Sie sind auf Seite 1von 9

ON THE MECHANISM OF SERRATED GRAIN

BOUNDARY FORMATION XN Ni-BASED SUPERALLOYS

A. K. KOUL’ and G. H. CESSINCER’


‘Structures and Materials Laboratory, Montreal Road, National Research Council, Ottawa,
Canada KIA OR6 and “Central Laboratory, Brown Boveri & Cie, CH-5401 Baden, Switzerland

(Recehed I5 Dccrn3hrr 1982)

Abstract-Grain boundary serration is studied as a function of cooling rate in a number of Ni-based


superalloys. e.g. IN 738, Nimonic 115 and Nimonic 105. The mechanism is related ta y’ precipitation in
these alloys and a higher y’ solvus temperature than the M,C, solvus temperature appears to be a
prerequisite for the development of serrations. A model based on the grain boundary primary y’ particle
movement causing the displacement of the local grain boundary segment and initiating serrated grain
boundary formation has been proposed.

R&m&--Nous avons ttudii la formation de joints de grams dentelis en fonction de la viteaae de


refroidissement dans un certain nombre de superalliages a base de nickel, comme par exemple IN 738.
Nimonic I IS et Nimonic 105. Dans ces alliages, le mecanisme est lie a la precipitation de y ’ et une
temperature du solvus de y’ plus Clev& que celle du solvus de M& semble une condition prcalable a
la formation des deutelures. Nous proposons un mod&k dam ie quel le mouvement des partioules
intergranulaires primaires de y’ provoque le d&placement local d’un segment du joint de grains et entraine
la formation d’un joint de grains deatelt.

Zuaammeuf~An einer Rcihe van Superlegierungen auf Nickelbasis, LB. IN 738, Nimonic 115und
Nimonic 105, wird die Korngmnzenfacettierung in Abhlngigkeit von der Abkiihlmtc untersucht. Der
Mechanismus hHngt mit dem y’-AusseheidungsprozeB in diesen Legierungen zusammen. Es seheint, daR
Vorbedingung !Iir die Ausbildung der Famtten eine y’-Solvustemperatur ist. die h&her als die
M&Z,-Solvustemperatur ist. Ein Modell wird vorgesehlagen, das von den y’-Teiichen an der prim&n
Komgrenze ausgeht. Die Rewegung dieser Teilchen an der prim&n Komgrenze ausgeht. Die Beweguug
dieter Teilchen an der prim&en Korngrenze ausgeht. Die Bewegung dieser Teikhen emeugt eine
Verschiebung des lokalen Komgrenzbereiches und verursacht die Bildung der facettierten Komgrenee.

1. ~~RODU~ION the grain boundary segment bb is shorter than the


segment aa in Fig. i, the localized boundary dispkce-
Furnace cooling treatments designed to generate
ment occurs due to the lower energy configuration of
serrated or zig-zag grain boundaries have long been
the segment bb, thus forming a serrated grain bound-
used to improve the stress-rupture properties of ary. However, no quantitative measurements of
20-I IP austenitic stainless steel [l-3]. More recently,
controlled furnace cooling treatments have been used
to develop serrated grain boundary structures in
wrought and powder processed Ni-base superalloys,
e.g. Nimonic 115, IN 792, Rem! 95 and Astroloy, in
order to improve the high temperature impact
strength and creep ductility in these materials [4-g].
The beneficial elfects of serrated grain boundaries on
the stress-rupture properties of investment cast Ni-
base superall superalloys, e.g. IN 738. have also been
pointed out in the literature ]9]. These grain boundary
serrations are said to impede grain boundary
sliding, thereby improving the overall high tem-
perature creep properties [4-91.
In 20-I II’ austenitic stainless steel, the precip-
itation of coarse grain boundary M2,C6 carbides was
held responsible for the creation of grain boundary Fig. I. A model for serrated brain boundary formation
based on M&, precipitation. Straight grain boundaries
serrations [l-3). Grain boundary M,,C, precipitates
(a-a) formed on solution treatment and serrated grain
arc assumed to grow from one grain into the matrix boundaries (b-b) formed after M2& precipitation (Refs
of the adjacent grain such as indicated in Fig. I. Since H-31).
1061
1062 KOUL and GESSINGER: ON SERRATED GRAIN BOUNDARY FORMATION

