ON RIGIDITY FOR THE FOUR-WELL PROBLEM ARISING IN THE CUBIC-TO-TRIGONAL PHASE TRANSFORMATION ANGKANA R ULAND AND THERESA M. SIMON

2025-05-02 0 0 504.1KB 19 页 10玖币
侵权投诉
ON RIGIDITY FOR THE FOUR-WELL PROBLEM ARISING IN THE
CUBIC-TO-TRIGONAL PHASE TRANSFORMATION
ANGKANA R ¨
ULAND AND THERESA M. SIMON
Abstract. We classify all exactly stress-free solutions to the cubic-to-trigonal phase trans-
formation within the geometrically linearized theory of elasticity, showing that only simple
laminates and crossing-twin structures can occur. In particular, we prove that although this
transformation is closely related to the cubic-to-orthorhombic phase transformation, all its
solutions are rigid. The argument relies on a combination of the Saint-Venant compatibil-
ity conditions together with the underlying nonlinear relations and non-convexity conditions
satisfied by the strain components.
1. Introduction
Shape-memory alloys are materials with a thermodynamically very interesting behaviour:
They undergo a diffusionless, solid-solid phase transformation in which symmetry is reduced.
More precisely, a highly symmetric high temperature phase, the austenite, transforms into a
much less symmetric low temperature phase, the martensite, upon cooling below a certain critical
temperature [Bha03]. Mathematically, these materials have very successfully been described by
an energy minimization [BJ92] of the form
ˆ
W(u, θ) dxmin .(1)
Here R3denotes the reference configuration, which often is chosen to be the austenite state
at the critical temperature θc>0, the deformation of the material is u: Ω R3, the temperature
is denoted by θ: Ω [0,) and W:R3×3
+×[0,)[0,) corresponds to the stored energy
function. This function encodes the physical properties of the material and is assumed to be
(i) frame indifferent, i.e. W(F, θ) = W(QF, θ) for all FR3×3
+and QSO(3),
(ii) invariant with respect to the material symmetry, i.e. W(F, θ) = W(F H, θ) for H∈ Pa
where Padenotes the symmetry group of the austenite phase, which we assume to strictly
include the symmetry group of the martensite phase.
Here (i) can be viewed as a geometric nonlinearity, while (ii) encodes the main material nonlin-
earity which, for instance, reflects the transition from the highly symmetric austenite to the less
symmetric martensite phase. Both structure conditions imply that the energies in (1) are highly
non-quasiconvex and thus give rise to a rich energy landscape. As a result, minimizing sequences
can be rather intricate, which physically leads to various different microstructures.
In this note, it is our objective to study a specific phase transformation for which experimen-
tally interesting microstructures are observed. Seeking to capture “crossing-twin structures” in a
fully three-dimensional model (see Figure 1, left), we focus on the cubic-to-trigonal phase trans-
formation in three dimensions. This deformation, for instance, arises in materials such as Zirconia
or in Cu-Cd-alloys but also in the cubic-to-monoclinic transformation in CuZnAl. We refer to
[Sim97,PS69] for experimental studies, to [HS00] for an investigation of special microstructures
in a geometrically nonlinear context and to [BD00,BD01] for mathematical relaxation results for
1
arXiv:2210.04304v1 [math.AP] 9 Oct 2022
2 ANGKANA R ¨
ULAND AND THERESA M. SIMON
strains possible normals (aij , nij )
e(1), e(2) [1,0,0],[0,4,4]
e(1), e(3) [0,0,1],[4,4,0]
e(1), e(4) [0,1,0],[4,0,4]
e(2), e(3) [0,1,0],[4,0,4]
e(2), e(4) [0,0,1],[4,4,0]
e(3), e(4) [1,0,0],[0,4,4]
Table 1. The possible normals arising in simple laminate constructions. For
these the strain alternates between the two strain values given in the left column
of the table.
the associated geometrically nonlinear problems. Since the study of the minimization problem
(1) can be rather complex, in this note we make the following three simplifying assumptions
which are common in the mathematical analysis of martensitic phase transformations:
