REVIEW 126

by S. Floreen

The physical metallurgy of

marag ing steels

MARAGINGSTEELSARE A CLASSof high-strength steels The 200/0nickel steel, Alloy B, has an Ms temperature
characterised by very low carbon contents and the use of the order of 2000 C (4000 F) and completely transforms
of substitutional elements to produce age-hardening in to martensite after solution-annealing. In this case
iron-nickel martensites. The term 'maraging' was hardening requires only a maraging heat-treatment in
coined from a combination of martensite and age- the 425-4850 C temperature range. The general
hardening. characteristics and properties of both steels are given in
The initial development of these steels was carried Ref. (I).
out by C. G. Bieber of The International Nickel Com- In Ig60, Decker, Eash, and Goldman2 discovered the
pany in the late 1950s. This work resulted in the first very effective hardening produced in Fe-Ni martensites
two grades of maraging steel, the so-called 25 % and by cobalt + molybdenum. A number of alloy compo-
20% nickel steels. The compositions of these steels are sitions based upon this mode of hardening have been
given in Table I as Alloys A and B, respectively. These developed. Alloys C, D, and E in Table I are the so-
alloys depend on a combination of titanium, aluminium, called 18Ni (200), 18Ni (250), and 18 Ni (300) grades of
and niobium to produce age-hardening. maraging steel. These alloys all have an 18% nickel
Alloy A, the 25 % nickel steel, is a semi-austenitic base with varying levels of cobalt, molybdenum, and
composition; that is, after solution-annealing the Ms titanium, to produce nominal strength levels after
temperature is sufficiently low for the alloy to remain maraging of200, 250, and 300 ksi, respectively. Titanium
austenitic upon cooling to room temperature. To is used in these compositions as a supplemental hardener,
achieve transformation to martensite, the alloy is first but the primary strengthening effect comes from the
heat-treated at temperatures of the order of 7000 C combination of cobalt and molybdenum.
(13000 F). This heat-treatment, called 'ausaging' (aus- In addition, several more specialised alloys have been
tenite + age-hardening), precipitates nickel-rich inter- developed. Alloy F is a composition with greater re-
metallic compounds in austenite, and the consequent sistance to austenite reversion which has proved of
removal of alloying elements from solid solution raises
the Ms temperature sufficiently for essentially complete
Table I. Nominal chemical compositions (wt.-%,
bal. Fe) of maraging steels
transformation to martensite to take place when the alloy
is then refrigerated at - 770 C (- 1060F). Alternatively, Alloy Ni Co Mo AI Ti Other
the alloy can also be transformed by cold working at
least 25 % and then refrigerating at - 77° C. In either A 25 0-3 1-4 0'4Nb
case, final hardening is accomplished by maraging at B 20 0·3 1-4 0-4Nb
425-4850 C (800-go00 F). C 18 8 3-2 0-1 0-2
D 18 8 5-0 0-1 0-4
E 18 9 5·0 0-1 0-6
F 15 9 5-0 0-7 0-7
S. Floreen, B.S., M.S., Ph.D_, is with the Paul D. Merica Research G 18 12 4-0 0-1 1-6
Laboratory, The International Nickel Coo, Inc_, Sterling Forest, H 17 10 4-6 0-1 0-3
Suffern, N.Y., U.S_A. I 12 3-0 0-3 0-2 5Cr

METALLURGICAL REVIEWS 115

 Floreen: The physical metallurgy of maraging steels

interest for elevated-temperature magnetic applications.
Alloy G is a recent composition that achieves a 350 ksi
strength level. Alloy H is a cast alloy. Alloy I is the
so-called 12-5-3 maraging steel. This latter alloy differs
from the recent maraging compositions in that it does
not use a combination of cobalt and molybdenum for
hardening; it is intended primarily for applications re-
quiring very high fracture· toughness at strengths of the
order of 180 ksi. In all the alloys the carbon content is
specified as 0 '03 wt.- % maximum. Titanium in most
cases plays a double role of hardener and refining agent
to tie up residual carbon.
The heat-treatments and properties of these alloys
have been described in a number of publications.
Typically the alloys are solution-annealed for I h at
8150 C (1500° F), although other annealing tempera-
tures or multiple annealing heat-treatments are used in
certain cases. After annealing the alloys transform com-
pletely to martensite upon cooling to room temperature.
Because of the high nickel content and the virtual absence
of carbon, hardenability is not a problem and the cooling
rate after annealing is unimportant. In the as-annealed
condition the alloys have a hardness of the order of
Rockwell C 30, and can be readily machined or fabri-
cated. Hardening is then achieved by maraging,
generally using a treatment of 3 h at 485° C (gooOF).
Dimensional changes during age-hardening are very
small, and in many cases completely finished pieces
can be machined in the annealed condition and then
hardened by maraging. The alloys are readily weldable,
and possess resistances to hydrogen embrittlement and
stress-corrosion cracking that generally are superior to
high-strength low-alloy steels. Perhaps their primary
attribute, however, and the main reason for the interest
in them, is the distinctly better toughness they exhibit
at high strength levels.
Following the development of maraging steels a
considerable amount of effort has been devoted to
studying the characteristics of alloys of this kind. Some
of the earlier results have been discussed in previous
reviews,3-7 but a considerable amount of work has been
done since then. This review will be concerned primarily
with the various phase transformations and age-harden-
ing reactions that occur in these alloys. No attempt will
be made to discuss welding, corrosion, machinability, or
other technological considerations. Also no attempt will
austenite will not decompose into the equilibrium aus-
tenite and ferrite compositions, even if held for very
long times in the two-phase region. Instead, with further
cooling, the austenite transforms to martensite that has
a b.c.c. crystal structure. The martensite transformation
temperatures as a function of nickel content are shown
in the metastable equilibrium diagram of Jones and
Pumphrey9 (Fig. 2).
If the martensite is reheated one of two things may
happen. If the alloy is brought to a temperature below
the As temperature (i.e., the start of the ex -*y transforma-
tion shown in Fig. 2) the martensite will decompose into
the equilibrium austenite and ferrite compositions, i.e.
the martensite reverts to the equilibrium structures given
by Fig. I. The rate of this reversion reaction depends
upon the temperature and, fortunately for maraging
steels, the rate at temperatures of the order of 485° C
(gooOF) is slow enough for considerable precipitation-
hardening to be achieved before the reversion reaction
predomina tes.
If, on the other hand, the alloy is heated above the
As temperature, the martensite transforms by a shear
reaction back to an austenite of the same composition.
In practice, it is quite often found, even with relatively
fast heating rates, that some reversion occurs during
heating which influences the subsequent shear reaction.
Fig. 2 indicates only that alloys containing up to ro-.I
33 % nickel will transform martensitically. This diagram
is actually rather oversimplified, because recent work
has revealed that a surprising variety of transformations
can take place in these alloys.l0-16 In the range 0-
100/0 Ni, ferrite will form at slower cooling rates and
martensite at higher cooling rates.I7 Increasing nickel
2500
y
400 a+y
a 500

be made to review precipitation-hardening or maraging \

\

stainless steels. o
o 10 20 30 40 50
WEIGHT PERCENT. NICKEL

[Courtesy Iron Steellnst.

Iron-rich portion of the iron-nickel equilibrium diagram.
(Owen and Liu.8)
Martensite formation
1000
Several basic characteristics of maraging steels are
1500
directly related to the characteristics of the iron-rich end
of the iron-nickel phase diagram. Actually two dia- ;' 600
1000 ~
grams must be considered. The first is the equilibrium w
7g~ ~
diagram, shown in Fig. I, as given by Owen and Liu.s ~
~
400
500 ~w
Recently, some modifications to this diagram have been ~ 200 a.
::!: fj
proposed but these changes need not concern us here. w
l- 0 /0% 0 l-

The essential point to be noted is that at lower tempera~ 90%

tures the· equilibrium phases for alloys containing 3-300/0 -200

10 20 30 40

Ni are ferrite and austenite. WEIGHT PERCENT. NICKEL

In practice it turns out that upon cooling an alloy [Courtesy Iron.Sleellnst.

containing, say, 180/0 nickel from the austenite field the 2 Metastable iron-nickel diagram. (Jones and Pumphrey. g)