Fig. 3. Experimental technique employed to determine


Migratory forces octlng on 0 groin serration wavelength (i) and amplitude (A).
(0 1 boundary during slow cooling from
solution treofment temparotura

SECONDARY y’ forces involved in the creation of these serrated grain


PRIMARY y’ PAR’ICLES
boundaries will be dealt with at a fundamental level.

2.‘EXPERIMENTAL MATERIALS AND METHODS

L In order to broaden the scope of the investigations,


ORIGINAL GRAIN BOUNOARY
the materials used in this study were investment cast
Sarroted grain boundaries formed on IN 738, wrought Nimonic I15 and Nimonic 105.
(bl cooling from solution treatment to Samples were solution treated at a specific tem-
portiol solution treotment temperatures
perature for a given length of time and furnace cooled
Fig. 2. A model showing serrated grain boundary formation at a rate of O.S”C/min, 1.33”C/min or 2.33”C/min to
on cooling from solution treatment through y’ precipitation
a selected intermediate temperature below the y’
range to PST (Refs [5] and [q).
solvus and then immediately water quenched to arrest
the precipitation reactions. One sample from each of
M& carbide length and the amplitude of serrations the three materials was used for microstructural
have been presented to demonstrate the viability of analysis. The details of various solution treatment
this hypothesis. temperatures and times, including the intermediate
In contrast, slow cooling through the y’ precip cooling temperatures for each material are given in
itation range is considered essential to form serrated Table 1. The selected solution treatment temperature
grain boundaries in Ni-base superalloys [7-IO]. The for each material was well above their y’ and M&
mechanism of serrated grain boundary formation is solvus temperature, whereas the intermediate cooling
thought to be associated with the heterogeneous temperature was chosen well below the y’ solvus
nucleation of y’ at the grain boundaries, and the temperature. All heat treatments were carried out at
migration of grain boundary sections in between the atmospheric pressure in a muffle furnace fitted with
primary y’ particles as shown in Fig. 2(a) and (b). a proportional controller and Data Trak pro-
These grain boundary segments supposedly continue grammer.
to migrate until y’ forms homogeneously throughout IN 738 and Nimonic 115 samples were etched in
the grain at a lower temperature. If the material is 8% Br-92% CH,OH solution, whereas Nimonic 105
cooled rapidly through the y’ precipitation range, the was etched in 40% HCl-20% Hr0,-400/, H,O solution
heterogeneously formed y’ particles are small and the to reveal the grain boundaries for quantitative metal-
boundary migration is limited [7-81. However, it is lography. In specimens showing serrated grain
not clear why the grain boundary segments in be- boundaries, the serrations were quantified in terms of
tween the primary 7’ particles should be mobile when the amplitude (A pm) and a wavelength (A pm) such
cooling from a high temperature. The observations as defined in Fig. 3. These serration parameters were
indicate that the basic mechanisms of serrated grain measured manually off the grain boundary tracings
boundary formation are but imperfectly understood. made from the optical micrographs. A centre line was
This paper therefore aims to present some system- drawn through a chosen set of tracings and parame-
atic observations and explanations on the influence of ters A and A were determined with reference to this
varying cooling rates on the serrated grain boundary centre line alone, Fig. 3. A number of micrographs
formation in Ni-base superalloys. Also, the driving per heat treatment condition were examined and

Table 1. Selected solution treatment and cooling conditions used on difkent alloys

