.:.

marte nsite

343

tions which

are not diffusionless

. .Thesejransformations

still involveihe

..

change of the crystal structure by shear and thus a macroscopic shape change, though they are not usually termed "martensitic." The growth of ferrite plates in bainite is a good example of this. A shape change accompanies the transformation (se e as far as the iron and substitutional solute atoms are concerned, but the rate of growth of the plates is determined by the rate of flow of the rapidly diffusing carbon atoms through the austenite 3way from the growing ferrite. (Carbon diffuses roughly 10!2 times faster in austenite at these temperatures than does iron or a substitutional solute.) The free-energy decrease for the precipitation of a carbon-poor ferrite from austenite is appreciably greater than for the transformation of the higherdefinite macroscopic

Fig. 6-20). It is diffusionless

carbon austenite martensitically. Thus in the bainite range the lattice transforms by shear at arate determined by the rate of removal of diffusion of carbon away from the region being transformed. The growth of Widmanstatten plates at high temperature seems to be a continuation of this process since the orientation relationship between a and 'Y is the same for martensite, bainite, and Widmanstatten, although the ha bit plane changes. There is also a surface relief involved in forming a Widmanstatten, whereas no surface relief is found with grain-boundary allotriomorphs of ferrite formed at the same temperature. Thus, at higher temperatures a Widmanstatten appears to form from austenite by a diffusion-accompanied shear, while a allotriomorphs forrn behind an a'Y interface at which no shear occurs.' Still otti'H examp1es are found in the coherent precipitation of transition phases such as the hcp 'Y' phase from fcc AI-Ag alloys as was discussed in Seco 6-12. Surface relief is found where these particles meet the surface. This implies some sort of cooperative movernent of the matrix atoms across the interface, though again the precipitation must be accompanied by a local change in composition. The point to be remembered is that martensitic transformations and precipitation reactions usually are distinct and different reactions. However, at lower temperatures it is not uncommon to find precipitates whose formation involves a shape change plus a composition change.

8-7.

strength

of rnartenslte? heat treated to give a marby the first set

Commercially tensitic

the highest volume of materials

transformation

are the low-alloy steels represented

! See articles by Christian, Speich, and Kaufman, el al. in "Decomposition of Austenite by Diffusional Processes," V. Zackay and H. Aaronson (ed s.), Interscience Publishers, New York, 1962. 2 M. Cohen, Trans. AIME, 224:638 (1962); W. Leslie and R. Sober, Trans. A.1I1. Soco Melals, 60:459 (1967).

344 transformations

in metals

--

1.100

1.000

900

Martensitic structure (quenched)

65

800 60
<Il

- 250 U

700

>
.
<Il
Q)

Q:; .x: .~

c;;
600 50 500

;: .x:

o a:::
.
<Il
Q)

~ o
'iJ?
N

Q)

~ '" I

o
sn .:t: ..c

400

40

~ '" I

-120

o
e

300

Pearltic structure
200

30 20 10

v;
.~

~
>-

100

Spheroidized carbide structu re

0.20 0.40 0.60 0.80

Carbono % 8-16. Hard,ss 01 various iransjormaiion producis in sleel. tFroni E. Bain and H. Paxlon, Alloyiru Elemenis in Sleel;" 2d ed., 3d rev. priniinq, p. 37, American Sociely [or Mclals, Meials Park, Ohio, 1966.) figure
r ,

of steels in Table 8-1. These combine high strength and toughness with a relatively low cost. The strength of such martensite depends primarily on their carbon content. Figure 8-16 shows the marked variation of hardness, or ultimate strength, with carbon content as well as the large difference in hardness between martensite and medium-fine pearlite or a spheroidized carbide structure of the same carbon content. The hardness of martensite is independent of the concentration of substitutional alloying elements, these elements being added primarily for their effect in shifting the TTT curve for pearlite and bainite to longer times. High strength in an alloy means that there are many barriers throughout the piece which inhibit dislocation motion through the lattice. In one sense the martensitic reaction in steel can be looked upon as the instantaneous change of austenite, in which the carbon is soluble, into ferrite, in

_\

martensite

345

wh'ich the' carbn precipitation strength this obvious is in'~'6Iubl-~This can and do~s ultimately H00e~r;sfeel' to hardening ~artnsites and consider of carbide, carbon -'ead to the bypass of very of a carbide. contribution have co-nsiderable other contributions.

before the precipitation in low-to-medium

so let us temporarily

The martensite

low-alloy steels consists

fine needles or lath.! Each of these laths is a separately formed martensite olate, with pairs of sets forming so that the shear strains accompanying transformation in each tend to cancel one another. The lattlce invariant shear of the transformation occurs by slip. The resulting martensite has a dislocation density comparable to that of a highly deformed metal (1011 to 1012 dislocationa/cm"), in addition to a mean plate or needle thickness of about 0.1 L. These lath boundaries and the tangle of dislocations provide formidable barriers to dislocation motion in and of themselves. One can't produce carbn-free martensites in pure iron that give a comparable structure, but a structure similar to this can be produced in iron by shock loading, and it has a flow strength of roughly 100,000 psi.? This is far short of the 300,000 psi strength attainable in quenched steel but is still far above the roughly 10,000 psi flow strength of annealed iron. The mechanisms by which carbon in solution in martensite contributes to hardness are built around the fact that the carbon atoms strain the ferrite lattice and this strain can be relieved by carbon atoms redistributing to relieve the strains set up by dislocations. This carbon redistribution sets up a substantial binding between the dislocations and the carbon atoms so that a large stress is required to l1-o~ve the dislotations away. This binding results from segregation to the dislocation core region, as well as from a local ordering in the stress field of the dislocation many atom diameters away. This ordering or rearrangement well away from the core can be understood with the help of Fig. 8-12, which shows the distortion along a (IDO) axis +hat is set up by a carbon atom. If the atom jumps to an adjoining interstitial position, note that the axis of the distortion shifts to a (100) direction normal to that of the original position. The dislocations also set up preferential distortion along certain directions. The carbon interstitial can thus become "bound" to the dislocation by making only one jurnp.! Carbon can and does diffuse in martensite at room temperature. One result is the binding to dislocations. Another result is the formation of carbon atom clusters. The uniaxial or dipole distortion referred to above leads to an attraction between the atoms which gives clustering on the l00 type planes. That is, the distortional energy of the carbon atoms is reduced if they cluster on a cube plane so that their axes of distortion are all parallel.
1

