the heterogeneous nucleation of the
ferrite dendrites in detriment to the growth of the
same (which is reflected in the increase in the fraction
of equiaxed grains and a decrease in the fraction of
columnar grains in the macrostructure). These refiners
are strong carbide, nitride, and carbonitride formers,
such as Ti, V, Zr, and Nb.15,33,50,54,72,75–78 In fact,
these compounds have high temperature of formation
and a crystalline structure similar to ferrite, which
are the basic, though not sufficient, requirements for
an efficient nucleant.80 Some authors have shown that
the efficiency of the addition of Ti, V, Zr, and Nb in
the nucleation of the ferrite increases if they are
preceded by the addition of strong deoxidants such
as Al and Mg.17,50,76,77
International Materials Reviews 2001 Vol. 46 No. 2
 106 Boccalini and Goldenstein Solidification of high speed steels
a presence or absence of non-metallic inclusions playing
Published by Maney Publishing (c) IOM Communications Ltd
b this case, MC carbide is formed through the eutectic
a plain M2 steel; b M2+0·05%N
19 Effect of addition of nitrogen on morphology
of M C eutectic in M2 steel solidified at
2
1·5 K s−1; etching, electrolytic Cr O (Ref. 45)
2 3
2. Intense segregation in the liquid phase during
the precipitation of the dendrites of primary ferrite,
creating a ‘barrier’ that impedes the growth of these
dendrites owing to the low solubility of the element
in this phase; in this way, nucleation prevails, increas-
ing the fraction of equiaxed grains. There is evidence
that the rare earth metals73,74 particularly cerium,74
play this role.
Eutectic morphology
The morphologies of the M C and MC eutectics are
2
influenced by the presence of minor elements.
Additions of nitrogen and calcium promote the forma-
tion of the complex-regular M C eutectic, Fig. 19,
2 rates in producing the as solidified microstructure of
while additions of aluminium give rise to the forma-
tion of the irregular M C eutectic.45,53,81,82 The nature
2 the liquid metal into droplets before solidification, so
of the effect of these elements is not well established.
In the case of nitrogen, for example, Fredriksson and
Brising29 proposed that its effect is due to the forma-
tion of carbonitrides of V at temperatures above those
of the eutectic reactions, since this causes a decrease
in the level of V in the interdendritic liquid and,
consequently, favours the formation of the complex-
regular M C eutectic. This hypothesis was contra-
2 represent a third family of rapid cooling processes.1
dicted by Boccalini,45 who showed that nitrogen, in
addition, favours the formation of the M C eutectic
2
in relation to the M C eutectic (if the effect of a
6
diminished level of V predominated, the opposite
should occur). According to this author, nitrogen
increases the supercooling for the eutectic reaction,
favouring the metastable eutectic M C.
2
International Materials Reviews 2001 Vol. 46 No. 2
The application of the concept of modification of
the morphology of the carbides through the addition
of surface active elements or elements with low solu-
bility in the eutectic phases is the subject of studies
in high speed steels,45,54,83 in a way similar to that
which occurs in the Fe–Cr–C alloys.84,85 Chaus and
Rudnitskii54 studied the influence of various surface
active elements, including Bi, Ca, Sr, and Sb, on the
as cast microstructure of M2 steel, observing the
modification of the morphology of the M C eutectic,
2
which transformed from irregular to complex-regular.
Lichtenegger et al.59 showed that the morphology
of the MC carbide in M2 steel depends on the
the role of suitable sites for nucleation. Isolated
massive MC carbides form as a result of precipitation
on Ti(C,N) directly from the remaining liquid from
the peritectic reaction, not as a result of eutectic
decomposition of this liquid. The precipitation occurs
through a mechanism known as dual stage
nucleation,59,77 in which simple or mixed Al and Ca
oxides (deoxidation products) are nuclei for the pre-
cipitation of Ti(C,N), which are nuclei for the precipi-
tation of MC. If complex oxysulphide inclusions are
present instead of oxides, the first stage – hetero-
geneous nucleations of Ti(C,N) – is hindered and the
precipitation of massive MC carbides is avoided. In
decomposition of the residual liquid and the petallike
or branched petallike morphology prevails.
Tamura and Uchida83 showed that the addition of
rare earth metals diminished the temperature of the
formation of MC eutectic, transforming it from irregu-
lar to complex-regular. Boccalini et al.66 also observed
that the MC eutectic precipitates along with the M C
2
eutectic or the M C eutectic, forming a structure
6
they called duplex eutectic (Fig. 20). Curiously,
Fredriksson and Nica31 have observed the formation
of duplex M C–MC eutectic when adding 1·5%Si to
6
M2 steel.
The studies unanimously affirm that the morph-
ology modifying effect is due to the changes in eutectic
growth brought about by the build-up of modifying
elements at the eutectic carbide/liquid interface.
Rapid cooling
Processes characterised by the use of very high cooling
the high speed steels normally involve atomisation of
as to obtain powder for subsequent consolidation of
the material by powder metallurgical routes, or spray
deposition of a layer onto a solid substrate – centrifu-
gal spray deposition or Osprey are examples of the
latter.2 High intensity focused heat sources ( laser and
electron beam) allow for the remelting of surface
layers on semifinished high speed steel parts, and
Powder metallurgy
The major commercial rapid cooling process for the
production of high speed steel parts is powder metal-
lurgy, atomisation being the predominant process
used.86,87 The subsequent consolidation of the powder
particles may be conducted by several routes, such as
 Boccalini and Goldenstein Solidification of high speed steels 107

a b
Published by Maney Publishing (c) IOM Communications Ltd

c d

a, b optical, etching KOH+K Fe(CN) ; c, d SEM

3 6
20 a, c duplex M C–MC eutectic and b, d duplex M C–MC eutectic in rare earth modified M2 steel45
6 2

hot pressing, or extrusion and sintering, injection
moulding and sintering, and hot isostatic pressing.