Selected solution Intermediate


Treatment cooling M,& ’
Temperature Time Temperature Solvus &us
Material (“0 (h) (“C) (“G) (“0
IN-738 I200 2 1090 IO00 1160-1175
Nimonic I I5 II75 1.5 1000 1080-l IO0 Il40-I I60
Nimonic 105 1170 2 loo0 I lo&l 150 1020-1060
KOUL and GESSINGER: ON SERRATED GRAIN BOUNDARY FORMATION 1063

treated and controlled furnace cooled at a rate of


0.5”-2.33”C/min, revealed very sharp but straight
rain boundaries at all cooling rates, Fig. 8. An
additional Nimonic 105 specimen was cooled at a
much slower rate of 0. I “C/min from 1170” to IOOO‘C,
with the same result.
Electron microscopy of the solution treated and
water quenched specimens of IN 738 and Nimonic
I I5 showed straight grain boundaries with some
cuboidal primary y’ particles precipitated at the grain
boundaries, Fig. 9. The controlled furnace cooled
specimens, however, revealed serrations with primary
y ’ particles nearly elliptical in shape associated with
all serration peaks, Figs 10 and 11. These primary y’
(a) particles were always present on the concave side of
the serrations in both alloys. The particles remained
confined to their parent grains, Figs 10 and 11. In
contrast, Nimonic 105 contained a fair proportion of
grain boundary M2J6 carbides on most of the grain
boundaries, together with a fine y’ distribution within
the grain interiors.

4. DI!KUS!SION

On closely studying the electron micrographs


presented in Figs 9-11, the serrated grain boundary

(bl
Fig. 4. Optical micrographs showing serrated grain bound-
ary formation in IN-738. (a) Cooling rate of 2.33”C/min (b)
cooling rate of 0.5”C/min.

approximately 100 serrations per sample were mea-


sured.
This foil electron microscopy was carried out on all
heat-treated samples, where discs 3 mm in diameter
and 0.25 mm in thickness were electrolytically pol-
ished using Struer’s Tenupol polisher. The discs were
polished in a solution comprising of 94 parts acetic (a)
acid-6 parts perchloric acid, at a current density of
4mA/mm2 at room temperature.

3. RESULTS

IN 738 and Nimonic 115 specimens that were


furnace cooled at a rate of 0.5”-2.33”C/min revealed
profuse serrations, Figs 4 and 5, whereas the speci-
mens water quenched immediately after solution
treatment showed a straight grain boundary struc-
ture, Fig. 6. The serration amplitude and wavelength
increased in both alloys on decreasing the cooling
rate from 2.33” to O.S”C/min, Table 2. An increase in
wavelength with a decreasing cooling rate should be
noted. In Nimonic 115, however, both serration (b)
amplitude and wavelength appeared to have reached Fig. 5. Serrated grain boundary formation in Nimonic 115.
an optimum at a cooling rate of O.S”C/min, Fig. 7. (a) Cooling rate of 2.33’C/min, (b) cooling rate of
In contrast, Nimonic 105 specimens solution O.S’C/min.
I064 KOUL and GESSINGEK: ON SERRATED GRAIN BOUNDARY FORMATION

* tUitl?onir II5
E 0 tN 738

ir>, ,
d 9

0.5 1.0 I.5 2.0 25 3.0

Cooling rote %/min

c
A Nimonir II6
(b) 0 IN 738
E, z5

Cooling mto lC/rein

Fig. 7. Effect of cooling rate on the amplitude and wave-


lengths of serratlons in IN 738 and Nimonic 1IS. (a)
Amplitude variations, (b) wavelength variations.

