Professional Documents
Culture Documents
2 11 Giugno 1980
I. Mt+LL~R
~aehbereich 9, Herman~-F6ttinger-Institut, T U Berlin - 1 Berlin 12
]~. ~VILB[A~SKI
Institut o] ~ u n d a m e n t a l Technological t~csearch
Polish Academy o] Sciences - 00049 Warsaw
1. - Introduction.
283
284 I. l~IOLL]~l% &rid K. WILMANSKI
motion. The elements have two stable equilibrium positions of equal energy
and a metastable one. Statistical mechanics of the model permits the deriva-
tion of explicit functions for the free energy. These functions have a form
characteristic for bodies with a phase transition of the first order and contain
qualitatively most of the properties t h a t are exhibited b y a pseudoelastic body.
Such properties, t h a t are not simulated b y the model, include strain hard-
ening and thermal expansion. This is due to the fact t h a t the model in its
present form does not explicitly consider internal stresses, nor does it take
the expansion of the lattice structure in the body into account.
Also for the present purpose the model has been constructed so as to simu-
late a body under uniaxial tensile or compressive loading. I t will have to be
modified, if we want it to describe more complex loads.
(1) J. P~i~KI~S, Editor: Shape Memory Ef]eets in Alloys (New York, N.Y., ~nd
London, 1975).
(2) L. DELA]~Y, R. U. KI~ISENA~, H. TAS and H. WAI~LIMO~T: J. Mat. Sci., 9,
1511 (1974).
A M O D E L F01~ PHASiE T R A N S I T I O N I N _PSEUDO~L~kSTIC ]BODI/~$ ~
m f
III /
/ -4~
?Dt'~/;~/t _
r,
D
a) =5
P~_-_: ?_~-]
py___A ~
D
),
--PY =TR,
o)
f /
~A 3
D
)
D
=rq "~-Ts2
Fig. 1.
Moreover, once the b o d y has left the virginal curve, it will not be able, under
dead loading, to assume any state (_P, ])) within the hysteresis loop.
At the higher temperatures TR1 and 2J,~ the load-deformation curve shows
two hysteresis loops under dead loading and these occur in the first and third
quadrant. Figures le) and d) show t h a t there exists an elastic branch at
small loads and small deformations whose slope grows with growing tempera-
ture. Also the yield load P r now increases with temperatttre. Here again the
2~ I. MULLER a n d ~:. ~VILMANSKI
body m a y be loaded b e y o n d the yield load and its states will then be given b y
points on the second elastic branch, along which we m a y move up or dow~l
when we increase or decrease the load. However, unloading will now not lead
to a residual deformation, rather at P -- PR the body recovers what it yielded
at s ~ P y and when P falls below PR the body moves back to the origin along
the initial elastic branch. PR increases with increasing temperature and the
difference between P~ and _P~ grows smaller. Under compressive dead loading
the body behaves like under a tensile load with the obvious differences.
The complete recovery of the initial configuration despite the intermediate
yielding has given rise to the t e r m pseudoelasticity. I t is also said t h a t the
body exhibits shape memory, because, if we leave it at a low temperature Tp~ (say)
in the deformed state (P, D) ~ (0, De), the body will creep back to the initial
shape D -~ 0 when h e a t e d to T.~ (say); this is so because, according to fig. lc)
and d), at the higher t e m p e r a t u r e there is only one possible unloaded state
and this occurs at D ~ 0.
At high temperatures Ts~ and Tz~ the hysteresis loops vanish and we ob-
serve load-deformation curves like those of fig. le) and ]), all parts of which
can be traversed in both directions. In the neighbourhood of the origin the
curve is steeper for a higher temperature.
While in dead loading it is impossible to reach any state within the hysteresis
loop of fig. la) through ld)~ this is well possible if we clamp the body to a de-
formation in the appropriate region. The load that is necessary to maintain
the clamp depends on the m a n n e r in which we apply the clamp~ but, of course,
it is such t h a t the state (P, D) lies within the hysteresis loop. Once such a state
is established, we m a y change it along the dashed elastic lines within the loop
by shifting the clamp.
(3) F. ~BAuMGART, J. JORDE ~nd H. G. R]~ISS: Krupp Forsch. Ber., 34, 1 (1976).