The products obtained by these routes exhibit
superior properties and performance to those manu-
factured by the classic route (i.e. by normal cooling),
mainly with respect to toughness, isotropy of
properties, size stability in heat treating and grind-
ability,70,88,89 owing to their fine scale micro-
structures.2,19,87 Moreover, the refinement of the
eutectic carbides makes it possible to produce high
speed steels with higher alloying content, mainly
carbon and vanadium, thus ensuring better wear
resistance of the products.18,86
The cooling rate during the solidification of gas
atomised high speed steel particles in the size range
below 500 mm is estimated to fall in the range
104–106 K s−1 (Ref. 87) – the smaller the particle size
the higher the cooling rate. Moreover, the solidifi-
cation of splat caps encasing some particles, formed
as a result of in-flight collision between completely
liquid droplets and fully solidified particles, is esti-
mated to occur at cooling rates exceeding 108 K s−1
(Refs. 87, 90). Since the resulting powder consists of
particles of widely varying sizes, and since collisions
are a common feature of the process, a spectrum of
size scale of the microstructure can be observed in a
single atomised charge (Fig. 21).91 The as solidified
microstructure of the atomised particles is two to
three orders of magnitude more refined than those
obtained for conventionally cast ingots.
Depending on the particle size, the morphology of
the microstructure of the high speed steel powder
particles is dendritic, cellular, or a mixture of both,
the fraction of cellular morphology being significant
only in the very small sized particles, i.e. those solidi-
fied at higher cooling rates.92,93 The morphology
changes to a globular one in the Nb-containing high
speed steels, which is explained by the presence of
primary NbC acting as nucleation sites for the pri-
mary solid solution phase.94
The existence of undercoolings high enough to
promote the formation of a cellular morphology is
attributed to the higher cooling rate and to the lesser
number of potential heterogeneous nucleation sites
for solidification of the particles.92 Figure 22 shows
micrographs of particles with dendritic and cellular
morphologies, as well as the relationship between the
size of the particle and the fraction of the particles
with cellular structure for a high speed steel (Fe–1·2C–
4Cr–6Mo–4W–1V–1Nb–12Co).
Except for its smaller scale, the solidified micro-
structure of the high speed steel powders is not
significantly different from that of the slow cooled
high speed steel ingots, since it is unlikely that the
particles are sufficiently undercooled to allow
diffusionless solidification, owing to the large freezing
range of those steels.95 Thus, the normal micro-
segregated structure constituted by solid solution
dendrites (or cells) delineated by a more or less
continuous interdendritic (or intercellular) network of
International Materials Reviews 2001 Vol. 46 No. 2
 108 Boccalini and Goldenstein Solidification of high speed steels
Published by Maney Publishing (c) IOM Communications Ltd
a 840–2000 mm; b 354–420 mm; c 62–74 mm; d <37 mm
21 Microstructure of nitrogen atomised Rex25 high speed steel for various particle size ranges, SEM91
carbides also characterises the powder micro-
structure.18,91,93,95–97 However, for cooling rates of
the order of 108 K s−1, as in the case of splat cooled
material (splat caps encasing some powder particles,
for instance), only supersaturated solid solution
occurs as a solidification product; this is interpreted
as a result of diffusionless solidification.18,97,98
The rapid solidification sequence of high speed
steel powders is not clearly described in the literature,
the available data allowing one to propose only an
overall picture. This is probably a consequence of the
difficulty of observing the fine scale microstructure
(the identification of the solidification products is
performed through X-ray diffraction techniques), as
well as of the impossibility of performing classical
solidification experiments, such as quenched direc-
tional solidification and thermal analysis.
Body centred cubic phases – martensite or delta
ferrite – and fcc phase – austenite – are identified in
the dendrites (or cells). The presence and the origin
of these phases depend on the chemical composition
of the high speed steel, mainly on its carbon content.
For carbon contents higher than roughly 1·3% (P/M
grades of high speed steels, like ASP30 and T15, for
instance), primary austenite is partially transformed
to martensite during cooling, thus leading to a matrix
with retained austenite and martensite. For carbon
contents lower than that (conventional T1 and M2
steels), delta ferrite is the primary phase and the
formation of peritectic austenite is partially sup-
pressed due to the high cooling rate; the resultant
matrix has delta ferrite, retained peritectic austenite,
and martensite.91,93,95–97
International Materials Reviews 2001 Vol. 46 No. 2
The remaining interdendritic (or intercellular)
liquid decomposes through eutectic reactions
(Lc+carbides) and the resultant eutectic austenite
is partially transformed to martensite. In the case of
the P/M grades, the MC and M C eutectics precipi-
2
tate, owing to the high carbon and vanadium contents
and to the high cooling rate.91,93,95 The total volume
fraction of the eutectic carbides is insensitive to
particle size (i.e. to cooling rate), but the volume
fraction of the MC carbide increases with decreasing
particle size at the expense of the M C carbide.93,95
2
For the conventional M2 and T1 steels, however, the
literature shows conflicting data: M C or M C, mixed
2 6
or not, were identified in powder particles solidified
under similar cooling rates.90,96,97 After high temper-
ature consolidation and final heat treatments of the
material, all metastable phases disappear, yielding an
isotropic and very fine distribution of carbides in a
martensitic matrix.
Laser and electron beam surface modification
Lasers and electron beams are direct sources of high
intensity heat that can be used to heat treat, remelt
(glaze), or modify the chemistry of tool steel surface
layers.1 Laser remelting of surface layers of semifin-
ished high speed steel parts has been studied since
1969.99 The only further modifications possible on
those layers are achieved by heat treatment, as it is
not possible to work them mechanically. Adjusting
the power density and the traverse speed of contin-
uous wave high power carbon dioxide lasers, it is
possible to obtain different molten pool temperatures
and shapes, resulting in cooling rates ranging from
 Boccalini and Goldenstein Solidification of high speed steels
a step process’, in which alloying elements are added
Published by Maney Publishing (c) IOM Communications Ltd
(b)
22 a Microstructures of powder particles with
dendritic and cellular morphologies
(53–63 mm), SEM and b relation between
particle size and fraction of particles with
cellular structure (data from Ref. 92)
103 to 106 K s−1 (Refs. 1, 100, 101), with little or no
cracking and porosity formation.