(b) mation in Ni-base superalloys, Nimonic 105 was


examined since its M2rCb solvus is higher than its y’
Fig. 6. Straight and sharp grain boundaries formed on
solution treatment and water quenching of (a) IN 738, (b) solvus temperature, Table 3. On cooling from, a
Nimonic I 15. solution treatment temperature of 1170°C to lOOO”C,
the grain boundary Mr3Cb carbide precipitation pre-
cedes y’ precipitation at all cooling rates [4]. There-
formation mechanism in IN 738 and Nimonic 115 fore, Nimonic 105 should have generated a serrated
appears vastly different from the one proposed for grain boundary structure in accordance with the
20-l 1P austenitic stainless steels [l-3] in Fig. 1. On mechanism proposed in Fig. 1. Instead, Nimonic 105
cooling from the solution treatment temperature, y’ developed a straight grain boundary structure at all
precipitation occurs well in advance of M.& precip- cooling rates, Fig. 8, where grain boundary M&
itation in both nickel-base alloys [4-g], and this is also particles were uniformly distributed. Grain boundary
indicated by their y’ and M& solvus temperatures M&Z6 precipitation would generally be expected to
presented in Table 3. The resultant grain boundary inhibit the boundary migration rather than create a
and the matrix y ’ precipitation would be expected to serrated grain boundary such as proposed in Fig. 1
inhibit the type of grain boundary migration required [ll]. This is perhaps why the grain boundary ser-
by the mechanism proposed in Fig. 1. Therefore, rations could not be generated in Nimonic 105, even
M&J6 precipitation may not necessarily participate in at a very slow cooling rate of O.l”C/min. If Nimonic
the serrated grain boundary formation in Nimonic 105 is considered a generalized reflection of serrated
115 and IN 738. grain boundary formation in other Ni-based super-
In order to elucidate the precise role of M& alloys, the mechanism is obviously very different
precipitates on the serrated grain boundary for- from the one proposed in Fig. 1.

Table 2. Effect on cooling rate on serration parameters


Serration amplitude and wavelength in
Nimonic I IS IN 738
Cooling rate Amplitude Wavelength Amplitude Wavelength
“C/min Otm) (rm) (rm) (rm)
_~_.-___-... . ..--- __-
2.3 I.0 3.15 1.5 4.0
I .33 1.8 6.0 1.6 4.4
0.5 2.0 6.7 2.35 6.5
KOUL and GESSINGER: ON SERRATED GRAIN BOUNDARY FORMATION 1065

Fig. IO. Elliptical primary y’ associated with the serrations


in IN 738. Solution treated and cooled at a rate of
2.33”C/min.

cooling of 3”Cfmin in Nimonic 115. It is, however,


interesting to note that in any alloy where grain
boundary serrations were successfully produced, the
y’ solvus temperature was .always higher than its
M,C, solvus temperature, Table 3. Therefore, it was
only on cooling from the solution treatment tem-
perature through the y ’ precipitation range that these
serrated grain boundaries were formed. The mech-
anism thus appears to be related to y’ precipitation
rather than M& precipitation in these materials.
Figure 2 illustrates the primary y’ precipitation
dependence of serrated grain boundary formation
f b) proposed by other workers in N&base superalioys
Fig. 8. Straight and sharp grain boundaries formed in [S, 7,9]. However, on cooling from a higher tem-
Nimonic 105. (a) Cooling rate of 2.33OC/min, (b) cooling perature to a lower temperature, the migration of the
rate of O.S”C/min. boundary segments in between the primary y’ par-
ticles appears unlikely, considering the logistics of the
time and temperature dependence of the grain size in
A brief literature search of the conditions utilized any alloy system. A representative grain coarsening
by different workers to produce serrated grain schedule for the ~ready-h~t-~t~ Nimonic 115 is
boundary structures in a variety of Ni-based super- presented in Fig. 12, and superimposed on it is a
alloys was carried out, and these results are presented typical cooling curve used in the present in-
in Table 3 [4-lo]. The cooling rates vary anywhere vestigation. In any alloy, the grain size is usually
from air cooling in IN 792 to very slow furnace

Fig. 9. Cuboidal grain boundary primary y’ precipitated in Fig. I I. Elliptical primary y’ associated with the grain
IN 738 on S.T. at lMo”C and water quenching. boundary serration in Nimonic II 5.
KOUL and GESSINGER: ON SEKRATELI GRAIN BOUNDARY FORMATION

Table 3. A brief lileraturc search on Ni-Base superalloys with serrated grain boundaries
..’ Serrated grain Minimum
MI&
Si~lVUS SOIVUS boundaries cooling
Alloy temperature temperature fOl7Wd rate
type (‘>C) (“C) or not required

Nimonic I I5 1080-i 100 1140-1160 Yes 3”C/min.


