Bhadeshia H.K.D.H., Honeycombe R. Steels: Microstructure and Properties

Подождите немного. Документ загружается.

118 CHAPTER 5 FORMATION OF MARTENSITE

Table 5.3 Values of difference in enthalpy of element X in austenite (γ) and in marten-

site (α

′

)

X CNMnNiCuCrWMoVTi

H

′

→γ

(kJ mol

−1

) −33.9 −22.4 −10.7 −8.4 −5.4 −5.0 +5.7 +5.7 +11.8 +37.7

rewritten, expressing the driving force of the reaction G

′

→γ

as:

G

′

→γ

= (1 − x)G

α→γ

+ xH

′

→γ

. (5.13)

After determining H

′

→γ

, the free energy change for the martensite reaction

G

′

→γ

can be calculated. Values for T

, the temperature at which γ and α

′

have the same free energies, can be calculated by putting G

′

→γ

equal to zero.

Alloying elements either expand the γ-loop, i.e. stabilize γ, or contract the

loop and encourage α-formation, and this will have different effects on H

′

→γ

(Section 4.1). Elements which expand the γ-loop will make this term negative

and lower T

, while elements which favour α-formation will make the term

positive and raise T

It is interesting to look at the values of H

′

→γ

which are available in the

literaturefor anumber ofcommon alloying elements(Table 5.3).There aresome

anomalies, e.g. chromium, which contracts the γ-loop, has a negative H value,

suggesting that H has been computed from data at too low a temperature.

Cohen and co-workers have provided detailed data for iron–carbon alloys

between 0 and 1.1 wt% carbon and in Fig. 5.19 the temperature dependence

of G

′

→γ

is plotted for several carbon levels. The intersection of the curves

with the H

′

→γ

=0 axis provides values of T

for the various compositions. It

was found that the driving force G

′

→γ

at the M

temperatures of the alloys

was practically constant, approximately 1250 J mol

−1

, independent of carbon

content. However, work on iron–nickel alloys has shown that the driving force

increases with increasing nickel content, i.e. as the M

is depressed.

5.6.3 The effect of deformation

The effect of uniaxial stress on the martensitic transformation is normally to

raise the M

temperature. The superimposed stress ﬁeld from the plastic, or

elastic, deformation reinforces that caused by the nucleation of a martensite

plate, and in one sense the subsequent shape change is a further plastic deform-

ation process. We can deﬁne a temperature M

, greater than M

, above which

deformation of the parent phase does not form any martensite. However, it is

likelythat deformation ofthe austenite aboveM

willalter the M

onsubsequent

cooling through the martensitic range. Usually in these circumstances, the M

5.6 KINETICS OF TRANSFORMATION TO MARTENSITE 119

Fig. 5.19 Free energy change for the austenite–martensite reaction as a function of

temperature and carbon content (Kaufmann and Cohen, Progress in Metal Physics 7, 165, 1958).

lowered, and the resultant increased stability of the austenite is referred to as

mechanical stabilization.

5.6.4 Stabilization

Stabilization means a reduction in the amount of transformation of austenite

to martensite, as a result of processes which interfere with the nucleation and

growth of the plates. Plastic deformation above the M

temperature can achieve

this. However, the term stabilization is normally applied when the cooling of a

120 CHAPTER 5 FORMATION OF MARTENSITE

steel is arrested in the M

−M

range. The transformation, when resumed by

lowering the temperature, does not result in as complete a transformation to

martensite as would have been the case if no isothermal pause had occurred.

At the chosen delay temperature, the degree of stabilization increases to a

maximum with time, and as the temperature approaches M

, the extent of sta-

bilization increases. It appears that stabilization is at a minimum when only a

small amount of martensite is present in the matrix.