For the higher cooling rates, cellular/dendritic mor-
phologies are obtained, with dendrite arm spacing
between 0·5 and 3 mm. Increasing the cooling rate
leads to a substantial increase in the amount of
retained austenite and an increase of up to 130 HV
on the final hardness (after triple tempering of M42
steel solidified with 106 K s−1). This is due to the
increased solubility of carbon and carbide forming
elements in the austenite.100 The amount of interden-
dritic eutectic microconstituent is also greatly reduced
and its structure refined.
Similarly, for electron beam remelted M2 steels the
amount of eutectic microconstituent decreases with
increasing cooling rate, until it disappears altogether
for cooling rates in excess of 106 K s−1 (Ref. 102)
resulting in intense segregation of substitutional
alloying elements on dendrites or cells and large
109
amounts of retained austenite. It is noteworthy that
up to those extreme cooling rates the microstructure,
though very fine, is morphologically similar to the
conventional ingot structure.
Laser remelting can also be used to create a surface
layer with a different composition than the substrate.
This technique, known as ‘laser alloying’, can be used
to create a layer of high speed steel on the surface of
a lean alloy substrate100 or to create an alloy or
carbide enriched layer on the surface of a conventional
high speed steel part.103 A variety of alloying addition
techniques have been used, based on the replacement
of alloy additions, using foils, powder cored strips,
plasma spray powders, and pack boronising.1,100
Recent papers have concentrated on the simpler ‘one
to the molten pool using an inert gas blow feeder, the
inert gas helping to protect against oxidation of the
alloy.101 Alloy additions of Co (Ref. 101), vanadium
and chromium borides,100 and various alloy carbides
(vanadium, titanium, niobium and molybdenum car-
bides) have been successfully demonstrated with this
technique.104
The main goal of those experiments was to increase
the carbide volume fraction at the cutting edge of
high speed steel tools without incurring the precipi-
tation of coarse proeutectic and eutectic carbides that
deteriorate the toughness and hot workability of the
steel. The high temperatures achieved by irradiation
heating allows the complete dissolution of MC (Nb,
Ti, and V) and M C (Mo) carbide powder additions
2
on the molten pool, while the high cooling rate results
in a fine solidification structure. Figure 23 shows the
microstructure maps obtained by laser alloying high
speed steel M2 with up to 20%Nb, up to 45%V, and
up to 50%Mo; all microstructure maps correspond
to conditions where the average cooling rate is
6×103 K s−1 (Ref. 105). Additions of more than
2 wt-%Nb and corresponding carbon to ingot cast
high speed steels promotes the formation of coarse
MC carbides directly from the melt. The micro-
structure map in Fig. 23a shows that for this cooling
rate (6×103 K s−1) it is possible to add up to
4 wt-%Nb to the molten pool without precipitation
of proeutectic carbides. In a similar fashion, Fig. 23b
and c show that it is possible to add up to 7 wt-%V
and 30 wt-%Mo by laser alloying.103,105
Although there are reports in the literature that
laser alloyed high speed steel cutting tools can
increase life and wear resistance101,106 and that the
process is considered stable, there is no commercial
utilisation of this process at the moment.
High speed steels for rolling mill rolls
In most cases, work rolls for the finishing stands of
hot strip mills are composite components made of an
outer shell of centrifugally cast wear resistant material
and a core of ductile iron or steel (centrifugally or
statically cast). The development of materials for the
outer shell has enjoyed rapid advances beginning in
the early 1980s, culminating in the application of cast
alloys of the Fe–C–Cr–W–Mo–V system, which
replaced high chromium cast iron and Ni-hard cast
iron with better performance.107–109 These alloys are
International Materials Reviews 2001 Vol. 46 No. 2
 110 Boccalini and Goldenstein
100
50
(a) 0
0 M2 4 8
Niobium content, wt-%
100
Volume fraction (%)
50
Published by Maney Publishing (c) IOM Communications Ltd
(b)0
0 M2 10 20
Vanadium content, wt-%
100
50
(c) 0
0 M2 10
Molybdenum content, wt-%
23 Microstructure maps obtained by laser alloying
high speed steel M2 with a NbC, b VC, and
c Mo C: average cooling rate 6×103 K s−1
2
(adapted from Ref. 105)
generically termed high speed steels or, more rarely,
multicomponent white cast iron.110,111
The idea of using these alloys for the manufacture
of work rolls for hot strip mills resulted from an
insight into the requirements involved in this type of
application: fundamentally, the capacity to retain a
high level of hardness even when submitted to high
temperatures, and also wear resistance. Both are
fulfilled by the classical high speed steels.
The alloy design of the high speed steels for rolls
is based on the composition of the M2 steel, the main
changes being the higher carbon and vanadium con-
tents. Thus, though the roll makers have developed
alloys specifically designed to the operational con-
ditions of each hot strip mill plant, their chemical
compositions often fall into the following ranges:
1·5–2·0%C; up to 5%W; up to 5%Mo; 3–7%Cr,
and 4–8%V.
Figure 24 shows the calculated isoplet Fe–5Cr–
5W–5Mo–5V–C (wt-%), a typical composition inside
International Materials Reviews 2001 Vol. 46 No. 2
Solidification of high speed steels
Temperature, °C
12 16 20
Carbon content, wt-%
24 Isopleth Fe–5Cr–5W–5Mo–5V–C (wt-%)
calculated using ThermoCalc112
30 40 those ranges.112 For carbon levels between 1·7 and
2·0%, the solidification sequence can be described by
the following reactions:
(i) primary crystallisation of austenite
(ii) eutectic decomposition of residual interden-
dritic liquid: liquidaustenite+carbides; this
reaction moves down a eutectic trough with
continuously changing phase compositions, so
as to form different eutectic carbides according
to the decreasing temperature.
