Analysis for phase field models of Cahn-Hilliard type [Elektronische Ressource] / von Mathias Wilke

Sujets

• Mathematics
• Physik

Informations

• Publié par: martin-luther-universitat_halle-wittenberg
• Publié le: 01 janvier 2007
• Nombre de lectures: 21
• Langue: English

Extrait

Analysis for Phase Field Models of
Cahn-Hilliard Type
Dissertation
zur Erlangung des akademischen Grades
doctor rerum naturalium (Dr. rer. nat.)
vorgelegt der
Naturwissenschaftlichen Fakult¨at III
der Martin-Luther-Universit¨at Halle-Wittenberg
von
Herrn Dipl.-Math. Mathias Wilke
geb. am: 04. Oktober 1979 in: Merseburg
Gutachter:
1. Prof. Dr. Jan Prus¨ s, Halle (Saale)
2. Prof. Dr. Reinhard Racke, Konstanz
Halle (Saale), 22. November 2007 (Tag der Verteidigung)
urn:nbn:de:gbv:3-000012766
[http://nbn-resolving.de/urn/resolver.pl?urn=nbn%3Ade%3Agbv%3A3-000012766]Contents
Introduction 1
1 Mathematical Preliminaries 8
1.1 Some notation, Function spaces, Laplace- and Fourier transform . . . . . . . . . . 8
∞1.2 Sectorial operators,H -calculus,R-boundedness . . . . . . . . . . . . . . . . . . . 9
1.3 Joint functional calculus, Sums of closed operators . . . . . . . . . . . . . . . . . . 11
1.4 Model problems, Maximal L -regularity . . . . . . . . . . . . . . . . . . . . . . . . 12p
2 Conserved Penrose-Fife Type Models 15
2.1 Derivation of the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 The Linear Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Local Well-Posedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4 Global Well-Posedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5 Asymptotic Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.6 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3 The Non-Isothermal Cahn-Hilliard Equation with Dynamic Boundary Condi-
tions 43
3.1 Derivation of the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.2 The Linear Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.3 Local Well-Posedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.4 Global Well-Posedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.5 Asymptotic Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.6 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4 A Generalized Cahn-Hilliard Equation based on a Microforce Balance 67
4.1 Derivation of the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
n4.2 The Linear Cahn-Hilliard-Gurtin Problem inR . . . . . . . . . . . . . . . . . . . 69
n4.3 The Cahn-Hilliard-Gurtin Problem inR . . . . . . . . . . . . . . . . . . . 76+
4.4 Localization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.5 Local Well-Posedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.6 Global Well-Posedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.7 Asymptotic Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.8 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Bibliography 97Introduction
Since the last century there exists an active interest to model and analyze phase transitions math-
ematically. Phase transitions arise within the most diverse ranges of the daily life. We present
some examples in order to give a first impression.
• The transition between solid, fluid and gaseous phases or in other words vapor-
izing/condensing (fluid ↔ gaseous), melting/freezing (solid ↔ fluid) and sublima-
tion/resublimation (solid↔ gaseous);
• Thetransitionbetweenferromagneticandparamagneticphasesinmagneticmaterialsatthe
Curie-temperature;
• The transition of some metals to superconductors at very low temperatures;
• The Bose-Einstein-condensate; a state in which the matter is cooled down, almost up to the
◦lowest absolute temperature (0 K=−273,15 C).
The well-known classical two-phase Stefan problem, a model problem for the analytic description
of a phase transition, received especially much attention in the past.
i i iκ ∂ θ −d Δθ =0, in Ω (t),i t i
1 1Bθ =b, on ∂Ω (t),
i
θ =0, on Γ(t),
[d∂ θ]=‘V, on Γ(t),ν
i i iθ (0)=θ , in Ω ,0 0
Γ(0)=Γ .0
nHere Ω⊂R is a homogeneous material, consisting of two separated phases. The initial state of
1 2these two phases at t = 0 is given by Ω and Ω , respectively, which are separated by a sharp0 0
interface Γ . It is assumed that Γ does not intersect the boundary of Ω, to avoid so-called contact0 0
angle problems whosemathematicaltreatmentisachallengingtask. WedenotebyΓ(t)theposition
1 2of the moving interface at time t and Ω (t), Ω (t) denote the two phases, separated by Γ(t). κi
andd aretheheatcapacitiesandtheheatconductivitiesofeachphase, respectively. Thequantityi