The explanation of these complex effects lies in the fact that the formation

of martensite plates leads to accommodating plastic deformation in the sur-

rounding matrix, which can result in high concentrations of dislocations in the

austenite. Interaction of some of these dislocations with the glissile dislocations

in the martensite plate boundary will then cause it to be no longer mobile, so

that the plate cannot grow further. Any phenomena which help to encourage

this process will achieve stabilization. Resting at an intermediate temperature

gives time for plastic relaxation, i.e. movement of dislocations, as well as the

locking of interfacial dislocations by carbon atoms.

5.7 THE STRENGTH OF MARTENSITE

The high hardness and brittleness of rapidly quenched steels is the result of the

formation of martensite, yet many shear transformations in non-ferrous alloy

systems do not produce this dramatic hardening. Indeed, if carbon is eliminated

from the steel the resulting hardness is very much lower. Figure 5.20 shows the

large effect of carbon content on the hardness of martensite compared with the

relatively small effect of carbon on the strength of austenite, retained to room

temperature by the addition of nickel.

The strength levels reached depend also on the detailed structure of the

martensite, e.g. whether it has remained stable during quenching and testing at

room temperature. By addition of nickel to iron carbon alloys, Winchell and

Cohen depressed the M

temperature to −35

◦

C,so that martensite formed only

at low temperatures and auto-tempering was eliminated (Chapter 8). In addi-

tion, the samples were deformed at 0

◦

C, with the results shown in Figure 5.21,

indicating that the ﬂow stress of martensite increases with carbon content up

to about 0.5 wt% C. Allowing the martensite to rest for 3 h at 0

◦

C, resulted in

the upper curve (Fig. 5.21), demonstrating that martensite can age harden at

ambient temperature or below.

The question of the origin of the high strength of martensite is a difﬁcult one,

compounded by the complexity of the structure, a tetragonal lattice with inter-

stitial carbon in solid solution, formed by shear which leads to high densities of

dislocations and ﬁne twins. There are, as a result, several possible strengthening

mechanisms:

(a) substitutional and interstitial solid solution;

(b) dislocation strengthening, i.e. work hardening;

5.7 THE STRENGTH OF MARTENSITE 121

(d) grain size;

(e) segregation of carbon atoms;

(f) precipitation of iron carbides.

Fig. 5.20 The effect of carbon on the hardness of martensite and austenite (Winchell and

Cohen,Transactions of the Metallurgical Society of AIME 224, 638, 1962).

122 CHAPTER 5 FORMATION OF MARTENSITE

Fig. 5.21 Ageing of martensite at 0

◦

C in Fe–Ni–C alloys (Winchell and Cohen, Transactions

of the Metallurgical Society of AIME 224, 638, 1962).

The interstitial solid solution of carbon which results in the tetragonality of

martensite is a prime candidate for the role of major strengthening factor. The

work of Winchell and Cohen enabled the determination of the yield stress as a

function of carboncontent under conditionswhen thecarbon atoms wereunable

to diffuse to form atmospheres and precipitates. The ﬂow stress was shown to

vary as c

1/3

, where c =carbon and concentration, but later it was found that the

strength could be shown equally well to vary as c

1/2

5.7 THE STRENGTH OF MARTENSITE 123

Fleischer examined the situation theoretically with a model of a dislocation

bending away from interstitial solute atoms with short range interactions, and

using a parameter ε, the difference in longitudinal and transverse lattice strain

caused by an interstitial carbon atom in martensite (ε ≃0.38). He found the

following expression for the ﬂow stress τ:

τ = τ

2Gεc

1/2

, (5.14)

predicting that the ﬂow stress is proportional to c

1/2

. The curve has a slope of

G/15 to G/20. Other experiments on martensites with low M

temperatures sup-

port the c

1/2

relationship, with slight differences in slope depending on whether

the martensite is of lath type or twinned (Fig. 5.22).

The proposal that the ﬁne twins characteristic of higher carbon marten-

sites make a major contribution to strength has not received wide acceptance.