The high carbon content leads to the primary crystal-
lisation of austenite and makes the composition of
the high speed steels for rolls less hypoeutectic than
20 30 compositions of the high speed steels for tools ( lower
volume fraction of proeutectic phase).
The residual interdendritic liquid decomposes
through different eutectic reactions as it moves down
a eutectic trough, leading to the formation of up to
three eutectics: c+MC; c+M C or c+M C; and
6 2
c+M C . Although the isoplet in Fig. 24 predicts
7 3
the formation of the c+M C eutectic, this eutectic
6
seldom forms since the solidification normally occurs
under non-equilibrium conditions and the formation
of the metastable c+M C eutectic is favoured.
2
The c+MC eutectic always precipitates first, owing
to the high vanadium content of these alloys. The
sequence by which c+M C and c+M C precipitate
2 7 3
depends on the overall chemical composition, the
former being favoured by high W, Mo, or V contents
and the latter by high Cr or C contents.
The total volume fraction of eutectic carbides and
the volume fraction of each different eutectic carbide
depend mainly on the chemical composition, the effect
of the cooling rate being less significant.111,113,114
Table 9 gives the volume fraction of the MC, M C ,
7 3
and M C eutectic carbides in high speed steels for
2
rolls. The total volume fraction of eutectic carbides
ranges from 9% to more than 15%, which is one of
their main characteristics (as cast conventional high
speed steels for tools have up to 8%).
 Boccalini and Goldenstein Solidification of high speed steels 111
Published by Maney Publishing (c) IOM Communications Ltd

a Fe–5Cr–5W–5Mo–5V–5Co–C; b Fe–5Cr–2W–2Mo–5V–5Co–C; c Fe–5Cr–2W–2Mo–8V–5Co–C

25 Characteristics of eutectic carbides in microstructure of high speed steels for rolls (by kind permission
of Sasaguri et al.111)

Figure 25 shows the features of the eutectic carbides
in the microstructure of typical high speed steels for
rolls with carbon contents ranging from 1·7 to 2·9%.
The morphology of the eutectic carbides, mainly
MC and M C carbides, are influenced by the chemical
2
composition.110,111 Figure 26 shows the relationship
between the contents of the main alloying elements –
C, V, Mo, and W – and the morphologies of the
eutectic carbides.
The as cast microstructure has a matrix with prod-
ucts of austenite decomposition (normally martensite
or bainite), globular secondary carbides precipitated
in the matrix and eutectic carbides distributed both
in the interdendritic or intercellular regions (M C,
2
M C , and nodular or flaky MC) and inside the
7 3
eutectic cells (coral-like MC). Figure 27 shows
Table 9 Volume fractions of MC, M C , and M C
7 3 2
eutectic carbides in high speed steel for
rolls111,113
the overall microstructure of a typical high speed
steel for rolls (Fe–5V–5Cr–5Mo–5W–2C).112
The performance of work rolls for hot strip mills
is basically evaluated by the extent of their life and
by the surface quality of the rolled products, both
related to properties of the roll, including strength,
toughness, and wear resistance. Work roll wear is a
process characterised by the simultaneous operation
of several wear mechanisms, such as abrasion,
oxidation, adhesion (‘sticking’), and thermal
fatigue.114–117 These properties are determined by the
microstructural characteristics of the high speed steels
for rolls, principal ones being the type, morphology,
size, volume fraction, and distribution of the eutectic
carbides.114,118–122 Research on the development of
these steels for rolls has thus been focused on the
improvement of the microstructure by means of alloy
design113,114,118–122 and by improvements in the
manufacture process.107,123–127

Chemical composition
Volume fraction, % Summary
(main elements), wt-% MC M C *
7 3
M C
2
This review has mainly examined the effects of the
1·95C–5·1V–1·6W–1·7Mo–5·5Cr 4·4 4·7 ... chemical composition and cooling rate on the devel-
2C–6·1V–1·8W–1·1Mo–5·6Cr 8·2 3·2 ... opment of the as solidified microstructure of high
2C–4V–1·8W–2·5Mo–5·5Cr 4·4 6·9 ... speed steels. The major observations are as follows.
2C–4·3V–6·5W–4·1Mo–5·6Cr 4·9 ... 10·6 1. The as cast microstructure of high speed steels,
* Small amounts of the M C eutectic carbide are included.
2
despite the diversity of compositions resulting from
International Materials Reviews 2001 Vol. 46 No. 2
 112 Boccalini and Goldenstein Solidification of high speed steels
(a)
Carbon content, wt-%
Vanadium content, wt-%
(b)
Carbon content, wt-%
Published by Maney Publishing (c) IOM Communications Ltd
Tungsten equivalent (Weq), wt-%
a Fe–5Cr–2W–2Mo–5Co–V–C; b Fe–5Cr–5V–5Co–W–Mo–C
# divorced MC; C irregular MC;
irregular M C; & complex-regular M C; c complex-regular M C
2 2
26 Influence of V and C contents and W on type
eq
and morphology of eutectic carbides (adapted
from Ref. 111)
27 Overall microstructure of high speed steel
for rolls (Fe–5Cr–5W–5Mo–5V–2C): etching,
electrolytic Cr O (Ref. 112)
2 3
alloy design development, maintains its basic features:
dendrites surrounded by a more or less continuous
interdendritic network of eutectic carbides. The solidi-
fication sequence of the high speed steels can be
qualitatively described by isoplets of the Murakami–
Hata–Kuo quaternary Fe–W–Cr–C diagram, or more
accurately by isoplets of the Fe–Cr–W–Mo–V–C
diagram calculated using Calphad’s thermodynamic
International Materials Reviews 2001 Vol. 46 No. 2
data base. The solidification sequence may be
described by the following reactions: (a) precipitation
of primary delta ferrite from liquid; ( b) formation of
austenite through peritectic reaction; (c) formation of
eutectic carbides through eutectic reaction; and
(d) solid state delta eutectoid transformation, which
occurs at the end of the solidification.