2 1 1 2[d∂ θ] := d ∂ θ −d ∂ θ represents the jump of the normal derivatives of θ and θ across theν 2 ν 1 ν
interfaceΓ(t)and‘isthelatent heat,whichisneededforthephasetransition. Thenormalvelocity
of Γ(t) is denoted by V and B means Dirichlet or Neumann conditions on the boundary ∂Ω.
TheStefanproblemhasbeenextensivelystudiedbyanumberofauthorsduringthelastdecades
and it is still in the focus of mathematical analysts. In this model one assumes that the interface,
which separates the two phases of the system, is infinitely thin. However, instead of such a sharp
interface one observes smeared interfaces in experiments, which have a thickness of approximately
−8 ˚10 cm = 1A, the atomic radius. So, in the fifties of the last century, mathematicians started to
derive models, which take into account a certain width of the interface between the phases. In
these models one or more extra variables are introduced, to describe the state of the system, the
so-called order parameters. An order parameter is a measure for the degree of order in a system
1Introduction 2
with extremes -1 for total disorder and +1 for complete order. Otherwise the order parameter is
assumed to take values between -1 and +1. Examples for such parameters are the mass density of
the system under consideration (often assumed to be a conserved quantity) or the magnetic flux
in ferromagnetism. But there are quite more possibilities to define order parameters.
Due to the large variety of such models we want to mention here two classical and very famous
ones, namely
• the non-isothermal Cahn-Hilliard equation and,
• the Penrose-Fife model.
In contrast to the Penrose-Fife Model, the Cahn-Hilliard equation is based on the assumption
that the absolute temperature θ of the system is far from zero and has only a small deviation
∗ ∗˜from a fixed value θ . Then one introduces the relative temperature function θ :=θ−θ and the
nonlinearities in the differential operators may be approximated by linear terms, such that the
quasilinear Penrose-Fife Model becomes a semilinear system.
In this thesis we will study the following models for phase transitions.
0 0∂ ψ−Δμ=0, μ=−Δψ+Φ (ψ)−λ (ψ)ϑ,t
(0.1)
∂ (b(ϑ)+λ(ψ))−Δϑ=0,t
and
∂ ψ−div(a∂ ψ)=div(B∇μ)t t
(0.2)0μ−c·∇μ=β∂ ψ−Δψ+Φ (ψ).t
In (0.1) the function ϑ is the reciprocal of the absolute temperature of the system, if one sets
b(s) = −1/s. In this case we obtain the classical conserved Penrose-Fife equations which were
proposed by Penrose & Fife in [32]. Conversely, if we set b(s) = s, the result is the classical
non-isothermal Cahn-Hilliard equation, proposed byCahn & Hilliard in [8]. The second model
(0.2) was proposed by Gurtin [16] in order to model the action of forces that are associated with
microscopic configurations of atoms which are not considered in the derivation of the classical
Cahn-Hilliard equation. In this connection one often speaks of microforces. The equations (0.2)
are a generalization of the classical Cahn-Hilliard equation with constant temperature and they
are known as the Cahn-Hilliard-Gurtin equations.
Letusexplaintheequationsindetails. Thefunctionb(ϑ)isacontributiontotheinternalenergy
e. In fact, it holds thate=b(ϑ)+λ(ψ). It is possible to choose other functions thanb(s)=−1/s
or b(s) = s for b, provided that they satisfy certain assumptions, which are introduced below, in
order to guarantee the mathematical well-posedness of the system. The nonlinearity Φ is the so
called physical potential which characterizes the two different phases of the physical system. A
prominent and often used example is the double-well potential
2 2Φ(s)=Φ (s −1) ,0
with some positive constant Φ > 0. The two distinct minima of Φ correspond to each of the two0
phases. We remark here that there is no maximum principle for (0.1) available, since the equation
for ψ is of fourth order, hence the interval [−1,1] is not an invariant set for (0.1), in general.
Therefore, some authors use logarithmic physical potentials of the form
θ θc 2Φ(s)= ((1+s)log(1+s)+(1−s)log(1−s))− s ,
2 2
to ensure that the order parameter takes values between -1 and +1. For results on problem (0.1)
with logarithmic potentials we refer to Abels & Wilke [1], Bonfoh [5] and the references cited
therein. Next, the functionλ represents the latent heat, which is crucial for appearance of a phase
transition. Two examples are given by
2 2λ(s)=λ (s−s ) and λ(s)=λ (s −s ),0 ∗ 0 ∗Introduction 3
whereλ ,s >0. The chemical potential μ is responsible for the mass transport inside the system0 ∗
and it is g