AVERAGE STRESS IN MATRIX AND AVERAGE ELASTIC ENERGY OF MATERIALS

WITH MISFITTING INCLUSIONS*

T. MORIt and K. TANAKA7 $

Hrtving noted an important role of image stress in work hardening of dispersion hardened materials,‘*.3)
the present paper discusses 8 method of calculating the average internal stress in the matrix of a material
containing inclusions with transformation strain. It is shown that the average stress in the matrix is uni-
form throughout the material snd independent of the position of the domain where the average treatment
is carried out. It is also shown that the 8ctual stress in the matrix is the average stress plus the locally
fluctuating stress, the average of which vanishes in the matrix. Average elastic energy is also considered
by taking into account the effects of the interaction among ths inclusions and of the presence of the free
boundary.

COSTRAIXTE MOYENNE DANS LA MATRICE ET ENERGIE ELASTIQUE MOYENNE DES

XATERIAUZi CONTEKATU’T DES INCLUSIONS IMPARFA~S
Ayant remarque que la force image joue un &le important dans la consolidation des mat&iaux
durcls par dispersion, les auteurs proposent ici une m6thode de calcul de la contrainte inteme moyenne
dam la metrice d’un mat&iau conbnant des inclusions pr&entant des d6formations dues $ une trans-
formation, et montrent, que 18 contrainte moyenne dans la matrice est uniforme 8. travera le matbriau
et indbpendsnte de 18 position de la zone dans laquelle le traitement moyen est effect&. 11s montrent
aussi que ia contra&e r&&e dans 18 matrice est &8le $, 18 somme de 18 contrainte moyenne et de la
contfsinte locale variable dont 18 moyenne pour toute 18 metrice tend vers z&o. L’&ergie &stique
moyenne est Bgalement calculb en tenant compte des effets d’interaction entre les inclusions et de 18
presence du joint libre.

DIE JlITTLERE SPANNUNG IN DER MATRIX UND DIE MITTLERE ELASTISCHE

ENERGIE VON MATERIALIEN MIT EINSCHLUSSEN
rv’schdem die gro&. Bedeutung der Bildkraft fiir die Verfestigu~ von dispersions geh~e~n
Materi81ien’s+SL beront wurde, diskutiert die vorliegende Arbeit eine Method8 zur Beschreibung der mitt-
leren inneren Spannung in der Matrix eines Materials, das Einschliisse mit UmW8ndlUngSVen3p8nnUngen
enthitlt, Es wird gezeigt, d8B die mittlere Spannung in der Matrix im ganzen Material gleichfiirmig und
unabhitngig van der Lage des Bemichs ist; fiir den die Behandlung durchgefiihrt wurde. AuRerdem wird
gezeigt, d8B die aktuelle Spennung in der Matrix gleich der mittleren Spannung plus einer lokat
fluktuierenden Spannung ist, deren Mittelwert iiher die gesamte Matrix verschwindet. Die mittlere
eIastische Energie wird ebenf8lls diskutiert unter ~~c~~ehti~ng der W~hseIwjrkung‘~~ekte
znischen den Einschliissen und der Gegenwart der freien Oberfliiche.

