Abstract
We investigate limit models resulting from a dimensional analysis of quite general heterogeneous catalysis models with fast sorption (i.e. exchange of mass between the bulk phase and the catalytic surface of a reactor) and fast surface chemistry for a prototypical chemical reactor. For the resulting reaction–diffusion systems with linear boundary conditions on the normal mass fluxes, but at the same time nonlinear boundary conditions on the concentrations itself, we provide analytic properties such as local-in-time well-posedness, positivity, a priori bounds and comment on steps towards global existence of strong solutions in the class \(\mathrm {W}^{(1,2)}_p(J \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\), and of classical solutions in the Hölder class \(\mathrm {C}^{(1+\alpha , 2 + 2\alpha )}({\overline{J}} \times {\overline{\Omega }}; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\). Exploiting that the model is based on thermodynamic principles, we further show a priori bounds related to mass conservation and the entropy principle.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In chemical engineering, catalytic processes often play an important, if not predominant role: Certain chemical reactions taking place within a chemical reactor are supposed to be accelerated, whereas other, usually undesirable, side reactions should be suppressed. This aim can be accomplished by adding substances which catalytically act in the fluid mixture (homogeneous catalysis), or e.g. by using suitable coatings for the reactor surface (active surface) which may then act as a catalytic surface to accelerate the desirable reactions on the surface. In many cases, such heterogeneous catalysis mechanisms are actually more efficient by several orders of magnitude than homogeneous catalysts, and, moreover, one may often avoid the need for filtration technology to separate the desired product from the catalyst. Heterogeneous catalysis mechanisms and sorption processes may be modelled starting from a continuum thermodynamic viewpoint by reaction–diffusion systems in the chemical reactor and on the active surface which are coupled via sorption processes, i.e. the exchange of mass between the boundary layer of the bulk phase and the active surface, cf. [20]. In accord with their purpose, catalytic accelerated chemical reactions on the surface are very fast, i.e. both the surface chemistry (at least for the desired reactions) as well as sorption processes take place on very small time scales. Hence, it is natural to consider limit models, for which the surface chemistry and sorption are taken to be infinitely fast, i.e. surface chemistry and sorption processes are modelled as if they would attain an equilibrium configuration instantaneously. Using a dimensionless formulation of such coupled reaction–diffusion–sorption bulk-surface systems, several of such limit models have been proposed in [5], including a general formulation of such a fast sorption and fast surface chemistry model. The mathematical analysis of such systems has been accomplished for the case of a three-component system with chemical reactions of type \(A_1^\Sigma + A_2^\Sigma \rightleftharpoons A_3^\Sigma \) on the surface, neglecting any bulk chemistry (the latter being no severe obstacle, and for the construction of (uniquely determined) strong solutions not a highly relevant assumption). In the present manuscript, the mathematical analysis is continued for limit systems of the same structure, but for general bulk and surface chemistry. In particular, the results on local-in-time existence of strong solutions, positivity, first blow-up criteria as well as a priori estimates for the solutions will be extended to the generic case.
The paper is organised as follows: In Sect. 2 some basic notation is introduced and some preliminary results are recalled. Thereafter, in Sect. 3, the class of heterogeneous catalysis models considered in this manuscript is introduced and the underlying modelling assumptions recalled from the article [5]. Section 4 constitutes the core of this article and is split into subsections on \(\mathrm {L}_p\)-maximal regularity of a linearised version of the fast-sorption-fast-surface-chemistry model, on local-in-time existence of strong \(\mathrm {W}_p^{(1,2)}\)-solutions and classical solutions in the Hölder class \(\mathrm {C}^{(1,2) \cdot (1+\alpha )}\), on an abstract blow-up criterion as well as a priori bounds, e.g. entropy estimates, on the solution. There is a vast amount of literature on reaction–diffusion-systems or general parabolic systems in the bulk phase with homogeneous or inhomogeneous, linear or nonlinear boundary conditions, e.g. [14, 17] for a start, and quite recently thermodynamic principles have become a resourceful driving force for entropy methods, e.g. [11,12,13]. Astonishingly, however, up to now (at least to our knowledge) combined type boundary conditions, i.e. systems where at a fixed boundary point \(\varvec{z} \in \partial \Omega \) Dirichlet type boundary conditions are imposed on some of the variables (or, a linear combination thereof), whereas the remaining variables satisfy Neumann type boundary conditions, have only been considered rarely in the literature, see, for example, [14, Chapter 7, Section 10] and the references given therein. Instead, so far most authors focussed on other types of generalisations: for example, in [1, 16, 21] parabolic systems with nonlinear boundary conditions have been considered, but these were always assumed to be of a common order, cf., for example, the non-tangentiality condition in [16]. In, for example, [6, 10, 19], general parabolic systems or reaction–diffusion-systems with dynamic boundary conditions have been considered, i.e. typically two parabolic systems in the bulk phase and on the surface are coupled. In [9], on the other hand, the authors consider more general structures leading to parabolic systems based on the notion of a Newton polygon.
2 Notation and preliminaries
Throughout, all Banach spaces appearing are Banach spaces over \({{\,\mathrm{{\mathbb {F}}}\,}}\), the field of real numbers \({{\,\mathrm{{\mathbb {R}}}\,}}\) or complex numbers \({\mathbb {C}}\), and \(\left| z \right| \) denotes the modulus of a real or complex number z, \(\mathfrak {R}z\) its real and \(\mathfrak {I} z\) its imaginary part. Real or complex vectors (or, vector fields) \(\varvec{v} = (v_1, \ldots , v_N)^{\mathsf {T}} \in {{\,\mathrm{{\mathbb {F}}}\,}}^N\) will be typically denoted by small, Roman letters in boldface and have Euclidean norm \(\left| \varvec{v} \right| = \sqrt{\sum _{i=1}^N \left| v_i \right| ^2}\), whereas matrices \(\varvec{M} = [m_{ij}]_{i,j} \in {{\,\mathrm{{\mathbb {R}}}\,}}^{n \times m}\) (or \({\mathbb {C}}^{n \times m}\)) most of the time are written in Roman capitals and boldface. The set of natural numbers or integers are denoted by \({{\,\mathrm{{\mathbb {N}}}\,}}= \{1, 2, \ldots , \}\) (or \({{\,\mathrm{{\mathbb {N}}}\,}}_0 = \{0, 1, 2, \ldots \}\)) and \({{\,\mathrm{{\mathbb {Z}}}\,}}= \{\ldots , -1, 0, 1, \ldots \}\), respectively, and vectors of integers \(\varvec{\alpha }= (\alpha _1, \ldots , \alpha _N)^{\mathsf {T}} \in {{\,\mathrm{{\mathbb {Z}}}\,}}^N\) by small Greek letters in boldface, and with length \(\left| \varvec{\alpha } \right| = \sum _{i=1}^N \left| \alpha _i \right| \), but \(\varvec{\nu }= [\nu _{i,j}]_{i,j} \in {{\,\mathrm{{\mathbb {Z}}}\,}}^{N \times M}\) may also denote integer-valued matrices.
\(\Omega \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}^n\) typically denotes an open and nonempty subset of \({{\,\mathrm{{\mathbb {R}}}\,}}^n\), \({\overline{\Omega }}\) its closure, \(\partial \Omega \) its boundary, and \(J \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}\) an interval. Function spaces that are frequently used are \(\mathrm {C}(\Omega )\) and \(\mathrm {C}({\overline{\Omega }})\) (continuous functions on \(\Omega \) and \({\overline{\Omega }}\), resp.), \(\mathrm {C}^k(\Omega )\) and \(\mathrm {C}^k({\overline{\Omega }})\) (\(k \in {{\,\mathrm{{\mathbb {N}}}\,}}_0\) times continuous differentiable functions on \(\Omega \) and \({\overline{\Omega }}\), resp.), \(\mathrm {C}^{k+\gamma }({\overline{\Omega }})\) (\(k \in {{\,\mathrm{{\mathbb {N}}}\,}}_0\) times continuously differentiable functions with bounded and \(\gamma \in (0,1]\) Hölder continuous derivatives of order k), \(\mathrm {L}_p(\Omega )\) (Lebesgue spaces of order \(p \in [1, \infty ]\), where as usual function classes are identified with its representatives), \(\mathrm {W}_p^k(\Omega )\) (Sobolev spaces of differentiability order \(k \in {{\,\mathrm{{\mathbb {N}}}\,}}_0\) and integrability order \(p \in [1, \infty )\)), \(\mathrm {W}_p^s(\Omega )\) (Sobolev–Slobodetskii spaces, \(s \in {{\,\mathrm{{\mathbb {R}}}\,}}_+\), \(p \in [1, \infty ]\)). Similarly, one also has their corresponding boundary (for sufficiently regular boundary), Banach space E-valued and anisotropic versions, e.g. \(\mathrm {L}_p(\Omega ; E)\) (E-valued Lebesgue spaces), \(\mathrm {L}_p(\partial \Omega )\) (Lebesgue spaces w.r.t. surface measure) and
etc.
Remark 2.1
(Sobolev–Slobodetskii spaces and Besov spaces) Recall that for sufficiently regular domains \(\Omega \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}^n\), one has \(\mathrm {B}_{pp}^s(\Omega ) = \mathrm {W}_p^s(\Omega )\) for \(s \in {{\,\mathrm{{\mathbb {R}}}\,}}_+ \setminus {{\,\mathrm{{\mathbb {N}}}\,}}_0\), but \(\mathrm {B}_{pp}^k(\Omega ) \ne \mathrm {W}_p^k(\Omega )\) for \(k \in {{\,\mathrm{{\mathbb {N}}}\,}}\) and \(p \ne 2\).
For the definitions, basic properties and more information on these spaces, the interested reader is referred to the literature, e.g. [2, 3, 15].
