• 検索結果がありません。

Transmission Electron Microscopic Study of the Phase Transformations in Iron and Steel

N/A
N/A
Protected

Academic year: 2021

シェア "Transmission Electron Microscopic Study of the Phase Transformations in Iron and Steel"

Copied!
13
0
0

読み込み中.... (全文を見る)

全文

(1)

Transmission Electron

M

icroscopic Study

of

the Phase Transformations

in Iron and Steel*

By Zellji Nishiyama**

Synopsis

This /Ja/Jer is fI rCl'iew of the IrollSfnisJioll electron lIIicroJco,bic sludies carried out maill!V ill Ihe aUlhor'sformer laboratOl:V (at Osaka Vlliveni()') cOllcernillg Ihe IJlwse Irallifornwtioll ill iroll alld steel.

(I) The marlellsile ClJ'slal alw{!),s colliailis sOllle illllmjectiolls, e.g., illternal twillS, slackillg faulls, abulldalll dis/ocaI iOIlS, or Ihe mixtllres ~f Ihem. These iIllIJe(/ecliolls l'erib lhe occurrellce q/ Ihe secolld shear ill the double shear lIIechallism of martellsite trall~/or/llatjoll. The 1IIarlen"ite lattice occasionally has a slacking modulaliolls.

(2) III the lower bainile, Ihe IJrecilJilaled carbides are /llate-like alld are distributed ill Ihe directiolls of Ihe lraces of some Illalles (jJrobably of stackillg faults). III the ulJper baillite, the acicular ferrite cOlltaills a large number of dislocatiolls, Ihe carbide i),jllg outside. These structures can be illte/jHeted b), Ihe idea that in Ihe baillile reactioll a sUjJersaturated ferrite grows martellsiticali)' alld carbides precijJitate during the growth.

(3) All the celllelltile lamellae ill a colDlev of the Ilearlite luwe lIearly the sallie crystallographical orielltatioll. The same is true for Ihe case of the ferrite lamellae. The cemelltite in the IJenrlile .frequelltly has jJlalle de-fects: stackillgfnulls or sequtllcefaults all (DOl), alld sequencefaulls all

(010), (100), (01 I), (103), (021), (/1 1) alld (212).

(-I) ill the cold-rolled alld all Ilea led Fe-25°oCr-20~oNi alia)', Ihe sigma jJhase precipilales jHiferelltial1y at Ihe laltice illl/mjectiolls, e.g., at the elld of the slackillg faull, al Ihe illienectioll of Iwo Iwins of diflerellt orielltatioll alld at the grain bOUlldal)'.

I. Introduction

Recently the technique

or

transm.lsslon electron microscopy has been developed and appl ied to the stud y or the structu re or metals and alloys. Th is technique enables us to reveal the ultramicroscopic structure which markedly controls the physical and mechanical properties of' metals and alloys. About the phase transrormation, it brings variotls inrorma -tions which are important for understanding of the mechanism or the transformation.

In this article descriptions will be given mainly of the experimental results obtained in the author's former laboratory, The Institute of Scientific and In-dustrial Research, Osaka University.

II. Substructures of Martensites

It is generally approved that the martensitic trans-formation occurs by co-operative moveluent or atoms which can be described as a derormation of lattice. The deformation, however, should not be homogeneous throughout one martensite crystal 1'01' the relaxation of the strain due to the lattice change ill. tl-ansrormation. This is the origin of the double shear mechanism on the martensite transformation. I n this theory it is assumed that ultramicroscopic twinning or slipping must be induced during the transformation. These

occur by the movemf'nt or partial or perfect di sloca-tions, respectively, and there are possibilities that the rormer leave stacking faults behind and some of the latl.er remain within the martensite crystal. For the in-vf'stigation or the mechanism or martensi te lI'allsrorma -tion it is desirable to observe these remained lattice imperfections by electron microscopy.

About firteen years ago the author et ali. I), 2) carried ou t this ex perimen t by th e replica tech niq ue and they observed striations of 100~200A in spacing on the surrace rei ier arised rrom the martensi te transrormation in P -:'\i alloys and Kovar alloys. These striations were considered to give an evidence or the presence of stacking faults, but it was difficult to determine whether they were twin faults or derormation raults. For that determination, the transmission technique was em-ployed arterwards. The results obtained will be given below.

1. Trall.iformalion from Face-Centred Cubic Lattice into Body-Centred Cubic (or Telragollal) Lattice

Fe-Ni Alloys:lJ

A Fe-30.64%Ni alloy was prepared by vacuum mclting and homogenization just below its melting point. Several rolled sheets or 0.2 mm in thickness were heated in vacuum 1'01' 30 min. at 800°C, rollowed by rurnace-cooling. Subsequently they were tra ns-rormed into martensite by sub-zero cooling for 25 min. in an acetone bath cooled with solid carbon dioxide. From these sheets thin roils for transmission electron microscopy were prepared by the Bollmann's method using an electrolyte containing 200 cc of orthophos-phoric acid and 100 g of chromic acid. The electron microscope used is a Hitachi HU-Il and operated at 100 kY.