We fix temperature below the transition temperature,
we consider only the material nonlinearity while linearizing the geometric nonlinearity,
and we study only exactly stress-free structures.
Instead of investigating the full minimization problem (1), we thus study the differential inclusion
e(u)∈ {e(1), e(2), e(3), e(4)}in T3,(2)
where
e(1) =
d11 1
1d21
1 1 d3
, e(2) =
d111
1d21
1 1 d3
,
e(3) =
d111
1d21
11d3
, e(4) =
d11 1
1d21
11d3
,
(3)
and d1, d2, d3are material-specific constants. In order to avoid additional mathematical difficul-
ties, we assume that the reference configuration is given by the torus T3:= T1×T2×T3, where
Ti:= [0, λi) for some λi>0 and i∈ {1,2,3}.
We observe that all the matrices in (3) are symmetrized rank-one connected, i.e. for each
i, j ∈ {1,2,3,4}there exist aij R3\ {0}, nij S2such that
e(i)e(j)=1
2(aij nij +nij aij ).
It is well-known that, as a consequence, the differential inclusion (2) thus allows for so-called twin
or simple laminate solutions, i.e. solutions u(x) = u(nij ·x) with nij S2denoting the vectors
from above. These are rather rigid, one-dimensional structures, which are frequently observed
in experiments [Bha03], see also Figure 1, right. The possible pairs (aij , nij )R3×S2for the
cubic-to-trigonal phase transformation are collected in Table 1.
Contrary to other materials such as alloys undergoing a cubic-to-tetragonal phase transfor-
mation, simple laminates are not the only possible solutions to (2). As in the (more complex)
cubic-to-orthorhombic phase transformation, also in the cubic-to-trigonal phase transformation
“crossing-twin structures” can emerge. These are two-dimensional structures involving “lam-
inates within laminates” (see Figure 1, left). In particular, these patterns locally consist of
zero-homogeneous deformations which involve specific “corners” which are formed by four dif-
ferent variants of martensite.
RIGIDITY FOR THE FOUR-WELL PROBLEM 3
1.1. The main result. As our main result, we classify all solutions to the differential inclusion
(2) and prove that in addition to the simple laminate solutions only crossing twin structures
arise.
Theorem 1. Let e(u) := 1
2(u+ (u)t)with e(u) : T3R3×3
sym be a periodic symmetrized
gradient. Assume that
e(u)∈ {e(1), e(2), e(3), e(4)}.
Then the following structure result holds:
(i) There exists j∈ {1,2,3}such that je(u) = 0.
(ii) Assuming that j= 2, there exist functions f1:T1Rand f3:T3Rsuch that
f1, f3∈ {−1,1}and either e13(u)(x) = f1(x1)for a.e. xor e13(u)(x) = f3(x3)for
a.e. x.
(iii) Assuming that e13(u)(x) = f3(x3), consider Φ(s, t) := (tF3(s), s), where F3(s)is such
that F0
3(s) = f3(s)a.e. and F3(0) = 0. Then, there exists g: Φ1(T3)R,(s, t)7→ g(t)
such that
(e12 Φ)(s, t) = g(t),(e23 Φ)(s, t) = f3(s)g(t).
Remark 2. All cases not listed result from the symmetries of the model under permutation of
the space directions, as these only permute the side lengths of the torus T3and the constants d1,
d2, and d3, the precise values of these constants do not enter the argument. The permutations
play the following roles: If an index i∈ {1,2,3}has been fixed, the other two can be exchanged
via transposition. A fixed index can be transformed into a different fixed index by a full cyclic
permutation.
Let us comment on this result: From a materials science point of view, it gives a complete
classification of exactly stress-free solutions for the cubic-to-trigonal phase transformation in the
geometrically linear framework. Mathematically, Theorem 1provides a rigidity result for a phase
transformation which leads to more complex structures than simple laminates. While a similar
classification and rigidity result had been obtained in [R¨ul16a] for the cubic-to-orthorhombic
phase transformation in three dimensions, this required strong geometric assumptions on the
smallest possible scales. These assumptions were also necessary as a result of the presence of