3 The model
In this paper, the following, rather general fast sorption and fast surface chemistry limit model will be considered:
where the appearing variables and coefficients have the following physical interpretation and relations with each other. Thermodynamic and geometric variables and vectors:
-
\(\varvec{c} = (c_1, \ldots , c_N)^{\mathsf {T}}: {{\,\mathrm{{\mathbb {R}}}\,}}\times {\overline{\Omega }} \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^N\) denotes the vector field of molar concentrations, i.e. \(c_i(t,\varvec{z}) \in {{\,\mathrm{{\mathbb {R}}}\,}}\) is the molar concentration of the chemical substance \(A_i\) at time \(t \in {{\,\mathrm{{\mathbb {R}}}\,}}\) in position \(\varvec{z} \in {\overline{\Omega }}\), for \(i = 1, \ldots , N\);
-
\(\varvec{J} = [\varvec{j}_1 \cdots \varvec{j}_N]: {{\,\mathrm{{\mathbb {R}}}\,}}\times {\overline{\Omega }} \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^{n \times N}\) for \(\varvec{j}_i: {{\,\mathrm{{\mathbb {R}}}\,}}\times {\overline{\Omega }} \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^n\) the vector field of individual mass fluxes of species \(A_i\), \(i = 1, \ldots , N\);
-
\(\varvec{n}: \partial \Omega \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^n\), the outer normal vector field to \(\Omega \) on \(\partial \Omega \);
-
\(\varvec{r}(\varvec{c}) = \sum _{a=1}^m R_a(\varvec{c}) \varvec{\nu }^a\), the vector field of total molar reaction rates in the bulk phase, modelling chemical reactions given by the formal (reversible) chemical reaction equations
$$\begin{aligned} \sum _{i=1}^N \alpha _i^a A_i \rightleftharpoons \sum _{i=1}^N \beta _i^a A_i, \quad a = 1, \ldots , m, \end{aligned}$$where \(\varvec{\alpha }^a = (\alpha _1^a, \ldots , \alpha _N^a)^{\mathsf {T}}\), \(\varvec{\beta }^a = (\beta _1^a, \ldots , \beta _N^a)^{\mathsf {T}} \in {{\,\mathrm{{\mathbb {N}}}\,}}_0^N\) and the stoichiometric vector of the a-th reaction is given by \(\varvec{\nu }^a = \varvec{\beta }^a - \varvec{\alpha }^a \in {{\,\mathrm{{\mathbb {Z}}}\,}}^N\). Moreover, \(R_a(\varvec{c})\) denotes the molar reaction rate (forward minus backward rate) for the a-th chemical reaction in the bulk phase;
-
\(\varvec{\nu }^{\Sigma ,a} = \varvec{\beta }^{\Sigma ,a} - \varvec{\alpha }^{\Sigma ,a} \in {{\,\mathrm{{\mathbb {Z}}}\,}}^N\), \(a = 1, \ldots , m^\Sigma \), are the stoichiometric vectors of the surface chemical reactions
$$\begin{aligned} \sum _{i=1}^N \alpha _i^{\Sigma ,a} A_i^\Sigma \rightleftharpoons \sum _{i=1}^N \beta _i^{\Sigma ,a} A_i^\Sigma , \quad a = 1, \ldots , m^\Sigma , \end{aligned}$$where \(\varvec{\alpha }^{\Sigma ,a} = (\alpha _1^{\Sigma ,a}, \ldots , \alpha _N^{\Sigma ,a})^{\mathsf {T}}\), \(\varvec{\beta }^{\Sigma ,a} = (\beta _1^{\Sigma ,a}, \ldots , \beta _N^{\Sigma ,a})^{\mathsf {T}} \in {{\,\mathrm{{\mathbb {N}}}\,}}_0^N\) and the adsorbed versions \(A_i^\Sigma \) of species \(A_i\).
Modelling assumptions:
-
The bulk concentrations are assumed to be very small (compared to a characteristic reference concentration \(c_\mathrm {ref}\) of some solute which is not included in the model, \(0 \le \nicefrac {c_i(t,\varvec{z})}{c_\mathrm {ref}} \ll 1\), dilute mixture), and the fluid in the bulk is at rest (the basic problem with vanishing barycentric velocity field which will be considered here) or \(\langle \varvec{v}, \nabla c_i \rangle \), i.e. the inner product between the barycentric velocity field \(\varvec{v}\) and the gradient of individual concentrations, is considered as a perturbation, so that Fickian diffusion,
$$\begin{aligned} \vec {j}_i = - d_i \nabla c_i \quad \text{ with } \text{ some } \text{ diffusion } \text{ coefficients } \, d_i > 0, \, i = 1, \ldots , N \end{aligned}$$is a reasonable (though, not thermodynamically consistent) model for the diffusive fluxes;
-
the chemical potentials \(\mu _i\) in the bulk phase are modelled as those of an ideal mixture with
$$\begin{aligned} \mu _i(\varvec{c}, \vartheta ) = \mu _i^0(\vartheta ) + R \vartheta \ln x_i, \quad i = 1, \ldots , N, \end{aligned}$$where \(\mu _i^0(\vartheta )\) corresponds to some temperature-dependent chemical equilibrium and \(x_i = \frac{c_i}{c}\) is the scalar field of molar fractions, where \(c = \sum _{i=1}^{N+1} c_i\) is the total concentration in the bulk phase, including the concentration of some solvent \(A_{N+1}\). Instead of including the solvent \(A_{N+1}\) in the model, we replace c by some constant approximation \(c_\mathrm {ref}\) to the actual total concentration c, so that we may consider the vector \(\varvec{x} = (x_1, \ldots , x_N)^{\mathsf {T}} = \nicefrac {\varvec{c}}{c_\mathrm {ref}}\) and its dynamics instead of \(\varvec{c}\). Formally assuming \(c_\mathrm {ref} = \mathrm {e}^{R \vartheta }\), we then have \(\mu _i(\varvec{c}, \vartheta ) = \mu _i^0(\vartheta ) + \ln c_i\). Additionally, an isothermal system is assumed; hence, \(\mu _i^0(\vartheta ) = \mu _i^0 \in {{\,\mathrm{{\mathbb {R}}}\,}}\) is simply a constant;
-
the molar reaction rate \(R_a(\varvec{c})\) of the a-th reaction is modelled (consistently with the entropy principle) as \(R_a(\varvec{c}) = k_a^f \varvec{c}^{\varvec{\alpha }^a} - k_a^b \varvec{c}^{\varvec{\beta }^a}\) with \(k_a^f, k_a^b > 0\) satisfying the relation
$$\begin{aligned} \frac{k_a^f}{k_a^b} = \exp (\langle \varvec{\nu }^a, \varvec{\mu }^0 \rangle ), \end{aligned}$$for \(\varvec{\mu }_0 = (\mu _i^0)_i\).
-
Throughout, we assume that all equilibria of the surface chemistry are detailed-balanced, i.e.
$$\begin{aligned} \varvec{\nu }^{\Sigma ,1}, \ldots , \varvec{\nu }^{\Sigma ,m^\Sigma } \quad \text {are linearly independent.} \end{aligned}$$Then, \(\varvec{e}^k \in {{\,\mathrm{{\mathbb {R}}}\,}}^N\), \(k = 1, \ldots , n^\Sigma := N - m^\Sigma \), characterises linearly independent conserved quantities under the surface chemistry, spanning the orthogonal complement of the surface stoichiometric vectors \(\{\varvec{\nu }^{\Sigma ,a}: a = 1, \ldots , m^\Sigma \}\).
Moreover, we use the notation \(\varvec{a}^{\varvec{b}} := (a_1^{b_1}, \ldots , a_m^{b_m})\) for any two vectors \(\varvec{a} \in (0, \infty )^k\), \(\varvec{b} \in {{\,\mathrm{{\mathbb {R}}}\,}}^k\), \(k \in {{\,\mathrm{{\mathbb {N}}}\,}}\).
Under these assumptions, and the additional assumption that the sorption processes and surface chemistry take place very fast, i.e. on much smaller time scales than the bulk diffusion and the bulk chemistry, it has been demonstrated in [5] that (GFLM) is a reasonable limit model for the limiting case of infinitely fast surface chemistry and sorption processes (actually, independent of whether one of these two fast thermodynamic mechanism is even ultra-fast), and the condensed form of the limit model (including initial values) reads
where \(\Sigma = \partial \Omega \) denotes the boundary of \(\Omega \), acting as an active surface, \(\varvec{D} = {\text {diag}}(d_1, \ldots , d_N)\) is the diagonal matrix of (Fickian) diffusion coefficients, and some initial values \(\varvec{c}^0: {\overline{\Omega }} \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^N\) are given.
4 Local-in-time well-posedness for general bulk and surface chemistry
This section is devoted to the local-in-time well-posedness analysis for generic fast sorption and fast surface chemistry limit models of the form (1)–(4).
Under the imposed modelling assumptions, the system (GFLM) under consideration takes the form:
with the equilibrium constant \(\kappa _a = \exp (- \langle \varvec{\nu }^{\Sigma ,a}, \varvec{\mu }^0 \rangle )\). A possible linearised (around some sufficiently smooth reference vector field \(\varvec{c}^*: (0,\infty ) \times {\overline{\Omega }} \rightarrow (0, \infty )\)) version of this nonlinear system is obtained by taking the partial derivatives
for \(c_i|_\Sigma \ne 0\), and reads
or, for short,
where \(\varvec{C}_a^*= (\varvec{c}^*|_\Sigma )^{\varvec{\nu }^{\Sigma ,a}} {\text {diag}}(c_i^*|_\Sigma ^{-1})_{i=1}^N: \, (0,\infty ) \times \Sigma \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^{N \times N}\).
Remark 4.1
Since only concentrations \(c_i, c_i^\Sigma \ge 0\) have physical significance, only linearisations around states for which all components are (uniformly) strictly positive, i.e. only uniformly strict positive initial values, will be feasible by this approach towards linearisation. Regularisation effects for reaction–diffusion systems (cf. the strict parabolic maximum principle), however, suggest that this is no severe restriction as (under slight structural assumptions on the chemical reaction network) for any initial values \(c_i^0 \ge 0\), but not identically zero, the solution immediately becomes strictly positive, cf. the strict maximum principle for reaction–diffusion equations (see, for example, [18]).
The program of the remainder of this section is the following:
-
Show \(\mathrm {L}_p\)-maximal regularity of the linearised problem, provided sufficient regularity of the reference function \(\varvec{c}^*\). This can be done based on abstract theory in a slightly extended version of the results in [7, 8], or with techniques presented in [14, Chapter 7, Section 10], so that mainly the validity of the Lopatinskii–Shapiro condition and regularity properties have to be checked.
-
Use \(\mathrm {L}_p\)-maximal regularity of the linearised problem and the contraction mapping principle to establish local-in-time existence for the fast sorption and fast surface chemistry limit, provided the initial values are regular enough, have uniformly strictly positive entries and satisfy suitable compatibility conditions.
-
Moreover, this procedure will give a “natural” blow-up criterion for global-in time existence, where by “natural” it is meant that this norm appears in the contraction mapping argument for the local-in-time existence.
4.1 \(\mathrm {L}_p\)-maximal regularity for the associated linear problem
To show that the linearised problem (LP) possesses the property of \(\mathrm {L}_p\)-maximal regularity, let us first consider the case of constant coefficients (i.e. a constant reference function \(\varvec{c}^*\in (0, \infty )^N\)) and a flat boundary (i.e. consider the special case of a boundary \(\Sigma = {{\,\mathrm{{\mathbb {R}}}\,}}^{n-1} \times \{0\}\) for the half-space domain \(\Omega = {{\,\mathrm{{\mathbb {R}}}\,}}^{n-1} \times (0, \infty )\)). The corresponding linear initial-boundary value problem to be investigated, on the half-space, then takes the general form of a parabolic reaction–diffusion system with boundary conditions of inhomogeneous type. For technical reasons (unboundedness of the domain \(\Omega \)), we include a damping constant \(\mu \ge 0\). The system then reads as
Here, we write \(\varvec{z} = (\varvec{z}', z_n) \in {{\,\mathrm{{\mathbb {R}}}\,}}^{n-1} \times {{\,\mathrm{{\mathbb {R}}}\,}}_+\) for the spatial variables, and the right-hand sides—as the analysis of the linearised problem will reveal—have to satisfy the following regularity conditions:
-
(1)
\(\varvec{f} \in \mathrm {L}_p(J \times \Omega ;{{\,\mathrm{{\mathbb {R}}}\,}}^N)\);
-
(2)
\(\varvec{g} \in \mathrm {W}_p^{(1,2) \cdot (\nicefrac {1}{2}-\nicefrac {1}{2p})} (J \times \Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^{n^\Sigma })\);
-
(3)
\(\varvec{h} \in \mathrm {W}_p^{(1,2) \cdot (1 - \nicefrac {1}{2p})}(J \times \Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^{m^\Sigma })\);
-
(4)
\(\varvec{v}^0 \in \mathrm {W}_p^{2-2/p}(\Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\).