Photos I and 2 are examples of the electron micro-graphs taken: Photo I shows a number of dark parallel fine bands of about 100 A. wide, and Photo 2 a number of short bands of parallelepi ped shape. To know the entity of these structures, selected-area diffraction patterns were taken as shown at the upper right corner of each photograph. Either of these patterns indicates the existence of crystals of two twin related orientations, the twinning plane being (112) a' perpendicular to the foil plane. From this fact, together with dark field images taken with encircled twin spots, it is evident that both the long fine striations and the short bands are images of thin twin plates existing within the matrix

*

This article is rewrillen sharply in reference to the technical review, Japanese lext of which was received on February 5, 1964and

printed in Tetsu-to-Hagalli (Journal, Iron & Steel Inslitute, Japan), 50 (1964), 13, 2215-2243.

**

Dr. Sci., Prof. Emeritus of Osaka University, Yawata Iron & Steel Co., Ltd.

(2)

Photo 1. Interior of a martensite crystal in Fe-30%Ki alloy, showing long fine bands of internal twins. A photo inserted at the upper right corner is a selected-area electron diffraction pattern. A photo inserted at the upper left corner is an enlargement of the whire-rramed area in the micrograph.

Photo 2. Interior of a martensite crystal in Fe-30~bNi alloy, showing short bands of internal twins. A photo inserted at the upper right corner IS a selected-area electron diffraction pattern.

of the martensi te crystal. The di rfractiol1 spots In Photo I are elongated to the direction

<

112)., which is perpendicular to the fine striations. This elonga -tion will be caused by the thinness of the twin plates and also by the variation of the lattice spacing at the faults. The appearance of the twin plates depends on their orientation to the foil plane. The image of such a twin plate could be interpreted as the projec -tion of a slice of the twin plate having the shape of a thin ribbon whose longitudinal direction is parallel to the twinning direction <III).,. In Photo I, the twin plate looks to be a long striation, since the surface of the twin is almost perpendicular to the foil and its longitudinal direction approximately lies on the foil plane. While, in Photo 2, both the surface and direc -tion of the twin are inclined and so the projection of the slice of the twin plate appears as a parallelogram.

Each of the striations in Photo I consists of three or four lines, as shown in the upper left corner, which is an enlargement of the white-framed area in the figure.

Tetsu-to-Hagane Overseas Vol. 5 No.4 Dec. 1965 (331)

These lines are interference fringes due to the existence of one twin plate inclined to the foil plane. From the spacing of two inner lines, the thickness of the plate is estimated to be 30- 70A.

The distribution of the twin bands varies consider -ably from place to place. The overlapped regions are strongly dark. The bands are densely distributed in the central part (probably midrib) of the martensite crystal as seen in Photo 3.

Photo 3. fnterior of a martensite crystal in Fe-30%Ni alloy, showing internal twins crowded along a midrib (dark region). (By K. Shimizu)

In some cases, the short band has longitudinal In -terference fringes as seen in the upper part of Photo 4.

Photo 4. Interior of a martensite crystal in Fe-30%Ni alloy, showing short bands of internal twins with longitudinal interference fringes. (By K. Shimizu)

In Photo 5, short bands exhibi ting the so-called " moire fringe" are observed. This fringe is c on-sidered to be due to the interference of reflections from two overlapped twin plates.

Detailed inspection of Photo 2 enables us to find the deviation of the interface between a twin plate and matrix from one of the twinning plane {I12} a' as i llus-trated in Fig. 1. The amount of the deviation is almost constant throughout one martensite crystal but fluctuates from crystal to crystal in the range of 3 _ 21 o. This fact suggests that the deformation for the marten-site transformation is accompanied by additional slips.

(3)

(332) Tetsu-to-Hagane Overseas VoL 5 No. 4 Dec. 1965

Photo 5. Interior cf a martensite crystal in Fe-30%Ni alloy, showing short bands of internal twins

\Vith moire pattern. (By K. Shimizu)

Fig. 1. Schematic figure showing a twin plate, which is divided into slices by slipping along a plane parallel to [111] axis, the mean interface being deviated from (112)

Slightly uneven edges of the parallelogram in Photo 2 may indicate the inhomogeneolls distribution of such slips. If those additional slips actually take place, it is

very important for the establishment of the mechanism of the transformation. Some of the fine striations

within the parallelogram except the images due to overlapping of two or more parallelograms SeelTI to be

the traces of those slips. The quantity of the a ddi-tional deformation due to these slips will depend on the boundary condition of the manensite crystal. Fe-Ni-C Alloys

Photo 6 is an electron micrograph of a quenched 29.7%Ni-0.42%C steel, reproduced from a paper by

Tamura et ali4). Lamellae seen in this micrograph

are also internal twins. In every lamella a complex structure is found. Tamura et ali. explained that this might be due to the lattice strain caused by the segreg a-tion of carbon atoms.

Research Articles

Photu 6. Unausformcd Fe-29.7%Ni-0.42%C alloy, sho w-ing a martensite crystal with internal t\Vins. (By l. Tamura et ali.'»

Photo 7. 30% ausformed Fe-29.7%Ni-O.42%C alloy, sho\Ving a martensite crystal with broad bands. (By T. Tamura et ali.)