1. INTRODUCTION by uniform plastic deformation occurring only in the

-4s Brown pointed out in describing the present
situation,(f) some papers have recently appeared which
discuss work-hardening of dispersion hardened mate-
rial~.(*-~) All these papers deal with internal stress
which developed as a result of plastic deformation
occurring only in the matrix. References (2) and (4)
pursue energy consideration and agree with each other
wit’h respect to the hardening rate, while references
(3) and (5) discuss the role of int,ernal stress in the
matrix in work hardeni~~~. Although Brown presented
a comprehensive vipn- of what was discussed in these
papers, (1) we would like nonetheless to report our
own understanding. for Brown’s explanation and
derivation of some stresses and strains were, in details,
not completely acceptable from our point of view and
we feel that an alternative treatment is possible.
Therefore, the results of some of our calculations
of the internal stress and of elastic energy will be
presented here. Since the internal stress developed
* Received July 22, 1972: revised September 8, 1972.
t Department of Metallurgic81 Engineering, Tokyo Institute
of Technology, Ookayama, Meguro, Tokyo, Japan.
$ Now at: National Research Institute for Metals, Nakame-
guro, Meguro, Tokyo, Japan.
-.
ACTA METALL~RG~~A, VOL. 21, MAY 1973
matrix can be duplica~d by giving uniform tranafor-
mation strain to inclusions alone, internal stress will
be charscterized in the following sections by inclusions
with uniform transformation strain. Thus, some ofthe
results will be directly applicable to a material with
misfitting precipitates.
2. AVERAGE INTERNAL STRESS IN MATRIX
For simplicity, the case where elastic constants
Ciinl are uniform throughout a specimen V, will be
considered. Suppose the specimen has N inclusions
which are, on the macroscopic scale, uniformly dis-
tributed in the matrix. When an inclusion V acquires
uniform transformation strain EijT, total strain eij’ is
introduced into the specimen. E<~’is divided into
two parts; cijm and eijim. Eij * represents constrained
strain* (total strain) when the inclusion is in an infinite
body, and E~:“’ represents image strain introduced
into the actual specimen by the presence of the free
boundary 1V,l of the specimen. &ijim is elastic in
nature throughout the specimen. is elastic
&ii *
outside the inclusion, but ~~~~~- cijT) is elastic
* In Eshelby’s notation eija is expressed 8s size.
571
572 ACTA
inside the inclusion. cij m is given by
Epj* = E,, ’
+ 4j, li(xa~‘)1/2}
dDtx’)>
where Gki is Green’s function for an infinite homogen-
eous body with elastic constants Cijkl.@) It can be
shown t,hat when aU inclusions acquire identical
transformation strain pijT, the specimen as a whole
undergoes a shape change and the average strain,
(Q~~)~~, that descibes the specimen shape change is
given by
(EijF)gTb = fe*jTt
where f is a volume fraction of the inclusions. From
this, let us assume the following : (1) the average of
total strains, in domain V,, from all the inclusions,
is also equal t,o fciiTif VR is a representative domain
for the specimen. This implies that V, has a sufficient
number of inclusions, say, Y inclusions, MV/V, = f,
and VR is not in a neighborhood close to the boundary
1V,,l. Assumption (I) is realistic. As an example, let
us consider the lattice constants of martensite. The
macroscopically determined ltverage lattice constants
of a martensite single crystal (if it were present) are
believed to be equal to those measured by a narrow
beam X-ray technique if interstitial atoms are uni-
formly distributed. Assumption (1) was justified by
Eshelby when &ijT = &aij (bij is the Kronecker
delta).‘?)
Let us consider the case where V, has a shape
similar to that of the inclusions. Then total strain
cijF is expressed as
M
P>M
s length of the fluctuation is apparent.ly of the order of
-j- 2 e,,iyx.xPf,
I=1
where .zijm(x, xp) is the quant,ity defined by equation
(1) when the P-th inclusion is at xp, ~~~~‘~(5,sp) is t.he
image strain due to the P-th inclusion, and VR is
assumed to contrtin inclusions 1 to i@. Apparently,
the second and the third sums are eIast,ic and do
not fluctuate much in VR. Therefore, it, is meaningful
to define the average of the second and the third
sums in equation (3) in V, as t,he average elastic strain
(%j)Y,3
NETALLURGICA, VOL. 21, 1973
As shown in a previous paper,(*) when a single inclusion
V with uniform transformation strain is within an
infinite homogeneous body, volume integrals of total
stmin and stress vanish if the integration is carried
(1) out, in the region V’ - I’, where V’ is a domain
surrounding V and is of shape similar to that, of I’.
Applying this conclusion, the average of the first sum
of equation (3) in V, is calculated as
ilf
where XijKlare Eshelby’s tensors and are equal to the
integrals in equation (1) when x is within I’ in equa-
(2)
tion (l).(6) Assumption (1) and equat.ions (3) through
(5) give
(&ij)yn = -f K%jmnemnT - ejjT)- (6)
Thus, the average elastic strain defined above is
independent of the posit,ion and the size of I’,.
The average internal &ress, (~i~~o)~-~,
defined by
(Oi$O>va= ~i$&&-R’ (7)
is also independent of the position and the size of
V,, insofar as VR can be regarded as a representative
volume. (aijo) v, is the average in V, of the sum of
the image stresses of all the inclusions and the stresses
of the inchtsions outside V, when they are in an
infinite body. However, it should be noted that. this
sum itself is nearly constant in V,.
Next, let us consider the following sum,
where aij m(x, zp) is internal stress due to the P-th
inclusion at xp within V, when it, is in an infinite
body. This sum, ai,* (1, M) cannot be assumed nearly
constant in V,; instead, it fluctuates and t,he wave
(3) the inter-inclusion spacing. cijm (1, X) is called 1ocaIIy
fluctuating stress, to which nearby inclusions obviously
contribute predominantly. However, the average of
this locally fluctuating st,ress in the matrix of Vn can
be shown to be zero. Instead of averaging equation
(8) directly, let us take the following approach.
Consider a fixed point in V,. First imagine V,
without inclusions. Next introduce an inclusion into
V, such that the fixed point is always outside the
inclusions and record (T$~~~V. Here, uilm is the stress
due to the inclusion when it is in an infinite body snd
SV is a small volume element around the fixed point.
If the introduction of the inclusion is repeated many
times in a random manner, Caij”6V becomes propor-
tional to the integral, Sv,_ v aiim6 V, where the center
of the inclusion is conversely fixed at the opposite
 NOR1 AND TAKAKA: MATERIALS WITH MIISFITTISG INCLUSTOSS 673