The corresponding maximal regularity result on the half-space reads as follows:
Proposition 4.2
(\(\mathrm {L}_p\)-maximal regularity on the half-space for constant coefficients). Assume that \(\varvec{c}^*\in (0, \infty )^N\) is a constant and let \(p \in (1, \infty )\). Then, there is \(\mu _0 \ge 0\) such that for all \(\mu \ge \mu _0\), the half-space problem (6) admits a unique solution \(\varvec{v} \in \mathrm {W}_p^{(1,2)}(J \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) if and only if \(\varvec{f} \in \mathrm {L}_p(J \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\), \(\varvec{g} \in \mathrm {W}_p^{(1,2) \cdot (\nicefrac {1}{2}-\nicefrac {1}{2p})}(J \times \Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^{n^\Sigma })\), \(\varvec{h} \in \mathrm {W}_p^{(1,2) \times (1 - \nicefrac {1}{2p})}(J \times \Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^{m^\Sigma })\) and \(\varvec{v}^0 \in \mathrm {W}_p^{2-2/p}(\Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\), and, moreover, the following compatibility conditions are satisfied, if the respective time traces exist:
In this case, there is \(C = C(p, \mu ) > 0\), independent of the boundary and initial values, such that
Proof
Starting with the system of PDEs (6), taking the partial Laplace–Fourier transform \({\mathcal {F}}\) for \((\lambda , \varvec{\xi }') \in \overline{{\mathbb {C}}_0^+} \times {{\,\mathrm{{\mathbb {R}}}\,}}^{n-1}\), and setting \(y = z_n\), leads (formally) to the following parameter-dependent initial value problems
where \({\hat{v}}_i = {\mathcal {F}} v_i\), \({\hat{f}}_i = {\mathcal {F}} f_i\) etc. For fixed \((\lambda , \varvec{\xi }) \in \overline{{\mathbb {C}}_0^+} \times {{\,\mathrm{{\mathbb {R}}}\,}}^{n-1} \setminus \{(0,\varvec{0})\}\), the general solution to the ODE in the first line of this system is
As we look for a solution in the class \(v_i = {\mathcal {F}}^{-1}({\hat{v}}_i) \in \mathrm {L}_p((0,T); \mathrm {L}_p({{\,\mathrm{{\mathbb {R}}}\,}}^{n-1} \times (0,\infty )))\), one necessarily needs to have \(\lim _{y \rightarrow \infty } {\hat{v}}_i(\lambda , \varvec{\xi }', y) = 0\) and hence, (provided the square root \(z^{\nicefrac {1}{2}}\) is chosen such that \(\mathfrak {R}z^{\nicefrac {1}{2}} > 0\) for all \(z \in {\mathbb {C}}_0^+\)) \({\hat{v}}_{i,+}(\lambda , \varvec{\xi }') = 0\) for a.e. \((\lambda , \varvec{\xi }') \in \overline{{\mathbb {C}}_0^+} \times {{\,\mathrm{{\mathbb {R}}}\,}}^{n-1}\), and in that case
To match the solution with the boundary conditions at \(\Sigma = {{\,\mathrm{{\mathbb {R}}}\,}}^{n-1} \times \{0\}\), we thus need to solve the linear system
where \(\varvec{R} = {\text {diag}}\left( (\tfrac{\lambda + \mu }{d_i} + \left| \varvec{\xi } \right| ^2)^{\nicefrac {1}{2}} \right) _{i=1}^N\). This system is uniquely solvable for all \((\lambda , \varvec{\xi }') \in \overline{{\mathbb {C}}_0^+} \times {{\,\mathrm{{\mathbb {R}}}\,}}^{n-1}\) if and only if the matrix
All matrices \(\varvec{D}\), \(\varvec{R}\) and \(\varvec{C}_a^*\) are diagonal and the matrices
coincide with the \(N \times N\) diagonal matrix \(\varvec{{\tilde{C}}} = {\text {diag}}(c_i^*|_\Sigma )_{i=1, \ldots , N}^{-1}\) up to a nonzero factor \((\varvec{c}^*|_\Sigma )^{\varvec{\nu }^{\Sigma ,a}}\). The matrix \(\varvec{M}\) is invertible if and only if
where \(\tilde{\varvec{D}} = \varvec{D}^{-1} \varvec{R}^{-*} \tilde{\varvec{C}}\) is an \(N \times N\) diagonal matrix with entries \([\tilde{\varvec{D}}]_{ii} \in {\mathbb {C}}_0^+\) on the diagonal. Due to Lemma 4.3, this is the case. Using the inverse partial-Fourier–Laplace transformation \({\mathcal {F}}^{-1}\), system (6) for \(\varvec{f} = \varvec{0}\) thus has a unique solution which is given by
By Duhamel’s formula, the general solution of (6) is then given by
where \(*\) denotes convolution with respect to the variable \(y > 0\), and \(\varvec{v}\) lies in \(\mathrm {W}^{(1,2)}_p({{\,\mathrm{{\mathbb {R}}}\,}}_+ \times {{\,\mathrm{{\mathbb {R}}}\,}}^n)\), provided \(\mu \ge \mu _0\) and \(\mu _0 \ge 0\) is chosen sufficiently large. This solves the problem of \(\mathrm {L}_p\)-maximal regularity for the constant coefficient case on the half-space. \(\square \)
For validation of the Lopatinskii–Shapiro condition, we employed the following:
Lemma 4.3
Let \(s,m \in {{\,\mathrm{{\mathbb {N}}}\,}}\) and \(N = s + m\). Let \(\varvec{v}^1, \ldots , \varvec{v}^m \in {{\,\mathrm{{\mathbb {R}}}\,}}^N\) and \(\varvec{w}^1, \ldots , \varvec{w}^s \in {{\,\mathrm{{\mathbb {R}}}\,}}^N\) be linearly independent, real vectors such that
Let \(\delta _j \in {\mathbb {C}}\), \(j = 1, \ldots , N\), be such that \(0 \not \in {\text {conv}} \{\delta _j: j = 1, \ldots , N\}\) and the matrix \(\varvec{M} \in {\mathbb {C}}^{N \times N}\) be defined as
Then, \(\varvec{M}\) is invertible, i.e. \(\varvec{D} \varvec{v}^1, \ldots , \varvec{D} \varvec{v}^m\), \(\varvec{w}^1, \ldots , \varvec{w}^s\) form a basis of \({\mathbb {C}}^N\).
Proof
As \(\varvec{M} \in {\mathbb {C}}^{N \times N}\) is a square matrix, it suffices to demonstrate injectivity of \(\varvec{M}\). Let \(\varvec{u} \in \mathrm {N}(\varvec{M})\). Then, in particular,
Therefore, there are \(\varvec{\gamma }_i \in {\mathbb {C}}\), \(i = 1, \ldots , m\), such that
Writing \(\varvec{V} = \left[ \begin{array}{ccc} \varvec{v}^1&\cdots&\varvec{v}^m \end{array} \right] \in \varvec{{{\,\mathrm{{\mathbb {R}}}\,}}}^{N \times m}\), from \(\varvec{M} \varvec{u} = \varvec{0}\) it follows \(\varvec{V}^{\mathsf {T}} \varvec{D} \varvec{u} = \varvec{0}\), thus
In particular, since \(\varvec{V}^{\mathsf {T}} = \varvec{V}^*\) (as \(\varvec{V}\) has real entries), for the inner product on \({\mathbb {C}}^m\) one finds
As \(\left| (\varvec{V} \varvec{\gamma })_i \right| ^2 \ge 0\) and \(0 \not \in {\text {conv}} \{\delta _i: i = 1, \ldots , m\}\), this can only hold true if \(\varvec{u} = \varvec{V} \varvec{\gamma }= \varvec{0}\), hence \(\varvec{M}\) is injective. \(\square \)
The general \(\mathrm {L}_p\)-maximal regularity theorem (for bounded \(\mathrm {C}^2\)-domains \(\Omega \)) can then be derived via the standard technique, i.e. first a generalisation to the bend-space problem and, thereafter, a localisation procedure. For these techniques to work properly, one needs additional conditions on the (then non-constant) reference function \(\varvec{c}^*: {\overline{\Omega }} \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^N\). In our case, the bulk diffusion operator \(- \varvec{D} \Delta \) does not depend on the spatial position \(\varvec{z} \in \Omega \). Therefore, there is no need to consider perturbations of it, i.e. \({\mathcal {A}}^{sm} = 0\) in the language of [7, 8]. Neither do the conserved quantities \(\varvec{e}^k\), \(k = 1, \ldots , n^\Sigma \), but only the matrix \(\varvec{{\tilde{C}}}(\varvec{z}) = {\text {diag}}(\varvec{c}^*|_\Sigma )^{-1}(\varvec{z})\) (which is \(\varvec{C}_a^*\) up to an a-dependent factor \((\varvec{c}^*|_\Sigma )^{\varvec{\nu }^{\Sigma ,a}}\)) depends on the spatial position \(\varvec{z} \in {\overline{\Omega }}\). Using the same strategy as in [7], one may write
where \(\varvec{C}^\mathrm {small}_a(\varvec{z})\) corresponds to a small perturbation of \(\varvec{C}_a^*(\varvec{z}_0)\). As in [7], one may then consider the problem
Letting \(A_0 = - \varvec{D} \Delta \) on the domain
one then derives a fixed point equation of the form:
Proceeding as in [8], cf. the upcoming paper [4], one finds that one should demand the following regularity of \(\varvec{{\tilde{C}}}\), hence of \(\varvec{c}^*\):
There are \(s,r \ge p\) with \(\tfrac{1}{s} + \tfrac{n-1}{2 r} < 1 - \tfrac{1}{2p}\) such that
$$\begin{aligned}&\vec {{\tilde{C}}}\in {} \mathrm {W}_{s,r}^{(1,2) \cdot (1-\nicefrac {1}{2p})}(J \times \Sigma ; {{\,\mathrm {{\mathbb {R}}}\,}}^{N \times N}) := \mathrm {W}_s^{1-\nicefrac {1}{2p}}\\&\quad \qquad (J;\mathrm {L}_r(\Sigma ;{{\,\mathrm {{\mathbb {R}}}\,}}^{N \times N})) \cap \mathrm {L}_s(J; \mathrm {W}_r^{2-\nicefrac {1}{p}}(\Sigma ; {{\,\mathrm {{\mathbb {R}}}\,}}^{N \times N})), \end{aligned}$$hence, the reference function \(\varvec{c}^*\) should be uniformly positive and
$$\begin{aligned}&\varvec{c}^*\in \mathrm {W}_{s,r}^{(1,2) \cdot (1-\nicefrac {1}{2p})}(J \times \Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^N) \\&\quad := \mathrm {W}_s^{1-\nicefrac {1}{2p}}(J;\mathrm {L}_r(\Sigma ;{{\,\mathrm{{\mathbb {R}}}\,}}^N)) \cap \mathrm {L}_s(J; \mathrm {W}_r^{2-\nicefrac {1}{p}}(\Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)). \end{aligned}$$
Details will be provided in [4].