Photo 7 shows the structure of the martensite in a 30% au formed specimen of the same material. With-in the martensite crystal, parallel diffused broad bands are seen. If these diffused bands are due to the la t-tice strain, it is worth noticing, because they may be

related to the strengthening of the steel by ausforming. 2. Transformation ji-om Face-Centred Cubic Lattice to

Close-Packed Hexagonal Lattice Fe-Mn-C Alloys

In order to find the cause of the abnormal wo rk-hardening in high manganese steel, the author et ali.5 ) made a study some years ago by electron microscopy

using the !'eplica technique. From this preliminary e x-periment it was suggested that there might be a very

thin plate of h.c.p. Co:) phase passing through each of strain markings. Afterwards it was confirmed6) by

transmission electron microscopy, as described below. A steel plate containing 0.97%C and 9. 75%Mn was heated for four hI'. at 1,000DC in vacuum, followed by

water-quenching. Subsequently it was beaten with

a hammer to cold-work, and then electrolytically

polished to a thin foil by the Bollmann's method using an electrolyte containing 200 cc orthophosphoric acid

and 100 g chromic acid.

(4)

Photo 8. Parallel bands of ,,' phase In a beaten sheet of 9.75%Mn-0.97%C steel

a number of dark parallel bands are observed. By

selected area diffraction, it was found that they were

h.c.p. (0:')* phases produced on {Ill} planes of the

f.c.c. matrix. Photo 9 is a photograph of another

region, in which two kinds of bands are observed:

bright bands marked by " c' " are due to 0' plates and

parallel fringes marked by" S.F." are due to a stac

k-ing fault.

Photo 9. Parallel bands of ,,' phase and stacking faults (S.F.) In a beaten sheet of 9.75%Mn-0.97%C steel

As shown in the micrographs described above,

stacki ng faults are observed in some regions, whereas

c' plates are developed in other regions. Such an inhomogeneity in structure is caused by

nonuni-formi ty of the degree or rate of cold-working.

In a cold-rolled Hadfield steel (1.1 7%C, 12.4S%Mn), a somewhat different figure as shown in Photo 10 was

obtained in add i tion to 0' phase bands and stacking

faults fringes. The electron diffraction pattern of the

white-framed area in Photo 10 is shown in Photo 11.

The distribution of strong spots in this pattern shows

that the foil plane is parallel to (1TO) plane of a f.c.c. crystal which is considered to be the matrix. Other

weak spots are symmetrical in arrangement with the

spots of the matrix about the trace of the (Ill) plane

perpendicular to the foil. Therefore, the fine bands

Tetsu-to-Hagane Overseas Vol. 5 No. 4 Dec. 1965 C 333)

Photo 10. Parallel-line structure of deformation twins in a 30% cold rolled Hadfield steel (12.41%Mn, 1.19%C)

Photo 11. Electron diffraction pattern of the framed area in Photo 10, showing matrix and twin spots and streaks

observed in Photo 10 may be interpreted as twins. In the case of an alloy having rel-rods**, if the Ewald sphere cuts the rods at a large angle, the sections of the rods may often appear likely as twin spots even when there are no true twin reflections. In Photo 11, how-ever, the cutting angle is zero and therefore, the o b-tained spots are considered to be true twin spots.

As mentioned above, the cold-worked high man-ganese steel has c'-phase bands, stacking faults and fine

deformation twins. The existence of these lattice

imperfections will contribute to the abnormal harden

-ing of the steel. 18-8 Stainless Steels

Nagashima et ali.S) observed the 0' martensite in IS-S stainless steels. The author et ali.9) also observed it in 304-type stainless steels (0.06%C, O.5%Si, 1.03%Mn,

O.042%P, 0.23%Mo, IS.l %Cr and 9.7%Ni), heat

-treated and strained by tension, as shown in Table 1.

1. Structure of Specimens A and B (Furnace-Cooled

and Subsequently Deformed at Room

Tempera-ture)

Photo 12 (a) and (b) are representative electron

*

Thc prime mcnns the phase produced by a martensitic transformation.

*

*

Rods at the reciprocal lattice point'.

(5)

(334) Tetsu-to-Hagane Overseas Vol. 5 No.4 Dec. 1965 Specimen A 13 C D E F

Table 1. Treatments of the specimens \\'ay of cooling from I,OSO'C Furnace cool* Quench Deformation 3°~ tension at room temp. 7% 3.6% .. 7% 00' ,0

* 'T'hc cooling ratc ncar 700°C is about SO/min.

Martensite contained None None

"

~/+a' t.'+a' micrographs of specimen A subjected to 3 % tension

at room temperature. In (a), images of dislocation

lines tangled and piled up against the grain boundary

(--»are observed in an austenite grain. In (b), which

was taken from another portion, fringes due to stacking

faults are found sporadically. In this way, the aus -tenite grains contain only lattice defects such as

dislocations and stacking faul ts and there are no

mar-.-

~

..

Research Articles "\

\

tensi te crystals.

Photo 13 (a) and (b) are micrographs of speci men

B for which the degrec ol tension was increased up to

7%.

From (a) the stacking faults are found to be

increased both in number and width compared with

those in specimen A. In addition, dark bands are

observed in another area as shown in (b). These dark bands are s'-martensites which have been produced

on {Ill} planes of the austenite matrix by deformation

at room temperature. The diffcrence between (a)

and (b) in structure may be regarded as clue to the

difference of the true degree of strai n in each crystal

grain.