116 METALLURGICAL REVIEWS

 Floreen: The physical metallurgy of maraging steels

contents lower the cooling rate necessary to form

martensite, and at ,......, 10% nickel a completely marten~
sitic structure is formed even with very slow cooling.
This particular martensite has generally been called
massive martensite.
Massive martensite is found over a composition range
10-25 % Ni. The composition limits are not exact,
of ,......,
and vary not only with cooling rate but also the anneal-
ing temperature18 and interstitial content.10,12
Beginning at,......, 25% Ni two new martensite mor-
phologies start to form, one containing an internally
twinned structure, and a surface martensite. The reasons
for the formation of these various structures is a matter of
conjecture at this time. Fortunately, however, only the
massive martensite transformation is of importance to the
present story.
Figure 3 is a typical example of the structure of
massive martensite. This consists of a series of elongated
laths or platelets that contain a high density of dis-
locations. Patterson and Wayman19 report that the
dislocations in structures of this kind tend to lie in
[I I I ]b.c.c. directions and are predominantly screw in
nature. The platelet interfaces are usually rather wavy.
At one time it was believed that adjoining platelets
were twin~related, but this does not appear to be gener~
ally true.12 Recent work20 has indicated that the habit
plane is [0.463,0.532, 0.708].r (100 from (111)1')' The
orientation relationship is believed to be Kurdjumov-
Sachs, i.e.,
(111)1'11 (OIl)"
[I 1o] l' II [i I I]"
To date, no formal microscopic theory has been pro~
posed to describe the transformation. Indeed, there is
considerable controversy at present about many aspects
of this martensite transformation.
Structures of the type illustrated in Fig. 3 have been
observed in a variety of iron-base alloys containing both
substitutional and interstitial solutes. In the case of the
maraging-type compositions, Miller and Mitche1l21
have examined the structures of a series of Fe-7%
Co-S % Mo-O'4 % Ti alloys with various nickel contents.
They found that massive martensite was formed with
nickel contents up to 23 %, and twinned martensite at
higher nickel contents. In some instances22 the lack of
visible surface shears on prepolished samples has raised
doubts as to whether the 18 Ni (250) steel always trans~
forms to massive martensite. All the transmission micro~
scopy evidence indicates, however, that-with the ex-
ception of the 25 % nickel steel-all maraging steels
normally have massive martensite matrix structures.
The suggestion has been made that the major factor
determining whether massive or twinned martensite is
formed is the Ms temperature, with lower Ms tempera-
tures favouring the twinned structure. 23 This general
rule is of interest as regards maraging steels, because, as
discussed below, there is some evidence that after ageing
a massive martensite matrix gives better toughness than
a twinned martensite. The desire to avoid forming a
twinned martensite, and also to avoid the presence of
untransformed austenite, places rather definite limits
upon the alloying additions that can be made to an
Fe-Ni base. The alloying elements commonly used
generally lower the Ms temperature, but the effect of
any individual addition often is not constant but depends
METALLURGICAL REVIEWS
[Courtesy Amer. Soc. Test. Mat.
3 Transmission electron micrograph of massive martensite
structure in Fe-21'9 % Ni alloy. (Mihalisin.'°)
upon the total composition of the alloy. One rather
significant exception is cobalt, which raises the Ms
ternperature in Fe-Ni alloys24 and maraging steels in
general. Goldman and Robertson25 have suggested that
this effect is due to the lowering of the austenite shear
modulus by cobalt. In practical terms, this effect of
cobalt has been very helpful in allowing higher alloy
contents to be added to the steels.
The properties of massive martensites
Studies of the plastic deformation of an 180/0 Ni binary26
and several Fe-Ni-Cr ternary 27 massive martensites
showed that the alloys all behayed in a rather similar
fashion. The yield strengths of these alleys were of the
order of 100 ksi. Their work-hardening behaviour, the
temperature- and strain-rate dependences of the flow
stress, and the derived values of the activation volumes
and energies for deformation, were all reasonably con-
sistent with the behaviour observed in iron or low-alloy
steels. The Petch slope values, based on the prior
austenite grain diameters, were on the low side of the
results reported for low-alloy steels. In most respects,
however, the deformation characteristics of the alloys
were about what might be expected of highly alloyed
b.c.c. iron.
The 18% nickel alloy showed quite good tensile
ductilities even when tested in liquid hydrogen. In
the Fe-Ni-Cr ternary alloys, higher nickel levels con-
siderably improved the impact properties at sub-zero
temperatures. Chromium was also slightly beneficial to
the toughness.
One rather noticeable deformation characteristic
found in the 18°~ Ni alloy was the propensity for slip to
be concentrated in localized bands. This may result
from the dislocation structure introduced by the mar-
tensite transformation. Basinski and Jackson28 observed
similar slip behaviour in copper single crystals that had
been initially deformed by torsion or compression and
then strained in tension. They attributed the localised
fi7

slip during tensile straining to the 'alien' dislocation dis-
tribution introduced by the initial deformation. It
seems likely that the martensite transformation could
also introduce an 'alien' dislocation structure and con-
sequently deformation behaviour analogous to that
described by Basinksi and Jackson.
Heat-treating the massive martensites at temperatures
up to 4000 C (7500 F) produced fairly large increases in
the elastic limits, and small increases in the yield and
ultimate tensile strengths. Other investigators have
noted small decreases in residual microstress29 and
electrical resistivity, so after heat-treating 25 % nickel
alloys. In the case of the Fe-Ni-Cr alloys the increases
in the elastic limits were inversely related to the Ms
temperatures of the alloys. The reasons for the observed
change in properties with tempering are not certain.
Part of the effects may be due to the precipitation of
residual amounts of carbon or nitrogen. It seems likely,
however, that a recovery reaction to relieve the residual
stresses generated by the martensite transformation is
the primary reason. 27
Ausaging reactions
Several studies have been made of the precipitation
reactions during ausaging. Reisdorf31 states that
ausaging the 25 % Ni steel for 4 h at 7050 C (13°00F)
produced small, spherical precipitates that by microprobe
analysis were found to contain 70 % Ni, 20 % Ti, and
small amounts of Fe, AI, and Si.
A more extensive study was made by Pitler and
Ansells2 of an alloy containing 28 instead of 25 % Ni,
and also higher Ti, AI, and Nb contents. They observed
that ausaging for 0·1 h at 7050 C (13°00 F) produced
50-A-dia. precipitates. The precipitates increased in
diameter at a rate proportional to (time)l up to ,......, ed that mechanical stabilisation, or destruction of
150 A dia., and then grew at a slower rate. After 64 h
precipitation within the grains was uniformly distributed
as 250-A-dia. spheres. In addition a cellular precipi-
tate was observed at the grain boundaries. Both pre-
cipitates were identified as y' -Nis(Ti, AI). Surprisingly,
these authors found that the spherical particles were
randomly oriented within the austenite grains.
~
N HARDNESS AFTER
AND REFRIGERATING
AUSAGING
AT
AT
- 196 ° C
600°C (1115° F)
i 800 (-320° F)
(/)
(/)
~ 400
a
II:
et
J:
o
O~I I 10
AGEING TIME, HOURS
[Courtesy
4 Variation in the Ms temperature after ausaging a 29-7 % Ni,
6% AI alloy at 600° C (1115° F) or 700 C (1295° F). (Hornbogen
0 During maraging of samples that had previously been
and Mayer.37)
118
Floreen: The physical metallurgy of maraging steels
Garwood and Jones33 too have examined ausaging
in a 25 % Ni steel. They also observed spherical y'
precipitates within the matrix, and cellular y' as well
as "I), the hexagonal form ofNi3 Ti, in the grain boundaries
after overageing. As opposed to Pitler and Ansell, they
reported that the spherical y' particles distinctly showed
a preferred orientation. Hornbogen and Mayer34,35
also found oriented y' precipitates after ausaging Fe-Ni-
Al and Fe-Ni-AI- Ti alloys. In view of the close match-
ing available between austenite and y' one would
expect a preferred orientation.
The variations in Ms temperature with ausaging are
rather interesting. Since alloying elements are being
removed from solution during the process, one would
expect the Ms temperature to rise. If, however, ausaging
is done at temperatures of the order of 6000 C (I I 150F)
the Ms temperature initially decreases, and only starts
to rise after fairly long ageing times.34-39 Figure 4
shows an example of this from the recent work of Horn-
bogen and Mayer. The figure also shows that ageing at
7000 C (12950F) produced only a continuous increase
in the Ms temperature, as has been noted by other
workers.3,32,39
Hornbogen and Mayer reported that small particles
of f.c.c. e: copper or y' precipitated in austenite were
sheared to a b.c.c. crystal structure when the austenite
was transformed to martensite. They attributed the
decrease in Mstem pera ture after short ageing times at
6000 C (I I 15 F) to the excess energy required to shear
0
these particles to a new crystal structure. After longer
ageing times the precipitates in the austenite became
coarser and no longer were transformed on cooling.
When this happened the effect of removing alloying ele-
ments from solid solution began to predominate and the
Ms temperature rose.34,36,37 It has also been suggest-
martensite embryos by precipitation, might influence
the Ms temperature. 3,38
Variations in Ms temperature can produce correspond-
ing variations in the resulting martensite structure.
Hornbogen and Mayer34,36 established that when the
Ms temperature was Iowa twinned martensite was
produced, but when the Ms rose after further ausaging,
a massive martensite was formed. This change is in
agreement with the predicted influence of Ms on mor-
+400 u.. phology discussed earlier.
With the 25 % nickel steel Decker et al.! found that
o.
+ 200 ~
:J
~
when martensite was formed by cold work re- +
frigeration the resultant mechanical properties were
~
~ superior to those of martensite formed by ausaging +
refrigeration. Garwood and Jones33 showed that in the
-200 I-
-400 former case a massive martensite structure was pro-
:Eo'"
duced, while the latter treatment gave a twinned
martensite. They attributed the differences in mechan-
cal properties to the difference in martensite structure.
This seems quite likely, since Bonizewski and Baker40
have shown, for example, that a twinned martensite
structure had a much poorer resistance to fracture than
massive marten sites in hydrogen-charged samples of
various steels.
100 1000
'Acta Metallurgica'. Maraging after au~aging
ausaged and refrigerated, Pitler and Ansell32 discovered
METALLURGICAL REVIEWS
 Floreen: The physical metallurgy qf maraging steels

