Thermal stabilization of austenite in iron-carbon-nickel alloys

THERMAL STABILIZATION IRON-CARBON-NICKEL

OF

E. R. MORGANt

AUSTENITE ALLOYS*

IN

and T. KO$

The thermal stabilization of austenite above and below rlf,? during cooling and during isothermal holding has been investigated. It has been found that X, is depressed by stabilization above X., and the effect is reflected in an increase in the amount of retained austentite when the steel is slowI) quenched to a reference temperature. The amount of martensite lost during stabilization may be recovered by sufficient further cooling. A mechanism of stabilization based on the concept of an increase in the resistance to martensite formation has been suggested. In nickel steels, this increase could be caused by the formation of a Cottrell atmosphere around dislocations. LX ST.~BILIS.~TIOS

‘I’HERMIQI-E DE L’i\lJS’I‘ENITE FER-CARBONE-NICKEL

DANS

LES

ALLIAGES

La stabilisation thermique de l’austenite a et6 invcstiguCe au-dessous et au-dessus de Y& lors du refroidissement et lors du maintien isotherme. II a 6th constat que ,l$ est abaissP par stabilisation au-dessus de dt et I’effet se manifeste Dar un accroissement de la quantltC d’austenite retenue quand I’acier est lentement trempC jusqu’B une temperature de rCf&ence: La quantitb de martensite p&due pendant la stabilisation, peut Ctre r6cupCr6e en refroidissant suffisamment par aprts. Un mCcanisme de stabilisation bash sur le concept d’un accroissement de la rCsistance B la formation de la martensite a et6 propose. Dans les aciers au nickel, cet accroissement pourrait &re causC par la formation d’un “nuage” (atmosphc\re) de Cottrell autour des dislocations. THERMISCHE STXBILISIERl~NGEN EISEN-KOHLEXSTOFF-NICKEL

VON AUSTENIT LEGIERCNGEN

IN

Die thermische Stabilisierung von ,\ustenit wird oberhalb und unterhalb & (d.i. die Temperatur, bei der die LIartensit-Transformation wzhrend des Xbkiihlungsvorganges beginnt) wghrend der Abkiihlung und w%hrend isothermer Haiteperioden untersucht. Es stellte sich heraus, dass A& durch Stabilisierung oberhalb van dV8herabgedriickt wird; dieser Effekt zeigt sich in einem Anwachsen der. ?vIenge des verbleibenden Austenits, wenn der Stahl langsam auf eine Bezugstemperatur abgeschreckt wit-d. Der Martensit, der wghrend der Stabilisierung verloren geht, kann durch hinreichende weitere Kiihlung zuriickgewonnen werden. Ein Mechanismus fiir die Stabilisierung, der auf der \70rstellung eines r2nwachsens des Widerstandes gegen die Martensit Formation basiert, wird vorgeschlagen. Dieses Xnwachsen kann in Nickelst~hlen zur Bildung von Cottrell-Bereichen urn die \7ersetztmgen (“Dislocations”) hervorgerufen werden. 1. INTRODUCTION It

has long been

knolvn

austenite-martensite fluenced [l]

more

retained

that

transformation

showed

that

oil-hardening when

water-quenched.

certain

and

period,

It

the

result

instance,

steels been

held

after a certain

Mat-

than

observed range is

isothermally

the transformation

in-

contain

oil-quenched also

of the be

the martensitic

steel

s;lrill- resume immediately but only

has

through

can

For

austenite

if the cooling

interrupted

in two ways:*

the progress

by the rate of cooling.

hews when

that

for a

does not neces-

when cooling is continued temperature

lag, which may

in a l&s

of martensite when the steel is temperature. This quenched to a fixed reference phenomenon, first reported b>TGulyaev [2], is known as “stabilization “thermal

differentiate effect

of

of austenite.”

stabilization” between plastic

transformation,

However,

this

effect

deformation which

stabilization.” Deg_ree of thermal

the term

is to be preferred

may

and on

the

the

be called

in order to retarding

martensitic “mechanical

METALLL7RGIC.-1,

amount

can be expressed

\-OL.

1, J-IS.

1953

(i)

in terms

of retained

is quenched

to

of the

austenite

a reference

ture at which cooling which

when cooling The increase high

[2;

with

[6;

temperature

on the degree

amounts

for tool steels

111. However,

.then for a given

holding

of prior transformation.

tion to yield conflicting

[4] indicated

red throughout,

This

investigations results

of [j],

carbon

with increasing

when

the complexity

has led the two principal

plain

the

of stabilization

steel,

introduces

the amount

[7] and

increases

temperature

temperatures

T,, at

has been found to

time

61, low alloy

steels [8-lo]. The rate of stabilization

sidered,

and

the temperarecommences

when

is constant,

progressively

alloy

steel

[3];

[4].

of stabilization,

prior transformation

6, in

the

and that,

transformation

is resumed

extent

0, between

is interrupted

the martensitic

increase, when

temperature

(ii) in terms of the difference,

Petrosyan stabilization

*Received August 8, 1952. tDepartment of Metallurgy, LJniversity of Birmingham, England. Now at Scientific Laboratory, Ford Motor Company, Dearborn, Michigan, U.S.A. $Department of bletallurgy, University of Birmingham, .\CTA

the

effect

of

is con-

at different of different complication on stabiliza-

[3; 41.

that

the transformation

stabilization range,

occurand that

the amount of stabilization, in terms of 0, increased with decreasing holding temperature for a fixed time, passed through a maximum, then a minimum, and *‘rhe notation used in this paper is summarized in ‘I’ablc I.

iollo\\~itl~ Itohlitlg ior Ixv-iotls Ciftc.r tll~s i5oth~~rtlt~il tt-,ttisfot-m;itiotl had cx~;iwci 1151. I ligh I&itv c~hromiittn high cxrboti allo!-s Il,l\~c Ixwt iotttld 10 1x I);~t-tic.itlarl\- sctisiti\-c to cooliti~ trates [20] :ltl(l in cert;tin c;isc's AIL,<\v;is dcyrcssv(l I)! t~cciltcitl~ f IlC coolitig

r;itc

l’rotii

the

attstctiitiAng

2. EXPERIMENTAL

t~tll~~c~t~~ttltt~~‘.