Al though a spcci men rolled by 30

'/0

red uction in thickness was also examined, no other noticeable struc

-tures could be observed than stacki ng faults and s' -martensites. Hence it can be supposed that in 3 04-type stainless steel :x' -martcnsi tes arc not produced by

,~

J

...... i.. 1: Photo 12. Specimen A of 30+ stainless steel (3% tension at room temperature) having disloca -tions and stacking faults (b)

Photo 13.

Specimen B of 30+ stainless steel (7% tension at room

temperature), having E.'

(6)

defOl'mation at room temperature even if the

deforma-tion is severe.

II. Structure of Specimens C, D and E (Furnace

-Cooled and Subsequently Deformed at - 195°C)

Specimen C, which vvas not deformed but only

dipped in liquid nitrogen, was examined by optical microscopy, but neither c'- nor et.'-martensites were observed. Specimen D, which was elongated by 3.6% at liquid nitrogen temperature, had the structures similar to those in specimen B elongated by 7% at room temperature.

Increasing the degree of tension up to 7% (s peci-men E), et.' -martensites were frequently observed in addition to the above structures, as shown in Photo

14. In (a), ct.'-martensites crowd in the region be-tween two sheets of neigh bou ring c' -martensi tes. From this it seems probable that these martensites have

been induced by the c'-martensites (<--» which had been produced at the initial stage of deformation.

All these CI.'-martensites have no twin faults as ob-served in Fe-Ni alloy and high carbon steels but COI1 -tain many dislocations. Some of the dislocation lincs

arc straigh t and arranged regularly, as shown by an arrow (<-) in Photo 14 (a).

III. Structures of Specimen F (Water-Quenched and Subsequently Sub-zero-Cooled)

Specimen F, which was water-quenched and then

cooled in liquid nitrogen, has such an optical micro

-graph as shown in Photo 15, in which both c'-and 'l.'

-martensites are found. Thus, the water-quenching promotes the formation of CI.' -martensi teo Th is can be

considered to be caused by the q uenchi ng stress.

Photo 16 is an example of electron micrographs taken of specimen F. This structure is similar to that

in Photo 14· (a) taken of specimcn E, that is, et.' -ma

r-tensites also crowd in the region betwcen two sheets (A and B) of neighbouring c'-martensites. From this

similarity, it can be conelllded that in thc water

-Tetsu-to-Hagane Overseas Vol. 5 No. 4 Dec. 1965 [335

J

Photo 15. Optical micrograph of specimen F of 304 stainless steel (water-quenched and cooled in liquid nitrogen), showing £'- and rr'-marte

n-sites in the austenite

Photo 16. Electron micrograph of specimen F (same as

Photo 15), sho"'ing (t '-martensite crowded between two parallel £' -martensite plates

Photo 14.

Specimen E of 304 stainless

steel (7% tension at - 195°C), having a'-martensites in ad-dition to the 2'-martensites

(7)

(336) Tetsu-to-Hagane Overseas Vol. 5 No. 4 Dec. 1965

quenched specimen ,,'-martensites are fOI"med prior to the formation of ("f.'-martensites, as in the fur nace-cooled specimen.

In Ol'der to examinc the details of the substnlctures in ("f.' -martensi tes, electron m.icrographs were taken in various orientations by tilting a specimen, examples of which are shown in Photo 17 (a) and (b). (b) was tilted from (a) by 4° around an axis marked with an arrow ( ). Wi thin the 0:' -martensite marked by , M ' in (a), the images of dislocations and stacking faults are scarcely recognized. On the other hand, in the same martensite plate in (b) many dislocation lines are apparently revealed. Some of these dislocations appear to be straight and parallel to the direction shown by an arrow ( i). By analyses of the electron diffraction patterns, it was found that they lie on slip

planes, {OIl}, [112) or (123) as usual.

The wide shape of the 0:' -martensite marked by , M 1' and' M ' in Photos 16 and 17, can be explained as the projection of a plate, which is nearly parallel to the plane of the foil. If the martensite crystal had the form of a needle, the projection of its section would

not be observed to be of such a wide parallelogram in shape.

IV. Relation between ,,'- and o:'-.1V[artensite

The ("f.' -martensites were found in the region between two sheets of neighbouring c' -martensites, suggesting that Gf.' -martensite can preferentially develop in the vicinity of c'-martensite already formed. However, it does not mean that c'-martensite is a nucleus and an

intermediate stage in the

r->

0:' transformation. I t can be preferably considered that 0:' -martensi te nucleates at the austenite matrix near c'-martensite, since the austenite is strained due to the formation of the ,,'-martensite. So long as such a strain is produced in the matrix, even some other phase may stimulate the formation of ("f.'. Recently Dash and OttelO) studied the morphology of the martensite, in 18%C

r-Research Articles

12%Ni stainless steel cooled to - 196°C without prior cold-working. They found ,,'-phase between two 0:'

-crystals and concl uded that ("f.' was firstly formed and

fe' appeared to be a consequence of Gf.'. This result

at a glance seems to be reverse to that obtained in the present experiment, but it does not contradict the

latter. Because their specimens were different in

composition from the author's and s' was easily formed only by sub-zero cooling but c' vvas not formed fir tly as thei r specimens were not cold-worked and had a different composition. The formation of rx' will stimulate to produce c' -phase by the transformation stress. Such a phenomenon will take place at the region between two Gf.' crystals, as they found.