A MODEL FOB, P H A S E TRANSITION IN PSEUDOELASTIC BODIES 287
I n particular, one should then be able to predict the shapes of the deformation-
t e m p e r a t u r e curves for different loads. The m e m o r y p r o p e r t y of the body is
v e r y m a r k e d in fig. 2: the body alternates between large and small deforma-
Fig. 2.
b) Lattice structure.
a) b) c)
Fig. 3.
~l 9
@
i~
@
/] --
iI ~, M+---
@ @
9 ~ 9
$ $ 9
a) b) c) a)
Fig. 4.
because they separate lattice cells of type M+ from their twins M . The process
of twinning consists of shuffling the a t o m + to a new positive within the
lattice cell and this results in a change of the shape of the cell.
I t is obvious from a comparison of fig. 4b) and c) t h a t the fraction of
twins M+ and M determines the length of the body. I n the present instance,
therefore, a tensile load will promote the formation of M+, while a compressive
load will be favourable for M_. Experiments show t h a t at high temperature
A MODEL FOR PHASE TRANSITION IN PSEUDOELASTIC BODIES 289
3. - Model.
a) Basic element.
a)
,')~ /]
b) I
i
- - - ~ J I~
c)
Fig. 5.
and denote it b y A, while the configurations of fig. 5a) and c) are called mar-
tensitic and are denoted b y M and M+, respectively. The lattice layer M_
is shorter b y J t h a n A, while M+ is longer b y J. E x c e p t in configuration A
the lattice cells are not rectangular; we accept this drawback of the model
for the sake of simplicity.
We m a y imagine t h a t the three configurations of the lattice layer can be
converted into each other b y applying a shear stress to the top and b o t t o m
surface. I n fact, all intermediate configurations can thus be obtained and
their displacements wili be characterized b y the variable • (see fig. 5b)).
F r e q u e n t l y we shall make this displacement d nondimensional b y choosing
d = A / J as a measure of displacement.
Figure 6a) represents the postulated potential energy r of a lattice layer
as a function of & This potential energy has minima at 8 ---- 0 and 8 ---- • 1.
We conclude from fig. 6a) t h a t the layer configurations in fig'. 5 are equilibria
and, in particular, t h a t the martensitie layers are stable, while the austenitic
one is metastabte. There are energy barriers between the minimal energies
and for IN i > J the potential energy tends to infinity rapidly.
19 - I I N u o v o Uimento B.
290 ~. ~i)LLm~ and x. WIL.~a.~SKI
q5
>
"-" 0 i 0 1
a) b)
/ % " 9' d2 (3po + 2~v,)64 . (2q% + ~vt) a '; for i,~[ ";: 1 ,
l fro' [,~!> 1 .
Fig. 6.
b) l?seudoelastic body.
~) C o n s t r u e l i o n of the body. W e m o d e l the b o d y as a stack
of la.ttiee layers. T h e layers are a r r a n g e d at an angle of 45 degwees to t h e
A MODEL FOR PHASE TRANSITION IN PSEUDOELASTIC BODIES 291
loading surfaces ('). Figures 7a) through 7c) show this a r r a n g e m e n t for the
three cases when the b o d y is either austenitie or martensitie of types M+
and M_ throughout. I t is clear t h a t raider a tensile or a compressive load the
surfaces of the lattice layers are surfaces of m a x i m a l shear stress. Thus the
layers of t y p e A m a y be sheared into layers of t y p e s M+ or M or vice versa
and the configurations shown in fig. 7b) and c) m a y be the result of com-
pressive and tensile loading, respectively.
c~)
S
I / / /
"A
9 A
/
"
A
b) " c)
O
ct)
Fig. 7.
(*) We consider only uniaxial loading in this paper and the model is constructed
accordingly.
292 L M/JLLEI% a~nd_ K. W~LMtkNSKI
where A~ is the displucement of the i-th lattice layer. The factor 1/~/2 results
from the arrangement of the layers at an angle of 45 degrees.
The potential energy of the b o d y is assumed to be the sum of the po-
tential energies of the lattice layers (*)
2/
(3.3) E ot = ~ qP(Ai).
i=1
(*) We neglect the potential energy of the incomplete layers which do not contribute
to the deformation.