that the y' precipitates started to dissolve, and become
replaced by '11 phase. The '11 phase formed as thin plates
in a Widmanstatten pattern.
In the work of Garwood and Jones, 33 however, the
originally spherical y' particles became elongateq. during
maraging. They proposed that a Ni3Ti precipitate with
the £c.c. structure of y' was nucleated at the original
y' particles. Reversion of the martensite matrix to
austenite also appeared to be nucleated at the y'
particle surfaces. Precipitation of '11 was observed only
in segregated regions in which there were fewer y'
particles. Garwood and Jones have suggested that the
differences between their results and those of Pitier and
Ansell might have been due to the somewhat different
chemical compositions of the alloys studied.
Prior ausaging can increase the Vickers hardness of
as-formed martensite by up to r-...J 100 points.3,33,34,36
During subsequent maraging the hardness differences
between ausaged specimens and those that have not
been ausaged generally grows smaller, and in some
cases the samples that were not ausaged may eventually
become harder. Normally, however, prior ausaging
makes a significant contribution to the final strength
after maraging.
Maraging reactions
Many different substitutional alloying elements can
produce age-hardening in Fe-Ni martensites. The
degree of hardening, however, may vary appreciably
with changes in the nickel content,41 or with variations
in the ageing kinetics or the reversion reaction. In
addition, there can be rather strong interactions between
particular combinations of elements, such as that between
cobalt and molybdenum. Qualitatively, though, one
can classify the alloying additions roughly as follows:
elements that produce 'strong' age-hardening (Be and
Ti) ; 'moderate' hardeners (AI, Nb, Mn, Mo, Si, Ta, V
and W) ; and 'weak' hardeners (Co, Cu, and Zr).
In recent years a considerable amount of effort has
gone into studying maraging reactions and identifying
the precipitates formed. Table II summarises the results
of the various phase-identification studies.
A number of comments must be made on the results
in Table II. The first concerns the procedures used to
identify the precipitates. The most common technique
has been the study of diffraction patterns from extracted
precipitate particles. Quite often it has proved difficult
to extract the very small particles present after normal

Table II. Results of precipitate identification studies in maraging alloys

___ ,
Composition (wt.- %, bal.I Fe) Heat·
Investigator Treatment, Precipitate Ref.
Ni Co Mo Ti Other °C(OF)Jh
Reisdorf 17·9 7·6 4·9 0·35 485(900)/3 Ni3Mo 31
20·0 1·38 0·52 Nb 485(900)/1 Fe2Ti
0·23 AI
25·8 1·20 0·52 Nb 455(850)/1 F.c.c. ppte., ao = 3·58 A
0·31 AI

Speich 20 3 500(930)/1 l1-~iaTi 42

500(930)/24 Fe2 ri
20 5 500(930)/100 NiaMo
20 10·7 Cu 400(750)/100 E-copper

Pitler & Ansell 28·2 1·93 1·07 Nb 425(800)/16 ll-NiaTi 32

0·92 AI

Baker & Swann 18 Ni(250) maraging ? Ni3Mo; 43

possibly Ni3Ti

y 670(1240) /? NiTi 44

-1==
_K_a_rd_o_n_S_k _&_p_e_rk_a_s
.__ 8__
r
1 I 1'5 (esCI structure)

Kardonsky & Perkas 8·2 1·6 AI 600(1115)/1 NiAI and Ni3AI 45

Borrelly 1;"61== 4·9 _ 480(895)/14 Ni3Mo 46

Reisdorf & Baker 17·4 7·6 4·8 0·43 485(900)/8 NiaMo; possibly NiaTi 47
18·8 9·0 4·8 0·80 485(900)/3 Ni3Mo; also possibly Ni3Ti after
longer times
17·7 7·5 4·7 485(900)/80 a-FeMo
17·7 7·5 0·46 485(900)/8 Possibly y'-NiaTi
18·0 7·5 0·42 AI 400(750)/3 F.c.c. ppte., ao = 4·16 A
18·0 4·7 485(900)/30 a-FeMo
-
Goldman & Manenc 5·0 4·1 11·6 Mn 500(930)/1 Ordered tetragonal ppte.; ordered 48
cubic ppte., ao = 3aou
500(930)/16 ,,+ Fe2Ti
Gavrilova & Tyapkin 8 1·5 AI 400(750)/4 (Fe,Ni)AI (CsCI structure) 49
29 2AI 480(895)/9
29 3·5AI 400(750)/4
-
Malyshev & Vasilevskaya 27
~I 500(930)/1/3 n-Ni Ti
3 50

continued
METALLURGICAL REVIEWS 119
 Floreen: The physical metallurgy of maraging steels

Table II continued

Composition (wt.- %, bal. Fe) Heat-

Investigator treatment, Precipitate Ref.
Ni Co Mo Ti Other °C(OF)/h

Miller & Mitchell 18·7 - - 1·4 - 600(1115)/3 ll-Ni3Ti 51

18·1 8·2 - 1·6 - 690(1115)/3 ll-Ni3Ti
18·5
18·1
-
7·9
7·1
6·6
-
- -
-
550(1025)/16
550(1025)/16
Fe2Mo
Fe2Mo
14·4 7·5 5·1 0·4 - 600(1115)/3 NiaTi+ Fe2Mo

Banerjee & Hauser 17·7

17·3
7·8
-
4'0
-
-
1·12
-
-
540(1000)/6
540(1000)/6
cr-FeMo
ll-NisTi
52

17·7
17·4
-
-
4'6
-
-
-
-
2·5 AI
540(1000)/6
540(1000)/3
cr-FeMo
NiAI
---
Garwood & Jones 25·7 - - 1·68 0·23 AI 525(980)/3 ll-Ni3Ti 33
0·4 Nb
-
Mihalisin & Bieber ,8 18 14 - - 485(900)/20 [1.-Fe7Mo6 53
Marcus, Schwartz, & Fine 17·2 8·0 4·9 0·52 - 480(895)/3
480(895)/41
Probably NisMo
Fe2Mo
54

---
Mihalisin 19·0 28·5 - - - 485(900)/3 Coarse and fine austenite pptes. 55

Bane~e~Capenos,& 18·5 9·1 5·1 0·62 - ? NiaMo6 56

Hauser possibly NiaTi

Kardonsky & Perkas 8 - - - 4 Mn 510(950)/? NisMn (CusAu structure) 57

Hornbogen & Mayer 30 - - - 6AI 400(750)/140 (Fe,Ni)sAI 34
800(1470)/? (Fe,Ni) AI
26·5 - - 2·8 4·3 AI 500(925)/135 (Fe,Ni)s AI 35
ll-NisTi
20 - - - 10 Cu 400(750)/20 &-copper & a (b.c.c.)-copper 36
Chilton & Barton 18·4 9·0 4·9 - 1·06 AI 485(900)/6 NisMo and FesAI 58
18·1 7·7 5·1 - 1·65 V 485(900)/3 NiaMo and O'-FeV
17·3 8·3 4·9 0·42 - 485(900)/8 NisMo and O'-FeTi

Detert & Moon 15 25 5 - - 550(1025)/100 Fe2Mo 59

12 20 5 - - 550(1025)/100 Fe2Mo
12 20 - - 2 Nb 550(1025)/100 Fe2Nb
15 25 - - 2 Nb 550(1025)/100 Fe2Nb
12 20 - 1 - 550(1025)/100 ll-NisTi
15 25 - - 5 Ta 550(1025)/100 ll-NiaTa
12 20 - - 1·6 Ta 550(1025)/100 Orthorhombic NisTa
14·8 29·4 - 0·52 3·5 Ta 550(1025)/100 ll-NisTa
0·64 AI
15 25 - - 2 Cr 550(1025)/100 r' -NisSi
2 Si
15 25 - - 2V 550(1025)/100 r'-NisSi
1 Si
15 25 - - 0·5 Be 550(1025)/100 r'-NisSe
12 20 - - 0·5 Be 550(1025)/100 r'-Ni3Se I

15 25 - - 5W 550(1025)/100 F.c.c. ppte.