IN 738 loo0 1160-I 175 Yes F.C.’
IN 792 1000-1050 IIOO-II50 Yes A.C.2
Rene 95 LOW 1120-1160 Yes F.C.
Astroloy L. C. 1000 1120 Yes F.C.
Nimonic I05 IIW-1150 1020-1060 No -

‘F.C.-furnace cool.
zA.C.-air cooling.

established by the solution trea~ent temperature, either side of the grain boundary, Figs 10 and Il. It
but it is possible for the boundaries to migrate at a appears that the primary y‘ particles precipitated
lower temperature when enough time, of the order of adjacent to the grain boundary have moved under the
a few hours, is allowed [I 11. Accounting for the influence of some driving force during cooling. The y ’
cooling schedules used in Nimonic 115 and IN 738 particle movement would also displace the local grain
and those used by other workers in different N&based boundary segment and create a serration such as
superalloys, it is evident that the times spent at a shown in Fig. 13. It is suggested that y ‘ nucleates on
given temperature are far too short to influence the one side of the grain boundary on cooling through
boundary migration in these materials. Therefore, the the y’ precipitation temperature range. Due to the
mechanism proposed in Fig, 2 may not be responsible presence of a grain boundary, having a certain width
for serrated grain boundary formation in N&based adjacent to the y’ precipitate, the coherency strains
superalloys. Nevertheless, the mechanism is somehow on the boundary side of the y’ particle are easily
related to the y’ ~re~pitation on cooling from the accommodated, Fig. 13. A net strain energy
solution treatment temperature. Any proposed mech- differential between the interface A and B of the y’
anism should account for an increase in the serration particle, Fig. 13, provides a driving force to move the
amplitude and wavelength with a decreasing cooling y’ particle in the direction of the boundary. The y’
rate, Table 2 and Fig. 7. Also, the mechanism should particle movement will however, be opposed and
explain the variations in the required cooling rates to finally stopped by the line tension of the boundary..
initiate the formation of serrated grain boundaries in A model of this nature woutd predict the effect of
different Ni-based superalloys, Table 3. cooling rate to be similar to that observed in Fig. 7.
Also, the variations in the minimum required cooling
5. MODEL rates to produce serrations in different Ni-base super-

The primary y’ particles are elliptical and always


associated with a concave portion of the serration on t=o
~
““1 y’nucleoted on one side
of the groin boundory
270

250
E
a

P2=
,:
El
t
‘si
.E
PI0
tfionB
e
e 100

I70

AG between A and B provides the drivingforea


Ieo 12eo in the direction of the boundory
1000 lO!tO ,ioo I200
Tempemture PC)
Fig. 13. A schematic sprinting the model proposed for
Fig. 12. Grk coarsening characteristics of previously heat- serrated grain boundary formation in Ni-basedsu~ralloy~
treated Nimonic 115samples. based on y’ particle migration.
KOUL and GESSINGER: ON SERRATED GRAIN BOUNDARY FORMATION 1067

alloys could be attributed to the variations in the y’ in excess of 700 A in radius. The free energy of the
particie-mat~x misfit from one alloy to another, thus coherent side A is higher than the semicoherent
changing the magnitude of the driving force available portion B in Fig. 13, once the particle radius exceeds
to produce serrations. To simplify the analysis, we 700 A.
have assumed the presence of a spherical y’ particle The equilibrium state obtained on cooling to room
of a constant diameter at a constant temperature temperature after applying a controlled cooling treat-
causing a serration, but the model will be modified ment, gives a primary y’ particle radius of 35OOA
later to include growth of the y’ particle during when cooling at a rate of - 1.S”C/min. The coherency
translatory motion on cooling. The precise details strain energy difference of a primary y’ particle of
and the calculations of the different energy parameter radius 3500 %r.between sides A and B in Fig. 13 can
are given below. be calculated by using the expression AG,<= 4 ~6’ V
AC,, (total) = AC, of side A + AG,r of side B.
5.1. Force available fory'particle movement due to
strain energy d$erential Due to strain accommodation along side B of the
y’ particle in Fig. 13, the AG, of side B can be
The free energy of a crystal containing a fully
approximated in zero. Substituting for parameters p,
coherent spherical precipitate, e.g. y’, has con-
S and V, the coherency strain energy difference of
tributions from (i) the coherency strain energy AG,V,
7.75 x 10e6 ergs is obtained, which approximates to a.
and (ii) the chemical interfacial energy y,,, [12]. The
force FE of -0.2 dynes acting on the y ’ particle.
sum of these two terms is given by