PROCEDURE

2.1 Materials

Used

.\ svrics 0i allo~~5 c~otitaitlitl~ front 0.0 to 1.3 1x21c~3it c;whon antI 0 to 10 lw- wn t iiickel have becw iii\-estix:itcd. .All killo!.s except a coinmet-cial c;trhon steel \vere nicl ted under xgon from c.lcc.trol!.tic. iron, cat-bon. C‘hcmical an;ilvses carIJon\-I tiicl;el and intlicated that t hew uxs no longitrtditi;tl segregation in the chill-cxst ingots and the radial segregation \V;LS remo\*ecl I)\- subsequent liomogetiiz;ttioti. ‘i‘llc cwmpositiotis of the ailo\.s are sho\vn iii ‘I‘;ibk I I. ‘l‘hc alloy contnitied a small amount 0i impitrit!-, ;mti a t>.pical ,ittal!xis is: 0.04 per wnt 1111, 0.07 Iwr (wit Si, 0.020 lxx- cent S and 0.020 Iwr wnt I’.

MO!: 1oc 1OCl Y IOCZY IOC5N lOCION 12C2N l3C3N C;trlx)11 Steel

For ;t scrics of four steels containing 1.1 per cent ot carbon and \-ariotts amounts oi chromium or nickel (11) to 5.5 per cent, Harris and (‘ohen [3] iound that the maximum ;m~omt t of stabilization \\‘;Is xi\-eii I,\-:

l’cr cent C 1.01 I .Ol 0.92

I .os 0.98 I 26 1.32

I oi

‘I‘he ingots homogenized ior eight hours at 1 100°(‘ (2010°F) under ;tyon anti nickel plated, NW-~ forg:ed (4: 1 reduction), and the central portions oi the l)ars were iurther homogenized at 1 l.WY (2100°F) iii1 t il test samples shouwl no signs of heterogetieit!~. 6,,,;,, = cY~tlstntlt . (CT,- I-,), ?I‘he bars uwe ground betxveen v;~ch stage. Aftet annealing at 475°C’ (887°F) for- 18 hours, the!- I\YY-C~ \\.hct-e I‘, is t-he holding temperature and cs is a ground to 0.5 in. diameter and machittc~ti to 0.1 25 temperature above whtch no stabilization UYIS in. discs. ‘I’hesc discs were stored iii il dessir;ttot- to tlct-ecteti, and prexwit surface rusting. ‘I‘he use oi ti;t t specimens rs = 0.57 (MS) + 26°F. ensured that the sec+on examined. which \V;IS I kpression of pat-t of the martensitictranslorm;tIxu-allel to the flat sitrfxe, \vas Irev irotii sttc-h st w-5 tion range has also been observed for iron-nickel xntl temperature gradients ;Is \\.otllti have IKYYl prociitcetl in ;I cylindrical qxcimcti. alloys 112; 131 when the quench was interrupted and the temperature held constant. Partial transforma2.2 Austenitizing and Quenching I ion in the lower bainite range, above _I&, has been found [ll ; 14-191 to iotver JI, although in at least -4 high austcnitizing temperature IV;LS used. as one case [t 61 parameter measurements suggested results oI)taineci by Sheehan, Julien and ‘Troiano that there UYIS no compositional change in the [21] indicat-e that in certain allo\-s of the present austenitc. JIoreover a lo\vering of .U, \\-a~ found type there is ;I possibiiit?- of graphitization during

the anneal before machining. In order to avoid either carburization or decarburization of specimens, austenitizing was carried out in a modified salt bath, bvhich was neutral to 1.1 per cent alloys at 1100°C (2OlO’F). All the specimens were austenitized at this temperature. The specimens were suitably mounted on a wire to avoid contamination and lowered into the quenching media in a horizontal position. The equipment was so arranged that the transference of specimens from the austenitizing furnace to the quenching bath and from the quenching bath to the tempering bath could be made in less than 2 seconds. The specimens were retained in the quenching medium for 10 seconds beyond the time required to reach stead\- temperature in that medium. For quenching below 0°C the maximum quenching rate uxs obtained b\- quenching into brine at 0°C followed immediately by refrigeration. Cooling rates in all the media were determined at selected temperatures by means of a standard specimen on to which were welded thermocouple’ elements, one on each flat side of the specimen. For measurement of the quenching rate in mercury, an insulated alumel wire was sealed into the specimen while a chrome1 wire was welded to one flat face. The cooling- curves for brine showed that there was a difference of l/40 second in the respective times required to cool through the range 400-350°C in these tlvo arrangements. Figure 1 illustrates, for saturated brine, the effect of raising the quenching bath temperature upon the quenching effkienc\r when the temperature of the medium is raised from 23°C (73.4’F). There is a marked falling off above 66°C (150.8”F). This effect, as will be seen later, is reflected in the progress of the martensitic transformation as determined b! quenching into brine. Similar quenching curves are

shown in Figure 2 where mercury is the quenching medium. In this case there is a smooth change in the quenching efficiency of mercury as the bath temperature is raised.

FIGURE 2. The effect of raising the quenching bath temperature upon the cooling rate of 0.500 in. diameter 0.125 in. thick steel discs quenched from IlOO’C (2012°F) into mercury.

2.3 Estimation

of the Amount

The individual specimens were examined for macroscopic heterogeneity and for carburization or decarburization at the edges after heat-treatment. The actual surface examined was exposed by the removal of 0.025 in. from one flat surface. No specimen was accepted if it showed any heterogeneity. The amount of martensite was estimated b> counting after tempering for 30 seconds at 300°C (572”F), a time and temperature established as satisfactory for the present alloys. Specimens in a few series were tempered at other temperatures, as indicated on the diagrams. The reproducibility from specimen to specimen was of the order of A 1.5 per cent martensite. All results for individual specimens had a 95 per cent probability of an error not more than f 1.5 per cent. The etchant used was 4 per cent nital with the addition of 1 per cent cetJ.1 dimethyl benzyl ammonium chloride. 3. EXPERIMENTAL

3.1 Transformation am

of Martensite

RESULTS

Curves

Curves indicating the progress of transformation (M-r curves) are given in Figures 3 to 10 inclusive. These show that: (i) For the alloys studied except 10(*10X and 13CSN, M,Y is dependent upon the quenching medium used, an indication that the rate of cooling through the austenitic range must have an effect upon the martensitic transformation.