I

maraging heat-treatments and so overaged specimens
were used. In many cases, the diffraction patterns
obtained do not closely agree with the standard pattern
of the phases believed to be present. Also, in some
instances the standard patterns of two phases that might
be present are very similar to each other, and distinguish-
ing between them is extremely difficult. Finally, there
are always problems inherent with extraction methods
in that lines from inclusions, oxides formed by the ex-
traction solution, or other extraneous sources may confuse
the results. Because of these difficulties, there is fre-
quently an element of uncertainty in the phase identifi-
cations.
Several additional or supplemental identification
120
techniques have been used. In some cases, it has been
possible to obtain selected-area electron-diffraction
patterns from thin foils. In others, the chemical compo-
sitions of the extracted phases have been determined by
microprobe analysis. Notable in recent studies are the
use of Mossbauer spectroscopy 54 and the analysis of
the angles between diffraction spots. 58
A study of the results summarised in Table II quickly
reveals a number of discrepancies between the precipi-
tates that were identified in some cases in alloys of almost
identical compositions. Part of the reason for these
variations is no doubt the uncertainties inherent in the
phase identifications. In spite of the variations, however,
certain trends in the results deserve comment.
METALLURGICAL REVIEWS
 Floreen: The physical metallurgy qf maraging steels
In the case of molybdenum precipitation, the results
generally show that ageing for several hours at tempera-
tures of the order of 485° C (gooOF) produced NisMo
in alloys containing 180/0Ni. In alloys aged for longer
times, or at higher temperatures, either Fe2Mo Laves
or (1 phases were detected. Also, alloys with higher
molybdenum and/or lower nickel levels tend to form
Fe2Mo or Fe7M06. While there are exceptions to these
results, tHe data generally suggest that the conventional
maraging heat-treatment of several hours at 485° C
produced a metastable NisMo precipitate, and after
longer times and/or higher temperatures this precipitate
was replaced either by Fe2Mo or by (1.
Perhaps the best evidence for a sequence consisting of
metastable and then a more stable molybdenum pre-
cipitate is the M6ssbauer study of Marcus et al.54
Maraging an 18 Ni (250) steel for 3 h at 480° C (895° F)
produced a precipitate believed to be Ni3Mo. After
41 h this initial precipitate had disappeared and a new
precipitate that definitely could be identified as Fe2Mo
was present. These authors also found that the depletion
of alloying elements from the martensite matrix was com-
plete after 3 h at 480° C.
Detert60,61 has pointed out that the good lattice fit
available between AsB precipitates with an h.c. p.
crystal structure and the b.c.c. martensite matrix might
favour the precipitation of AsB compounds. He further
pointed out that, on the basis of the equilibrium phase
diagrams, NisMo would not be expected to be the
equilibrium precipitate. He suggested that while the
better fit with the matrix might favour the initial pre-
cipitation of NisMo, it was possible that increased co-
herency stresses could limit the growth of Ni3Mo and
lead to the nucleation of the equilibrium precipitate.
The identification of the titanium precip!tate in the
various maraging steel compositions also poses some
problems. In the 20 and 25°A>nickel steels titanium is
the principal hardener. In compositions of this type the
results in Table II are in reasonably good agreement
that ll-NisTi is the precipitate formed during maraging.
In alloys of the 18 Ni (250) type titanium is used as a
supplemen tal hardener. In these alloys the role of
titanium is not so certain. Several investigators have
indicated that the titanium precipitate was probably
ll-NisTi. The problem in this instance is that the very
similar structures of NisMo and ll-NisTi make the
simultaneous identification of both precipitates very
difficult. Chilton and Barton,58 on the other hand,
found that titanium precipitated as a (1 phase. It is
also possible that part of the titanium may be present
in the molybdenum precipitate, i.e., as Nis(Mo, Ti).
The exact role of titanium during maraging of the
molybdenum-containing alloys is therefore rather diffi-
cult to assess. A titanium precipitate is probably formed,
but the identity of the precipitate and how much of the
titanium is present in it is not certain.
One other trend that is suggested by the precipitate-
identification studies is that AsB-type compounds appear
to be favoured by higher nickel or nickel + cobalt
contents. At lower levels of these elements the precipi-
tates are usually A2B(Laves) or A7B6(!L) phases. Here
again there are exceptions to this generalisation, but it
is interesting to note that in the case of precipitation of
substitutional elements from austenite this same trend
of stabilisation of AsB precipitates with higher nickel
and/or cobalt levels has also been observed.5 The trend
METALLURGICAL REVIEWS
might be related to the higher electron: atom ratios
with higher nickel and/or cobalt, or perhaps to changes
in the lattice parameter that might favour coherent
precipitation of AsB compounds.
Chemical analyses of the precipitates extracted from
180/0 Ni alloys have shown that the actual precipitate
compositions are not simply those indicated by the
structural formulas given in Table 11.6,31,47,52,56
Generally, the precipitates. contained a considerable
amount of iron. Small amounts of other alloying
elements were also sometimes co-existent. Cobalt, on
the other hand, was usually present only in small
quantities, if at all. The M6ssbauer study54 also indi-
cated that no significant precipitation of cobalt occurred
during maraging.
Numerous transmission electron-microscope studies
have been made of maraged alloys. Figure 5 is an ex-
ample of the structures seen: in this case the specimen
was an 18Ni(250) alloy that was aged for 8 h at 4850 C
(gooOF). The structure consists of rod or ribbon-like
Ni3Mo precipitates and spherical particles believed to
be a titanium-containing (j phase. 58
Many investigators have noted that some dislocation
rearrangement, and usually a decrease in the dislocation
density, takes place during the early stages of maraging.
These changes are presumably part of the recovery re-
action within the martensite matrix mentioned earlier.
The onset of precipitation probably prevents further
dislocation motion.
Some evidence has been found suggesting that pre-
precipitate zones may form during the initial stages of
age-hardening. Garwood and jonesS3 observed dif-
fraction streaks in a sample of the 25 % Ni steel that was
cold worked, refrigerated and then aged for 2 min at
4750 C (8g00F). They suggested that there was a
G.P.-zone stage in which segregates formed parallel to the
[I I I ]b.C.C. direction. Diffraction pattern streaking was
also observed in the titanium-containing alloys stud-
ied by Miller and Mitchell. 51 They suggested that
titanium initially formed metastable, ordered h.c.c. zones
of Ni3Ti having a DO s crystal structure. These zones
were subsequently replaced by ll-NisTi. In the case of
molybdenum, they detected several additional diffrac-
tion lines which indicated that molybdenum formed
zones having a f:c.c. crystal structure with a lattice para-
Courtesy Amer. Soc. Metals.
5 Transmission electron micrograph of 18 Ni (250) maraging
steel aged for 8 h at 485 C (900° F). (Chilton and Barton.58)
0
121
 Floreen: The physical metallurgy of maraging steels
meter of.-, 4"1 A. After overageing, Fe2Mo was found.
The formation of copper-rich G.P. zones in Fe-200/0
Ni-Io% eu alloys is discussed by Speich42 and by
Hornbogen and Mayer. 36 Some of the results from the
studies of maraging kinetics are also indicative of zone
formation. These data will be considered later.
When precipitates are formed during maraging, they
quite often appear to have been nucleated at dislocations
or at the martensite platelet boundaries. In general, the
precipitates are distributed in a relatively uniform
fashion, and coarse precipitates at grain boundaries or
zones denuded of precipitates are not normally seen.
No doubt the massive martensite dislocation structure is
helpful in providing a quasi-uniform distribution of sites
for precipitation. From a strength-toughness point of
view a uniform precipitate distribution would be ex-
pected to be very helpful.
After ageing for several hours at .-, 4850 e (gooOF)
the precipitate particles generally reach a size of several
hundred Angstroms in their largest dimension. A a number of 180/0Ni-X ternary and 180/0Ni-8% Co-X
number of different precipitate morphologies, e.g.,
needles, plates, spheres, have been seen. Sometimes
quite small changes in composition or heat-treatment
rather markedly altered the morphology.
Often the precipitates have a preferred orientation.
They are frequently believed to be coherent with the
matrix, and in many cases strain fields have been
detected around the particles.32,33,36,45-47,51,58 Here
again, however, there are variations between the results
of different investigations. An interesting effect noted
by Chilton and Barton58 was that the orientation of the
Ni3Mo precipitate was changed by the secondary pre-
cipitate. When the secondary precipitate was either a
Ti- or V-bearing cr phase, the Ni3Mo had one orienta-
tion, but when an Fe3AI precipitate was present the
Ni3Mo had a different orientation. As pointed out by
Chilton and Barton, these results illustrate the difficulty
of generalising about the orientation relationships of pre-
cipitates in different alloys.
Garwood and jones33 have called attention to the
fact that a preferred precipitate orientation could arise
because of the preferred orientation of the dislocations
within the martensite matrix.19 If precipitation occurs
along the lengths of the dislocations the precipitates
would tend to lie in [I II]b.c.c. directions. In several
cases this type of orientation has been observed.
The cobalt/molybdenum interaction
Perhaps the most interesting strengthening effect in
maraging steels is that due to the combination of cobalt
and molybdenum. Decker, Eash, and Goldman2 found
that the hardening when these two elements were
present together was much greater than the sum of the
strength increments produced when they were added
individually. Although Banerjee and Hauser52 did not
observe this effect, the data of at least three other
investigators47, 62, 6'3, show this general result.
Age-hardening can be produced in ternary Fe-Ni-Co
alloys, and at higher nickel and cobalt levels very high
strengths can be achieved. 55, 64,65 Figure 6 shows the
curves of yield strength vs. cobalt content for two dif-
ferent nickel levels and two different maraging heat-
treatments. The strengths appear to increase approxi-
mately linearly with increasing cobalt concentration.
Using neutron diffraction, Mihalisin55 states that a
122
B2-type long-range ordering developed during maraging
a 22 '7% Ni-19·3% Co ternary alloy. He also observed
two types of precipitate that were structurally identical
to austenite after maraging a Ig'O<yoNi-28·5 % Co
composition. Presumably then, strengthening of the
higher-cobalt alloys is due to a combination of ordering
and precipitation-hardening. In alloys ofI8°~ Ni-80/0 Co
composition, however, no evidence of either long-range
ordering or precipitation-hardening has been discovered.
Nevertheless, maraging produces a strength increment of
.-, 25 ksi in alloys of this composition. 66 Since the
strengthening due to cobalt increases in a continuous
fashion (Fig. 6) one would suppose that the same
hardening mechanisms could operate over the range of
cobalt contents studied. Conceivably then, the harden-
ing at the 18% Ni-8°/0 Co level may be due to a low
degree of long-range ordering that may escape detection
in neutron-diffraction analysis.
Floreen and Speich62 have compared the strengths of
quaternary alloys, where X is AI, Be, Mn, Mo, Nb, Si,
or Ti. With the exception of molybdenum the addition
of 80/0 Co increased the yield strength by .-, 20-45 ksi;
i.e., of the order of the strengthening produced by
cobalt discussed above. Thus, in these alloys cobalt
supplied only a relatively small, roughly additive
strengthening contribution. In the molybdenum-con-
taining alloys, however, the strength was increased up
to 75 ksi by the cobalt addition. Examination by
transmission electron microscopy suggested that the
addition of cobalt resulted in a finer dispersion of pre-
cipitates in the molybdenum-bearing alloys. Since, as
discussed previously, cobalt has not been found to be
present to any extent in the molybdenum precipitates,
it was considered that cobalt might have lowered the
solubility of molybdenum in the matrix, and thus in-
creased the amount of molybdenum precipitate. Miller
and Mitchell also came to this conclusion. 51
Peters and CUpp63 studied the precipitation behaviour
of a number of ternary and quaternary alloys by
electrical-resistivity measurements. This technique
proved very useful because it was shown that the resis-
tivity w~s directly proportional to the molybdenum in
solid solution, but that changes in cobalt content had
a negligible effect on the resistivity. The measurements
showed that the resistivity changes due to the recovery
reaction in the matrix were very minor compared to
those produced by precipitation.
Figure 7 illustrates the resistivity changes observed by
Peters and Cupp in an 18°10 Ni-5 % Mo ternary and an
18% Ni-5 o~ Mo-8°/o Co quaternary alloy during
30
~ 25
~
J:
l-
e>
~ 20
Fe -25Ni
Q
BASE
..J
~ 15
Fe-IBNi
BASE
100
o 10 15 20
WEIGHT PERCEflT. COBALT
[Courtesy Iron Steel Insf·
6 Yield strength of Fe-Ni-Co ternary alloys. (Floreen.65)
• Maraged 24 h at 425° C (800° F)
• Maraged 3 h at 4850 C (900° F)
METALLURGICAL REVIEWS
 Floreen: The physical metallurgy of maraging steels