2. The dendrite arm spacing (DAS) can be related
to the cooling rate by the classical power law
expression DAS=A(dT /dt)−b, with 32∏A∏43 and
−0·28∏b∏−0·38. Small changes in carbon level
cause significant changes in parameter A, while
changes in vanadium content have little effect on A.
On the other hand, there are conflicting data about
the effect of the peritectic reaction on DAS.
3. The residual interdendritic liquid decomposes
through different eutectic reactions, leading to the
formation of up to three eutectic carbides: M C,
2
M C, and MC.
6
The morphology of the M C eutectic carbide is
6
characterised by the presence of a central platelet
from which arise secondary platelets (termed a fish-
bone morphology). This morphology is not influenced
by chemical composition nor cooling rate, except that
at faster cooling rates the distance between platelets
is decreased.
The morphology of the M C eutectic is substan-
2
tially influenced by chemical composition, by cooling
rate and also by minor additions of some elements.
In order to unify the many, often conflicting, different
complex-regular MC; % morphological classifications found in the literature,
7 3 a new terminology based on Crocker’s general
description of eutectics is proposed: M C eutectic can
2
be either irregular or complex regular. In the former,
M C carbide has a platelike morphology and a
2
tendency to assemble as radiating clusters (often
referred to as ‘fanlike’ or ‘feathery’), with extensive
branching. Complex-regular M C eutectic shows a
2
regular distribution of the M C carbide over small
2
areas, forming cells with macrofacets. Complex-
regular M C eutectic grows as a spiral eutectic with
2
degenerate or interrupted spirals. A third type of
morphology characterised by the presence of short
rod shape M C carbides, associated with faster cool-
2
ing rates, was shown to be a ‘broken complex-regular’
morphology coming from frequent interruption of the
spiral growth. The formation of the M C eutectic is
2
a metastable reaction, not present in the equilibrium
diagram of most commercial compositions. The M C
2
carbide decomposes into MC and M C when heated
6
to between 900 and 1150°C.
Three types of MC eutectic can be distinguished:
divorced (MC carbide occurs as isolated massive
crystals), irregular (poorly coupled eutectic in which
the MC carbide has a petallike morphology), and
complex regular (coupled eutectic in which MC car-
bide has a branched petallike morphology). In the
literature, the latter has been improperly classified as
regular or lamellar. The difference between the peri-
tectic temperature and the temperature of eutectic
formation determines the prevailing MC eutectic type:
the larger this difference, the greater the tendency for
the formation of the divorced eutectic; consequently,
the higher the cooling rate, the higher the tendency
to the formation of less coupled MC eutectic.
 Boccalini and Goldenstein Solidification of high speed steels
4. Generally, the total eutectic volume fraction
increases with increasing C content (mainly), increas-
ing W/Mo ratio, increasing V content, and decreasing
cooling rate. The volume fraction of each of the three
eutectic types is determined by competition between
them. M C eutectic is favoured with respect to M C
2 6
eutectic by increasing the Mo/W ratio, increasing the
V content or the C content and by increasing the
cooling rate (an interdependence of these variables
must be considered in determining quantitative limits
of their influence on eutectic formation). MC eutectic
is favoured with respect to M C or M C eutectic by
6 2
increasing the V content. Increasing overall V content
makes the residual liquid from the peritectic reaction
richer in V, increasing the driving force for MC
eutectic formation, as well as increasing both the
temperature of its formation and its volume fraction;
consequently, the fractions of the M C or M C
6 2
eutectics decrease. Besides the effects of the nominal
chemical alloy composition, the competition between
eutectics is affected by the presence of minor alloying
elements. The formation of M C eutectic is favoured
6
by silicon, niobium, rare earth metals, and yttrium at
Published by Maney Publishing (c) IOM Communications Ltd
the expense of M C eutectic, while the opposite
2
happens for additions of nitrogen, calcium, and alu-
minium. The order in which the precipitation of M C,
2 2
M C, and MC eutectics occurs is a function of the
6
segregation profiles resulting from the ferrite and
austenite formation and even of the eutectic decom-
position reactions themselves. Those profiles are also
influenced by the overall chemical composition and
cooling rate.
5. The as cast morphology can be modified or
controlled during processing, by refining the grain
size or the dendritic structure, as well as through
modifying or refining the morphologies of the eutec-
tics. The first goal is achieved mainly by controlling
the cooling rate and the thermal gradient in the liquid
during solidification. For large sized ingots, remelting
consumable electrodes (ESR) can produce directional
structures containing large oriented columnar grains
characterised by the uniformity of the size of eutectic
cells throughout the cross-section of the remelted
ingot. Refining elements are used also, mostly to
decrease the macroscopic grain size (dendrites of
primary ferrite), notably through the formation of a
higher fraction of equiaxed grains. The decrease in
the size of the dendrites of primary ferrite increases
the interdendritic solid/liquid interface and, therefore,
for a given volume fraction of this liquid, the colonies
resulting from its eutectic decomposition are more
homogeneously distributed in the microstructure.
6. The morphologies of the M C and MC eutectics
2
are influenced by the presence of minor elements.
Additions of nitrogen and calcium promote the forma-
tion of the complex-regular M C eutectic while
2
additions of aluminium give rise to the formation of
the irregular M C eutectic. Applications of the con-
2
cept of modification of the morphology of the carbides
through the addition of surface active elements or
elements with low solubility in the eutectic phases
have been attempted with some elements, including
Bi, Ca, Sr, and Sb. The addition of rare earth metals
diminished the temperature of formation of MC eutec-
tic, transforming it from irregular to complex regular,
113
while the MC eutectic precipitated along with the
M C eutectic or the M C eutectic, forming a duplex
2 6
eutectic. Most of the morphology-modifying effect of
these solutes can be explained by the changes in
eutectic growth brought about by the build-up of
modifying elements along the eutectic carbide/liquid
interface.