Ft(;C’Kt: 1. The effect of raising the quenching bath temperature upon the cooling rate of 0.500 in. diameter 0.125 in. thick steel discs quenched from 1100°C (2012°F) into brine.

(ii) For a given alloy the results from an individual quenching medium lie on a smooth curve, but the curves for the different media do not coincide.

w

x

0”

I

1

FIGURE 7. Martensite transformation curves cent carboll 10 per cent nickel-iron alloy.

for

I.0

per

FIGURE 9. Martensite transformation curves cent carbon 4.7 per cent nickel-iron alloy.

for

1.3 per

FIGURE 8. i\Iartensite transformation cenJ carbon 2 per rent nickel-irott alloy,

cwvcs

far

1.3 per

FIGURE 10, Martensite transformation curves for 1.1 per cent carbon steel (0.4 per cent Mn, “0.1 per cent Ni).

\iol<(;

\\’

I\():

.\I,,

tiI~I;.\i

I

\I.

\ I IO\

sI’.\I~II.l%

OF .\I~S~IlC\I

II

I I,’

_ISI‘11 10. h to 2, had no e1’fcc.t011 the prop-ess the rnartcnsitic. 1r;lnsformatiotl.

of

(iii) Tlir ;miti)uiit 0i mu-tensile iOl-l~~etl iii 110th inet-cur! - ;~ntl oil-clueii~‘hetl slwciincns is linearI\rcl,ttcd to temlmxture fr-om _\I,\ to almut 70 1)c.r wit d martensite, except iii the wsc’ 0i the niercur!-c~~~w~ckxlallo! 12(‘2S (I=ig. 8) for which the WI-\-~ is linear up to almit 50 lwr cent martensite. In the GISC of brine quenching there is a di\-erg-ence from linearit\-

aho\-e

marlml

;is

cnving

70°C

the

to the r;tpitl

of the brine. linearit\*

(158°F)

x\-hich

lmvering

rn;ul~~ other

curves

Imvcr

of departure

nex-

from

M,, ;LS reported

\vho wed commcrrial

ma-kers

more

untloul~tetll~

of the cIucnching

‘I‘hei-e is no evidence

of the J/-T

hecomes

rises,

temperature

I)\.

steels.

(iI*) The JImI‘ curves tail elf at martensite contents higher than 85 per cent, antl the transiorm~~( - 315’1-3. The tion is not complete at - lW(‘ ~IIIOLII~t of retained austenite at this t-emperaturc varies

from

per cent The

ahout

fact

72°C

there

uxs

.lI, (brine).

with

verified

i1.h en and

F) it containe(l co]-rcspo”(ling Grange

.l/,v for

ninnner:

\vas

quenched

in oil to

this

same

specimen

Iwine-quenched

to f4”C

0i mxtensite,

of a specimen

[22] has pointed near

;L more

following

14 per cent to that

curves

of JI,y during

the

lies between T& (oil) and after temlm-ing sho\vetl no

Esm~ination

re-austcnitized

lO(‘ to 17

in the

lO(‘SS

Ivhich

martensite.

J1p-T

is a depression

of allo?-

(161.6”1;),

dark

for allo!-

out

M,

vxs

(165.2’ ;I value

usecl only

that

tailing-off

increases

with

once. of the

increasing

\\-ith Iaborator~~ melts I\-here care has heen tnlien to reduce the segregation. During the prelxu-ation of alloy for the present \vork, special prcr.autions \verc taken to minimize the sty-qat iou oi :rllo)kg elements and this prohabl?accounts h- the resultant .GI‘ curves linear

.-\llo\Figure

lO(‘2S

So.

has ;I mixed

.G’I‘1I

grains

So.

\vhich

size,

5 to 6, but there

ha\~

size is not clear.

the grain

grain

of the grains an effective

;\s sho\vn

.Actual

size had no effect

transformation. mined

the

found

that

Grange .I1,v of a 0.65

pxin

per

of =\STlI

of the mixed

counting

cent

size \xiation,

to

are occasional size

shomwl

upon the progress

and Ste\vart

in

coniorming

1. The cause of the formation

0 to -

g-ain

M,.

11, the nmjorit!-

the range large

from

that of the

[23], \vho detercarbon

steel,

within

the

also

range

Thermal

ol )ser\-:t(ion

that

Stabilization

the

.I[,

t~niperaturcs

in

on the quenching medium used indicates th;lt certain changes which take place during quenching affect the martensitic transformation. This difference in AI,< must c.ontrihute signiticantI?- towards the difference in retained austenitc in the brine- antI oil-quenched specimens.

allo\-s tlq~~nd

AIoreover. the present .\I --I‘ curves produced h\. quenching into different media \vcw not parallel, and the ditiercnce uxs largest at almut 70°C (158’F). The decreasing divergence helms fO”C (158°F) can he partI\ accounted for in ;~llo~s lO(‘, lO(‘lS, 10(.2X, and the carbon steel, I)> the non-linearit!of

Ixundin~ ;md is smaller

being

‘l‘he these

compared

quench,

_A specimen

per cent

13C5S.

that

oil quenching, rapid

tjvo

for allo\.

3.2

the

Hommw-,

.11-7‘

curves

at

high

in ,IIIOJX 10(‘.5N and

martensite 12(‘25

contents. there

is still ;L

belo\v 70°C‘ c158”1;), whereas had signiktnt stabilization during quenching been occurring l~elo\~ JI,y as suggestetl 1)?- Harris and Cohen [A], the divergence I\-oultl be expected to increase as the quenching tcmper;~tur~ decreased, so long as the linear portions onl! oi the curves \vel-e considerctl. This suggests that t hc clitference hetwecn the amounts of retained mstenite found after oil and xvater quenching to the sanw temperature is due mainI!- to an effect. occurring ;~l)o\-e X,. If this is so, the tliti‘erence between the r,ites of cooling produced h\- cluenching into oil and brine must he ;I maximun~ \vhcn the quenching- mctlia are held at approximateI!. 70°C (158’1;). (‘omparison of the cooling curves in Figure 12 sholvs that there is ;t marked diiference l>et~Veen t hc cooling rates pi-o-

decrease

in divergence

42

ACTA

AIETALLURGICi\,

\.OL.