position of the above fixed point from the center of
Ti,. Because of the statement following equation (4),
this integral vanishes. Since this is t,rue for every
inclusion, the average of ~iirrnfrom ali the actually
present, inclusions in VR at a point within the matrix in
V, becomes zero. That is,
@ij”(ll M)), = 0.
(9)
The total average stress, (CT&~, in the matrix within
1B is the sum of {G,~O}~~and (~,~“(l, M)jnl. From
equations (6), (7) and (9),
when it is flexible. Brown and St,obbs also considered
t,he role of the locally fluctuating stress in work
hardening in a somewhat di~erent manner.(3~ It is
important to note that, hardening due to this factsor
operates as an energy dissipation mechanism.“*5)
Because of its fluctuating and position-dependent
character, estimation of hardening due to the locally
~uctuating stress involves certain approximations and
seems to depend on the choice of particular situations
to be considered; thus, it would not seem to be as
uniquely and simply performable as estimation of
hardening due to the average stress in the matrix,
taij)M*

with 3. ~NTE~CTIO~ AMONG INCLUSIONS AND

Csijr* = %?JL&nnT ENERGY CONSIDERATION
- 4% (11)
That the average internal stress in the matrix is
where oiila? is stress within a single inclusion when it is
equal to -fa..I”” naturally indicates that t’he actual
in an infinitely extended body. {aJIM is independent
stress within cn inclusion is, on the average, equal to
of the size and position of V, and of the shape of the
aijlm - foijIco. Thus, elastic energy per unit. volume
specimen.
of the specimen, EeEft,is given by
In summary, the internal stress at any point in the
matrix is the uniform average stress (aJM plus the E,1” = -f(l - f)oii%ilT/2. (12)
locally fluctuating stress from nearby inclusions. In Reference (2), the term (1 - f) was omitt,ed on the
This locally fluctuating stress is averaged to be zero assumption that the interaction among inclusions can
in the matrix. It is important to note that (a&X is be ignored. However, equation (12) is a correct
the average, within the matrix, of the sums of the expression which takes into account interaction among
st’resses of the inclusions when they are present in an inclusions together with the effect of the presence of
infinite body (Cbijm) and of the image stresses of all the free boundary of the specimen. If f is small,
the inclusions (X(rijim). The relative contributions of (1 - f) in equation (12) can be, for all practical
Cuijg and of Ceijim are not generally determined. purposes, replaced by unity. However, whenf is large,
However, as shown in the appendix, the contribution ignoring the effect,s of the interaction among inclusions
from each term can be calculated, if the specimen is of and the presence of the free boundary, as was done
ellipsoidal shape. Especially when the specimen is previously,@) may lead to a significant error.
similar in shape to the inclusions, (u~~);,~is solely due As already mentioned, Hart clarified,t5) in the case
to LT,,‘“. of n-ork hardening of dispersion hardened materials,
Bs Brown mentioned,“) when it moves in the mat- that a. hardening rate derived from consideration of
rix as a straight line, a dislocation feds, as a whole, energy balancet2) is essentially equivalent to that
only t,he average stress-(oij)4,f, since the average of the derived from consideration of the average internal
locally fluctuating stress in the matrix is zero. In stress in the matrix.cxv3’ If equat,ion (12) is used for
such a case. an extra applied stress equal to - (u,~)~~ an elastic energy calculation, exact agreement of the
ia needed to move a dislocation from the case where hardening rates oalculat,ed by the two approaches is
there is no internal stress. If &ijTis regarded as pro- verified.
duced by plastic deforma~,ion, t,he above hardening is ACKNOWLEDGEMENTS

work hardening which Brown and Stobbs identified We appreciate discussion with Professor T. Mura of
as due to the image stre#) and to which energy Northwestern Universit,y and the encouragement, he
balance consideration was applied.f2) The equivalency gave us. Dr. L. M. Brown not only supplied us with a
of both treat,ment>shas been discussed by Hart.t5) copy of his paper before publication but also corn.
From the present understanding of internal st.ress, mented on our &udy, and we greatly appreciat,e his
Hart’s treatment of work hardening of dispersion kindness.
REFERENCES
hardened materiaW5) is identified as that which
1. L. M. BROWN, Private communication (to be publidwd in
emphasizes and estimates the role of the locally Acta Met.).
fluctuating stress. As t’reated by Hart, a dislocation 0. K. TAKAKA and T. MORI, Acta Net. X8, 931 (1970).
3. L. M. BROWN rend W. M. STOBBS, Phil. Mw. eS, 1185
feels obstruct,ion from t,he locally fluctuating stress (1971).
574 ACTA METALLURGICA, VOL. 21, 1973