A MODEL FOR PIIhSE TI~,ANSITION IN PSNUDOELASTIC BODIES 293
4. - Statistical m e c h a n i c s o f the m o d e l .
is equal to
o [i ]
A1=--co ~/~,=--co i=1
The sttmmation in the exponent extends over all lattice layers, while the sums
in the denominator are sums over all values zJ of each layer. I n all sums proper
regard must be given to the condition t h a t the deformation D is fixed, so t h a t
we have
2g
(4.2) ~: ~ , = V ~ I ) .
i=l
zJ l = - - - c a ZJN----- co i=l
n
o M /ll
(4.3)
k In W ,
holds. W e use the Stirling f o r m u l a for expressing the factorials and obtain
for the e n t r o p y of the ensemble
semble is given b y
Tpot = E~ot-- T H ,
2(
(4.5)
co co
(4.8) ~ N~ = ~ and ~ /15T~ = V 2 D ,
A=--a~ A=--co
296 I. MULLER and K. WILMANSKI
where (4.8) reflects condition (4.2). The potential energy of the body can be
expressed by the formula
co
(4.9) E o~ =
W--
~c [ NA.
o(
=_ N/
co co
I n this ease N obviously vanishes for [A]> J and we obtain for IA I~<J (*)
(*) The subscript 0 refers to the box potential for which q~(A)=0 for IAI<~J.
A MOD:EL F O R P H A S E TRANSITION IN PSEUDO:ELASTIC BODIES 297
where c< and fl are Lagrange multipliers which rosy be determined by inser-
tion of (4.15) into the constraints (4.14). In the determination of the Lagrange
multipliers it is necessary to convert the sums in (4.14) into integrals. There-
fore, we assume t h a t the number of displacements between ~1 and ~1 ~- d3
is equal to zdLJ, where z is some constant of proportionality. All the integrals
can then be evaluated and we obtain 1 ~ g = ]n 2zJ -~ In (sinh fiJ/flJ), so t h a t
holds. Insertion of this expression for ~ z into (4.14) and calculation of the
integrals leads to ~ relation between the Lagrange multiplier fl and the de-
formation, viz.
(4.17) s = -- ~ 2 ~D- j ,
The index 0 on -0z(~') indicates t h a t this partition function applies to the box
potential where qs(d) = 0 for ]sJ [ < J . I t is now possible to calculate Epot, H
and [P,ot for this simple potential by use of eqs. (4.7). This leads to the re-
lations
[ --~N2? 1 sinhfiJ]
(4.193 o~OO~=0, q =~Vl~ l n g z J + f i J + n--7~ ] .
holds, where the &function under the integral takes care of constraint (4.2). We
recall t h a t ~b(d) is of the form shown in fig. 6b).
The following approximate calculation of the partition function is reminis-
cent of the first step in the virial expansion of the partition function of a real
gas and it was m o t i v a t e d b y t h a t expansion. The calculation proceeds in
four steps.
I n the first step we decompose the exponential in (4.20) additively into
two parts, viz.
0 for I A I < J ,
~o(A) = ~ for I A [ > J ,
so t h a t
1 for ]A[ < J ,
(4.22) Fo(A) = 0
for ]A I > J ,
while F~ is the remainder of exp [-- (I)(A)/kTJ, viz.
(4.24) Z(~,(D, T) = ~
;/ " ,=1
leI (Fo(d~) + FI(~,)) ~ 1)
( V~l d , )d& ... d ~ , .
--r --co
I n the second step we carry out the multiplication and obtain after some re-
an'angement of terms
co A
9 "
--co --co
(
where Z~v-A'D- (1/V~)E ~, has been introduced for
g=l
= • D 1 1 A
j "~176 / 9 .
--co --co
A MODEL FOR PI-IASE TRANSITION IN PSEUDOELASTIC BODIES 299
(4.27) Z(N)(D, T) =
A
ZA ... ,
--co --co
W i t h (4.18) we thus arrive at the final f o r m of the partition function for big
~00 I. MU~LLtiil~ and K. WILMANSKI
values of N :
J
fi) F r e e energy. The free energy ~,ot follows from insertion of (4.29)
into eq. (4.7)3:
(4.30) T~o~(D,T) =
J
[ q~(A)1 r
kT
The free energy formed with this partitioa function according to (4.7)3 reads
5'
- - T { N k ( l n 2 z J + # JTNVJ~-D + an sl--n~Jfi--J)}
'
A,d=--m
where ~ o t and H o are energy and entropy for the box potential (see (4.19))
and N00~Jis the distribution function (4.16) of elements in box potentials. Thus
in the limiting case (P(A)<<kT, the free energy ~Ppot= E ~ o t - - T H m a y be
formed with an e n t r o p y calculated for the box potentials and with an energy
t h a t uses distribution functions corresponding to the box potentials.