Remark 4.4
Note that for regularity of \(\varvec{{\tilde{C}}}\), one has to consider the regularity of the functions \((c_k^{*})^{-1}\), hence of
Provided \(c_k^*\ge \varepsilon > 0\) is (a.e.) uniformly positive, this implies that
where, provided \(\mathrm {W}_p^2(\Omega ) \hookrightarrow \mathrm {L}_{2p}(\Omega )\) (which is true for \(p \ge \frac{n}{4}\)), \(\big \Vert \left| \nabla c_k^* \right| ^2 \big \Vert _{\mathrm {L}_p(J \times \Omega )} \lesssim \big \Vert c_k^* \big \Vert _{\mathrm {L}_p(J;\mathrm {W}_p^2(\Omega ))}^2\). Hence, for \(p \ge n\), it follows from \(c_k^*\in \mathrm {W}_p^{(1,2)}(J \times \Omega )\) and \(c_k^*\ge \varepsilon > 0\) that \(\nicefrac {1}{c_k^*} \in \mathrm {W}_p^{(1,2)}(J \times \Omega )\) as well. Since the reference function should lie in the function class \(\varvec{c}^*\in \mathrm {W}_p^{(1,2) \cdot (1-\nicefrac {1}{2}p)}(J \times \Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) and be uniformly positive, one naturally may take \(s = r = p\) (other choices are possible as long as the parabolic Sobolev index stays the same, namely \(2 - \frac{n+2}{p}\)), and then the condition on the values of s and r (here, both equal p) reads
Note that in this case the embeddings \(\mathrm {W}_p^{(1,2)}(J \times \Omega ) \hookrightarrow \mathrm {C}(\overline{J} \times {\overline{\Omega }})\) and \(\mathrm {W}_p^{2-2/p}(\Omega ) \hookrightarrow \mathrm {C}({\overline{\Omega }})\) are continuous.
Theorem 4.5
Let \(p > \frac{n+2}{2}\) and \(\varvec{c}^*\in \mathrm {W}_p^{(1,2)}(J \times \Omega ; (0,\infty )^N)\), where \(J = (0,T)\) for some \(T > 0\) and \(\Omega \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}^n\) is a bounded \(\mathrm {C}^2\)-domain. Then, the linearised problem
has a unique solution in the function class \(\varvec{v} \in \mathrm {W}_p^{(1,2)}(J \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) if the data are subject to the following regularity and compatibility conditions:
-
(1)
\(\varvec{f} \in \mathrm {L}_p(J \times \Omega ;{{\,\mathrm{{\mathbb {R}}}\,}}^N) = \mathrm {L}_p(J; \mathrm {L}_p(\Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N))\),
-
(2)
\(\varvec{g} \in \mathrm {W}_p^{(1,2) \cdot (\nicefrac {1}{2}-\nicefrac {1}{2p})} (J \times \Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^{n^\Sigma }) = \mathrm {W}_p^{\nicefrac {1}{2} - \nicefrac {1}{2p}}(J;\mathrm {L}_p(\Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^{n^\Sigma })) \cap \mathrm {L}_p(J; \mathrm {W}_p^{1-\nicefrac {1}{p}}(\Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^{n^\Sigma })\),
-
(3)
\(\varvec{h} \in \mathrm {W}_p^{(1,2) \cdot (1 - \nicefrac {1}{2p})}(J \times \Sigma ) = \mathrm {W}_p^{1-\nicefrac {1}{2p}}(J;\mathrm {L}_p(\Sigma ; {{\,\mathrm{{\mathbb {R}}}\,}}^{m^\Sigma })) \cap \mathrm {L}_p(J; \mathrm {W}_p^{2-\nicefrac {1}{p}}(\Sigma ;{{\,\mathrm{{\mathbb {R}}}\,}}^{m^\Sigma }))\),
-
(4)
\(\varvec{v}^0 \in \mathrm {W}_p^{2-2/p}(\Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\),
-
(5)
\(\langle \varvec{C}_a^*\varvec{\nu }^{\Sigma ,a}, \varvec{v}^0|_\Sigma \rangle = h_a|_{t=0}\),
-
(6)
\(\langle \varvec{e}^k, \varvec{D} \frac{\partial }{\partial \varvec{n}} \varvec{v}^0|_\Sigma \rangle = g_k|_{t=0}\) (if \(p > 3\)).
In this case, the solution depends continuously on the data, i.e. for some \(C = C(n,\Omega ) > 0\) independent of the data it holds that
A proof of this result will follow from a subsequent paper on the extension of the abstract results in [7, 8] to the case of combined type boundary conditions, see [4].
More precisely, the following result will be demonstrated in [4].
Theorem 4.6
(\(\mathrm {L}_p\)-maximal regularity) Let E be a Banach space of class \({{\mathcal {H}}}{{\mathcal {T}}}\), \(m \in {{\,\mathrm{{\mathbb {N}}}\,}}\) and \(\Omega \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}^n\) be a domain with compact boundary of class \(\partial \Omega \in \mathrm {C}^{2m}\). For \(j = 1, \ldots , m\) let \(m_j \in {{\,\mathrm{{\mathbb {N}}}\,}}_0\) and linear, continuous projections \({\mathcal {P}}_{j,k} \in {{\,\mathrm{{\mathcal {B}}}\,}}(E)\) such that \({\mathcal {P}}_{j,k} {\mathcal {P}}_{j,k'} = 0\) for \(k \ne k'\) and \(E = \oplus _{k=0}^{m_j} {{\,\mathrm{\mathrm {R}}\,}}({\mathcal {P}}_{j,k})\) for every \(j = 1, \ldots , m_j\), be given. Let \(p \in (1, \infty )\) and suppose that assumptions (E), (LS), (SD), (SB) and (RB) hold true:
Let linear differential operators \({\mathcal {A}}(t,x,D)\) and \({\mathcal {B}}_j(t,x,D)\) and their principle parts be defined via their symbols
-
(E)
Ellipticity of the interior symbol: For all \(t \in J\), \(\varvec{z} \in {\overline{\Omega }}\) and \(\vec {\xi }\in \S ^{n-1}\) it holds that
$$\begin{aligned} \sigma ({\mathcal {A}}(t,\varvec{z}, \varvec{\xi })) \subseteq {\mathbb {C}}_0^+, \end{aligned}$$i.e. \({\mathcal {A}}(t,\varvec{z},D)\) is normally elliptic. If \(\Omega \) is unbounded, the same condition is imposed at \(\varvec{z} = \infty \).
-
(LS)
Lopatinskii–Shapiro condition: For all \(t \in J\), \(\varvec{z} \in \partial \Omega \) and all \(\varvec{\xi }\in {{\,\mathrm{{\mathbb {R}}}\,}}^n\) with \(\varvec{\xi }\cdot \varvec{n}(\varvec{z}) = 0\), and all \(\lambda \in \overline{{\mathbb {C}}_0^+}\) such that \((\lambda , \varvec{\xi }) \ne (0, \varvec{0})\), the initial value problem
$$\begin{aligned} \lambda v(y) + {\mathcal {A}}_\#(t,\varvec{z},\varvec{\xi }+ \mathrm {i} \,\varvec{n}(\varvec{z}) \tfrac{\partial }{\partial y})v(y)&= 0,&y > 0,\\ {\mathcal {B}}_{j,\#}(t,\varvec{z}, \varvec{\xi }+ \mathrm {i} \,\varvec{n}(\varvec{z}) \frac{\partial }{\partial y})v(0)&= h_j,&j = 1, \ldots , m, \end{aligned}$$has a unique solution in the class \(v \in \mathrm {C}_0({{\,\mathrm{{\mathbb {R}}}\,}}_+;E)\).
-
(SD)
There are \(r_l, s_l \ge p\) with \(\frac{1}{s_l} + \frac{n}{2m r_l} < 1 - \frac{l}{2m}\) such that
$$\begin{aligned} a_{\varvec{\alpha }}&\in \mathrm {L}_{s_l}(J; [\mathrm {L}_{r_l} + \mathrm {L}_\infty ](\Omega ;{{\,\mathrm{{\mathcal {B}}}\,}}(E))),&\left| \varvec{\alpha } \right| = l < 2m, \\ a_{\varvec{\alpha }}&\in \mathrm {C}_l(J \times {\overline{\Omega }}; {{\,\mathrm{{\mathcal {B}}}\,}}(E)),&\left| \varvec{\alpha } \right| = 2m. \end{aligned}$$ -
(SB)
There are \(s_{j,k,l}, r_{j,k,l} \ge p\) with \(\frac{1}{s_{j,k,l}} + \frac{n-1}{2m r_{j,k,l}} < \kappa _{j,k} + \frac{l-k}{2m}\) such that
$$\begin{aligned} b_{j,k,\varvec{\beta }} \in \mathrm {W}_{s_{j,k,l},r_{j,k,l}}^{(1,2m) \cdot \kappa _{j,k}}(J \times \partial \Omega ; {{\,\mathrm{{\mathcal {B}}}\,}}(E)), \quad \left| \varvec{\beta } \right| = l \le k \le m_j. \end{aligned}$$ -
(RB)
For every \(j = 1, \ldots , m\) and \(k = 0, 1, \ldots , m_j\) it holds that
$$\begin{aligned} b_{j,k,\varvec{\beta }}({{\,\mathrm{\mathrm {R}}\,}}({\mathcal {P}}_{jk})) \subseteq {{\,\mathrm{\mathrm {R}}\,}}({\mathcal {P}}_{jk}), \quad \left| \varvec{\beta } \right| = k. \end{aligned}$$
Then, the problem
has a unique solution in the class
if and only if the data f, \(\varvec{g}\) and \(u_0\) are subject to conditions (D) as follows:
-
(D)
Assumptions on the data:
-
(i)
\(f \in \mathrm {L}_p(J \times \Omega ; E)\),
-
(ii)
\(g_{j,k} \in \mathrm {W}_p^{(1,2m) \cdot \kappa _{j,k}} (J \times \partial \Omega ;E)\), where \(\kappa _{j,k} = \frac{2m - k - \nicefrac {1}{p}}{2m}\), and we then set \(g_j = \sum _{k=0}^{m_j} g_{j,k}\).
-
(iii)
\(u_0 \in \mathrm {W}_p^{2m(1-\nicefrac {1}{p})}(\Omega ;E)\),
-
(iv)
if \(\kappa _{j,k} > \nicefrac {1}{p}\), then \({\mathcal {B}}_j(0,\varvec{z}) {\mathcal {P}}_{j,k} u_0(\varvec{z}) = g_{j,k}(0,\varvec{z})\) for \(\varvec{z} \in \partial \Omega \).