ageing to 4550 C (8500F). Both curves show an initial w

=>
.J
~ 100 _
loss in resistivity due to the precipitation of molybdenum Fe -/8% Ni- 5% Mo
followed by an upward turn due to the formation of a
w
austenite. The cobalt addition tended to accelerate the <t 90
w
z
ageing reaction and to remove more molybdenum from Z

solution. Peters and Cupp concluded that adding 8°1o 4

u.
80
o
Co was equivalent to adding another 20/0Mo, and that
~ 70
the principal effect of cobalt was thus to increase the
>-
amount of molybdenum precipitation. t-
Fe-/8%Ni-5%Mo- 8%Co
~ 60
It has also been suggested that cobalt may alter the t-
(/)

(ij
dislocation structure of the martensite matrix to provide w
a: 50
a higher, more uniform, number of nucleation sites for
subsequent precipitation.6,52 While this may be possible, ANNEALED ·001 ·01
AGEING
0'1
TIME J
1·0
HOURS
10 100 1000

it seems unlikely that this mechanism could be the main [Courtesy Amer. Inst. Min. Met. Eng.
reason for the cobalt/molybdenum interaction. Since 7 Variations in electrical resistivity of two Fe-18 % Ni-base
precipitation often occurs on dislocations, one would alloys with ageing at 4550 C (850° F). (Peters and CUpp.63)
expect that, if changes in the dislocation distribution
were important, then cobalt would have significantly
50
raised the strength ofa number of the 18% Ni-8% Co-X
quaternary alloys studied by Floreen and Speich.62 ~
That study, however, made it clear that cobalt exercised
~a 46 Fe-/8%Ni-5%Mo-8% Co
a:
4
a strong hardening effect only in conjunction with I
42
molybdenum, and that with the other hardeners cobalt :u
produced only a small, approximately additive, .J
.J 38
w
strengthening effect. For this reason the hardening ~
contribution due to cobalt changing the martensite u~
0 34
a:
dislocation structure would seem to be minor.
30
•
ANNEALED ·001 .01 ·0 I 1·0 10 100 1000

Maraging kinetics AGEING TIME, HOURS

[Courtesy Amer. Inst. Min. Met. Eng.

A number of studies has been made of the age-hardening
kinetics, particularly with alloys of the 180/0nickel type.
As can be gathered from the previous discussion, a
number of reactions can take place during maraging,
and complete unravelling of these individual reactions
has not been possible. Up to the time when the reversion
of the martensite matrix to austenite becomes a major
factor, or roughly up to the time when peak hardness is
attained, the kinetics can be studied reasonably well.
Fortunately, these times are greater than those of the
standard maraging heat-treatments.
As noted before, many investigators have seen some
clislocation rearrangement and loss during the initial
stages of mar aging. The kinetics of this recovery reaction
has been studied by Miner et al.67 by means of internal-
friction techniques. They found a rapid initial drop in
clamping that was temperature-independent and similar
8 Variations in hardness of two Fe-18 % Ni-base alloys with
ageing at 4550 C (850° F). (Peters and CUpp.63)
The resistivity changes occurring during ageing of these
same alloys are shown in Fig. 7. Again, small but real
changes in resistivity were noted after the shortest ageing
times. Results such as these led Peters and Cupp to
conclude that the incubation time for age-hardening
was zero. Miner et al.67 came to a similar conclusion
from their internal-friction study.
Peters and CUpp63 also found evidence of a molyb-
denum precipitate reversion. This reversion reaction
involved only the precipitate and should not be confused
with the reversion of the martensite matrix to austenite.
Peters and Cupp showed that there was a pronounced
discontinuity in the ageing behaviour of the quaternary
Fe-180/0 Ni-7'8% CO-5'7°~ Mo alloy at 4500 C !"J

in general appearance to the recovery observed in cold-
worked metals. They concluded that the recovery was
due to athermal dislocation glide, and was ended by
precipitation.
Quite often precipitation-hardening takes place very
rapidly. In the 18 Ni (250) steel the Vickers hardness
was raised by the order of 100 points after several
minutes' ageing at 4850C (gooOF). 68 Comparable hard-
(8500F). This discontinuity appeared in Arrhenius plots
of the times to reach constant hardness or resistivity
values, or in the peak-hardness vs. ageing-temperature
curves. Brief pre-ageing at 5350 C (10000F) removed
the discontinuities.
More recent work by Peters69 has shown that a con-
siderable fraction of the molybdenum first precipitated
at temperatures < 4500 C (8500 F) would go back
f"o-.I

ness increases were observed in a number of Fe-18°/0
Ni-base ternary alloys66 after similar heat-treatments.
Figure 8 illustrates the hardness changes observed by
Peters and CUpp63 after ageing an 18% Ni-5 % Mo
ternary and an 180kNi-8% CO-5%>Mo quaternary alloy
at 4550 C (8500F). The results show that hardness in-
creases in both alloys were detected after ageing for
< I min. The small bumps in the curves at 0'1 h
f"o-.I
were believed to be due to the precipitation of residual
carbon.
METALLURGICAL REVIEWS
into solution when the alloy was then subsequently aged
at temperatures above the first ageing temperature.
Peters interpreted the evidence to indicate that below
4500 C there were two forms of molybdenum precipitate,
and speculated that there might be a dislocation-
nucleated precipitate and a matrix precipitate that was
possibly a molybdenum-rich zone. When the tempera-
ture was raised, the smaller particles of the matrix pre-
cipitate tended to dissolve and the molybdenum then
precipitated only at dislocations. The matrix precipi-
123
 Floreen: The physical metallurgy of maraging steels

tation was believed to be due to the high supersaturation
of molybdenum. Pre-ageing at 535° C (1000° F) low-
ered the supersaturation and eliminated the precipitate
reversion and thus, evidently, the matrix precipitate.
A much smaller amount of precipitate reversion was de-
tected in a Fe-180/0 Ni-5'4% Mo ternary alloy, presum-
ably because of the lower degree of molybdenum super-
saturation in the absence of cobalt.
In many cases it has proved that the isothermal
ageing kinetics could be expressed quite well by a
relationship of the type: ~x/xo = Ktn, where x is either
the hardness or electrical resistivity at time t, Xo the as-
annealed value, and K and n are constants.61,63,66,68
Quite often the values of the time constant n were of
the order of 0 '2-0'4; i.e., well below the n value of
0'5 for the idealised case of diffusion-controlled growth
of platelets. Arrhenius plots to determine activation
energies have not always given completely linear curves.
In all the 180/0 Ni-type alloys studied, however, the
nominal values of the activation energies were fairly
low, typically of the order of 30-40 kcal/mole. The
values were thus well below those commonly observed
for substitutional-element diffusion in ferrite.
On the basis of these results, it has generally been con-
preted in this way, and it seems possible that the tem-
perature-dependence of nucleation might be responsible
for the low activation-energy values observed. Any inter-
pretation must remain speculative, however, until the
precipitates can be more positively identified.
Effects of working before maraging
Many investigations have been made of the effects of
warm or cold working the 180/0Ni steels before marag-
ing. The warm-working experiments have generally been
of the ausforming type, i.e. def~rmation of the austenite
preceding the transformation to martensite. These
results have shown that ausforming produced only very
minor improvements in the final strength.56,68,73,74
Furthermore, a rather noticeable loss of toughness in a
direction transverse to the original working direction is
apparent. 56 The lack of any significant hardening as a
result of ausforming has commonly been interpreted as
due to the very low carbon contents of the maragin
steels.
One interesting effect noted by Bush73 was that when
the material was rolled at 200° C (400° F) so that f',,)

cluded that the maraging kinetics was primarily deter- 65% of the structure transformed to martensite during
mined by the martensite matrix structure. The lack of rolling, the final strength was increased by 50 ksi. f',,)

an incubation time has been attributed, after Cahn,70
to the elimination of the free-energy barrier to nuclea-
tion under conditions of high supersaturation and pre-
cipitation at dislocations.63 The low values of the time
constant n and the activation energy have commonly
been interpreted in terms of pipe-diffusion through the
This strength increase was much greater than that observ-
ed in material that was warm rolled in the entirely
austenitic condition, and was also somewhat greater
than that found in material that has been cold rolled
in the entirely martensitic condition. The explanation
for the higher strengths after rolling at 200 C is not 0

dislocations present in the martensite matrix. clear at this time.