We have used the limiting form (4.32) of the free energy when we f r s t
presented this research in (5).
With the potential q~(A) in the form (3.1) the free energy ~P~o~(D,T) of (4.31)
m a y be evaluated graphically and fig. 8 is a graphical representation
of the result for Fo----0.3125 and F 1 - 11.875. The figure shows ~Poj2V%--
--~/Jot(0, s as a function of d - - ~ / 2 ( D / N J ) for 5 typical values of
0 = k T / % (01 = ].21, 0 i i : J , 2 5 , 0 i i I ~ 1.26, 0iv = 1.32, 0v = 1.42).
At a low temperature 0~ the free energy has a m a x i m u m in the origin and
minima on the two sides for positive and negative values of d. At ~n inter-
mediate t e m p e r a t u r e 0 n > 0~ a minimum appears in the origin, while the lateral
minima persist. A somewhat higher t e m p e r a t u r e 0 m makes the minimum in
the origin deeper and the lateral ones shallower. At 0~v > 0H~ the lateral
minimg disappear and we have only the central minimum. B u t at 0~v the
curves ~ vs. D still show two points of inflection both for positive and negative
values of D. I f the t e m p e r a t u r e is still higher, at 0v, the slope of these curves
increases monotonically.
This rather drastic change of character of the (~P, D)-curves occurs in a
fairly narrow range of temperature as can be read off from the numbers in fig. 8.
The dashed parts of the curves correspond to unstable equilibria (D, T),
as we shall see in subsect. 6 a), while the solidly drawn parts represent stable
equilibria. At low temperatures, in the range of 0~, there exist only two such
stable branches, at large negative and positive values of d. These obviously
correspond to equilibria in which the model is in phase M_ or M+, respectively.
In the intermediate t e m p e r a t u r e ranges of 0~, 0 m and 0~v there exist three
stable branches of which the one in the middle represents equilibria in the
O~ < 1.2143)
1.21.13 < 011 < 1.2553 )
t.-1195 -( 0 v .
1.0
0.8
0.6
J
I 0.4
).-~ 0.2
-0.2
d
--1
F i g . 8.
A ~ [ 0 D E L F O R 1)HASE T R A N S I T I O N I N F S E U D O E L A S T I C B O D I E S 303
a) Thermodynamic results.
1
(6.1) d H = ~ (dE - - P d D ) .
(6.2) d(P ~- -- H d T ~- t) dD
~T(D, T)
(6.3) P(D, T) - a_u
~T,o~(D, T)
(6.4) P(D, T) = caD
Therefore, we can obtain the load-deformation curves of our model for dif-
ferent t e m p e r a t u r e s f r o m the ~Vo~ vs. D curves of fig. 8 b y differentiation. The
resulting load-deformation curves are shown in fig. 9 as the solid and dashed
lines for typical t e m p e r a t u r e s in t e m p e r a t u r e ranges 0~ through 0v t h a t were
defined in (5.1). These curves will be discussed in subseet. 6 b) below.
(7
~) / e)
f"~\\1\\ ~ i
(if" .............
:)
I I I I
. . . .
\\
i o,:1.27
I I
--0.5
I
0
1
0.5
-L J r
--0.5
0n=1.25
0
I i
0.5 --0.5
I
0m=l 2S
0 0.5 --0.5
81~=1.32
0 0.5
I
--0.5
0g =1.42
I
0
I
0.5
I~_
Fig. 9.
Y) P h a s e e q u i l i b r i u m . A b o d y m a y sometimes be in a uniform
equilibrium (D, T) which satisfies the t h e r m o d y n a m i c stability condition (6.5)
and y e t it m a y not be in a stable equilibrium. This can happen when the
body is capable of assuming a free energy ~ lower t h a n ~P(D, T) b y decom-
posing into two different phases ~ and fi, thus becoming a nonuniform body.