-
(i)
In the concrete situation of this manuscript, we may choose \(m = 1\) and \(E = {{\,\mathrm{{\mathbb {R}}}\,}}^N\), \(a_{\varvec{\alpha }}(t,z) = \varvec{D}\) for \(\alpha = 2 e_i\), \(i = 1, \ldots , N\), and \(a_{\varvec{\alpha }}(t,z) = 0\) otherwise. Moreover \(b_{j,1,\varvec{e}_k}(t,z) \varvec{v} = \langle \varvec{e}^j, \varvec{\nu }\cdot \varvec{D} \varvec{v} \rangle \varvec{e}_k\) for \(k = 1, \ldots , n^\Sigma \), and \(b_{1,0,\varvec{0}}(t,z) \varvec{v} = b_{1,1,\varvec{0}} \varvec{v} = \sum _a \langle \varvec{C}_a^*(t,z) \varvec{\nu }^{\Sigma ,a}, \varvec{v} \rangle \varvec{e}_{n^\Sigma + a}\), \(a = 1, \ldots , m^\Sigma \). Furthermore, \({\mathcal {P}}_{1,1}\) is the (orthogonal) projection in \({{\,\mathrm{{\mathbb {R}}}\,}}^N\) onto \({{\,\mathrm{\mathrm {span}}\,}}\{ \varvec{D}^{-1} \varvec{e}^k: k = 1, \ldots , n^\Sigma \}\) and \({\mathcal {P}}_{1,0}\) is the (orthogonal) projection in \({{\,\mathrm{{\mathbb {R}}}\,}}^N\) onto \(\{\varvec{D}^{-1} \varvec{e}^k: \, k = 1, \ldots , n^\Sigma \}^\bot \). Thus,
As typical for many semi-linear systems, \(\mathrm {L}_p\)-maximal regularity, and (in this case) \(\mathrm {L}_p\)-\(\mathrm {L}_q\)-estimates can be employed to find a (unique) strong solution of the quasi-linear problem. This will be the aim of the next subsection.
4.2 Local-in-time existence of strong solutions, blow-up criteria and a priori bounds
Having maximal regularity for the linearised problem at hand, we may now come back to the semilinear problem. \(\mathrm {L}_p\)-maximal regularity will play the typically crucial role in the construction of strong solutions via the contraction mapping principle. Thereby, rather as a by-product, a condition for continuation of the solution to a maximal solution will be established, where in general the solution may be global in time (which might be expected for a large subclass of the systems considered here), or one may observe a blow-up in finite time. Whereas the boundedness of the \(\mathrm {W}_p^{2-2/p}\)-norm cannot be guaranteed in general, or more precisely, it is unclear whether boundedness holds true without any restriction on the bulk and surface chemistry, for slightly weaker norms a priori bounds are possible, indeed. The latter will be considered in the second part of this subsection.
4.2.1 Local-in-time existence and maximal continuation of solutions
Recall the form of the fast sorption and fast surface chemistry limit (5), and additionally consider given initial values \(\varvec{c}^0: {\overline{\Omega }} \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^N\) which should be regular enough (in a sense to be made precise later on):
Moreover, let \(T_0 > 0\) be any fixed, finite time horizon and \(\varvec{c}^*: (0,T_0) \times \Omega \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^N\) be a (sufficiently regular) auxiliary function which admits a time trace and \(\varvec{c}^*|_{t=0} = \varvec{c}^0\). Introducing \(\varvec{v}(t,\varvec{z}) := \varvec{c}(t,\varvec{z}) - \varvec{c}^*(t,\varvec{z})\) leads to the reaction–diffusion–sorption system for \(\varvec{v}\) as follows:
where \(\varvec{C}_a^*= (\varvec{c}^*)^{- \varvec{\nu }^{\Sigma ,a}} {\text {diag}}(\varvec{c}^*)^{-1}: (0,T_0) \times \Sigma \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}^{N \times N}\). Next, assume that \(p > \tfrac{n + 2}{2}\), and for \(\tau > 0\) consider the nonlinear solution operator
given by \(\varvec{v} \mapsto \varvec{w}\), where
and \(\varvec{w}\) is the unique strong solution to the linear problem
This problem can now be handled in the way typical for semi-linear parabolic problems, employing the maximal regularity of the linearised problem and the regularity of the nonlinear maps on the right-hand side, which allows for a fixed point argument via the contraction mapping principle. To establish the regularity properties which are needed, one first needs the following auxiliary result on embedding properties of \(\mathrm {W}_p^{(1,2)}(J \times \Omega )\) for bounded intervals J and bounded \(\mathrm {C}^2\)-domains \(\Omega \).
Lemma 4.7
Let \(p \in (\frac{n+2}{2}, \infty )\) and \(\Omega \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}^n\) be a bounded \(\mathrm {C}^2\)-domain. Fix \(T_0 > 0\). For \(T \in (0,T_0]\) and
the embedding constants for the continuous embeddings
can be chosen independently of \(T \in (0, T_0)\), e.g. \(C_p = 2^{\nicefrac {1}{p}} C_p(T_0)\) where \(C_p(T_0)\) is an embedding constant for \(T = T_0\).
Proof
Since for \(u \in \mathrm {D}_0(T)\) one has \(u(0) = 0\), it follows that
defines a function \({\tilde{u}} \in \mathrm {W}_p^{(1,2)}({{\,\mathrm{{\mathbb {R}}}\,}}_+ \times \Omega )\) and for its restriction to \([0,T_0] \times \Omega \) it holds that
From here the assertion follows easily. \(\square \)
Theorem 4.8
(Local-in-time existence of strong solutions). Let \(p > \frac{n + 2}{2}\) and assume that \(\Omega \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}^n\) is a bounded domain of class \(\partial \Omega \in \mathrm {C}^2\). Then, the fast sorption and fast surface chemistry limit problem (5) admits a unique local-in-time strong solution, if
More precisely, for every reference initial values \(\varvec{c}^\mathrm {ref} \in \mathrm {I}_p^+(\Omega )\), there are \(T > 0\), \(\varepsilon > 0\) and \(C > 0\) such that the following statements hold true:
-
(1)
For all \(\varvec{c}^0 \in \mathrm {I}_p^+(\Omega )\) with \(\big \Vert \varvec{c}^0 - \varvec{c}^\mathrm {ref} \big \Vert _{\mathrm {I}_p(\Omega )} < \varepsilon \), there is a unique strong solution \(\varvec{c} \in \mathrm {W}_p^{(1,2)}(J \times \Omega ; (0, \infty )^N)\) of (5’) for \(J = [0,T]\).
-
(2)
For any two initial values \(\varvec{c}^0, \tilde{\varvec{c}}^0 \in \mathrm {I}_p^+(\Omega )\) with \(\big \Vert \varvec{c}^0 - \varvec{c}^\mathrm {ref} \big \Vert _{\mathrm {W}_p^{2-2/p}}, \big \Vert \tilde{\varvec{c}}^0 - \varvec{c}^\mathrm {ref} \big \Vert _{\mathrm {W}_p^{2-2/p}} < \varepsilon \) and corresponding strong solutions \(\varvec{c}, \tilde{\varvec{c}} \in \mathrm {W}_p^{(1,2)}(J \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) one has
$$\begin{aligned} \big \Vert \varvec{c} - \tilde{\varvec{c}} \big \Vert _{\mathrm {W}_p^{(1,2)}(J \times \Omega )} \le C \big \Vert \varvec{c}^0 - \tilde{\varvec{c}}^0 \big \Vert _{\mathrm {W}_p^{2-2/p}(\Omega )}. \end{aligned}$$ -
(3)
Any strong solution \(\varvec{c} \in \mathrm {W}_p^{(1,2)}(J \times \Omega )\) can be extended in a unique way to a maximal (Hölder) strong solution \(\varvec{c}: [0, T_\mathrm {max}) \times \Omega \rightarrow (0,\infty )^N\) (with \(T_\mathrm {max} \in (0, \infty ]\)), where \(\varvec{c} \in \mathrm {W}_p^{(1,2)}((0,T) \times \Omega ; (0,\infty )^N)\) for every \(T \in (0, T_\mathrm {max})\).
Proof
Let \(\eta > 0\) and initial values
be given. Let \(\rho _0, T_0 > 0\) be such that
where \(\rho \in (0, \rho _0], \, T \in (0, T_0]\). (Such a choice is possible, e.g. by Lemma 4.7.) Moreover, let \({\mathcal {E}}\) be a linear, continuous extension operator from \(\mathrm {W}^{2-2/p}_p(\Omega )\) to \(\mathrm {W}^{(1,2)}_p((0,T_0) \times \Omega )\) and from \(\mathrm {C}({\overline{\Omega }})\) to \(\mathrm {C}([0,T_0] \times {\overline{\Omega }})\). Moreover, w.l.o.g. assume that \({\mathcal {E}} \varvec{v} \ge \frac{\eta }{2}\) on \((0,T_0) \times \Omega \) whenever \(\varvec{v} \ge \eta \) on \(\Omega \), and let \(\rho _0 \le \frac{\eta }{3C(T_0)}\) where \(C(T_0) > 0\) is a common embedding constant for the continuous embeddings \(\mathring{\mathrm {W}}_p^{(1,2)}((0,T] \times {\overline{\Omega }}) \hookrightarrow \mathrm {C}([0,T] \times {\overline{\Omega }})\) from Lemma 4.7. Then, \({\mathcal {E}} \varvec{c}^0 + \varvec{v} \ge \frac{\eta }{6}\) for all \(\rho \in (0, \rho _0]\), \(T \in (0, T_0]\) and \(\varvec{v} \in \mathrm {D}_{\rho ,T}\) on \((0,T) \times \Omega \).