3. Noriferrous Alloys

Similar striations are found in the (3' martensite of Cu-AI alloy.H),12) But in this case they are due to stacking faults (deformation faults) but not twin faults.

In the case of Cu-Sn alloy, two kinds of martensites, (3' and (3" have been found.13) (3' has stacking faults as in (3' ofCu-AI alloy, while (3" has twin faults similar to those of ferrous alloys. Besides (3" has stacking faults (probably) in its internal twins as well as in the matrix, the fault plane being oblique to the twinning plane. In ti tani um, the martensite contai ns comparatively large twins, in which many stacking faults parallel to (0001) are found.14I

4. Stacking Modulation

From the point of view of the dislocation theory F.C. Frank15) and H. Suzuki16) speculated that slippings would occur during the martensitic transformation, and they deduced the spacing of the slip planes to be six atomic layers or so. Such a small spacing is, however, difficult to be observed by the ord i nary method of electron microscopy and seems to have no relation to that of the twin faults mentioned above. But if the spacing is nearly constant throughout the crystal la

t-r-hoto 17. Specimen F (same as Photo 15). The dislocations in martensite M are clearly revealed in (b) which was tilted by 40 from (a).

(8)

tice, it is possible to be observed as the modified

elec-tron diffraction pattern 0(' the lattice. The

observa-tion may be easy when the parent lattice has a

super-lattice.

On this basis, the author et ali.17) studied

13'

mar-tensite formed from the f3,-phase ofCu-AI alloy. This

martensite had already been studied18), 19) by X-rays

using the powder method and yet it had not definitely

been determined, because the X-ray patterns obtained

were so complex. The reason is that the X-ray tech

-niq ue needs the considerable size of crystals of mar-tensite which is not attained in this case. On that

account the author et ali.17) studied this problem by

transmission electron microscopy, and they obtained

electron diffraction patterns of the single crystals of

13'

martensite, and the electron micrographs exhibited

fringes of stacking faults distributed in irregular spac

-ings of the order of 50

A.

.

From the electron diffrac

-tion patterns, the crystal structure was determined to

be monoclinic, whose unit cell consists of six

close-packed layers. The lattice has a structure which can be formed from f.c.c. lattice (expression ignoring the superlattice) by shearing in which the glide occurs in

the reverse direction on every three layers. This re

-verse glide is a mod ulation in atomic stacking and

may be considered to be contained in the homo

-geneous shear of the double shear mechanism of the

martensite transformation.

5. Conclusiolls

Summarizing the facts described above, it is con

-cluded that the martensite crystal always contains

some imperfections, e.g., internal twins, stacking faults,

abundant dislocations, or their mixtures. These de -fects can be considered to be traccs of the second

shear in the double shear mechanism of the marten

-site formation.

III. Structure of Bainite

Concerning the formation mechanism of the bainite,

several models have been proposed. It is, however,

difficult to decide which of those models is most

ade-quate, because of lack in detailed experimental data

concerning substructures, especially lattice imperfec

-tions contained in bainite. Metallographic

observa-tion with a hot-stage optical microscope shows that

the surface relief arises in bainite transformation. This

suggests that the transformation is accompanied by

a shear which is characteristic of the martensitic trans

-formation. Another important fact is that bainite is

a mixture of ferrite and carbide, and that its formation is controlled by the diffusion of carbon atoms. From these facts it is supposed that in the bainite reaction,

a supersaturated ferrite is first formed martensitically

(by the co-operative movement of atoms), and that

the ferrite immediately decomposes into a carbon-poor ferrite and carbides. The growth of a plate of the

supersaturated ferrite will be allowed as much as

the stress relaxation due to the precipitation of carbides

in the already-formed part of the ferrite plate.

When the initial supersaturated ferrite is assumed

to be formed martensitically, it is of interest to know

Tetsu-to-Hagane Overseas Vol. 5 No. 4 Dec. 1965 [337)

what kind of lattice imperfections exists in the ferrite.

For this reason, Shimizu et ali.19) observed the bainite

by transmission electron microscopy.

The material used in the experiment is a sheet of

0.7% carbon steel, 0.2-0.3 mm thick. Disks taken

from this sheet were heated in vacuum for 30 min. at

950°C, followed by dipping them into a lead-bismuth

bath. They were held at 300°C for 30 min. for the

formation of the lower bainite and at 450°C for 80 sec.

for the formation of the upper bainite.

1. Lower Bainite

Photo 18 is an electron micrograph of the lower

bainite. This structure consists of lens-shaped grains

which contain large number of internal fine prec

ipi-tates. Of course, the lens-shaped grains are ferrite

and the precipitates are cementite. The precipitates

have the form of a plate and they are arranged parallel to a lattice plane in a ferrite grain.