Phase c~ contributes the part D~ to the deformation D and phase fi contri-
butes D~ such t h a t
where x is the ratio of lattice layers in phase /3. The free energy is similarly
decomposed into two parts according to
(6.7)
~(D:T)..,/
gent for the deformations D~ and D~; hence their vMues follow, see fig. 10a).
k
P(D~T)
/ 1
I I
i D I
5
Fig. 10.
The slope of the common tangent is the load at which the phases can coexist
at the given temperature. ~
5Ioreover~ from (6.3) we have ~(D2~ T ) - T(1)1; T) :fP(D~ T)4l) for all
Dz
D1 and D2 and, therefore, in particular for D1 - D~ and D~ ~-- D~. Combining
this with (6.8)5, we obtain
DB
(6.9) ~ ----~v(D~, T) @
~T(D~,
~D
T)(D--D~).
~0 - II Nuovo Ciraento B.
306 i. M~LL~R and x. WILMANSKI
(6.10) P ( D , T) =
where, as always, W(~J) ---- -- %/2D/NJ holds. W i t h exp [-- +/kT] ~ 1 -- ~P/I~T
we get
-- kT ( ) ~(1--qb(A)/kT) coshflA dA ~
0
]'
(r I
LO M+
IM_,U§ ~ .
. i I I ; 1 I i I I I I I~
1.21 1.25 1.29 1.33 1.37 1./+1
Fig. 11.
c) De]ormation-temperatuve curves.
Obviously for the p r e p a r a t i o n of fig. :11 we h a d to h a v e load-deformation
curves for m a n y more temperattlres t h a n the 5 shown in fig. 9. A n d indeed
we calculated such curves for v e r y m a n y t e m p e r a t u r e s over the interesting
i n t e r v a l of t e m p e r a t u r e s . Thus we were also able to construct graphically the
dependence of the deformation on the t e m p e r a t u r e for various prescribed
loads. T h e result is shown in fig. :122 whose curves refer to positive loads
except fig. 12a), which corresponds to zero load. There are various branches
in the d e f o r m a t i o n - t e m p e r a t m . e diagrams of the model which we h a v e denoted
A I~OD~][, }'OR PIiA8~ TRANSITION IN PS~U][)O}gLASTIC BO]DI:NS 309
I I
I I
0.8 a)
0-=0
i i
i I ~ t I t
:x x i
: ~. I
-0.8
I
l
F I
I
I I I
0.8 b)
i lJ-~ z
---0.2
i i i i
I I:
I
i
I
--0.8- I
I
I I
1
i
l I
I c)
0.8- I I
I
I I t :f" u =0.5
r
0 I i : ~I I r ~ i I i i
e._i
-I
-- 0.8-
I
a)
=1.0
o il r i I I I F I I I
,-: ._7
I 1 r i i I I I [. ] ]
1.21 1.25 1 29 L33 1.37 1.41
l~ig. 12.
d) Elastic modulus.
The elastic modulus :Y of the model is the slope of the load-deformation
curve at zero load, or the curvature of the free energy-deformation curves at
their minima. B u t there are three minima of free energy in the interesting
range of temperature, namely two identical ones for the martensitie phases M_
and M+ and another one for the austenitie phase A. The corresponding elastic
moduli are drawn in fig. 13. The values t h a t have been used to draw t h a t
figure were calculated numerically from eq. (4.32).
\
" 31
10
8
0
1.21 11.25 1.29 L33
I
1.24-31
Fig. 13.
7. - The n o n u n i f o r m body.
a) Constrained equilibrium.
Figure 10 and the discussion of load-deformation curves of fig. 9 show
w h a t hysteresis means. The b o d y assumes the states on the solid branches of
the free-energy curves r a t h e r t h a n the states on the t a n g e n t where it would
be in phase equilibrium a n d h a v e a smaller free energy.
I n t e r m s of the model the b o d y fails to establish phase equilibrium, be-
cause the lattice layers are constrained in their motion. Two different reasons
can be conceived as possible causes for this constraint:
i) The energy barrier between two equilibrium positions m a y p r e v e n t
the lattice layer to flip to a new equilibrium position.
ii) I n a polycrystalline b o d y the lattice layers will t e n d to inhibit
each other's free motion.