31O’C

1,

1953

(590°F).

This means

SO, above

ture,

which

It lies between

310°C

for the alloy lOC5N.

(590’F)

155°C

(311°F)

cannot

and 405°C

The progress

below M, was virtually below

that there is a tempera-

stabilization

of transformation

independent in the carbon

T;\BLE

occur. (761’F)

of cooling

rate

steel.

I\*

ALLOY lOC5N Reference

FIGURE 12. Cooling curves of a steel specimen 0.500 in. diameter and 0.125 in. thick in saturated brine, mercury and oil at 75°C (lF7’F).

duced by brine and oil quenching media held at 75°C

(167°F).

Moreover,

75°C

shows that

the cooling

into brine

is the greater

by quenching approximately quenching

650°C

the

same

process

(1200°F).

of

than

quenching

temperature

stabilization

only

(167°F)

must

52

below

50 47

as

produces

into mercury

(Figs.

Quenched into lead bath at 405°C (761°F) held 30 seconds and then (a) brine-quenched to reference temperature (b) oil-quenched to reference temperature (c) air-cooled to reference temperature Quenched into lead-tin eutectic alloy at 310°C (590°F) held 30 seconds and then (a) brine-quenched to reference temperature (b) oil-quenched to reference temperature (c) air-cooled to reference temperature Quenched into lead-tin eutectic alloy at ‘200’C (392°F) and held 15 minutes and then (n) brine-quenched to reference temperature (b) oil-quenched to reference temperature Quenched into lead-tin eutectic alloy at 200°C (392°F) held 30 seconds and then brine-quenched to reference temperature Brine-quenched to reference temperature Oil-quenched to reference temperature

held at

rate achieved

Consequently,

into brine held at 75°C

more martensite at

for brine and mercury

4,

5 and

occur

below

held 8)

31 50

the.

650°C

(1200°F).

52

(i) Efects qf cooling varying

the cooling

retained

austenite

specimens

rate above

of

M, on the amount

of

rates

through

are shown in Tables T-ABLE

(Specimens

tempered

lOC5N M, - 80°C

Carbon Steel 31,s - 140°C

72°C

14

20°C (FVF)

88

2O’C 20°C

ranges.

for 30 secondsj Cooling procedure

that

(761°F).

and

results

clearly

show

that

stabilization

was

occurring rapidI\- between 405’C (761°F) and 310°C (590°F) because the amount of stabilization is independent of the cooling rate above the first temperature,

and

onl?,

slightly

dependent

belo\+

stabilization

above A&.

depression

of

effect

which

by holding specimens cooling

curves

ing to these heated

Specimens

occurred

due

to a

below

the

previous

shown

that

temperatures,

the

to a lower temperature

405°C at

results complete

to the required

holding

specimens

and immedtemperatures.

of alloy IOC5N were mercury-quenched

(248°F)

(395.6”F)

and

and immediately 280°C

periods of between quenched in brine (167°F).

was begun

Since had

holding

to 12O’C

75°C

It was thus

was

would occur during slow direct quench-

were quenched iately

M,

of alloy lOC5N isothermally

temperatures.

stabilization

estimated These

the

The rate of this process was then measured

different

hIercurJ--quenched to 110°C (230°F) and held for 65 min. before brine quenching to 73°C Direct saturated brine quench Salt - quenched to 155’C (311°F) and aircooled to 20°C Direct oil quench Salt-quenched to 155’C and brine-quenched to 20°C

88 87

(ii) Isothermal shown

stabilization

(572°F)

12

GO 50

by quenching different

I I I and I\-.

Per cent martensite

73°C (163.4”F)

The

III

at 300°C

Reference temperature

All0y

effects

were investigated

at various

The results

rate above M,.

22°C (71.6”F)

Cooling procedure

60 51 47

comparison

curves

(167’F)

Per cent martensite

into the respective

of the cooling

temperature

(536°F)

raised

+ 2°C

to 202°C (4”F),

for

0 and 195 minutes and then to a reference temperature of

Measurement

of the holding

at the time at which to be at steady

period

the specimen

temperature

was

as shown by

cooling curves. The results are given in Figures 13 and 14, and it is evident that the depression of M,, calculated from the amount of tnartensite lost at the reference temperature due to the isothermal holding, in-

c‘re;w5 w.itfi increasing time. f’or ;Itlo!. 12C2S similar tests showd that n 4”(’ (7”Fj depression in AI, could be achieved in 20 li~illiftes at 200°C (392°F) anti further holding Lfp to 65 minutes did not protlifcx~ an!. signiticanT: fifrther depression. _kwrat.e measurements of act iv&ion energ. for stabilization are in prop-ess; pretiminar\~ estimates yiritt an itpt)roximate t-atife of 30.~0~ cal. per mole. (‘t~~m~tuc~~tly il this process is act-wring at temperatures up tf) -MfI”(- (752°F) it \vitt owur at the rates gi\-eff lwto\\. : I‘emt~c~r;ft~frr “(Y It; E3rr 3M 32.i 3.50 375 J&Cl

‘t‘inie required ior 6°C (1lOFj depression of AI, of atlo\. 10f5S

5.16 572 61; 662 707 7.52

38 se0x& 14 4.6 1.6 0.6 0.25

1:ram Ipigures 13 anti 14, it is apparent that the rate of std~ilization ctcpends ul)on the amount oi st~fl~itization that has atread\- occurred, the process

TIME

(HOURS)