Recent evidence suggests, however, that this inter- Cold working the martensite before maraging has
pretation, i.e. that the kinetics is dictated solely by the been found to give more substantial strength in-
martensite structure, is not correct. Peters and Floreen 71 creases.41,56,60,61,68,73-77
Figure 9 shows the tensile and
have studied the ageing kinetics of an Fe-8 % Ni- I 3 o~ toughness changes due to prior cold working observed
Mo alloy. By suitable heat-treatment this alloy could by Kula and Hickey.74 Cold working tends to increase
be made either ferritic or martensitic. The ageing be- the initial rate of hardening somewhat,60,61,68and the
haviour was then evaluated in both the ferritic and largest strength gains are usually found after short ageing
martensitic matrices, and also in a ferritic matrix that
had been cold rolled 50% before ageing. The hardness
curves of the martensite and the cold-worked ferrite stru-
tures were essentially identical. More important, all 260
three structures exhibited noticeable incubation times,
and the activation-energy values for age-hardening of
all three structures were of the order of 65 kcal/mole; -x." 240
i.e., of the order of the diffusion values for the substitu- I-
l!)
Z
tional element. l1J
a:
In addition to these results, Marcus et al.72 have t;
220
determined an activation-energy value of 60 kcal/mole
for the ageing of a martensitic stainless steel. These
findings show that activation-energy values commen-
surate with diffusion of the substitutional alloying ele- TENSILE ELONGATION

ment can be obtained during ageing in a massive marten-

site matrix. It follows, therefore, that low values of the z
activation energy, and also the absence of an incubation ~
enlie:: 100
period, cannot be interpreted simply in terms of a mar-
If)
tensite matrix structure. If)
l1J

The question then is why do the 18% Ni alloys Z

~ 80
harden as fast as they do? At present this question can- ::>
o
0 20 40 60
I- PERCENT. REDUCTION
not be answered. Conceivably, the behaviour may be
[Courtesy Amer. Inst. Min. Met. Eng.
related to the tendency of the higher-nickel alloys to
9 Changes in tensile properties and fracture toughness of
form A3B precipitates and to the close fit with the b.c.c. 18 Ni (250) maraging steel with cold rolling and subsequent
matrix that is generally possible with structures of this maraging at 485 C (900 F) for 3 h. Toughness specimens all
0 0

type. The absence of incubation times could be inter- ground to 0·038 in. thickness after ageing. (Kula and Hickey. 74)
124 METALLURGICAL REVIEWS
 Floreen: The physical metallurgy of maraging steels
times. In the case of the 15 % Ni steel (Alloy F in Table
I), Detert78 found that cold rolling 5 % lowered the
ductility and also the coercive force. He suggests that
these changes may have been due to the elimination of
a small amount of untransformed austenite.
The effects of cold working after maraging and of
subsequent re-ageing, were also explored by Kula and
Hickey.74 The strength changes produced by these
treatments were relatively small.
Effects of neutron irradiation
Neutron irradiation of the 18 Ni (250) and the 12-5-3
maraging steels has been studied by Cupp et al.79
Irradiation with integrated flux levels of 1'5-3'1 X 1019
n/cm2 (> I MeV) produced a 4% loss in strength in
maraged 18 Ni (250) samples, and a 5% strength in-
crease in maraged 12-5-3 samples. The notched impact
transition temperature of the 12-5-3 steel was increased
only 45-50 degC (80-go degF) upon irradiation, and the
upper shelf energy decreased by only a small amount.
These property changes were much smaller than those
observed in other high-strength steels.
Austenite formation
As indicated by the phase diagrams in Figs. I and 2, the
iron-nickel martensite matrix is metastable and after pro-
longed holding at elevated temperatures will eventually
decompose by a diffusion-controlled reaction to ferrite
and austenite. Depending upon the temperature and
composition, the austenite formed may be so enriched
in nickel that upon subsequent cooling its Ms tempera-
ture is well below room temperature and consequently
austenite will remain in the microstructure.
The rate of austenite formation is quite sensitive to
composition. In binary iron-nickel alloys increasing
nickel levels generally tend to accelerate austenite forma-
[Courtesy Amer. Soc. Metals.
10 Replica electron micrograph showing austenite formation in
an 18 Ni (~50) maraging steel heat-treated for 1 h at 6500 C
(1200° F). (Floreen and Decker.68)
METALLURGICAL REVIEWS
tion.80,81 Additional alloying elements can markedly
affect the reversion reaction. Part of the reason for the
effect of an alloying element on reversion may be ex-
plainable in terms of whether the element tends to
stabilise austenite or ferrite.82 A much stronger effect
probably results from the change in matrix composition
that accompanies precipitation. Titanium, for example,
has been found to markedly retard reversion, evidently
because the formation of Ni3Ti lowers the nickel content
of the matrix. 81
Molybdenum additions, on the other hand, enhance
reversion.81 In this case reversion was believed to have
been associated with the dissolution of Ni3Mo and the
formation of Fe2Mo.61,81 Presumably, when Ni3Mo
goes into solution the local enrichment of the matrix
with nickel leads to austenite formation.
Austenite usually starts to form at the martensite
platelet boundaries, and after large amounts of austenite
have formed the structure is lamellar in appearance with
elongated austenite ribbons strung out along the
boundaries, as shown in Fig. 10. In addition, smaller
austenite pools form within the platelets. The precipitate
particles sometimes have been observed to be dissolved
by the austenite.
With prolonged holding at temperatures of the order
of 400-700° C (750-1300° F), very appreciable amounts
of stable austenite can be formed. When this happens
the strength is considerably reduced. Generally, the
austenite formed in this temperature range cannot be
transformed to martensite by refrigeration. Cold
working, however, will transform the austenite to
martensite. 83,84The presence of austenite in the struc-
ture considerably increases the uniform elongation
during straining, and Legendre84 has shown that this
can be very helpful in operations such as deep drawing.
He also showed that subsequent maraging of the cold-
worked material produced quite good combinations of
strength and ductili ty.
Austenite can also be formed by heating the alloy
above the As temperature. In this case the martensite
tran~forms by a shear reaction to an austenite of the
same composition. Studies of this reaction in iron-
nickel binary alloys have shown that the resultant
austenite contains a high density of dislocations. 85-89
Orientation relationships indicate that the crystallo-
graphic nature of this transformation is similar to the
austenite ~ martensite transformation.
If the heating rate is slow enough, some reversion of
the martensite matrix to austenite + ferrite by the dif-
fusion-controlled reaction may occur. When this hap-
pens the subsequent shear reaction may be affected. The
general effect seems to be that the reversion reaction
tends to stabilise the martensite and raise the As tem-
perature. This effect might be due to mechanical
stabilisation by the austenite particles formed during
heating, and/or to the corresponding decrease in the
nickel conten t of the martensite.
When additional alloying elements are present the
changes in the shear reaction with composition and
heating rate can become very complex indeed. In the
case of the 180/0 Ni steels the general trend still appears
to be that decreasing the heating rate raises the As
temperature, 90,91but the multiplicity of reactions makes
interpretation of the data difficult.
As the temperature is increased above the As tempera-
ture a homogenisation reaction becomes predominant.
125
 Floreen: The physical metallurgy of maraging steels

That is, precipitates dissolve and concentration gradients
due to precipitation or reversion are removed. 3, 81,92
The time and temperature for complete homogenisation
will, of course, depend upon the composition and history.
In most cases an hour at 815° C (1500° F) has been
adequate for the various maraging steel grades. Very
prolonged exposures in the 400-700° C (750-1300° F)
range, however, would probably require considerably
higher temperatures and longer times for homogenisa-
tion.
Recovery, recrystallisation, and grain growth of the
austenite can also all take place during austenitising.
In the case of the strengthening mechanism, the yield
strengths of unaged maraging steels are typically r'-I 100
ksi. After maraging, the yield strengths of the various
180/0 Ni steels range from 200 to 300 ksi, so that the
values are raised 100-200 ksi by age-hardening. Several
investigators have pointed out that precipitation-
strengthening of this magnitude can be accounted for
quite reasonably by an Orowan relationship, viz.
rxGb
(j = (jo + -)..-
where is the total strength, 0 the matrix strength, ex
(j (j

As would be imagined, these reactions are very depend- a constant, G the shear modulus, b the Burgers vector,
ent upon the specimen history. To date, no systematic and ).. the interparticle spacing. After maraging the
study has been made of these reactions in maraging-
type compositions.