convex integration solutions for the corresponding differential inclusion in the case of the cubic-
to-orthorhombic phase transformation. In contrast, in our model, such conditions are not needed.
Due to the smaller degrees of freedom that are present (four instead of six possible strains), in
fact any exactly stress-free solution must satisfy the structural conditions and “wild” convex
integration solutions are ruled out.
1.2. Relation to the cubic-to-orthorhombic phase transformation. Due to the outlined
rather different behaviour (in terms of rigidity and flexibility) of stress-free solutions of the cubic-
to-orthorhombic and the cubic-to-trigonal phase transformation, we explain the algebraic relation
between these two transformations: To this end, we recall that for the cubic-to-orthorhombic
phase transformation, the exactly stress-free setting in the geometrically linearized situation
corresponds to the differential inclusion
e(u)∈ {e(1), . . . , e(6)},
4 ANGKANA R ¨
ULAND AND THERESA M. SIMON
e(4) e(4)
e(4) e(4)
e(3) e(3
e(3) e(3)
e(1)
e(1)
e(2)
e(2)
[0,1,0]
[1,1,0]
[1,1,0]
e(1)
e(1)
e(1)
e(1)
e(2)
e(2)
e(2)
e(2)
Figure 1. Schematic illustration of a crossing twin structure (left) and a simple
laminate (right). Only very specific twins can be used to form crossing twin
structures. These are obtained as a consequence of the compatibility conditions
satisfied by the strain equations.
with
e(1) :=
1δ0
δ1 0
0 0 2
, e(2) :=
1δ0
δ1 0
0 0 2
, e(3) :=
1 0 δ
02 0
δ0 1
,
e(4) :=
1 0 δ
02 0
δ0 1
, e(5) :=
200
0 1 δ
0δ1
, e(6) :=
2 0 0
0 1 δ
0δ1
.
Here δ > 0 is a material dependent parameter. If now one assumes that a microstructure only
involves the infinitesimal strains {e(1), . . . , e(4)}and if one carries out the change of coordinates
x7→ ˆx:= Ctx,u7→ ˆu:= Cu with
C:=
1
30 0
03
2δ0
0 0 1
3
·
0 1 1
2 0 0
0 1 1
,
using that the (infinitesimal) strain transforms according to e(ˆu) = Ce(u)Ct, one exactly arrives
at the differential inclusion (2) for the cubic-to-trigonal phase transformation with the parameters
d1=1
3,d2=3
δ2,d3=1
3. This shows that the differential inclusion for the cubic-to-trigonal
phase transformation indeed corresponds to a subset of the differential inclusion for the cubic-
to-orthorhombic phase transformation. Due to the fewer degrees of freedom, in contrast to the
full cubic-to-orthorhombic phase transformation, it however displays strong rigidity properties.
1.3. Main ideas. The arguments for the proof of Theorem 1rely on a combination of the
linear compatibility conditions for strains in the form of the Saint-Venant equations and the
nonlinear constraints in our model. More precisely, the Saint-Venant conditions imply structural
conditions on the possible space dependences of the strains. Furthermore, the full classification
result requires a breaking of symmetries that can only be deduced in combination with the non-
convexity of the problem, i.e., the fact that for all i, j ∈ {1,2,3}we have that eij (u) attains
at most three possible values and the nonlinear relation e23 e12e13 = 0. For a simplified
model with only two-dimensional dependences similar arguments had earlier been considered in
[R¨ul16a]. However, in contrast to [ul16a], in the present setting we do not need to make use of
摘要:

ONRIGIDITYFORTHEFOUR-WELLPROBLEMARISINGINTHECUBIC-TO-TRIGONALPHASETRANSFORMATIONANGKANARULANDANDTHERESAM.SIMONAbstract.Weclassifyallexactlystress-freesolutionstothecubic-to-trigonalphasetrans-formationwithinthegeometricallylinearizedtheoryofelasticity,showingthatonlysimplelaminatesandcrossing-twins...

展开>> 收起<<
ON RIGIDITY FOR THE FOUR-WELL PROBLEM ARISING IN THE CUBIC-TO-TRIGONAL PHASE TRANSFORMATION ANGKANA R ULAND AND THERESA M. SIMON.pdf

共19页,预览4页

还剩页未读, 继续阅读

声明:本站为文档C2C交易模式,即用户上传的文档直接被用户下载,本站只是中间服务平台,本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私,请立即通知玖贝云文库,我们立即给予删除!
分类:图书资源 价格:10玖币 属性:19 页 大小:504.1KB 格式:PDF 时间:2025-05-02

开通VIP享超值会员特权

  • 多端同步记录
  • 高速下载文档
  • 免费文档工具
  • 分享文档赚钱
  • 每日登录抽奖
  • 优质衍生服务
/ 19
客服
关注