AG inherent) = AG, + yEt,x precipitate area. (1) 5.2. Retarding force due to boggy energy
A net force (F,) of 0.2 dynes will move the y’
In the case of a spherical particle of volume Y and particle in the direction of the force and considering
radius r and a misfit 6, AG, is represented by the ordered nature of the y‘ precipitate, the moving
rr4,& V, where p is the shear modulus of the matrix. particle will also drag the localized grain boundary
Therefore, equation (1) can be rewritten as segment such as indicated in Fig. 14. The boundary
segment, however, will exert a retarding force F2 due
AG t~he~nt) = 4~2,4~3~r3 -f- 47tr2*y,,. (2)
to the grain boundary energy y and oppose the y’
If the same precipitate has a semicoherent or incoher- particle motion.
ent interface that completely relieves the effects of The y’ particle shown in Fig. 14 will be attached to
misfit (such as by the presence of a grain boundary the boundary along a length 2nr sin 8. The retarding
of finite thickness adjacent to the y ’ particle) there force component due to grain boundary energy is
will be no misfit energy. However, in the case of a given by y cos 8 and therefore, the y’ particle will feel
semicoherent interface there will be an extra struc- a back force of 27rr-ysin f3cos ff = nry sin 28,
tural contribution to the interfacial energy (y%,)due to or F2 = xry sin 20 (4)
the structural distortion caused by the misfit dis-
locations. The free energy of a semicoherent precip- where y’ is the grain boundary energy of the order of
itate is therefore given by N 300 ergs/cm2 I:121 and assuming a 6 value of w 45”
from Fig. 10, the retarding force (F2) approximates to
AG (s~i~herent) = 0 + 4%r2(y,&+ rd. (3) -0.033 dynes. It must, however, be pointed out that
the forces F, and F2 will be acting along two different
Therefore, for a given 6 there will be a minimum y’
planes pp’ and qq’ of the y ‘ particle in Fig. 13.
particle radius above which AG (coherent) will exceed
Therefore, in calculating the net force acting along y’
AG (semicoherent), thus creating an energy
particle, the forces F, and F2 can not be simply
differential between two sides (A and B) of the y’
subtracted, and a detailed analyses of appropriately
particle. These boundary conditions can be repre-
treating these forces will be presented in the follo~ng
sented by [12].
section.
4wr2(y, -l-y,?,)= 4pS2 )7rr3 t 4nr2y,.

Assuming the yrh term to be similar on both sides


of the y’ particle, then r = 3~,/4@~, and upon
substituting the IN 738 shear modulus value
u 10” N/m2 at 1lOO”C, a misfit value of -0.01 [I4
&d a Y,~,value of 0.10 N/m, an t value of 700 A
is obtained for the boundary conditions. Therefore
on furnace cooling the solution treated material
through the y’ precipitation range, a net y’ particle
free energy imbalance occurs between sides A and B
in Fig. 13, once the primary y’ particle precipitated Fig.14.A schematic representing the boundary y’-particle
adjacent to the grain boundary has grown to a size contact and the forces involved.
1068 KOUL and GESSINGER: ON SERRATED GRAIN BOUNDARY FORMATION

planes pp’ and qq’ respectively. Under the influence


of a normal stress, the chemical potential of vacancies
is changed by an amount equal to the product of the
magnitude of the stress and the atomic volume. Since
F, and F2 are acting in opposite directions, the
respective chemical potential gradients VP,, and V/J,
developed by these forces will also oppose one an-
other. Hence, the net chemical potential gradient VP
Fig. 15. A schematic showing the planes of action for forces
is defined by VP~, - VpF2and the equation (5) takes
F, and Fz on a y’-particle. the form