OF

ISOTHERMAI

HOLDING

thy maximum differenc*e iwt\veetl the amounts of nxlrtertsite proctuceci in brine and oil quenching is of the order of 20 per cent (depression of 20°C‘ or 36°F). The data +*en above for the time required to give a 6°C ’ (11°F) depression of AI, appi!. to the later portions of the prowss; it is to be uspwtcd that the iifll 20 txr ccn t stahitizntion ow(frs in less than three times the t-alffes quoted for the 6°C‘ depression. ~(~ll~t~~frison xvith the times required for oil- and txine-quenched spwifncfts to pfss through the \xrious ranges of tempxfture indicates that the stdGtiz:ftion must have bep~n in the tcmpernturr range 375” to 400°C (707”-752°F). The 20 per cent toss Of martensite is the integrated result of stabitization during quenching. The vatucs plotted in Figurc’s 13 nnti 14 thus represent the continuation of the stahitization I\-hich had ;ftread!. or*c.ifrred (luring inikfl mercur\- quenching before the temperature 5x1s r~fiscti ior iso&mxd stahitixf~iofi tests.

the osperimefits .\tthough descritwti in the prvI+xfs sections clearI\. sho\v that the ditferencc in the amount of nrnrtensitc at ;t given ttmpcr;fture in t tic l~i~lr~e~isitic range is due mainly to n prowess taking tdacc d)o\-e Al,,, stahitimtion c;fn stilt occffi if the steel is held t~eio~\ A/, ior ;I sfffTic.ient time. iIi\~estigations \vere, therefore. fiixt~~ into this t’l~cflonlenoIl.

AC’J‘;\

RJ~:‘J‘_-\LL1’fiGICr\,

\.OL.

saturated (Figs. For

AFTER

nr 70’C 30

brine,

“Ol_DlNG FOR

content,

Specimens

c

3so*c 20

30

I

0

-I

FOP

SECONDS

\

I

I

I

10

40

60

I 82

TEMPERATURE

I 100

\I

22°C

I

’ C

FIGURE 15. hktrtensite transformation curves for carbon steel during normal quenching and after holding at TOT (158°F) for 30 minutes.

for

or 74°C

transfor-

The)-

(165.2”F) this

therefore, which

periods before

been

at

and

J.et

transformation able

effect.

high

below

shown

The

maximum

austenite

is limited

reference

temperature.

is reached,

enough

T, from

to

to be

low to be

prevent

removing increase

the

an observof

retained

to 20 per cent by the choice Refore

the maximum

the degree of stabilization

100-

to

of 22°C was

enough to allow up to 20 per cent stabilization measured

43°C

quenching

temperature

had

first

produced

were then isother-

different

A reference

because

were,

(109.4”F)

of martensite.

(71.6’F).

selected

I20

lOC5.U

to 43°C

stabilized

(109.4”F)

70

a constant

as it is known that stabilization

of alloy

37 per cent mally AT

of stabilization

to maintain

[3].

brine-quenched

TEHPEFSD

and lO(‘2N

on the degree of prior martensitic

mation \

lOC5N

of the rate

A& it is necessary

depends

MINUTES

and with alloys

a determination

martensite

f

I!)53

17-20).

below

O-UENCHING

1,

I

I

I

I

I

increases ,

of

amount with

-0

ALLOY 12c

IN

-

10

-

20

-

70

FKXRIC 16. .knounts of austenite stabilized in carbon steel at different reference temperatures after isothermal holding for 30 minutes at 70°C (158°F) and 53°C (127.4”Fi.

was

completely

43°C

(109°F).

the austenite which

stabilized When

continued

was greater

curve obtained brine,

until

(68’F), removed. Krupp

it was

to transform

than

that

at

the

That

reference

by the

the

effect has

with

to

more

[24]. Thus it is evident tion observed depends

than

of 20°C

had almost

of stabilization also

AI--T

in 12 per cent

temperature

of stabilization

removed

steels

cooled

but at a rate

shown

from direct quenching

the effect

completely

until

cooled below this temperature

been

been

can

observed

0.9 per cent

be

‘IO

-

TEMPERED

AT

330k FOR 30 SECONDS %I-

60

40

20 TEMPERATVPE

20

0

I 40

60

‘C

FIGURE 17. bIartensite transformation curves for alloy 12C2N during normal quenching and after holding at 48°C (118.PF).

in

cm-bon

that the amount of stabilizaupon the reference tempera-

ture used; so does the temperature un above which no stabilization is observed at a given reference temperature. at 70°C

Figure

(158“F)

16 in which the effect of holding

is shown,

together

with that

for a

holding temperature of 53°C (127.4”F), illustrates the fact that the temperature at which the maximum amount holding with

of stabilization is observed is affected by the temperature. Similar results were obtained

specimens

of the carbon

steel

quenched

into

FIGURE 18. ;\mounts of austenite stabilized in alloy 12C2N at different reference temperatures after isothermal holding for 2 hours at 18°C (118.4”F).

i.i

TEMPERED 330’C

30

30 -

AT

FOR

SECONDS

I

70

I

-40

I

I

-20

0

40

20

60

80

Cohcti’s

[3]

cotic~lusion m,\, above

tcm~)vratut-cb, be observed

under

It is e\-iclcnt, mental ittcwktsin~

time,

5x1~1~ of

12 Ijet- cent

mittutes

;tt

i-lo<:

t-ccIttired

0.5

hours

24

teni~jeratut-e

vxw

;tllo!-

13C5S

ant1 held

to

sl)cc.imens

the

for

rvferenw

In no (xc suggested

occur to

X0(

0i -

32.8”F)

lioltlitig-

(

and belo\v at

tiott

77.0

(u,, 36°C.

a refereric-e

stahil-

Tltcw

the cffwt

2

120

0

in

It is unlikcl!.

that

trctiio\-vtl i,t(‘t

martensite

.l/,

there

tetiilx7-,ttitrc m;trtcnsite

intcrfxe.

effects

.V,,.

near

[4]. mipht

(brine‘)

Stabiliz~ttion

~~ossible stat)ilization m;trtensite

for

that

onI\. could

content

a small

amount

be observed, bf-as txb~

these

in vie\\- of the

at

between

allo>-s, of least

30 per cent.

it was

isothermal Ivhen

at temlwr;lamount

of

oi isotherm,tl near

Al/,,

utiotjser\-;thle unless

been

if

utidvt

all!-

those

x5

the

chosen,

oi I)ainitc

Thus

l,tq,ct-

tcmlleritig

;tt- thv austrnite-

small

st~ibilizatioti

twa-(led

I)!- I’etros\-;tti

I)>- tenil)ering.

occurs

the

during

y-ogress of characteristic cq~lxwtion es&tnatiott be

an

[251.

prowss

ageing gro\vth

The

Ijt-ogress place

influences

of mat-f ensite

tem~xxiturc

has !.et been of the phenomenon pat-t

the

takes

\I-hich

the martensitk of the Ijt-ocess,

intey-al

trunsiorm;ttioli

b>- alkcting-

: sonic’

cooling.