W e are not sure which of these two is the dominating effect~ b u t it seems
t h a t the geometric inhibition described in ii) could be the p r i m a r y reason in
case of the (A~ M+)-transition, while it is primarily the height of the energy
barrier t h a t prevents phase equilibrium between M a n d M+. I n a n y ease
b o t h effects m a y contribute to m a k e the b o d y s t a y in a constrained equilibrium
even though it would stand to gain energy b y the flipping of lattice layers.
(7.6) P(D, T; x ) =
~D
and this function determines the load-deformation curves t h a t lie inside the
hysteresis loop of fig. 9d). Such curves can be observed in pseudoelastic bodies,
see subsect. 2 a), fig. lc) and d).
8. - A n a l o g y to f e r r o e l e e t r i c i t y .
a) Phenomenology o/ ferroelectrics.
;Y
b) c)
Fig. 14.
T
a) b) c)
Fig. 15.
Pseudoelasticity Ferroelectrieity
P E
D
Y S
T~-'~(o,T)
y'2
Fig. 16.
similar to those of fig. 8 for the model of a pseudoelastie b o d y and this similarity
is the reason of the a b o v e - m e n t i o n e d analogy between ferroelectricity and
pseudoelastieity.
Once we k n o w the curves ~/J = ~ ( ~ , ~'), all other information a b o u t the
t h e r m o d y n a m i c b e h a v i o u r of ferroelectrics follows just as the t h e r m o d y n a m i c
properties of the model for pseudoelastie bodies followed f r o m the knowledge
of its free-energy function. Qu'ditatively f r o m here on all a r g u m e n t s are tile
s a m e in the two approaches and, therefore, we are not surprised t h a t L a n d a u
t h e o r y describes curves like those of the fig. 14 and 15, which, in pseudo-
elasticity, we saw to be consequences of free-energy curves of the shapes
of fig. 16.
(*) The notation is taken from an instructive review article by DEVONSHIRE (9).
(9) A. F. DEVO~SIIIRE: Adv. Phys., 3, 85 (1954).
316 ~. Mi~,LI,ER and K. WILMANSKI
These curves are draw]l in fig. 17 and it is clear thut, un(ler the @plieution of a
field, the polarization will not follow the dashed parts of the curves but rather
ff
:T" : \:
i"l : E _~2J E
i \\ . :"i ~-- K I " ~ ------
a) c)
:Fig. 17.
the straight dotted lines. Thus the observed (a, E)-diagrams of fig. 14 appear.
I t seem~s that FALl( in 0 ~ was the first to notice that the observed
(P, D)-diag~rams of a pseudoelastic body can be summarized b y a free-energy
function of the form (8.1) with ~ replaced by D.
F o r one of the authors (K. ~YII,SrA~-SKI)this research was carried out during
his tenure as a won l I u m b o l d t research fellow in die Gesamthochschule Pader-
born nnd die Teehnische Uni~-ersit~t Berlin.
The difference ~ - - / 3 .
We have
.4
(1o) ]?. FALK: Wie hdngt die F,reie Energie ei~ter Memory Legie,rung con der Verzerr,ung
u,ud der Temperatur ab?, ZA3IM: Sonderband on GAMM meeting 1979 Wiesbaden
(in press)9
A MODEL FOR PHASE TRANSITION IN PSEUDOELASTIC BODIES 317
N i=1
s N - - A ~f(flJ) + ( X - - A ) J "
F o r N > > A we get
~f(fixJ) ~ ~ ( f l J )
A
since ~ A C A can at most be equal to AJ. Hence there follows t h a t fiA ~ ft.
i=1
.APPENDIX B
Zy' tD ) <
Z~oZC'(D )
Because of the definition (4.22) of N0(A) we m a y write the n u m e r a t o r on the
right-hand side as
co
--co
Thus
Since, b y (4.23), ]~(A)] < 1 holds, this inequality proves t h a t the terms in
(4.25) or (4.27) for higher values of A c o n t r i b u t e b u t little to the value of
the sum.
318 i. MULLER a n d K. WILMANSKs
9 RIASSUNTO (*)
M o a e a b Aria ~ a a o n o r o n e p e x o a a n n c e n ~ o y n p y r a x Teaax.