Photo 18. Lower bainite produced by heating at 300°C

for 20 min., having acicular ferrite grains

in which fine cementites have been formed parallel to one kind of planes

For reference, electron micrographs were taken of

the tempered martensite, which had been obtained by tempering at the same temperature and for the same

minutes, after quenching from a temperature of the

austenite range. Photo 19 shows a region in which

twin faults were produced during the prior martensite transformation. The cementites in such a region are

precipitated mainly along the planes of the twin faults. But, as indicated by a double arrow, the precipitation

along another kind of lattice planes occurs in the region

where the twin faults are not apparently observed. In

(b), which has such structures over the whole mar

(9)

(338) Tetsu-to-Hagane Overseas Vol. 5 No.4 Dec. 1965

Photo 19. Martensite tempered at 300°C for 20 min. (a) shows a part having internal twins and (b) another part having no such twins.

tensite plate, the precipitated cementites lie on three

or more kinds of lattice planes. In this way, the pre

-cipitation behaviour of the carbides in the tempered martensite is different from that in lower bainite. This supports the speculation that in the bainite the precipitation occurs during the growth of the initial supersaturated ferrite, but not after its full growth. 2. Upper Bainite

Photo 20 i an electron micrograph showing the

structure of the upper bainite which consists of acicular

ferrite grains and cementite plates situated side by

side. Within the ferrite grains there are a number of

dislocation lines.

Photo 21 (a) and (b) are reference micrographs of the martensite tempered at the same temperature and for the same minutes as in the upper bainite. In (a)

are observed several dark bands which can be con

-sidered to be the traces of internal twins produced during the martensite transformation. (b) shows another region where such traces are not observed.

The precipitation behaviours of the cementite in such

regions are very similar to those in the corresponding regions in the martensite tempered at 300°C. The resulting structures are distinctly different from those of the upper bainite.

3. Discllssion

In the electron micrograph of the lower bainite,

Research Articles

Photo 20. Upper bainite produced by keeping for 80

sec. at 450°C, having acicular ferrite grains containing dislocations and large cementite plates

(10)

the author could not detect any bands which might be considered as the traces of internal twins as observed in thc usual martensite. It follows from this that the initial supersaturated ferrite produced martensitically must contain some defects other than the twin faults. On the other hand, it was found that the cementite in the lower bainite was precipitated along only one kind of lattice plane over the whole ferrite grain. This sug -gests the existence of plane defects extending over the whole ferrite grain and being parallel to one lattice plane (probably {l12}). These plane defects are considered to be the stacking faults similar to those observed in the

fJ'

martensite of Cu-AI alloy.

As for the upper bainite, the acicular ferrite has a large number of dislocations, which can be predicted from the following reason: - if the slipping as the second shear is performed by motion of perfect dis -locations, then some of these dislocations may remain in the ferrite as in the case of the martensite in low carbon steels having a high transformation tempera -ture.

As mentioned above, the initial supersaturated ferrite in the upper bainite formation contains no plane defects that can induce the precipitation of carbides. While, carbon atoms can difTuse out, due to the high temperature, by pipe diffusion along dislocation lines, which results in the formation of large carbide plates at the boundaries with the surrounding austenite.

IV. Morphology of Pearlite and Plane Defects in

the Cementite

1. Morphology

rif

Pearlite

It is well-known that a massive grain of pearlite is subdivided into several colonies, and that, in one c olo-ny, the pearlite lamellae are all nearly parallel but in some places lines of discontinuity exist in the lamellar pattern. Concerning the line of discontinuity, it has been believed that two colonies growing in the same direction in an austenite grain have met each other along this line. But Frank and Puttick20) have suggested that such a discontinuity is also possible to occur even in a single colony if a trough is formed in the growing colony surface by inhomogeneity in the growing austenite, or by anomalous interlamellar spacing. If their suggestion is appropriate, the dif-ference in lattice orienta tion between the regions on both sides of the discontinuity line must be small. This is a problem to be solved.

Whether each plate in the pearlite lamellae is a single crystal or not is another problem. Wever and Koch21) directly observed cementite foils in the residue

of electrolytic extraction by means of electron mic ro-scopy and found extinction contours in a cementite plate. The appearance of the extinction contours is an evidence that each cementite plate in pearlite is a single crystal.

The next problem is whether or not all the cemen-tite lamellae in a pearlite colony have the same orien -tation. Belaiew22 ) verified by observations on frac -ture-cleavage planes that they had the same orienta -tion. A similar conclusion was given to the ferrite

Tetsu·to·Hagane Overseas Vol. 5 No. 4 Dec. 1965 [339)

lamellae by Hull and Meh[23), who used etch pits for the determination of the orientation.

The author et ali.24 ) studied the above problems by transmlSSlOn electron microscopy. The specimens used were 0.7 -O.8%C steel annealed at 950°C for 30 min., followed by cooling at the rate of I-5°C/min near the eutectoid temperature.

Photo 22 is a typical example of electron micro -graphs. From this it is evident that rods and dots which are conspicuously revealed in the micrograph are sections of plates and rods of cementite, respectively. The matrix in this micrograph is ferrite. ear the grain boundary, the cementite lamellae have stopped their growth earlier than ferrite lamellae.

Photo 22. Pearlite colonies in O.7%C steel

Photo 23 shows the inner part of one colony. It is seen that the cementite plates have extinction c on-tours within them, and the contours in some plates are all similar in appearance. These facts support the view that each cementite plate in pearlite is a single crystal and that some cementite plates in a colony have approximately the same orientation of lattice.