(oil)

loo.,

4. DISCUSSION

further

AI,

and

time

that

have

lje cjbscitred

completeI\-

difference

is slow

and

found

such

ion or the

little

cwjling.

Iw observed.

stabilization

m;~\- le:ttl to thy lornialion

nuclwt

temperature;

0i

;ttttl time

austenite

also

the

conditions.

has twen

isothermal

was

to

of stabiliza-

iurthcr

ati :t1qweciablc

m;tJ7 be small

treatment.

had been

occur

is suffi&ntl\-

I=or ;I fisetl

csljct-imental It

t hc c.fht

\I-ill not

t hc atiiotttit

is lo\\,

after

such

c;tti

(‘;111 Iwtl

sufficient

unless

is present.

signiiicant

the austenite

;tt the rcferencc

that

no

13Cj.X

xllo!-

\\-hich

of austenitc

Iwloxv 150”(‘,

(c,

esl)ct-iIntlectl,

sta~jilizatioti

temlwratut-r

(b) St,tljilizatiott

cert;aiti

with

sigiifkince.

th;tt

b!

of st;tt~ilizttion

Ixr cent

23

alqxu-entl!-

IWS

sho\\n

retiiovetl

Ix

holding,

wnnot

of a ~7~:

transformatiott st;it)ilization

holditi~-

.I/,\.

(‘;lllsc‘s

ii the refercncc

~l‘hus,

that

I,ClO\\

;Il-c’ s~vet-al

tail

this

ized at - 36°C

Rlinutes

120.2

\vas

n-et-e as follows: ;lustenitcs

time

tcntl~c~;ttttre 0 “F

steel

(35.6”l;)

stabilized

(’ -

I Ioltlin~

;i

c-ondition.

c, \xries

has little

[4] has found

oljwr\~atioti

is 36°C’ or 06.8”F)

above

of austenite

tetit~w;tturc

isother-

.kxxx-dingi\-.

2°C

equation,

atttl then heltl at tcmpetatures Th<~ i~mounts

that

holding.

cluenvhcd

C’ohen’s

is

esl~erimental

that

and

tett1pwaturcs

I‘hcTv

\TYIS thcw

onI>- if the

‘T,~Iwfore

\\xv-e

in I f,tr-t-is antl

all

tures [3] had

however,

ia I It h,ls twen

(122°F)

cooling

could

c~~oletl Ijclow

there

1vhic.h st:tl)ilization

a gix-en

conditions

Petroq-an at

(109.4’Fi

2 minutes of

(71.6”F‘i.

(Cohen

tii,il st;iijiliz;ttion

49 2.7

43”(‘

7.5

stabilization.

to 50°C before

a

xid

stabilization.

I I:tt-ris xttl first

nncl

Slxciniens

22°C

of

olxt-\-;tble

minute

(165.2”F),

13C.51. cluenched

for it to reach

one

20 ~)er cent

\vcrc’ brine to

required being

i4”(1

to xi\-e

I Alloy

time

(165.2”F)

.it

reslxctivcl! (ii

the

that

“c

TEMPERATURE

either

no

of the the

cx->-st&

tlq~endcncc tt-;ttisiormatiorls and

in

on

of the is ;I

satisfactoq-

ad\-anced. That an oi stabilization must

of an!.

theor>-

has long been

IJointed

of

martensitic

out b>- Troiano

The following theories of stabilization have been previously proposed : (i) Stress relaxation theory. Various authors [26-281 have proposed that isothermal stabilization occurs by relaxation of stresses. Orowan suggested [29] that the stresses around one martensite plate assist in the formation of other plates in the vicinity. Such relaxation would probably cause stabilization in those alloys, such as In-T1 1301, in which martensitic transformation proceeds by alternating shears. However, in austenite, martensite plates do not always form in the vicinity of previous plates, so that stabilization would not be expected to prevent completely the subsequent transformation. (ii) Composition change theory. Fisher, Hollomon and Turnbull [31] suggest that austenite regions of statistically low carbon content, at the austenitizing temperature, act as nuclei at M, when they are retained during the quench. Fisher [32] suggests that stabilization may be due to a migration of carbon from the carbon poor regions, now in the form of ferrite, to the surrounding austenite, thereby lowering the temperature at which the austenite can transform. This theory is unacceptable for the following reasons: (a) A stabilization effect of as much as 20°C depression in Xlf7, has been observed in a 1.0 per cent carbon alloy. This depression would be equivalent to an increase of about 0.1 per cent carbon in the austenite. Results obtained in investigations on bainite formation show no indication that there was any change in composition of this magnitude prior to the formation of bainite. (b) It should not be possible to remove the effect of stabilization if this was due to an increase in the carbon content of austenite. Furthermore, the slope of the Al-T curves should be smaller than that of the unstabilized austenite. These are contrary to the present observations. (iii) Exhaustion of nuclei theory. Most of the recent theories of martensite transformation [31-3.51 depict that the temperature dependence of athermal martensite formation is the consequence of exhaustion of nuclei, which have a statistical distribution in size. Each nucleus grows to full length controlled by the boundaries of the austenitic regions and to a thickness determined by strain considerations. The stabilization could be due to some effect upon the nuclei of martensite plates either by relaxation [35] or by a redistribution of embryos due to statistical fluctuations [34]. The second possibility is unlikely because it has been found that there is an upper limit for stabilization, So, and Cohen and his co-