We are looking for a solution \(\varvec{c} \in \mathrm {W}^{(1,2)}_p((0,T) \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) of the reaction–diffusion system with linear, combined type boundary conditions
We set \(\varvec{c}^*:= {\mathcal {E}} \varvec{c}^0 \in \mathrm {W}^{(1,2)}_p((0,T_0) \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) and \(\varvec{v} := \varvec{c} - \varvec{c}^*\). Then, in order for \(\varvec{c}\) to be the solution to the reaction–diffusion–sorption system considered here, \(\varvec{v}\) has to be the solution to the following initial-boundary value problem:
To derive a semilinear formulation of this problem, we rewrite the nonlinear boundary conditions on \(\varvec{v}\) as follows. By Taylor’s series, we may write
for a function \({\mathcal {Q}}_a\) which is continuously differentiable of arbitrary order, at least on
Hence, we may express the nonlinear boundary condition \((\varvec{c}^*+ \varvec{v})^{\varvec{\nu }^{\Sigma ,a}} = \kappa _a\) in the equivalent semilinear form
Then, \(\varvec{v} \in \mathrm {D}_{\rho ,T}\) is a solution \(\varvec{v} \in \mathrm {W}^{(1,2)}_p((0,T) \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) of (*) if and only if
Therefore, \(\varvec{v} \in \mathrm {D}_{\rho ,T}\) is a solution to (*), if and only if \(\varvec{v} \in \mathrm {D}_{\rho ,T}\) is a fixed point of the (well defined, due to the choice of \(\rho _0\), \(T_0\)) map \(\Phi : \mathrm {D}_{\rho ,T} \rightarrow \mathrm {W}^{(1,2)}_p((0,T) \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) defined as follows: For \(\varvec{v} \in \mathrm {D}_{\rho ,T}\) let \(\Phi (\varvec{v}) := \varvec{w}\) be the unique solution to the inhomogeneous, parabolic initial-boundary value-problem
By \(\mathrm {L}_p\)-maximal regularity of the linearised problem, the solution \(\varvec{w} \in \mathrm {W}^{(1,2)}_p((0,T) \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) exists and is uniquely determined, for every \(\varvec{v} \in \mathrm {D}_{\rho ,T}\). That is, \(\Phi \) is well defined. Since the initial values \(\varvec{w}(0,\cdot ) = 0\) are zero for all the functions constructed in this way, the constant \(C_T = C_{T_0} > 0\) in the maximal regularity estimate
for \(\varvec{v} \in \mathrm {D}_{\rho ,T}\) can be chosen independently of \(T \in (0, T_0]\). This can be seen, for example, by using the following mirroring type argument: Set
and consider the problem with time horizon \(T_0\):
Then, there is a constant \(C = C(T_0)\) (from \(\mathrm {L}_p\)-maximal regularity) such that
Then, by construction (and uniqueness of solutions), \(\varvec{w}|_{[0,T] \times \Omega }\) is the solution to the problem with time horizon \(T \in (0,T_0]\) for the given data \((\varvec{f}, \varvec{g}, \varvec{h})\) and
We will demonstrate that \(\rho \in (0, \rho _0]\) and \(T \in (0, T_0]\) can be chosen such that \(\Phi \) is a contractive self-mapping on \(\mathrm {D}_{\rho ,T}\), and hence attains a unique fixed point by the contraction mapping principle. To this end, we show that for suitably small \(\rho \in (0, \rho _0]\) and \(T \in (0, T_0]\), \(\Phi \) is a strictly contractive mapping from \(\mathrm {D}_{\rho ,T}\) into \(\mathrm {D}_{\rho ,T}\). First, note that since \(\big \Vert \varvec{v} \big \Vert _\infty \le \frac{\eta }{3}\) for every \(\varvec{v} \in \mathrm {D}_{\rho ,T}\) with \(\rho \in (0, \rho _0]\), \(T \in (0, T_0]\), we may denote by \(\big \Vert \varvec{r} \big \Vert \) and \(\big \Vert {\mathcal {Q}}_a \big \Vert \) the respective norms as functions on \([\frac{\eta }{6}, \big \Vert \varvec{c}^* \big \Vert _\infty + \frac{\eta }{4}]^N\). Secondly, we will frequently use that \(\varvec{v}(0,\cdot ) = \varvec{{\tilde{v}}}(0, \cdot ) \equiv \varvec{0}\), \(\varvec{g}(\varvec{c}^*)|_{t=0} = \varvec{g}(\varvec{c}^0) \equiv \varvec{0}\) (as \(\varvec{c}^0\) satisfies the compatibility conditions) and \(\varvec{h}(\varvec{c}^*, \varvec{v}) = \varvec{h}(\varvec{c}^*, \varvec{{\tilde{v}}}) \equiv \varvec{0}\). Therefore, for every \(\varvec{v}, \varvec{{\tilde{v}}} \in \mathrm {D}_{\rho ,T}\) and \(\rho \in (0, \rho _0]\), \(T \in (0, T_0]\), and after fixing some auxiliary value \({\tilde{p}} \in \big ( \nicefrac {n+2}{2}, p \big )\), hence \(\mathring{\mathrm {W}}_{{\tilde{p}}}^{(1,2)}((0,T] \times {\overline{\Omega }}) \hookrightarrow \mathrm {C}([0,T] \times {\overline{\Omega }})\) with an embedding constant \(C(T_0)\) that can be used uniformly for all \(T \in (0,T_0]\), we obtain estimates
Here, the critical terms in the estimates for the norms of \(h_a\) can be estimated as follows:
and for the time derivative
for the second-order spatial derivatives,
where \(C^{**} = C^{**}(T_0) > 0\) is a common constant for the continuous embeddings \(\mathring{\mathrm {W}}_p^{(1,2)}((0,T] \times {\overline{\Omega }}) \hookrightarrow \mathrm {C}([0,T]; \mathrm {C}^1({\overline{\Omega }}))\) and \(\mathring{\mathrm {W}}_p^{(1,2)}((0,T] \times {\overline{\Omega }}) \hookrightarrow \mathrm {L}_{2p}((0,T); \mathrm {W}_{2p}^1(\Omega ))\) with \(T \in (0, T_0]\). Standard arguments now give that for suitable choice of \(\rho \in (0, \rho _1]\), \(T \in (0, T_1]\) (for some \(\rho _1 \in (0, \rho _0]\) and \(T_1 \in (0,T_0]\)), the map \(\Phi \) is a strictly contractive mapping from \(\mathrm {D}_{\rho ,T}\) into \(\mathrm {D}_{\rho ,T}\) (here, \(\Phi (\mathrm {D}_{\rho ,T}) \subseteq \mathrm {D}_{\rho ,T}\) as \(\varvec{0} \in \mathrm {D}_{\rho ,T}\)), thus by the strict contraction principle admits a unique fixed point \(\varvec{v}^\mathrm {fix} = \Phi (\varvec{v}^\mathrm {fix}) \in \mathring{\mathrm {W}}_p^{(1,2)}((0,T] \times {\overline{\Omega }})\). Then, \(\varvec{c} := \varvec{c}^*+ \varvec{v}^\mathrm {fix}\) is the unique solution of the fast sorption and fast surface chemistry limit reaction diffusion system. Here, the uniqueness is valid first only in the class \(\varvec{c} \in \varvec{c}^*+ \varvec{D}_{\rho ,T}\), then arguing by contradiction shows that \(\varvec{c}\) is actually unique in the class \(\varvec{c} \in \mathrm {W}_p^{(1,2)}((0,T) \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\). Moreover, for \(\varvec{c}^{*,0} \in \mathrm {I}_p^\eta (\Omega )\), \(\eta > 0\) sufficiently small and initial values \(\varvec{c}^0 \in \mathrm {B}_\eta (\varvec{c}^{*,0}) \subseteq \mathrm {I}_p^\eta (\Omega )\) close to \(\varvec{c}^{*,0}\), there is a common choice of parameters \(\rho _0\) and \(T_2\) to make the respective maps \(\Phi = \Phi ^{\varvec{c}^0}\) strictly contractive self-mappings for any \(\rho \in (0,\rho _0]\) and \(T \in (0, T_2]\), so that for all these initial values the solution exists and is unique at least on the time interval \((0, T_2)\). Also it can be seen that as the maps \(\Phi ^{\varvec{c}^0}\) continuously depend on the initial values, so do the fixed points, hence the solutions to the fast sorption and fast surface chemistry reaction–diffusion-limit system. \(\square \)
From the proof we may extract blow-up criteria for solutions which are not global in time.
Corollary 4.9
(Blow-up criterion). Either \(T_\mathrm {max} = \infty \) (global existence), or \(T_\mathrm {max} < \infty \) and \(\big \Vert \varvec{c}(t) \big \Vert _{\mathrm {W}_p^{2-2/p}(\Omega ;{{\,\mathrm{{\mathbb {R}}}\,}}^N)} \rightarrow \infty \) (blow-up) or \(\min _{\varvec{z} \in {\overline{\Omega }}} c_i(t,\varvec{z}) \rightarrow 0\) (degeneration) for some \(i \in \{1, \ldots , N\}\) as \(t \rightarrow T_\mathrm {max}\).
Remark 4.10
The inclusion of the case \(\min c_i(t,\varvec{z}) \rightarrow 0\) for some i is due to the chosen linearisation around the reference function. To have enough regularity for \(\varvec{C} = {\text {diag}}(\varvec{c}|_\Sigma )^{-1}\), one needs uniform positivity of the solution candidate \(\varvec{c}\), thus on the initial value \(\varvec{c}^0\). Therefore, this approach breaks down as \(\min c_i(t,\cdot ) \rightarrow 0\).
4.2.2 Local-in-time existence of classical solutions in Hölder class
Analogous to the case of strong \(\mathrm {W}_p^{(1,2)}\)-solutions, we may also deduce existence of local-in-time classical solutions in the Hölder class
Theorem 4.11
(Local-in-time existence of classical solutions in Hölder class). Let \(\alpha \in (0, \nicefrac {1}{2})\) and \(\Omega \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}^n\) be a bounded domain of class \(\partial \Omega \in \mathrm {C}^{2+2\alpha }\). Then, the fast sorption and fast surface chemistry limit problem (5’) admits a unique classical solution in the Hölder class \(\mathrm {C}^{(1,2) \cdot (1 + \alpha )}\), if
More precisely, for every reference initial value \(\varvec{c}_0^\mathrm {ref} \in \mathrm {I}_\alpha ^+({\overline{\Omega }})\), there are \(T > 0\), \(\varepsilon > 0\) and \(C > 0\) such that the following statements hold true:
-
(1)
For all \(\varvec{c}^0 \in \mathrm {I}_\alpha ^+({\overline{\Omega }})\) with \(\big \Vert \varvec{c}_0 - \varvec{c}_0^\mathrm {ref} \big \Vert _{\mathrm {I}_\alpha ({\overline{\Omega }})} < \varepsilon \), there is a unique classical solution in the Hölder class \(\varvec{c} \in \mathrm {C}^{(1,2) \cdot (1 + \alpha )}(J \times {\overline{\Omega }}; (0, \infty )^N)\) of (5’) for \(J = [0,T]\).
-
(2)
For any two initial values \(\varvec{c}^0, \tilde{\varvec{c}}^0 \in \mathrm {I}_\alpha ^+({\overline{\Omega }})\) with \(\big \Vert \varvec{c}^0 - \varvec{c}^\mathrm {ref} \big \Vert _{\mathrm {C}^{2+\alpha }({\overline{\Omega }})}, \big \Vert \tilde{\varvec{c}}^0 - \varvec{c}^\mathrm {ref} \big \Vert _{\mathrm {C}^{2+\alpha }({\overline{\Omega }})} < \varepsilon \) and corresponding classical solutions \(\varvec{c}, \tilde{\varvec{c}} \in \mathrm {C}^{(1,2) \cdot (1+\alpha )}(J \times {\overline{\Omega }}; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) one has
$$\begin{aligned} \big \Vert \varvec{c} - \tilde{\varvec{c}} \big \Vert _{\mathrm {C}^{(1,2) \cdot (1+\alpha )}(J \times {\overline{\Omega }})} \le C \big \Vert \varvec{c}^0 - \tilde{\varvec{c}}^0 \big \Vert _{\mathrm {C}^{2+2\alpha }({\overline{\Omega }})}. \end{aligned}$$ -
(3)
Any (Hölder) classical solution \(\varvec{c} \in \mathrm {C}^{(1,2) \cdot (1 + \alpha )}(J \times {\overline{\Omega }})\) can be extended in a unique way to a maximal (Hölder) classical solution \(\varvec{c}: [0, T_\mathrm {max}) \times \Omega \rightarrow (0,\infty )^N\) (with \(T_\mathrm {max} \in (0, \infty ]\)) with \(\varvec{c} \in \mathrm {C}^{(1,2) \cdot (1+\alpha )}((0,T) \times \Omega ; (0,\infty )^N)\) for every \(T \in (0, T_\mathrm {max})\).