The same is almost true for the ferrite. Strictly speaking, however, there is a discontinuity of the extinction contours between the neighbouring ferrites, as shown in Photo 23. This implies that there is a

(11)

( 3401 Tetsu-to-Hagane Overseas Vol. 5 No. 4 Dec. 1965

Photo 23. Pearlite lamellae in O.7%C steel. Note

ex-tinction contours appearing similarly in all

ferrite plates.

small difference in orientation between the neighbo ur-ing ferrite plates.

The ferrite regions on both sides of the discontinuity in a colony, e.g., C1 and C2 in Photo 22 have each different darkness and therefore, the different orienta

-tion. The difference seems to be as small as in the

case of a subgrain boundary. This will be a verifica

-tion of Frank and Puttick's idea regarding the origin of the line of discontinuity.

The ferrite region,

N

in Photo 22, is divided into

subregions by a network and a cementite rod exists

at the node of the network. Since such subregions occasionally exhibit a different contrast indicating the difference in orientation, as seen at Bl and B2 , they

may be called subgrains.

Photo 24 is an example exhibiting coherency be-haviours between the cementite plate and the fen'ite

one. The bands (0) appearing in this figure are

sec-tions of cementite plates in pearlite. These bands

exhibit complex figures probably due to moire pa t-tern affected by boundary lattice imperfections which

seem to have some relation to the dislocation appearing

in the neighbouring ferrite (IX). At the boundary be -tween the cementite plate and the ferrite plate a

featherlike structure of a certain width is seen. This

Research Articles

Photo 24. Pearlite lamellae in O.7%C steel. Note

squamous moire patterns and the feathe r-like structure at the boundary between the cementite plate and the ferrite plate.

boundary structure suggests that there are parallel

arrays of defects due to the lattice coherency between

cemen ti te and Ferri teo

2. Lattice Dejects in Cementite

The author et ali.25 ) examined the substructure

or

the cementite in pearlite.

The specimens used are cementite foil particles

extracted with 50% nital from a Swedish 0.8%C

steel which was heated for 30 min. at 950°C and

cooled at the rate of O.6°C/min through the eutec -toid temperature. They were put on collodion films for observation.

About 10% of the observed particles had parallel

striations as shown in Photos 25 and 26. From their

diffraction patterns, they were found to be plane

faults as noted in the figures. Analyzing all the pho

to-graphs taken, each of the plane defects was inter-preted as one of the following stacking faults or se

-I>hoto 25.

Interior of a cementite crystal, having

sequpnce faults on various planes

..

(12)

quence (i) (ii) (iii) (iv) (v) faults :-(001) stacking fault* (010) and (Okl) (k>l) sequence fault (100) sequence fault (001) and (hkl) (h~l) sequence fault (103) twin fault They are prcdicted on the assumption that thc tri

-angular prism consisting of six iron atoms around one carbon atom is not destroyed by the fault.

The striations observed in the photographs except the (001) fault have a complicated figul"e like a bouc -le, which indicates the presence of lattice strain ncar the fault. On the contrary, the striations due to the

(00 I) faul t, as shown in Photo 26, are yery mooth in appearance and sometimes each of them consists of fine parallel fringes as in the case of the ordinary stac k-ing fault. This behaviour of traces shows less irregular strain at the fault plane, as expected from the atomic arrangement on this fault plane.

The study of the plane defects described above is of importance, because they give us much informations about the history of the steel.

v

.

Sigma Phase Precipitation

It is well-known that a-phase precipitation makes a cause of temper- or anneal-brittleness. Guarnieri et ali.27) reported that the formation of a-phase in Fe-Cr-N alloy was accelerated by cold work prior to a forming heat-treatment.

Nenno et a1.28) made an experiment to obtain some

information about this problem in an Fe-25% Cr-20%Ni alloy. The specimens were cold-rolled up to about 20% reduction in thickness and annealed at 700°C.

Photo 27 is an example of electron micrographs showing the precipitation of the a-phase, which oc -curred on a stacking fault marked by A. From selected-area di ffraction patterns, two parallel rods marked by Band C are found to be precipitates of a-phase. The fine structure observed in regions

D and E may be some moire pattern due to the pre -sence of thin precipitates on the stacking fault planes. Since the stacking faults, A and A', seem to lie on differ

-ent planes parallel to (Ill), the precipitates, B and

Tetsu-to-Hagane Overseas Vol. 5 No. 4 Dec. 1965 [341)

Photo 26.

Interior of a cementite crystal, having sequence faults on two kinds of planes

Photo 27. Sigma phase precipitated at stacking faults.

Fe-25%Cr-20%Ni, annealed for 3 hr. at 700°C after 20% cold rolling.

Photo 28. Sigma phase precipitated at the intersection of two twin bands of different direction.

Fe-25%Cr-20%Ni, annealed for 10 hr. at 700°C after 20% cold rolling

*

On the (001) stacking fault, Ken26 ) published a paper during the preparation of the original paper.

(13)

(342

J

Tetsu-to-Hagane Overseas Vol. 5 No.4 Dec. 1965

Photo 29. Sigma phase precipitated at a grain boundary. Fe-25%Cr-20%Ni, annealed for 3 hr. at 700°C after 20% cold-rolling.