workers (36) have also shown that the progress of martensite transformation in a nickel-iron is insensitive to the prequenching temperature in the austenite range. The progress of martensitic transformation is not necessarily controlled by the exhaustion of nuclei, as can be seen from the following considerations: (a) In the case of an Au-Cd alloy containing 47.5 atomic per cent of Cd (27), the transformation from a single crystal of the high temperature phase to a single crystal of the low temperature phase occurs by the movement of a single interface and is temperature dependent. Stabilization has been observed in this single-interface transformation. (b) Kurdjumov and his coworker [37] have reported that, in a Cu-Al-Xi alloy, the martensitic transformation was reversible without any hystersis in the martensitic range, and a different set of martensite crystals was obtained only when the alloy had been reheated to above M, and cooled. (c) When both coarse and fine grains are present, it is to be expected from the exhaustion theory that the same number of nuclei will be available, so that, as the plates in the small grains are shorter and thinner, less martensite should form. According to Fisher and his coworkers [31] a difference of about 1.5 per cent is expected when the austenitic grain size varies from ASTM No. 6 to No. 0. In fact, the same volume percentage of martensite ( rt 1.5 per cent) has been obtained at the same temperature in the present work, and there is no indication in the literature that austenite grains of vastly different sizes give appreciably different amounts of martensite, although M, may be slightly affected owing to the additional constraint in fine-grained materials. An alternative point of view which is qualitatively satisfactory and worthy of consideration is that the progress of the transformation is partly controlled by the resistance offered by the matrix to the growth of the new crystals. This resistance is important in the case of shear transformations in which the rate of growth is relatively high so that thermal agitation plays a less important part in overcoming the resistance. A more detailed description of this concept will be given elsewhere; it is sufficient to mention here the following points. The growth of martensite can be viewed as taking place by the movement of an interface consisting of a set of fractional dislocations [38]. The movement of these dislocations will be impeded by the presence of obstacles, such as other dislocations in the parent phase. When plastic

>lOli(;.\S

;\>I)

KO:

‘I‘HE:li.\l.\l.

deformat ion takes place in the austenite during the formation of martensite, more disloc;rtions will be gcneratcd, and ~td~li~io]l~~I driving force must‘ be provided bcforct more martensite ~2n form, This offers an espkmation for the temperature depenclencc~ of the tr;msform;ltion and ;~lso for the phenomenon of niet~hanitxl stabilization. The therm;4 st~tbifi~~~~i(~licxn then be causctl by an increase in the resistance of austenite to the growth ot’ martcnsitc due to the formation of carbon clusters during slo\v quenching or during isothermal holding. The suggestion that carl~on may l)e responsible for stabilization in steels is consistent with the preliminar?. estimate of the activation energ>- for stabilization. ‘I’ht present v~~luc of about 3c),OOOcat. per mole is of the wme order of magnitude as the value of the acti\-ntion energ>. for the diffusion of carbon in :iusteiiite iree iron7 martensite. The cxbon grouping could occur atr!~\vhert {within the matrix austenite. In steels which contain no t;irl)itle-iormin!: elements, it is more likcl~V that it 0wu1-s at clisloc;tt,ions b\. forming C‘ottrell ;1tmos[hews. If this is correct, the time required for stal~ilization to take plx~ should be that required ior cxlton ;iroms to tnove a tlistnitcr greater than t-hat between carbon atoms but smaller than that lwtweeii clislot;itions. The difiusion constant A,, anti xtivat ion cwcrg~. Q for carbon clifiusion at 5 atomic per cent cxbon hit\-e been given b\y \\:ells, Batz and ~I~~lil 1391 as 0.1 cm’ sec. and 30,000 ~1. per molt reqtectivcty. If it is assumed that these data remain valitl in t-lid temperature range 2OOMOO”c’ (390750”F), the length of diffusion path at these tem~wtxtures c;m be estimated. At 280°C (536’F), n w 1Ok’“cm”, sec., the time required for stabilkttion is oi the order of 100 seconcls. therefore .f = \I(KVj or 10-” to 10-6cn~., lvhich seems of the right order of mag+tucle required b\. the mechanism. The formntion of a (‘ottrell atmosphere around dislocations xvi11 not result in ;t permanent loss of martensite. On cooling to T.+, after holding at 7, and producing isothermal st~~~)ili~~~tio~i, the driving force for the tr~ltlsfori~l~~tion wilt have increased suffic-iently to overcome the extra resistance awl the first martensite plates will form. The formation of these plates xvi11result in plastic deformation of the anstenite matrix nnd some of the disloc;ltions will be moved att-ny from their nnchoriq carbon groups. As long as rapid cooling is continued, these rcleascd dislocations \viil remain free. 7’hc~ lo\\-er resistance together \vith the increased driving force will en;lble the resumed transformation to proceed at a greater

S’f.\l~11_1%.\‘1’IO\

OF .\I‘S’1‘l:Sl’I‘I~

4;

rate, and the efic’ct 05 stabilizntion

\\.ill c:\witually

be

I-enlovtYi.

The conce~iti-~tt~oil of c;trhon atoms in ;I rlisloc;~tion xvi11decrease \vith increasing tcmpernture until at the tcmlwrature So it becomes so small that stabilixntion c;innot occur. ‘f-he tcm~wraturc S,, should be higher, the higher the c~;~rbon content. Thus in 111~ present. allo>-s oi hig+cr nickel ;tnd carbon contents, there should be ;I mtrkctl increase in the rate of stabilization during clucnching oning to an increase in So and in the rate oi diffusion of carbon in the presence of nickel [JO]. This otl’ers ;t possible erpkmntion for the fact that AII,y(brine) and Al, (oil) arc similar in the alloy lO(‘101 and 13f’5N, bewuse JI, (brine) \vill qrescnt ii markedl\ depressccl IzJq. I’urthermore, since the wmpletene~s of an atmosphere Ivill increase u,ith tlecreasing temper;tture. the mnsimum clegree of stabilizition will also incrc;tsc xvith decrwsing t~tii~~er~i~itre, ~~lt]loLt~ll this Ill~t~it~lLttil etiect may not ftt* obtained bl- ;I limited time of ageing at 10~~ tcntpcr~~ttires o\ving to the to\\. rate of diffusion.* 5. SUMMARY