P. M, Kelly and J. Nutting,

J. Irori Sieel Lnsi .. 197:199 (1961).

2
3

W. C. Leslie and R. Sober, Trans. Am. Soco Meials, 60:459 (1967). G. Schoeck, H. Wiedersich, and A. Cochardt, Acta Mei., 3:533 (1955).

346

transformations

in JT1etals

This. is in effecra

~Gezone "such as ~was discussed in connection with the age hardening of alurninurn alloys i the last chapter.: The effect of this early redistribution of carbon on the strength can be studied by working with Fe-Ni-C alloys with an M, well below room temperature. For example, an Fe-25Ni-O.4C alloy has an M, "" -35C. If the martensite is formed by cooling to about -lOQoC, the carbon is completely immobilized at the temperature where martensite forms. If this steel is heated to QOC and held for a few hours, the flow stress at a few percent strain is increased substantially (see Fig. 8-17). At OC the carbon atoms make roughlya jump a second, so the increase shown here results from carbon redistribution to form clusters and to pin dislocations. The two contributions to the strength of martensite discussed above
1 R. A. Johnson, (1968).

Acla Met., 15:513 (1967); J. Genin and P. Flinn, Trans. AIME,

242:1419

Ni, appraximate
31 320.000 27 23

weight percent
19
I

15

'"

a. 280.000 e ~(l) 240;000

e
a. '"
<D O O

'" ~

R,.
",-

..

200.000 I
/

/ /

--

_-----~(2)

8 Compressian
with
O'

I 160.000 I

results va riatian

o
<1)

ro
V

1 Tensile

results with <T variatian

Cornpo srtrons. wt
%

'" ~
;; o

120.000

Ni

w::

80.000

30.8 30.5 28.4 26.8 23.3 19.2 16.7

I

40.000

wt% 0.02 0.02 0.08 0.23 0.40 0.59 0.82 0.8

o o
0.2

0.4

0.6

LO

Carbono weight percent

figure 8-17_ Flow slress al -195C oj various Fe-1Vi-C alloys iohosc A1, is so loto (-35C) thai no carbon difJusion occurs during he lransjormalion. Curve 2 is lhe slrenqlli oj lhejresh tnartensile. Curve 1 shotcs Ihe increase given by a 3-lIr anneal al DOC. [From P. l1iinchell and M. Cohen, Trans. Am. Soco Melals, 55:995 (1962).]

martensite

347

-j
seem _to be the main ones. O_neis a structurai contribution which is largely independent of carbon content; the other, strengthening due to the rearrangement of carbon atoms after quenching. The magnitude of this latter term increases roughly linearly with carbon content and in a martensite with 0.4 % e can contribute roughly twice as much to the tensile strength of the steel as the dislocation substructure. The strength of low-carbon martensites thus stems primarily from substructure, while rnediurn-carbon martensites gain more of their hardening from carbon rearrangemenU

!
i

8-8.

tempering

in plain carbon steel

Before a piece of steel quenched to martensite is put into use, it is first heated (tempered) at some temperature above room temperature but well below the Al to give it some ductility. When a piece of steel is quenched, the outside cools and contracts while the inside is still hot and plastic. As the specimen center cools, it is restrained from shrinking by the rigid cold outside. In addition to the stresses set up in this way, a volume increase of the order of 1 percent accompanies the austenite to martensite transformation. If the steel still has some plasticity at the M. temperature, these strains can be relieved by local plastic deformation. If the M. is so low that the flow stress is very high, these transformation strains set up stresses in the piece which are near, or above, the fracture stress of the piece. If they exceed the fracture stress of the-piece, it crack_s. This quench cracking can be avoided by !Wing to a milder quench or by redesigning the piece. Even if no cracking occurs on quenching, the piece may crack on sitting at room temperature. Or, if a load is applied it will add to the high locked-in stresses in some areas and cracking at a low applied stress results. If the piece is heated to only 100 e, the flow stress is reduced enough to allow local flow to partially relieve the stresses and thus avoid spontaneous cracking. Such tempering increases the toughness of the alloy and doesn't reduce the hardness at all. Figure 8-18 indicates the decrease in hardness resulting from a l-hr tempering anneal at various temperatures. The dashed curve on the figure shows the variation in toughness with annealing temperature. Toughness is here measured by the energy adsorbed by a notched specimen in breaking by impacto The as-quenched martensite is somewhat like glass in that it is quite hard but adsorbs relatively little energy in fracturing. Tempering reduces this tendency to brittleness. The iron-carbon martensite formed in steel is highly supersaturated with carbono The changes that occur in this finely deformed, supersaturated alloy on heating are quite consistent with what has been covered earlier in
1

W. C. Leslie and R. Saber, Trans. Am_ Soco Meiols, 60:459 (1967).