C, are at the ends of the fault planes. Photo 28 shows the (I phase precipitated at the intersection of two

deformation twin bands. Photo 29 shows that pre

-cipitation of the (I-phase occurred in a row at a grain

boundary.

Thus the (I-phase precipitates preferentially at the

lattice imperfection. It is, therefore, concluded that

the cold work accelerates (I-phase precipitation.

REFERENCES

I) Z. Nishiyama: Symposium at the 3rcl Congress of the

International Union of Crystallography (1954); Z. j ishiyama, K. Shimizu & S. Sat6: A1em. Inst. Sci. Ind.

Res., Osaka Univ., 13 (1956), I.

2) Z. Nishiyama, K. Shimizu & R. Kawanaka: A1em. Inst.

Sci. Ind. Sci., Osaka Univ., 16 (1959),87.

Research Articles

3) Z. Nishiyama & K. Shimizu: Acta Met., 7 (1959), 432; 9 (1961),980; K. Shimizu: ]. Plzys. Soc., japan, 17 (1962), 508.

4) 1. Tamura, H. Yoshimura, M. lbaraki & M. Tagaya: l11em.

Inst. Sci. Ind. Res., Osaka Univ., 19 (1962),67; ]. japan Insl.

Metals: 27 (1963), 206.

5) Z. Nishiyama, K. Shimizu, M. Oka & T. Hiromoto: A1em.

Inst. Sci. Ind. Res., Osaka Univ., 16 (1959), 73.

6) Z. Nishiyama & K. Shimizu: ]. Plzys. Soc. ja/Jall, 15 (1960),

1963.

7) Z. Nishiyama, M. Oka & H. Nakagawa: Trans. JIM, 6

(1965), 88.

8) S. Nagashima & Y. Matsuo: The Annual Meeting of Japan

lnsl. Metals, (1962).

9) Z. Nishiyama, K. Shimizu & S. Morikawa: NIem. inst. Sci.

Ind. Res., Osaka Univ., 21 (1964), 41.

10) .1. Dash & H. Otte: Acta Met., 11 (1963), 1169. II) Z. Nishiyama & S. Kajiwara: Trails. JIM, 4 (1963), 127.

12) P. R. Swann & H. Warlimont: Acta Mel., 11 (1963),511.

13) Z. Nishiyama, K. Shimizu & H. Morikawa: To be pu b-lished.

14) Z. Nishiyama, 1'vI. Oka & H. Nakagawa: ]. japan Inst. Metals (.Japanese), 29 (1965), 133; 139.

15) F. C. Frank: Acta Met., 1 (1953), 15.

16) H. Suzuki: Sci. Rep., RITU, 6 (1954), 30.

17) Z. Nishiyama & S. Kajiwara: japan]. AjJJJl. Plzys., 2 (1963), 478.

18) N. Nakanishi: Trans. JIM, 2 (1961), 792.

19) K. Shimizu, T. Ko & Z. Nishiyama: Trans JIM, 5 (1964), 225.

20) F. C. Frank & K. E. Puttick: Acta Met., 4 (1956), 206. 21) F. Wever & W. Koch: Stahl u. Eisen, 74 (1954), 989. 22) N. T. Belaiew: Mineralogical Mag., 20 (1924), 173.

23) F. C. Hull & R. F. Mehl: Trans. /ISM, 30 (1942), 381.

24) Z. Nishiyama, A. Kore-ecla & K. Shimizu: ]. Elec

tron-microsco/ry,7 (1959), 41.

25) Z. Nishiyama, A. Kore-eda & S. Katagiri: TrailS. JIM, 5 (1964), 115.

26) A. S. Kehl: Acta lvlet., 11 (1963), Ill.

27) G . .1. Guarnieri, J. Miller & F. J. Vawter: TrailS. ASM, 42 (1950), 981.

28) S. Nenno, M. Tagaya, K. Hosomi & Z. Nishiyama: Trans.

Fig.  1.  Sc hematic  fig ure  showin g  a  tw in  plate ,  whic h  is  d ivided  in to  slices  by  s lip ping  a long  a  pla ne  pa ra llel  to  [111]  ax is,  th e  m ean  interface  be ing  deviated  from  (1 12)

参照

関連したドキュメント

In [1, 2, 17], following the same strategy of [12], the authors showed a direct Carleman estimate for the backward adjoint system of the population model (1.1) and deduced its

In the first section we introduce the main notations and notions, set up the problem of weak solutions of the initial-boundary value problem for gen- eralized Navier-Stokes

Furthermore, the upper semicontinuity of the global attractor for a singularly perturbed phase-field model is proved in [12] (see also [11] for a logarithmic nonlinearity) for two

— We introduce a special property, D -type, for rational functions of one variable and show that it can be effectively used for a classification of the deforma- tions of

Beyond proving existence, we can show that the solution given in Theorem 2.2 is of Laplace transform type, modulo an appropriate error, as shown in the next theorem..

V ulpe , Classification of cubic differential systems with invariant straight lines of total multiplicity eight and two distinct infinite singularities,

approah, whih is based on a step by step onstrution of the walks [6, 5℄.. We repeat in Setion 3 the proof

If the tree is oriented from the root to the leaves, the first corner of v is at the right of the head incident to v as shown in Figure 15.. v first corner