Changes in other physical properties with 200

TEMPERING
400 600
TEMP.,
800
0 F
1000' 1200
11·4
ageing

!
6·2
II· 0 /""'-- .•... ,
Figure I I shows the changes in the thermal-expansion a/degF 6.0 a/degC
a. ",/ "... 400
x 10-6 X 10-610.6
coefficient, X-ray characteristic temperature, elastic 5·8
---------, 4W

modulus, electrical resistivity, and hardness detected by 5,6

10·2 ()

Perkas and Potapov93 during ageing of an Fe-8% Ni-

""0
1'5 °10 Ti alloy. A similar series of measurements on a 200
binary Fe-8% Ni alloy revealed practically no change in 3·0

these quantities over the range 200-700° C (400-1300° F). % ~~ 2·0

Variations in the elastic modulus similar to those shown
1·0
in Fig. 10 were a!~f)found by Banerjee et al.56 during
ageing of the 18 Ni (300) alloy. They suggested that
the modulus changes were due to depletion of nickel from - - - - - - -'- -- - - ..
\
0'7 P
the matrix during ageing. Heat-capacity measurements
\
\ ,
of Fe-801o Ni-base alloys containing various levels of 0·6

aluminium are given by Itkin et al.94,95 350

The magnetic properties of various maraging alloys
300
have been examined in several recent studies. 59-61,78, HV HV
81,96-98 The room-temperature d:c. magnetisation 250

curves of the 20 Ni, 18Ni (200), and 18Ni (250) maraging 200
100 300 500 700
steels were similar to each other and to those of several TEMPERING TEMP" ~c
other steels, as shown in Fig. 12.
[Courtesy 'Physics Metals Metallography'.
The magnetic properties of various maraging compo-
sitions did not change appreciably during maraging, as 11 Variation in coefficient of linear thermal expansion (a),
X-ray characteristic temperature(6), Young's modulus
long as austenite did not form. With the start of austenite (/::;£/£), electrical resistivity (P), and hardness (HV) of an
formation, however, the coercive force increased mar- Fe-8 % Ni-1·5 % Ti alloy after tempering. (Perkas and Pota-
kedly.6o,61,81 This variation with ageing has been pOV.93)
attributed to the fact that large changes in coercive force
could be expected only when particles of the dimensions
of the domain walls, or !"oJ 1000 A in size, were present.
The precipitate particles were, thus, generally too small
to significantly affect the properties, but the austenite
particles formed by the reversion reaction were of the
right size.
The desire to maintain good magnetic properties after
prolonged elevated-temperature exposures has led to
several alloy modification studies.59,78,96,98Alloys with
higher cobalt contents, in particular, have proved to
have greater high-temperature stability.

The relationship of structure to strength

and toughness
o
The primary attribute of maraging steels is their excel- o 50 100 150 200 250 300
lent combination of strength and toughness. This seems FIELD STRENGTH J OERSTEDS

to be especially true of the 180/0nickel grades, which in [Courtesy Amer. Soc. Metals.
general have a toughness superior to that of the other 12 Room temperature d.C. magnetisation curves for selected
grades at high strength levels. high-strength steels. (Honeycutt and Steigerwald. 96)