7. Except for its smaller scale, the solidified micro-
structure of high speed steel powders is not signifi-
cantly different from that of more slowly cooled high
speed steel ingots. Thus, the normal microsegregated
structure constituted by solid solution dendrites (or
cells) delineated by a more or less continuous inter-
dendritic (or intercellular) network of carbides also
characterises the powder microstructure. Depending
on the particle size, the morphology of the micro-
structure of the powder particles is dendritic, cellular,
or a mixture of both, the fraction of cellular morph-
ology being significant only in the very small-sized
particles. Body centred cubic phases (martensite or
delta ferrite) and the fcc phase (austenite) are identified
in the dendrites or cells. In the interdendritic or
intercellular regions, owing to the high carbon and
vanadium contents of the P/M high speed steel grades
and to the high cooling rate of the atomisation
process, the MC and M C eutectics precipitate.
8. Laser and electron beams can be used to remelt
(glaze) or modify the chemistry of tool steel surface
layers, melting beads or pools at the surface of
semifinished parts which are subsequently self-
quenched, resulting in solidification at cooling rates
ranging from 103 to 106 K s−1. For the higher cooling
rates cellular/dendritic morphologies are obtained,
with DAS between 0·5 and 3 mm. Increasing the
cooling rate leads to a substantial increase on the
amount of retained austenite and an increase of up
to 130 HV on the final hardness after tempering. This
is due to the increased solubility of carbon and
carbide forming elements in the austenite. The amount
of interdendritic eutectic microconstituent is also
greatly reduced and its structure refined. Laser remelt-
ing can also be combined with laser alloying through
the addition of alloying elements to the molten pool,
using a powder inert gas blow feeder. Laser alloying
can be used to create a surface layer of high speed
steel on the surface of a lean alloy substrate or to
create an alloy or carbide-enriched layer on the
surface of a conventional high speed steel part. The
increased solubility allows the addition of carbide-
forming alloying elements and carbon at levels that
would embrittle ingot cast materials. As an example,
it is possible to add up to 4 wt-%Nb, 7 wt-%V, or
30 wt-%Mo to the molten pool, by laser alloying
with a cooling rate of 6×103 K s−1, without precipi-
tation of proeutectic carbides.
9. A recent development makes use of cast highly
alloyed high speed steels as the material for the outer
shell of centrifugally cast or continuously poured
composite hot strip mill rolls. These materials were
developed with compositions near that of M2 tool
steel, but with higher carbon and vanadium contents.
For carbon levels between 1·7% and 2·0%, their
solidification sequence can be described by the follow-
ing reactions: (a) primary crystallisation of austenite;
( b) eutectic decomposition of residual interdendritic
International Materials Reviews 2001 Vol. 46 No. 2
 114 Boccalini and Goldenstein Solidification of high speed steels
liquid, leading to the formation of up to three eutec-
tics: c+MC; c+M C or c+M C; and c+M C .
6 2 7 3
The total volume fraction of eutectic carbides in high
speed steels designed for rolls ranges from 9% to
more than 15%, which is one of their main
characteristics.
Acknowledgements
The authors acknowledge support from the Fundação
de Amparo à Pesquisa do Estado de São Paulo
(FAPESP), through Grant 95/0251–3. Professor
Goldenstein acknowledges partial support of his
research by the Conselho Nacional de Pesquisas
through Grant 30.0288/80–8.
References
1. . , . , and . : ‘Tool steels’, 5th edn;
1998, Materials Park, OH, ASM International.
2. . : ‘High speed steels’; 1988, Butterworth, Cambridge.
3. . . : T rans. AIME, 1996, 236, 950–963.
4. . , . , . , and . : Steel
Published by Maney Publishing (c) IOM Communications Ltd
Res., 1987, 58, (8), 339–352.
5. .  and . . : Met. Mater., May 1968, 2, 144–155.
6. . . : in ‘Towards improved performance of tool mater-
ials’, 45–51; 1982, London, The Metals Society.
7. . , . , and . . : in ‘Towards
improved performance of tool materials’, 37–42; 1982,
London, The Metals Society.
8. . .  and . . . : ‘Niobium in high speed
tool steels’, Report 319/4, Fulmer Research Institute, Stoke
Poges, 1971.
9. . , . , and . : Z. Metallkd., 1983,
74, 199–205.
10. . . : ‘Stahlschlüssel’; 1995, Marbach, Germany, Verlag
Stahlschlüssel Wegst GmbH.
11. . . . , . . , . . , . . . ,
and . : ‘Hard alloys’, US Pat. 4 2244 060, Villares
Metals, 23 Sept. 1980.
12. .  and . : Scand. J. Metall., 1973, 2,
(2), 83–86.
13. . : Metall. Rev., 1964, 9, (33), 49–91.
14. . .  and . : in ‘Solidification and
casting of metals’, 363–370; 1979, London, The Metals Society.
15. . .  and . : Br. Foundryman, July 1985,
78, 289–296.
16. .  and . : Hutnik, 1968, 18, (4), 165–167.
17. . , . , and . : Berg Hüttenmänn.
Monatsh., 1968, 113, (3), 135–141.
18. . . , . . , and . : Powder
Metall., 1982, 25, (1), 1–9.
19. . .  and . : in ‘ASM handbook’, 10th edn,
Vol. 1, 780–792; 1990, Materials Park, OH, ASM
International.
20. . : Proc. 4th Int. Tooling Conf. ‘Progress in tool steels’,
Bochum, Germany, September 1996, Ruhr University, 27–36.
21. . , . , and . : Proc. 4th
Int. Tooling Conf. ‘Progress in tool steels’, Bochum, Germany,
September 1996, Ruhr University, 3–16.