; = F .(Vpc, -V/J,)

5.3. y ’ Particle movement under the influence of forces


where
F, and F,
In order to check the validity of the model, it is VC(~,= stress due to force F, x atomic volume
necessary to examine the distance x pm moved by a
grain boundary primary y’ particle under the = F,
XSf
influence of two opposing forces F, and F,, acting at area of pp’ plane x 2r
an average approximated temperature of T K over a
time period of t seconds, and then compare the x vpF, 2$= 4.75 x 10-r’ dynes.
value with the experimentally determined serration
amplitude A value. Providing the calculated x value Similarly,
and the experimentally determined amplitude A value
are of the same order of magnitude, then only the V’Fz -- stress due to force F2 x atomic volume
feasibility of the present model can be acknowledged
with some degree of certainty. 4 x s;
The forces F, and F2 acting on the grain boundary = area of plane qq’ x 2r
primary y ’ particle create a chemical potential
F2G
difference VP, which in turn leads to a net flux of
= 2xr3 sin%
vacancies by diffusion [l4-161. As a result of this
diffusional flux, the y’ particle can move a distance x where 8 = 45” from Fig. 10 and VpF, = 1.57 x
pm in a given time t by three possible mechanisms: lo-” dynes. Therefore, VP = VpF, - Vcl,, = 3.18
x IO-” dynes.
(i) Interfacial-grain boundary diffusion
Substituting for VP and other variables in equation
(ii) Volume diffusion through the particle
(6). dx equals 4.9 x IO-scm/s. However, the time
(iii) Dissolution of y’ particle on one side, where
allowed for y’ precipitate to migrate within a tem-
volume diffusion of the solute takes place through the
perature range of 1160’-1000°C at a rate of
matrix and the diffused species reprecipitate on the
l.S”C/min equals 6400 s. Therefore, the y’ particle
opposite side of the particle.
would move a distance of 3.13 pm within a time
However, the occurrance of interfacial-grain period of 6400 s. An x value of 3. I3 pm compares
boundary diffusion is more likely since Dgb >>D. and favourably with the observed serration amplitude A
the diffusion path is much shorter via the particle of N 1.6 pm in both alloys.
interface than through the y-matrix. Therefore, the An increase in the serration wavelength with a
movement dx of a y’ particle in a time interval dt decreasing cooling rate, Table 2 and Fig. 7, can be
under the influence of a chemical potential gradient attributed to the precipitation of fewer primary y’
VP is described by particles adjacent to the grain boundary, since the
matrix supersaturation effects are maintained for a
ds 2 Dgb6,
-=-Vp (5) longer period of time. These particles further grow
dt kTr
and coalesce, where their coalescence will only
where 6, is the diffusion thickness assumed to be of influence the driving force for onward migration that
the order of lattice spacing -4 x IO-*cm, Tis the y’ determines serration amplitude rather than the wave-
precipitation temperature .range - 1430 K. r is the y’ length. On top of this outward particle migration, the
particle radius -3500 x IO-” cm, Dgb is the grain grain boundary primary y’ particles change their
boundary ditfusion coefficient similar of Ni
to that shape from a cuboidal to ellipical morphology with
_ I .34 x IO -’ Cm’/s and k is the Boltzmann’s con- a decreasing cooling rate, Figs 9 and IO. This shape
stant _ I.38 x IO.-“ergs/K. change decreases the overall energy of the system,
In calculating the 011 for they’ particle, it is evident even though its effects have not been calculated in the
from Fig. 15 that two opposing but normal stresses above mentioned model for simplicity purposes.
proportional to forces F, and F? are acting along the These observations clearly indicate that the grain
KOUL and GESSINGER: ON SERRATED GRAIN BOUNDARY FORMATION 1069