126 METALLURGICAL REVIEWS

 Floreen: The physical metallurgy qf maraging steels
values of A are of the order of 300-500 A, and the corres-
ponding strength increments are f'J 100-200 ksi. Thus
an Orowan-type hardening involving dislocation motion
between the precipitate particles can account for the
strengths observed.6,43,47
On the other hand, Detert60,61 has suggested that
hardening could be controlled by the stress necessary to
shear through the precipitate particles. Reisdorf and
Baker47 have observed, however, that after straining
1-2% the structure of the 18 Ni (250) steel showed
tangles of dislocations around the precipitate particles,
and also instances of dislocations bowing between the
particles. Moreover, the initially high work-hardening
rate of the steels is in general more consistent with an
Orowan mechanism.6,47 For these reasons an Orowan-
type hardening mechanism seems more likely in marag-
ing steels.
The temperature-dependence of the flow stress has
been analysed by Conrad. 99 He concluded that the
effective shear stress varied with temperature in the same
way as that of mild steel, and that the major strengthen-
ing was due to long-range internal stresses such as would
be developed by fine precipitate particles.
The reasons for the superior toughness of the 18% Ni
maraging steels are not clear. It has often been noted
that the elimination of carbon and other deleterious
impurities should be beneficial. Also the relatively uni-
form precipitate distribution achieved by age-hardening
in a massive martensite matrix would be helpful. In
this regard molybdenum appears to play an important
role in minimising embrittlement of the prior austenite
grain boundaries. Floreen and Speich62 reported that
low-energy intergranular fractures of Fe-I80;0 Ni-
8% Co-base alloys hardened with aluminium or titanium
could be eliminated by additions of 20;0 Mo. It should
be noted, however, that although the molybdenum
additions raised the Charpy impact energies from
"'" 3 ft-Ib to 17 ft-Ib, the latter values were still relatively
poor compared to the properties of the 18% Ni steels.
The possible role of the high nickel content of the
matrix in preventing fracture also has to be considered.
In low-alloy steels, nickel reduces the tendency to cleav-
age and lowers the ductile-brittle transition temperature.
In maraging steels this effect of nickel could be helpful
in minimising the effects of hydrogen embrittlement,
where quasi-cleavage fractures have generally been ob-
served. Normally, however, fracture in maraging and
other high-strength steels occurs by a localised ductile-
fracture process, i.e., by the nucleation of voids followed
by either coalescence of voids or the propagation of a
crack between the voids. Whether nickel would be help-
ful in preventing fractures of this type is not certain.
More directly, the fracture toughness of other marag-
ing compositions has generally been found to be signi-
ficantly lower than that of the 18Ni (250) steel when com-
pared at equal strengths.62,66 These other steels also
had low carbon and impurity contents, massive marten-
site matrices with fairly uniform precipitate distribu-
tions, and high nickel contents. It follows, therefore,
that while these variables may be necessary to achieve
good toughness, they are not sufficient to account for
the superior toughness of the 18 Ni (250) steel.
Evidently then there is something rather special about
the 18 Ni (250) composition. Since molybdenum is the
primary hardener in this steel, it thus seems that Ni3Mo
is a precipitate that gives better toughness. Why this
METALLURGICAL REVIEWS
should be so is difficult to see, because structurally there
is nothing particularly unique about NisMo compared
to other precipitates. The one distinctive aspect that has
been noted regarding molybdenum-hardening is that
the precipitate is metastable and eventually goes into
solution. Conceivably, this process may start during
the normal 3-h maraging heat-treatment to give a
structure of Ni3Mo particles with adjoining austenite
that might be more effective in preventing void forma-
tion at the precipitates, and thus delay ductile fracture.
Whether voids actually are nucleated at precipitate
particles in maraging steel is not known. It seems likely
that they could be, because, e.g. Palmer et al.lOO have
demonstrated that voids could be nucleated at particles
as small as 50A in internally oxidised copper alloys.
These questions of void nucleation and crack growth in
maraging steels have not been well explored. Such work
is needed if the reasons why certain precipitates give
superior toughness are to be made clear.
References
I. R. F. Decker, R. B. G. Yeo, J. T. Eash, and C. G. Bieber,
Mat. Design Eng., 1962, 55, (5), 106.
2. R. F. Decker, J. T. Eash, and A. J. Goldman, Trans. Quart.
ASM, 1962, 55,58. .
3. R. F. Decker, 'The Relation Between the Structure and Mech-
anical Properties of Metals,' p. 648. 1963: London (H.M.
Stationery Office).
4. W. Steven, 'Metallurgical Developmentsin High-Alloy Steels',
(Special Rep. No. 86), p. 115. 1964: London (Iron Steel
lnst.).
5. R. F. Decker and S. Floreen, 'Precipitation in Iron-Base
Alloys', p. 69. 1965: New York, &c. (Gordon and Breach).
6. G. Pellesier, 'Problems in the Load-Carrying Application of
High-Strength Steels', DMIC Rep. (210), p. 173, 1964.
7. G. P. Contractor, J. Metals, 1966, 18, 938.
8. E. A. Owen and Y. H. Liu,J. Iron Steel Inst., 1949, 163, 132.
9. F. W.Jones and W. 1. Pumphrey, ibid., 1949,163, 121.
10. J. R. Mihalisin, 'Fifty years of Progress in Metallographic
Techniques' (Special Tech. PubI. No. 430), p. 250. 1968:
Philadelphia, Pa. (Amer. Soc. Test. Mat.).
11. A. G. Haynes, Trans. Amer. Inst. Min. Met. Eng., 1957, 209,
1314.
12. W. S. Owen, E. A. Wilson, and T. Bell, 'High-Strength
Materials', p. 167. 1965: New York and London (John
Wiley).
13. W. S. Owen and E. Gilbert, Acta Met., 1962, 10,45.
14. C. M. Wayman, 'Physical Properties of Martensite and Bainite'
(Special Rep. No. 93), p. 153. 1965: London (Iron Steel
lnst.).
15. G. R. Speich and P. R. Swann,J. Iron SteeIInst., 1965,203,480.
16. R. Huizing and J. A. Klostermann, Acta Met., 1966, 14, 1693.
17. W. D. Swanson and J. G. Parr, J. Iron Steel Inst., 1964, 202,
104.
18. W. S. Owen and E. A. Wilson, Ref. 14, p. 53.
19. R. L. Patterson and C. M. Wayman, Acta Met., 1966, 14, 347.
20. K. H. Eckelmeyer and G. Krauss, to be published.
21. G. P. Miller and W. 1. Mitchell, J. Iron Steel Inst., 1965, 203,
895·
22. T. Bonizewski,Brit. Weld. J., 1965, 12, 557.
23. O. Johari and G. Thomas, Acta Met., 1965, 13, 1211.
24. R. B. G. Yeo, Trans. Met. Soc. A.I.M.E., 1963, 227,884.
25. A. J. Goldman and W. D. Robertson, Acta Met., 1965, 13,
391.
26. S. Floreen, Trans. Met. Soc. A.I.M.E., 1964,23°,842.
27. S. Floreen, ibid., 1966, 236, 1429.
28. Z. S. Basinski and P. J. Jackson, Appl. Phys. Letters, 1965, 6,
148.
29. V. M. Kardonsky, V. G. Kurdjumov, and M. D. Perkas,
Physics Metals Metallography, 1961, II, (4), 117.
30. L. F. Porter and G. J. Dienes, Trans. Met. Soc. A.I.M.E., 1959,
215,854.
31• B. G. Reisdorf, Trans. Quart. ASM, 1963, 56, 783'
32. R. K. Pitler and G. S. Ansell, ibid., 1964, 57, 220.
33. R. D. Garwood and R. D. Jones, J. Iron Steel Inst., 1966, 204,
512•
127
 Floreen~' The physical metallurgy of maraging steels
34. E. Bornbogen and W. A. Mayer, Z. Metallkunde, 1967, 58,
372•
35. E. Bornbogen and W. A. Mayer, ibid., p. 445.
36. E. Bornbogen and W. A. Mayer, ibid., p. 297.
37. E. Bornbagen and W. A. Mayer, Acta...Met., 1967, 15, 584'
38. K. A. Malyshev and M. M. Vasilevskaya, PhysicsMetals Metal-
lography, Ig65, 18, (2), 150.
39. C. G. Bieber, unpublished data.
40. T. Bonizewski and R. G. Baker, Brit. Weld. J., 1965, 12,349.
41. M. D. Perkas and V. 1. Snitsar, Physics Metals Metallography,
1964, 17, (3), 75·
42. G. R. Speich, Trans. Met. Soc. A.I.M.E., 1963, 227, 1426.
43. A. J. Baker and P. R. Swann, Trans. Quart. ASM, 1964, 57,
1008.
44. V. M. Kardonsky and M. D. Perk as, Physics Metals Metallo-
graphy, Ig65, 19, (2), 133.
45. V. M. Kardonsky and M. D. Perkas, Metal Sci. Heat-Treat.
Metals, 1964, (11),672.
46. R. Borrelly, 'Contribution a l'Etude au Microscope Electroni-
que des Aciers "Maraging" a 18% de Nickel', Thesis Univ.
Lyon, 1965.
47. B. G. Reisdorf and A. J. Baker, Wright Patterson AFB, Ohio,
Tech. Rep. (AFML-TR-64-390), 1965.
48. A. J. Goldman and J. Manenc, Trans. Quart. ASM, 1965, 58,
645'
49. A. V. Gavrilova and Yu. D. Tyapkin, Physics Metals Metallo-
graphy, 1965, 20, (2), 145.
50. K. A. Malyshev and M. M. Vasilevskaya, ibid., Ig65, 18, (2),
80.
51. G. P. Miller and W. 1. Mitchell, J. Iron Steel Inst., 1965, 203,
8gg.
52. R. B. Banerjee and J. J. Hauser, Wright PattersonAFB, Ohio,
Tech. Rep. (AFML-TR-6(l-166), Ig66.
53. J. R. Mihalisin and C. G. Bieber, J. Metals, 1966, 18, 1033.
54. B. Marcus, L. H. Schwartz, and 1\1. E. Fine, Trans. Quart.
ASM, 1966, 59, 468.
55. J. R. Mihalisin, ibid., p. 60.
56. R. B. Banerjee, J. M. Capenos, and J. J. Hauser, "Advances
in Electron Microscopy' (Special Tech. Publ. No. 396),
p. 115. 1966: Philadelphia, Pa. (Amer. Soc. Test Mat.).
57. V. M. Kardonsky and M. D. Perkas, Metal Sci. Heat-Treat.
Metals, 1966, (4), 254·
58. J. M. Chilton and C. J. Barton, Trans. Quart. ASM, 1967, 60,
528.
5g. K. Detert and D. Moon, to be published.
60. K. Detert, Trans. Quart. ASM, 1966, 59,262.
61. K. Detert, Arch. Eisenhiittenwesen,1966, 37, 579.
62. S. Floreen and G. R. Speich, Trans. Quart. ASM, 1964,57,714.
63. D. T. Peters and C. R. Cupp, Trans. Met. Soc. A.I.M.E., 1966,
236, 1420.
64. M. Tanaka, T. Suzuki, and J. Komura, J. Japan Inst. Metals,
1967,31, 1016.
© The Metals and Metallurgy Trust.
128
65. S. Floreen, to be published.
66. S. Floreen, Trans. Quart. ASM, 1964, 57, 38.
67. R. E. Miner,J. K.Jackson and D. F. Gibbons, Trans. Met. Soc.
A./.M.E., 1966, 236, 1565.
68. S. Floreen and R. F. Decker, Trans. Quart. ASM, Ig62, 55,
518.
69. D. T. Peters, Trans. Met. Soc. A./.M.E., 1967, 239, 1981.
70. J. W. Cahn, Acta Met., 1957,5, 169.
71. D. T. Peters and S. Floreen, to be published.
72. H. L. Marcus, J. N. Peistrup, and M. E. Fine, Trans. Quart.
ASM, 1965, 58, 177.
73. R. H. Bush, ibid., 1963, 56,885.
74. E. B. Kula and C. F. Hickey, Trans. Met. Soc. A.I.M.E., 1964,
230, 1707·
75. H. W. Maynor and C. C. Busch, ASM Metals Eng. Quart., 1966,
6, (1),20.
76. M. D. Perkas, Physics Metals Metallography, [963, 15, (4), 62.
77. M. D. Perk as, Metal Sci. Heat-Treat. Metals, 1964, (I 1),664.
78. K. Detert, Trans. Met. Soc. A.I.M.E., 1967, 239, 553.
79. C. R. Cupp, E. P. Sadowski, and L. P. Trudeau, ibid., p. 1468.
80. N. P. Allen and C. C. Early, J. Iron Steel Inst., 1950, 166, 281.
81. D. T. Peters, Trans. Quart. ASM, Ig68, 61, 62.
82. S. Floreen and R. F. Decker, ibid., 1963,56,4°3.
83. D. L. Corn, 'Structure and Properties of Ultra- High-Strength
Steels' (Special Tech. Publ. No. 370), p. 54. 1965: Phila-
delphia, Pa. (Amer. Soc. Test .•Mat).
84' P. Legendre, Cobalt, 1965, (29), 171.
85. G. Krauss, Acta Met., 1963, II, 499.
86. H. Kessler and W. Pitsch, ibid., Ig67, 15,4°1.
87. Y. A. Izmaylov and V. G. Gorbach, Physics Metals Metallo-
graphy, Ig65, 20, (I), 101.
88. S. Jana and C. M. Wayman, Trans. Met. Soc. A.I.M.E., Ig67,
239, 1187.
89. G. Shapiro and G. Krauss, ibid., p. 1408.
go. W. A. Petersen, Weld. J., 1964,43, 428S.
gl. A. Goldberg and D. G. O'Connor, Nature, 1967,213, 170.
92. A. J. Sedriks and J. V. Craig, J. Iron Steel Inst., 1965,2°3,268.
93. M. D. Perkas and L. P. Potapov, PhysicsMetals Metallography,
1965, 19, (4), 134·
94. V. P. Itkin, B. M. Mogutnov, and L. A. Shvartsman, ibid.,
Ig66, 21, (5), 85'
95. V. P. Itkin, B. M. Mogutnov, and L. A. Shvartsman, ibid.,
1966,21, (6), 72.
96. C. R. Honeycutt and E. A. Steigerwald, Trans. Quart. ASM,
1966, 59, 113·
97. D. M. Pavlovic and J. J. Clark, J. Appl. Physics, 1966, 37,
1208.
98. D. T. Peters, to be published.
99. H. Conrad, Trans. Quart. ASM, Ig64, 57, 747.
100. 1. G. Palmer, G. C. Smith, and R. D. Warda, 'The Physical
Basis of Yield and Fracture', p. 53. 1967: London (lost.
Physics and Phys. Soc.).
1968.
METALLURGICAL REVIEWS