22. . , . , . , . , and . -
: Proc. 4th Int. Tooling Conf. ‘Progress in tool steels’,
Bochum, Germany, September 1996, Ruhr University,
397–409.
23. .  and . : Sci. Rep. T ôhoku Imp. Univ.,
1st Ser., 1936, 882–895; cited in Ref. 24.
24. . : J. Iron Steel Inst., 1955, 181, 128–134.
25. .  and . : J. Iron Steel Inst., 1959, 193, 254–269.
26. . . , . . , and . . : Metall.
T rans., 1972, 3, 919–926.
27. ‘ASM handbook’, 10th edn, Vol. 3, ‘Alloy phase diagrams’,
3–31, 3–35; 1997, Materials Park, OH, ASM International.
28. . .  and . . : Metall. T rans., 1974, 5, 1727–1733.
29. .  and . : Scand. J. Metall., 1976, 5,
268–275.
International Materials Reviews 2001 Vol. 46 No. 2
30. . c, . . , and . . : Metall.
T rans. A, 1977, 8A, 1787–1792.
31. .  and . : Scand. J. Metall., 1979, 8,
243–253.
32. .  and . : DEW T ech. Ber., 1971, 11, 147–154;
cited in Ref. 33.
33. . . , . , and . : Metall. T rans. A,
1989, 20A, 2133–2148.
34. .  and . . : Steel Res., 1992, 63,
(8), 354–360.
35. . .  and . . : Z. Metallkd., 1993,
84, 557–562.
36. . .  and . . : Z. Metallkd., 1993,
84, 860–866.
37. .  and . : DEW T ech. Ber., 1971, 11,
139–146.
38. . , . , and . : Prakt. Metallogr., 1986,
23, 374–398.
39. ‘A guide to the solidification of steels’, 134; 1977, Stockholm,
Jernkontoret.
40. . : Met. Sci., 1976, 10, 77–86.
41. . : Scand. J. Metall., 1976, 5, 27–32.
42. .  and . : Met. Sci., 1982, 16,
575–585.
43. . , . , . , and . : T rans. Iron
Steel Inst. Jpn, 1974, 14, 257–266.
44. .  and . . : ‘Fundamentals of solidification’;
1984, Aedermannsdorf, Switzerland, Trans Tech Publications.
45. . : PhD thesis, Escola Politécnica da Universidade
de São Paulo, São Paulo, Brazil, 1996.
46. .  and . : Met. Sci., 1982, 16, 283–294.
47. . . , . . , . . ,
and . . : Met. Sci. Heat T reat. (USSR),
Nov.–Dec. 1982, 24, 777–784.
48. . . , . . , . . , and
. . : Met. Sci. Heat T reat. (USSR), Sept.–Oct.
1987, 29, 643–648.
49. . , . , and . : Mater. Charact., 1992,
29, 15–24.
50. . , . , and . . : 109th AIME
Annual Meeting, Symp. on ‘Processing and properties of high
speed tool steels’, Las Vegas, NV, February 1980, 1–18.
51. . . , . . , . . ,
. . , and . . : Met. Sci. Heat. T reat.
(USSR), Sept.–Oct. 1979, 21, 791–795.
52. . , . , . . , and . :
‘CBMM Charles Hatchett award papers’; 1986, London, The
Institute of Metals.
53. . . , . . , . . ,
. . , and . . : Steel USSR, May
1981, 11, 296–298.
54. . .  and . . : Met. Sci. Heat T reat. (USSR),
1989, 31, (1–2), 121–128.
55. .  and . : 109th AIME Annual
Meeting, Symp. on ‘Processing and properties of high speed
tool steels’, Las Vegas, NV, February 1980, 91–102.
56. . , . , and . : T rans. AFS,
1996, 104, 907–915.
57. . . , . . , and . . : Proc. R. Soc.
(L ondon) A, 1973, A335, 15–37.
58. . , . . . , and . : Acta
Microsc., Sept. 1997, 6, Suppl. A, 172–173.
59. . , . , . , and . : Proc.
4th Int. Tooling Conf. ‘Progress in tool steels’, Ruhr
University, Bochum, Germany, September 1996, 17–26.
60. . , . , and . : Scand. J. Metall.,
1979, 8, 115–122.
61. . . . : MSc thesis, Escola Politécnica da Universidade
de São Paulo, São Paulo, Brazil, 1982.
62. . . : PhD thesis, Max-Planck-Institut für
Metallforschung, Stuttgart, Germany, 1997.
63. . : Scand. J. Metall., 1973, 2, 233–241.
64. . , . , and . : Metall. T rans. A, 1993, 24A,
1265–1272.
65. . , . . . , and . : Mater.
Sci. T echnol., 1999, 15, 621–626.
66. u. : ‘Tool steels’; 1978, Moscow, Mir Publishers.
67. .  and . . : Neue Hütte, 1968, 13, (5),
268–278.
 Boccalini and Goldenstein Solidification of high speed steels
68. . . : Proc. Int. Conf. on ‘Cutting tool materials’, Fort
Mitchell, KY, September 1980, ASM, 133–158.
69. . .  and . . : Proc. Int. Conf. on
‘Cutting tool materials’, Fort Mitchell, KY, September 1980,
ASM, 111–131.
70. . . , . . , . . , and . . : J. Iron
Steel Inst., 1971, 209, 606–619.
71. . .  and . : ‘Electro-slag refining’; 1969,
London, Chapman and Hall.
72. . .  and . : Proc. 1st Int. Steel Foundry
Congr., Chicago, IL, 11–13 November, 1985, Steel Founders’
Society of America, 173–182.
73. .  and . : Denki Seiko (Electr. Furn. Steel),
1986, 57, (3), 161–169.
74. . . , . . , . . , . . ,
. . , and . . : Met. Sci. Heat T reat.
(USSR), Nov.–Dec. 1981, 766–770.
75. . : Proc. 6th Int. Iron and Steel Congr., Nagoya,
Japan, 1990, ISIJ, 721–728.