boundary y’ particle migration is responsible for Ackndwledgemenrs-This paper was written while one of
serrations in these alloys. the authors (Dr G. H. Gessinger) was on leave of absence
It is possible that a similar mechanism is operative from Brown Boveri & Co. at NRC, Ottawa, Canada. The
authors appreciate the helpful discussions with Professor H.
in 20-I 1P austenitic stainless steels as well, since Ni Gleiter, Professor E. Hornbogen, and Drs R. Singer and T.
combines with P to form compounds similar in Duerig. The authors are also grateful to Dr W. Wallace of
morphology and distribution to that of y ’ [l-3]. Such NRC for providing the materials and laboratory facilities to
compounds were not considered in the interpretation conduct this investigation.
of the results on high P austenitic stainless steels.
REFERENCES
5. CONCLUSIONS
I. M. Tamazaki. J. Japan Insf. Mcrul.s 30, 1032 (1966).
2. M. Kobayashi CI (II.. J. Iron Sleet Inst. Jupan 58, 859
The mechanism of serrated grain boundary for- (1972).
mation is related to y’ precipitation in Ni-base super- 3. 0. Miyagawa et al., Proc. Third Int. Symp. on the
alloys. A higher y’ solvus temperature than the M&, Metallurgy and Manyfacture of Superalloys. Seven
solvus temperature appears to be a prerequisite for Springs, Pennsylvania, p. 245 (1976).
the development of serrations. The serration ampli- 4. C. H. White, The Nominic Alloys (edited by W. Better-
dige and J. Heslop), 2nd edn, p. 82. Arnold, London
tude and wavelength increases with a decreasing (1974).
cooling rate until an optimum is reached. The min- 5. J. M. Larson, Mefall. Trans. 7A, 1497 (1976).
imum required cooling rate to initiate the formation 6. R. Thamburaj, Metall. Trans. A April (1983).
of serrations varies anywhere from air cooling to a 7. J. M. Larson and S. Floreen, M&tall. Trans. 8A, 51
(1977).
very slow furnace cooling from one alloy to another.
8. Y. Shimanuki er al., J. Japan. Sot. Powder Powder
A model based on the grain boundary primary y’ Metall. 25, 14 (1978).
particle movement causing the displacement of the 9. J. Beddoes and W. Wallace, Metallography 13, (85
local grain boundary segment and initiating the ser- (1980).
rated grain boundary formation has been proposed. IO. A. K. Koul and D. Morphy, Microsrrucf. Sci. Vol. 11
(1982).
The net strain energy differential between the matrix II. A. K. Koul and F. B. Pickering, Acfa metal/. 30, 1303
side and the boundary side of the particle matrix (1982).
interface provides a driving force for the movement I7 D. A. Porter and K. E. Easterling, Phase Truns-
.A.
of primary y’ particles in the direction of the bound- formations in Metals and Alloys, p. 161. Van Nostrand
Reinhold, London (1981).
ary. The simplified theoretical analysis and the ex-
13. H. F. Merrick, Precipitation Processes in Solids (edited
perimentaly determined serration amplitude values by K. C. Russel and H. 1. Aaronson), p. 161. The Metals
correlate well with the model presented. Previously Sot. A.I.M.E. (1978).
established models based on MI& precipitation and 14. C. Herring, Strucfure and Properties of Solid Surfaces
boundary migration in between the primary y’ par- (edited by R. Gromer and C. S. Smith), p. 5. Univ. of
Chicago Press (1953).
ticles are not considered responsible for the serrations 15. P. G. Shewmon, Trans. Am. Inst. Min. Engrs 230, 1164
on cooling from a higher temperature to a lower (1964).
temperature. 16. F. A. Nichols, J. nucl. Muter. 30, 143 (1969).

Das könnte Ihnen auch gefallen