4. T. B&IRA,&tat.Conf. Me&. Pmpetiies of Materials at K.yoto, Equations (A3) and (A4) do not depend on the position
Vol. 2. (1971).
5. E. W. HART, Aeta Met. 80,275 (1972). of the inclusion. From equations (Al) and (A3),
6. J. D. ESEIELBY, Proc. R. Sot. A!%& 376 (1957).
7. J. D. ESHELBY, J. appl. Phys. 25,255 (1954).
8. K. TANAKA and T. MORI, J. Blmticity. To be published. UiiimdD = -VCijKI[Xktllnn(V~Vo)EmnT- &klT]. (A5)
9. Ii. TANAKA and T. MORI, Phil. Mag. 25,737 (1971). s v0
10. N. KINOSRITA and T. MURA, Whys. Status Solidi. 5, 759
(1971). Consequently, the average of the image stress in the
APPENDIX body, V,, is calculated as
AVERAGE STRESS AND IMAGE STRESS IN A
UNIFORM ELLIPSOIDAL BODY CONTAINING (air”“>r-a= -(vlvotc,j~,Es*,,,(v~cl,)e,,T - &*?I* (A6)
ELLIPSOIDAL INCLUSIONS WITH TRANS-
FORMATION STRAIN
Equations (A5) and (A6) are independent of the shape
and the position of the inclusion. What is used in bhe
Suppose that an ellipsoidal body, VO, contains an
above derivation is uniformit’y of sbress and straiu
ellipsoidal inclusion I’ with transformation strain
within an ellipsoidal inclusion with uniform trans-
~~~~~Elastic constants Cijkl are uniform throughout
formation strain when the inclusion is in an infinite
the body. Let dijDobe stress, assuming the inclusion
body. This uniformity is assured in an anisot,ropic
to be in an infinite body, and let rrijim be image stress
case.(r”j Thus, the above conclusions are valid when
to reflect the presence of the boundary IV,,l in the
the body is anisotropic.
actual body. The total internal stress in the actual
Finally, the average stress in the matrix will be
body, oij, is uijm + aijim. Since for static equilibrium,
considered when the body has many inclusions, the
crij dD = (dijm + CT~;~)dD = 0, number of which is N, all of which are identical in
s I’0 tir r-0 shape and volume. When the body is similar in shape
qjin dD = - jjj- dD -s,_pj =‘dD. to the inclusions, the result of equation (A4) becomes
.rF-o zero and the average stress in the matrix becomes
(Al) identical to the sum of the average image stress
As u, m in V is uniform,‘@ the first term in equation given by equation (A5). However, when the body has
(Al) is expressed as a different shape, the average stress in the matrix is,
from equations (A4) and (Ati),

(ofj>Jf = -(Nv/v*)C;.jkl[Sklmn(VO)EmnT - &klTl

where &,,,(I’) are Eshelby’s tensors defined for the
+ (NV~~‘~)( I - V/VJ’Cii,,
shape of the inclusion. Reference (8) show that the
second term in equation (Al) is independent of the x 1h%n(VcJ - %%mAwmnT.
size of the body V, and of the position of the inclusion Here, NV/ V0 is equal tof, the volume fraction of the
and is given by inclusion, and is finite. Since N is a large number,
V/V, must be negligibly small compared to unity,
_ Consequently, the above value becomes
- 4&nn(v)l~?nnTt (A31
where ~~~~~(Va) are Eshelby’s tensors appropriate
((3ij)izI = -~c~j*~[~*~~~(v)&~~T - &*lTI*

for the shape of the body I’,. Thus, the average, in In terms of the stress, defined in equation (I I), which
the matrix, of clijac of a single inclusion is written as is theinternal stress within a single inclusion in an
infinitely extended body, (ni,) is written as
laii4).lI = (aij”>~o-v

= {v/(vCl - V)}Cijkl
Xaij)+lI = -f”iflm,

x [4c,,,CVo, - ~mAWmnT. (W which is, of course, ident8ical to equation (10).