AND

CONCLUSIONS

The ~~~~~t~ot~~etto~lof stxbilizat ion cti austenitc during continuous cooling and during isothermal holding above and belotv M,y has bceti studied, ITsing specialI\selected hotnogeneous m;tteri;tls, it has been sho\vn that, for steels containing al)out 1 per cent carbon am1 O-5 per cent iii&t: [i) Stal~ilization of wstenite owurs tlwing isothermal holding ahove and below _12(,.‘J’he rate of stabilization v:u-its with t.he comlwsition of the austenite, increases with increasing tempwtture and is great l>. inrrensed by the present-e oi m;trtensite. (ii) The s~~~bili~~tion above _I& causes ;L depression of .lf,Vand increases the amount of nustcnite at ;L reference tcniperature near &. (iii) ‘I‘hc tliBerenre in the amounts of retnined aust.enite in ;I steel quenched nt various rates is chiefly due t.o st~tt~iliz~tio~~ above .lY_tiurin~ quenrhing. (iv) The amount of martensite iost due to athermat and isothermal stabilization can be recovered h\. quenching to a sufficiently. low tempctxture. It. is suggested that st;tbilizntion is caused b>- an increase in the resistance of austcnitc to l~l~rtensite *.Yok c~ddadin pm& Further the authors (‘~.K.J and his other factors can be involved nickel contents in preventing nbove Al,,.

work carried oLtt by one of conorkers indicates that some in addition to the carbon and the observation of stabilization

.\C’I‘.\

-IS

;\IF,‘I‘.ALLI~KGIC.\,

iormation, and that this increase is mainly due to the formation of Cottretl iitmospheres around dislocations. 6. ACKNOWLEDGEMENTS The research was carried out in the MetalIurgy Department of the Universit). of Birmingham, under the direction of Professor D. Hanson, Director of the Departments of Metallurgy and Industrial >Ietallurgy, to whom the authors’ gratitude is due for his interest, encouragement and advice. The authors’ thanks are also due to Professor C. S. Barrett of Chicago, and Professor &A-\. H. Cottrell and Y1r. F. R. N. Xabarro of Birmingham for helpful discussions. The investigation has been supported bj- British Timken I,td., and the authors are indebted to Messrs. D. McNicoll, J. H. Evans, J. Mills and R. Treen of this firm for innumerable works facilities and to other members of staff for their co-operation. Thanks are due to Mr. H. W. G. Hignett of The &fond Nickel Co. Ltd., for the supply of nickel. Messrs. B. Edmondson, and S. G. Glover have given considerable help in the experimental work. REFERENCES J. -1. Trans. .%mer, SK. Steel Treating, 8 1. MATHEWS, (192%) 565. 2 GULYAISV, A. I’. Metallurg, 14 (1939) tX. 3. HARRIS, W. J., JH. and CC)HEN, M. Trans. Amer. Inst. 1lin. Met. Eng., 180 (1949) 447. 1. PETROSYAN,P. P. Dok. Akad. Nauk SSSR., 59 (1948) 6. 5, FLETCHEK, S. G. and &HI-:X, M. Trarx. Amer. Sot. XIetals, 34 (1943) 216. Trans. Amer. Sot. Metals, G. GORGON, P. and COHEN, X. 3Q (1922) 569. 7. SCHEIL, E. Z. Anorg. ;\llg. Chem., 183 (1929) 98. s. NIELSEN, H. P. and DOWDELL, R. L. Trans. Amer. Sot. Aletals, 22 (1931) 810. 9. SHTEINBERG, S. S. and ZY~ZIN, \-. I. Metallurg, 11

(1936) 3. 10. TAYYXANN,G. and SCHEIL, E. Z. Anorg. _Wg. Chem., 157 (1926) 1. 11. FLETCHIIR, S. G., AVERBACH, B. L., and Corre~, M. Trans. Amx. Sot. Metals, 40 (1918) 703. 28 (1936) 12. FORSTER, F. and SCHEIL, E. 2. Metallkwde, 2-l5. 13. JONES. F. 1%‘.and PCMPHREY, W. I. J. Iron SteeI Inst.! 163 (1949) 121.

1-01~.

1.

1!)53

14. I.,YMAS, ‘I‘. and ‘I‘AOIA~O. .\. R. 'l‘rans. .2mer. Netals, 37 (1946) 402. 13. G~L~Aw, zi. 1”. XIetalIurg, 15 (19&O) 13. 16. L\WAN, T. and ‘I‘KOI.~~O, A. R. Trans. .‘tmer. Ii~st. Alet.

Eng.,

Sec.

Atiu.

162 (IQ&Y) 196.

17. GORMW, I’., COHEN, RI., and ROSE, 1~. S. ‘l‘rans. ;\mcr. Sot. Aletals, 31 (1913) 161. 1s. TKOIANO, A4. R. Trans. ;\mer. Sot. Aletals, 41 (19-C9’1 1093. 19. SHTEITSERG, S. S. and Zmzrr, \.. I. Arch. Eisenhiitrenw., 7 (193_)) 537. 20. II. Trans. Amer. Sot. LIetals, 41 (1949) 33. 33. COHEN, M., MACHLIN, E. S., PARANJPE, Y. G. Thcrmodynamics in Physical Metallurgy (Cleveland, Amer. Sot. Metals, 1950), p. 242. E. S. Trans. Amer. Inst. &lilt. 36. COHEH, M. and i%;IACHLIN, Met. Eng., 189 (19Sl) 746. 37. WRDJVXOV, G. \r. and KHAP~DROS,L. G. Dok. Akaci. Nauk SSSR., 66 (1949) 211. 38. BARRETT, C. S. In “Phase Transformation in Solids” (New York, \Viley, 1950), p. 343. 39. ~Z’ELLS,C., BATZ, \2’., and MEHL, K. F. Trans. Amer. Inst. Min. Met. Eng., 188 (1950) 553. 40. WELLS, C. and MAIL, R. F. Trans. Amer. Inst. Xfin. Afet. Eng., 140 (1940) 279.