Proof
We may proceed as in the case of strong \(\mathrm {W}_p^{(1,2)}\)-solutions and consider a reference function \(\varvec{c}^\mathrm {ref} \in \mathrm {C}^{2+2\alpha }({\overline{\Omega }})\) and \(\varepsilon > 0\) such that \(0 < 2 \varepsilon \le \varvec{c}^\mathrm {ref}\) on \({\overline{\Omega }}\). Choosing \(\varvec{c}^0 \in \mathrm {C}^{2+2\alpha }({\overline{\Omega }})\) with \(\big \Vert \varvec{c}^0 - \varvec{c}^\mathrm {ref} \big \Vert _{\mathrm {C}^{2+2\alpha }} \le \varepsilon \) thus implies that \(\varvec{c}^0 \ge \varepsilon \) on \({\overline{\Omega }}\). Now, let \({\mathcal {T}}(\cdot )\) denote the strongly continuous semigroup on \(\mathrm {C}^{2\alpha }({{\,\mathrm{{\mathbb {R}}}\,}}^n)\) generated by the operator \(A = \varvec{D} \Delta \) with \({{\,\mathrm{\mathrm {D}}\,}}(A) = \mathrm {C}^{2+2\alpha }({{\,\mathrm{{\mathbb {R}}}\,}}^n)\), and set
where \({\mathcal {E}} \in {{\,\mathrm{{\mathcal {B}}}\,}}(\mathrm {C}^{2+2\alpha }({\overline{\Omega }}); \mathrm {C}^{2+2\alpha }({{\,\mathrm{{\mathbb {R}}}\,}}^n)) \cap {{\,\mathrm{{\mathcal {B}}}\,}}(\mathrm {C}({\overline{\Omega }}); \mathrm {C}({{\,\mathrm{{\mathbb {R}}}\,}}^n))\) is any bounded, linear extension operator (which exists as \(\partial \Omega \in \mathrm {C}^{2+2\alpha }\)). Then, \({\mathcal {E}} \varvec{c}^0, {\mathcal {E}} \varvec{r}(\varvec{c}^0) \in \mathrm {C}^{2+2\alpha }({{\,\mathrm{{\mathbb {R}}}\,}}^n) = {{\,\mathrm{\mathrm {D}}\,}}(A)\), hence, \(c^{*}\in C({{\,\mathrm{{\mathbb {R}}}\,}}_{+}; C^{2+2\alpha }({{{\,\mathrm{{\mathbb {R}}}\,}}^{n}}))\cap C^{1}({{\,\mathrm{{\mathbb {R}}}\,}}_{+};C^{2\alpha }({{{\,\mathrm{{\mathbb {R}}}\,}}^{n}}))\). Moreover, by maximal Hölder regularity [14, Theorem VII.10.3], it follows from \(\varvec{r}({\mathcal {E}} c^0) \in \mathrm {C}^{(1,2) \cdot \alpha }({{\,\mathrm{{\mathbb {R}}}\,}}_+ \times {{\,\mathrm{{\mathbb {R}}}\,}}^n)\) that, actually, \(\varvec{c}^*\in \mathrm {C}^{(1,2) \cdot (1 + \alpha )}([0,T] \times {{\,\mathrm{{\mathbb {R}}}\,}}^n)\) for every \(T > 0\). By the classical maximum principle (and suitable choice of \({\mathcal {E}}\)) we may find \(T_0 > 0\) such that \(\varvec{c}^*\ge \frac{\varepsilon }{2}\) on \([0,T_0]\), for every initial value \(\varvec{c}^0\) with \(\big \Vert \varvec{c}^0 - \varvec{c}^\mathrm {ref} \big \Vert _{\mathrm {C}^{2+2\alpha }} \le \varepsilon \). Let us fix \(\rho _0 \le \varepsilon /(3 \big \Vert {\mathcal {E}} \big \Vert _{{{\,\mathrm{{\mathcal {B}}}\,}}(\mathrm {C}({\overline{\Omega }}); \mathrm {C}({{\,\mathrm{{\mathbb {R}}}\,}}^n))})\) and \(T_0 > 0\) as above. Then, for every \(\varvec{c}^0 \in \mathrm {I}_\alpha ^+({\overline{\Omega }})\) with \(\big \Vert \varvec{c}^0 - \varvec{c}^* \big \Vert _{\mathrm {C}^{2+2\alpha }} \le \varepsilon \), a function \(\varvec{c} \in \mathrm {C}^{(1,2) \cdot (1 + \alpha )}([0,T] \times {\overline{\Omega }})\) is a classical solution to the reaction–diffusion–sorption problem
if and only if the difference function \(\varvec{v} := \varvec{c} - \varvec{c}^*\) solves the semilinear parabolic system
with \({\mathcal {Q}}_a\) defined as in the local-in-time existence proof for strong \(\mathrm {W}_p^{(1,2)}\)-solutions. We remark that \({\mathcal {Q}}(\varvec{v}, \varvec{c}^*)|_{t=0} = {\mathcal {Q}}_a(\varvec{0}, \varvec{c}^0) = \varvec{0}\), \(\varvec{r}(\varvec{c}^*)|_{t=0} = \varvec{r}(\varvec{c}^0)\) as well as \(\kappa _a - (\varvec{c}^*)^{\varvec{\nu }^{\Sigma ,a}}|_{\{0\} \times \Sigma } = \kappa _a - (\varvec{c}^0)^{\varvec{\nu }^{\Sigma ,a}}|_\Sigma = 0\) and \(\langle \varvec{e}^k, \varvec{D} \partial _{\varvec{n}} \varvec{c}^*|_{\{0\} \times \Sigma } \rangle = \langle \varvec{e}^k, \varvec{D} \partial _{\varvec{n}} \varvec{c}^0|_\Sigma \rangle = 0\). This allows to seek for \(\varvec{v}\) as a fixed point of the map
defined by \(\Phi (\varvec{v}) := \varvec{w}\), where \(\varvec{w}\) is the classical solution to the quasi-autonomous problem
We may now proceed as for the case of strong solutions, one helpful result being the auxiliary Lemma 4.12.
Details are left to the reader. \(\square \)
Let us write
Lemma 4.12
(Locally uniform maximal regularity constant). Let \(\alpha > 0\) and \(\Omega \subseteq {{\,\mathrm{{\mathbb {R}}}\,}}^n\) be a bounded domain of class \(\partial \Omega \in \mathrm {C}^{2+2\alpha }\). Moreover, let \(T_0 > 0\) and \(\varvec{c}^*\in \mathrm {C}^{(1,2) \cdot (1 + \alpha )}([0,T] \times {\overline{\Omega }}; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) with \(\varvec{c}^*\ge \varepsilon > 0\) be given. Then, there is a constant \(C = C(T_0,\varepsilon ,\big \Vert \varvec{c}^* \big \Vert _{\mathrm {C}^{(1,2) \cdot (1+\alpha )}}) > 0\) such that for every \(T \in (0,T_0]\) and every \(\varvec{f} \in \mathrm {C}^{(1,2) \cdot \alpha }([0,T] \times {\overline{\Omega }}; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\), \(\varvec{g} \in \mathring{\mathrm {C}}^{(1,2) \cdot (\nicefrac {1}{2} + \alpha )}((0,T] \times {\overline{\Omega }}; {{\,\mathrm{{\mathbb {R}}}\,}}^{n^\Sigma })\) there is a unique solution \(\varvec{v} \in \mathring{\mathrm {C}}^{(1,2) \cdot (1+\alpha )}([0,T] \times {\overline{\Omega }}; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) of the linear parabolic initial-boundary value problem
and it holds the estimate
Proof
As demonstrated (for the half-space case) in Proposition 4.2 in combination with Lemma 4.3, the Lopatinskii–Shapiro condition is satisfied, so that Hölder-maximal regularity follows by [14, Theorem VIII.10.4] (where for nonzero initial values complying with the compatibility condition, the maximal regularity constant may, in general, depend on \(T \in (0,T_0]\) as well). Let us demonstrate that for zero initial values, the maximal regularity constant may be chosen uniformly for \(T \in (0,T_0]\). To see this, let \(\chi \in \mathrm {C}^\infty ({{\,\mathrm{{\mathbb {R}}}\,}})\) be any function such that \(\chi \equiv 1\) on \((- \infty , \frac{4}{3}]\) and \(\chi \equiv 0\) on \([\nicefrac {5}{3}, \infty )\), say. Given any \(T \in (0,T_0]\), and \(\varvec{f}\), \(\varvec{g}\), \(\varvec{h}\) as in the statement of the theorem, we may now extend these functions by setting
Then, for \(\alpha \in (0, \nicefrac {1}{2}]\) one easily deduces that \(\varvec{{\tilde{f}}} \in \mathrm {C}^{(1,2) \cdot \alpha }({{\,\mathrm{{\mathbb {R}}}\,}}_+ \times {\overline{\Omega }};{{\,\mathrm{{\mathbb {R}}}\,}}^N)\) and \(\varvec{{\tilde{g}}} \in \mathrm {C}^{(1,2) \cdot (\nicefrac {1}{2} + \alpha )}({{\,\mathrm{{\mathbb {R}}}\,}}_+ \times {\overline{\Omega }};{{\,\mathrm{{\mathbb {R}}}\,}}^{n^\Sigma })\) with norms
For \(\varvec{{\tilde{h}}}\), we obtain that
Therefore, there is a constant \(C = C(\alpha )\), independent of \(T \in (0,T_0]\) and \(T_0 > 0\) such that
We may therefore consider \(\varvec{v} = \varvec{{\tilde{v}}}|_{[0,T] \times {\overline{\Omega }}}\) as the restriction to time \(t \in [0,T]\) of the solution of the parabolic initial-boundary value problem on \((0,T_0)\)
and may deduce that
\(\square \)
4.2.3 A priori bounds on the strong solution of the fast sorption and fast surface chemistry model
In the previous subsection, it has been noticed that a bound on the phase space norm \(\big \Vert \cdot \big \Vert _{\mathrm {W}_p^{2-2/p}}\) is enough for establishing global existence of a strong solution. To derive such a bound, however, is a delicate matter, and it is not clear whether global existence holds true in all cases. On the other hand, for some weaker norms at least a priori bounds can be established for free. The derivation of these a priori bounds is based on the parabolic maximum principle and entropy considerations, highlighting the fruitful interplay between mathematics and physics, and will be presented in this subsection.