Certainly, a large increase in strength is not found when the transition from

dislocated martensite to twinned martensite takes place. However, the high

dislocation densities of twin-free martensite must make some contribution to

strength, estimated to be not greater than 300 MN m

−2

, and there is reason to

believe that the ﬁne twinning makes a similar, but not additive, contribution.

Fig. 5.22 Effect of carbon on the strength of martensite (Chilton and Kelly, Acta Metallurgica

16, 637, 1968).

124 CHAPTER 5 FORMATION OF MARTENSITE

The austenitic grain size determines the maximum size of a martensitic plate,

so somedependence of strengthon grain size might be expected. In fact,a Petch-

type plot has been found for several alloy steels of different austenitic grain

sizes tested in the martensitic condition (Fig. 5.23). However, when the ﬁne

structure of martensite is examined other possible grain sizes much ﬁner than

the austenitic grain size can be considered as contributors to strength. Firstly,

there is the packet size in lath martensite, or the individual plate in lenticular

martensite, and beyond these there is the lath substructure which is usually well

below 1 µm in thickness. While many of these boundaries are really low angle

sub-boundaries, they do present obstacles to dislocation movement and must,

therefore, be considered to make some contribution to the overall strength.

It is also to be expected that carbon atoms segregate to the high disloca-

tion populations typical of martensite, bearing in mind the strong interactions

found in the case of ferrite. Internal friction measurements by Kurdjumov and

Fig. 5.23 The effect of prior austenite grain size on the strength of martensite (Grange,

Transactions of the American Society for Materials 59, 26, 1966).

5.7 THE STRENGTH OF MARTENSITE 125

Fig. 5.24 Comparison of the internal friction behaviour of low carbon martensites with that

of ferrite (Speich,Transactions of the Metallurgical Society of AIME 245, 2553, 1969).

co-workers have revealed the well-deﬁned temperature-dependent peak, the

Snoek peak (p. 6), which occurs as a result of the stress-induced movement of

carbon atoms in ferrite and martensite. Figure 5.24 shows that the Snoek peak is

much lower in a 0.026 wt% C martensite than in ferrite of the same composition.

This is a direct result of the reduction in free carbon atoms in the martensite

structure due to pinning by the high concentration of dislocations. These pinned

carbon atoms cannot contribute directly to the Snoek peak, the height of which

is proportional to the concentration of free carbon atoms in the lattice. In con-

trast, ferrite has a very low dislocation density and exhibits a much higher Snoek

peak (Fig. 5.24), because a greater concentration of carbon atoms is available

to move interstitially between the octahedral sites.

Work on the temperature dependence of the ﬂow stress of martensite in

Fe–Ni–C alloys has shown a strong temperature dependence, together with a

126 CHAPTER 5 FORMATION OF MARTENSITE

Fig. 5.25 Temperature dependence of the flow stress of Fe–Ni–C martensite (Owen and

Roberts, International Conference on Strength of Metals and Alloys,Tokyo, 1968).

peak in the curve associated with serrated ﬂow in the stress–strain curve (Fig.

5.25). Like the development of the yield point in α-iron, this has been attributed

to the Cottrell–Bilby interaction of carbon with dislocations.

However, this phenomenon leads to precipitation of iron carbide on the

dislocations which is responsible for the increase in strength shown by marten-

site aged at room temperature or just above. Also martensites with relatively

high M

temperatures will form cementite dispersions during the quench

(auto-tempering) which will also make some contribution to the observed

strength.

The yield strength of martensite, like that of ferrite, is markedly temperature

dependent, but this dependence is little affected by the presence or absence of

precipitate or by the amount of carbon in solution. It is, therefore, likely that

the temperature dependence arises from the basic resistance of the lattice to

dislocation movement, i.e. it is a result of the temperature dependence of the

Peierls–Nabarro force.

5.8 SHAPE MEMORY EFFECT

The shape deformation accompanying martensitic transformation can be

reversedby transforming backtothe parent phase. Suppose a crystalofaustenite