76. . , . , and . . : Arch.
Eisenhüttenwesen, 1965, 10, 725–736.
77. . , . , . , and . : Prakt.
Metallogr., 1987, 18, 467–478.
78. . , . . , and . . : T he Foundryman,
Feb. 1991, 84, 66–70.
79. h. , . . . , and h. : Mater.
Sci. T echnol., 1998, 14, 683–688.
Published by Maney Publishing (c) IOM Communications Ltd
80. . . : ‘Grain refinement in castings and welds’,
Warrendale, PA, 1983, TMS–AIME, 3–50.
81. .-. , . , . , and . :
Härt-T ech. Mitt., 1989, 44, 67–73.
82. . . : Manuf. Eng., June 1980, 83, 96–99.
83. .  and . : Proc. 6th Int. Iron Steel Cong.,
Nagoya, Japan, 1990, ISIJ, 649–656.
84. .-. , . , and . . : J. Mater. Sci., 1994, 29,
1083–1089.
85. . .  and . . : Cast Met., 1991, 4, (2), 83–88.
86. .  and . . : Powder Metall., 1978, 21, (2), 114–123.
87. . : Met. Powder Rep., 1983, 38, (4), 185–194.
88. .  and . . -: Prakt. Metallogr.,
1977, 14, 181–200.
89. . , . , . , and . : Powder
Metall., 1981, 24, (4), 196–202.
90. . . , . , . , and . . :
Met. Sci. Heat T reat. (USSR), 1984, 26, (3), 204–209.
91. . : ‘Atomization: the production of metal powders’;
1992, Princeton, NJ, Metal Powder Industries Federation.
92. .  and . . : Int. J. Powder Metall.,
1994, 30, (3), 301–311.
93. . , . , . . , and . : Metall. Mater. T rans. A,
1988, 29A, 1395–1404.
94. .  and . . : Metall. T rans. A, 1988,
19A, 1395–1401.
95. . . , . , and . . : Metall. T rans. A,
1991, 22A, 2733–2745.
96. . : Powder Metall., 1998, 41, (4), 269–273.
97. . .  and . : Metall. T rans. A, 1981, 12A,
1557–1568.
98. . .  and . . . : Met. Sci., 1979, 13,
269–279.
99. . . : J. Eng. Ind., August 1969; cited in Ref. 100.
100. . . , . , . . , and . : in
‘Heat treatment 87’, 189–195; 1988, London, The Institute
of Metals.
115
101. . : Proc. ‘Euromat ’91’, 2nd Eur. Conf. on ‘Advanced
materials and processes’, Cambridge, UK, July 1991,
Federation of European Materials Societies, 115–123.
102. . , . , . , . , . ,
and . : Prakt. Metallogr., 1988, 25, 465–487.
103. . , . , . , . , and
. : Proc. 4th Int. Conf. on ‘Tooling progress in
tool steels’, Bochum, Germany, September 1996, Ruhr
University, 397–409.
104. . , . , and . : Scr. Metall. Mater., 1994,
30, 253–258.
105. . , . , and . : Z. Metallkd., 1992,
83, 368–378.
106. .-.  and . . : Wear, 1988, 127, 253–268.
107. . , . , . , . , and
. : Nippon Steel T ech. Rep., July 1995, (66), 82–90.
108. . , . , and . : ISIJ Int., 1992, 32, (11),
1194–1201.
109. . , . , . , . , and . :
Proc. 38th Mechanical Working Steel Processing Conf., 1997,
34, 59–64.
110. . , . , . , . , and
. . : Proc. 1st Congr. on ‘Metallurgy and materials
technology’, São Paulo, Brazil, October 1994, Associação
Brasileira de Metalurgia e Materiais, Vol. 5, 1–10.
111. . , . , . , and . : T rans.
AFS, 1996, 104, 103–108.
112. . , . , and . : Proc. 37th
Rolling Seminar ‘Processes and rolled and coated products’,
Curitiba, Brazil, September 2000, Associação Brasileira de
Metalurgia e Materiais–ABM, 587–596.
113. . . , . , and . . : Mater. Sci. Eng. A, 1998,
254, 282–295.
114. . . , . . , and . : Metall. Mater. T rans. A,
1999, 30A, 399–409.
115. . , . . , . , and . :
Wear, 1994, 176, 261–271.
116. . : Rev. Métall., Aug.–Sept. 1986, 643–656.
117. . . , . , . . , and . . : ISIJ Int., 1992,
32, (11), 1221–1223.
118. . , . . , . . , . , and
. : Mém. Études Sci. Rev. Métall., June 1991, 88,
356–363.
119. . , . , . . , and . . : Metall. Mater.
T rans. A, 1997, 28A, 123–134.
120. . , . , . , . , and
. : Proc. 7th Int. Conf. on ‘Steel rolling’, Chiba, Japan,
1998, Iron and Steel Institute of Japan, 423–428.
121. . . , . , and . . : Mater. Sci. Eng. A, 1998,
254, 296–304.
122. . , . , . . , and . : Metall. Mater. T rans. A,
1999, 30A, 234–243.
123. . , . , . , and . : Sumitomo
Search, 1993, (51), 72–79.
124. . , . , . , . , and . :
Proc. 7th Int. Conf. on ‘Steel rolling’, Chiba, Japan, 1998,
Iron and Steel Institute of Japan, 411–416.
125. . . , . . , . . , . , and
. . : Proc. 4th Decennial Int. Conf. on
‘Solidification processing’, Sheffield, UK, July 1997, University
of Sheffield, 644–648.
126. . .  and . . : Proc. 7th Int. Conf. on ‘Steel
rolling’, Chiba, Japan, 1998, Iron and Steel Institute of
Japan, 429–434.
127. . , . , . , . , . . ,
. . , and . . : ISIJ Int., 1990, 30, (9), 756–763.
International Materials Reviews 2001 Vol. 46 No. 2