Theorem 4.13
(A priori bounds). Let \(\varvec{c}^0 \in \mathrm {I}_p^+(\Omega ) \cap \mathrm {C}^2({\overline{\Omega }}; {{\,\mathrm{{\mathbb {R}}}\,}}^N)\) and \(\varvec{c} \in \mathrm {C}^{(1,2)}([0, T_\mathrm {max}) \times {\overline{\Omega }};{{\,\mathrm{{\mathbb {R}}}\,}}_+^N)\) be a maximal classical solution to the fast sorption and fast surface chemistry limit problem
Further, assume that there is a conserved quantity with strictly positive entries, i.e. there is
Then, for every \(T_0 \in (0, T_\mathrm {max}] \cap {{\,\mathrm{{\mathbb {R}}}\,}}\) there is \(C = C(T_0) > 0\), also depending on the initial value \(\varvec{c}^0\), such that the following a priori bounds hold true:
-
(1)
\(\mathrm {L}_\infty ^t \mathrm {L}_1^{\varvec{z}}\)—a priori estimate:
$$\begin{aligned} \sup _{t \in [0, T_0)} \big \Vert \varvec{c}(t,\cdot ) \big \Vert _{\mathrm {L}_1(\Omega ;{{\,\mathrm{{\mathbb {R}}}\,}}^N)} \le C \big \Vert \varvec{c}^0 \big \Vert _{\mathrm {L}_1(\Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)}, \end{aligned}$$where the constant can actually be chosen independently of \(\varvec{c}^0\) and \(T_0\), but only depends on the ratio between the smallest and largest entry of \(\varvec{e} \in (0, \infty )^N\);
-
(2)
\(\mathrm {L}_1^t \mathrm {L}_\infty ^{\varvec{z}}\)—a priori estimate:
$$\begin{aligned} \sup _{\varvec{z} \in {\overline{\Omega }}} \big \Vert \varvec{c}(\cdot ,\varvec{z}) \big \Vert _{\mathrm {L}_1([0,T_0);{{\,\mathrm{{\mathbb {R}}}\,}}^N)} \le C; \end{aligned}$$ -
(3)
\(\mathrm {L}_2^t \mathrm {L}_2^{\varvec{z}}\)—a priori estimate:
$$\begin{aligned} \big \Vert \varvec{c} \big \Vert _{\mathrm {L}_2([0,T_0) \times \Omega ; {{\,\mathrm{{\mathbb {R}}}\,}}^N)} \le C; \end{aligned}$$ -
(4)
Moreover, provided the surface chemistry is constructed from thermodynamic principles, i.e. the affinity \({\mathcal {A}}_a = \mu _i^\Sigma \nu _i^\Sigma \) vanishes for chemical equilibria, and \(\mu _i^\Sigma = \mu _i|_\Sigma = \mu _i^0 + \ln c_i|_\Sigma \) for equilibria for adsorption and desorption, the following entropy identity holds true:
$$\begin{aligned}&\int _\Omega c_i(t,\varvec{z}) (\mu _i^0 + \ln \varvec{c}_i(t,\varvec{z}) - 1) \, \mathrm {d}\varvec{z} \\&\quad + \int _0^t \int _\Omega \sum _{i=1}^N d_i \frac{\left| \nabla c_i(s,\varvec{z}) \right| ^2}{c_i(s,\varvec{z})} \, \mathrm {d}\varvec{z} \, \mathrm {d}s + \sum _{a=1}^m \int _0^t \int _\Omega \big ( \sum _{i=1}^N \ln (c_i) \nu _i^a \big ) \\&\qquad \left( \exp \big ( \sum _{i=1}^N \ln (c_i) \nu _i^a \big ) - 1 \right) \, \mathrm {d}\varvec{z} \, \mathrm {d}s \\&\quad = \int _\Omega c_i^0(\varvec{z}) (\mu _i^0 \ln c_i^0(\varvec{z}) - 1) \, \mathrm {d}\varvec{z}, \quad t \in [0,T_0). \end{aligned}$$
Proof
\(\mathrm {L}_\infty ^t \mathrm {L}_1^{\varvec{z}}\)--a priori estimate: Since \(\varvec{e} \in (0, \infty )^N\) is a conserved quantity for both the bulk and surface chemistry, \(\langle \varvec{r}(\varvec{c}), \varvec{e} \rangle = \langle \varvec{r}^\Sigma (\varvec{c}), \varvec{e} \rangle = 0\) for all values of \(\varvec{c}\), where \(\varvec{r}^\Sigma (\varvec{c}) = k_a^f \varvec{c}^{\varvec{\alpha }^{\Sigma ,a}} - k_a^b \varvec{c}^{\varvec{\beta }^{\Sigma ,a}}\). Thus, using regularity properties of parameter-dependent integrals and the divergence theorem, for every \(t \in [0,T_0)\) it holds that
As a result,
and, since \(\varvec{e} \in (0, \infty )^N\), the map \(c \mapsto \int _\Omega \sum _{i=1}^N \left| c_i(z) \right| e_i \, \mathrm {d}\varvec{z}\) defines a norm which is equivalent to the standard \(\mathrm {L}_1\)-norm on the Lebesgue space \(\mathrm {L}_1(\Omega ;{{\,\mathrm{{\mathbb {R}}}\,}}^N)\). More precisely,
establishing the first a priori estimate with
independent of \(T_0 > 0\) and the initial value \(\varvec{c}^0\). \(\mathrm {L}_1^t \mathrm {L}_\infty ^{\varvec{z}}\)--a priori estimate: To derive the \(\mathrm {L}_1 \mathrm {L}_\infty \)-a priori bound, let us consider the function \(w: [0, T_0) \times {\overline{\Omega }} \rightarrow [0, \infty )\) defined by
As a parameter integral of a \(\mathrm {C}^{(1,2)}\)-function, w has the regularity \(w \in \mathrm {C}^2([0,T_0) \times {\overline{\Omega }})\) and using elementary results on parameter-dependent integrals, the evolution (5) and the assumption that \(\varvec{e}\) is a conserved quantity, we establish the estimate
Therefore, \(w \ge 0\) satisfies the system of differential inequalities
From the parabolic maximum principle for differential inequalities, it then follows that there is \(C > 0\) (depending on \(T_0\) and \(\varvec{c}^0\)) such that
and, consequently, one finds that
\(\mathrm {L}_2^t \mathrm {L}_2^{\varvec{z}}\)--a priori estimate: For the \(\mathrm {L}_2\)-estimate, let us fix \(T \in (0,T_0)\). Employing integration by parts, Fubini’s theorem, the no-flux boundary conditions on the conserved part and the fundamental theorem of calculus, we find for the integral
where in the last step it has been used that \(\varvec{c} \ge \varvec{0}\) and, therefore,
One may thus take
Entropy identity: By the theorem on derivatives of parameter-dependent integrals, and as the derivative of the function \((0,\infty ) \in x \mapsto x (\ln x - 1)\) is \(\ln x \) for all \(x \in (0, \infty )\), one finds that
The assertion will be established if \(\sum _i (\mu _i^0 + \ln (c_i)) d_i \partial _{\varvec{n}} c_i \, \mathrm {d}\sigma (\varvec{z}) = 0\) can be proved. From the boundary conditions \(\langle \varvec{e}^k, \partial _{\varvec{n}} (\varvec{D} \varvec{c}) \rangle = 0\), there are scalar functions \(\eta _a: [0,T_0) \times \Sigma \rightarrow {{\,\mathrm{{\mathbb {R}}}\,}}\) such that \(\partial _{\varvec{n}} (\varvec{D} \varvec{c})|_\Sigma = \sum _{a=1}^{m^\Sigma } \eta _a \varvec{\nu }^{\Sigma ,a}\). Hence,
by assumption since for all times \(t \ge 0\) and at all positions \(\vec {z} \in \Sigma \) the sorption processes and the surface chemistry are in equilibrium. Therefore, this contribution to the sum vanishes, and the entropy identity follows by the fundamental theorem of calculus. \(\square \)
References
P. Acquistapace, B. Terreni: Fully nonlinear parabolic systems. Recent advances in nonlinear elliptic and parabolic problems (Nancy, 1988), 97–111, Pitman Res. Notes Math. Ser. 208, Longman Sci. Tech., Harlow 1989.
R.A. Adams, J.J.F. Fournier: Sobolev Spaces. Second Edition. Pure and Applied Mathematics (Amsterdam) 140. Elsevier/Academic Press, Amsterdam 2003.
H. Amann: Anisotropic function spaces and maximal regularity for parabolic problems. Part 1. Function spaces. Jindřich Nečas Center for Mathematical Modeling Lecture Notes 6. Matfyzpress, Prague 2009.
B. Augner: \(L_p\)-maximal regularity for parabolic and elliptic boundary value problems with boundary conditions of mixed differentiability orders. In preparation, based on the appendix of arXiv:2006.12098v1.
B. Augner, D. Bothe: The fast-sorption and fast-surface-reaction limit of a heterogeneous catalysis model. Discrete Contin. Dyn. Syst. Ser. S 14 (2021), no. 2, 533–574.
D. Bothe, M. Köhne, S. Maier, J. Saal: Global strong solutions for a class of heterogeneous catalysis models. J. Math. Anal. Appl. 445 (2017), no. 1, 677–709.
R. Denk, M. Hieber, J. Prüss: \({\cal{R}}\)-boundedness, Fourier multipliers and problems of elliptic and parabolic type. Mem. Amer. Math. Soc. 788 (2003).
R. Denk, M. Hieber, J. Prüss: Optimal\({\rm L} _p-{\rm L}_q\)estimates for parabolic boundary value problems with inhomogeneous data. Math. Z. 257 (2007): 193–224.
R. Denk, M. Kaip: General parabolic mixed order systems in\({\rm L}_p\)and applications. Operator Theory: Advances and Applications 239, Birkhäuser/Springer, Cham, 2013.
R. Denk, J. Prüss, R. Zacher: Maximal\({\rm L}_p\)-regularity of parabolic problems with boundary dynamics of relaxiation type. J. Funct. Anal. 255 (2008): 3149–3187.
L. Desvillettes, K. Fellner: Exponential decay toward equilibrium via entropy methods for reaction–diffusion equations. J. Math. Anal. Appl. 319 (2006), no. 1, 157–176.
L. Desvillettes, K. Fellner, B.Q. Tang: Trend to equilibrium for reaction–diffusion systems arising from complex balanced chemical reaction networks. SIAM J. Math. Anal. 49 (2017), no. 4, 2666–2709.
P.-E. Druet, A. Jüngel: Analysis of cross-diffusion systems for fluid mixtures driven by a pressure gradient. SIAM J. Math. Anal. 52, no. 2, 2179–2197.
O.A. Ladyzenskaya, V.A. Solonnikov, N.N. Ural’ceva: Linear and Quasi-linear Equations of Parabolic Type. Trans. Math. Monographs 23, Amer. Math. Soc., Providence 1968.
E.H. Lieb, M. Loss: Analysis. Graduate Studies in Mathematics 14. American Mathematical Society, Providence, RI, 2001.
A. Lunardi: Analytic Semigroups and Optimal Regularity in Parabolic Problems. Progress in Nonlinear Differential Equations and Theor Application 16, Birkhäuser, Basel 1995.
M. Pierre: Global existence in reaction–diffusion systems with control of mass: a survey. Milan J. Math. 78 (2010), no. 2, 417–455.
M. H. Protter and H. F. Weinberger: Maximum Principles in Differential Equations. Prentice-Hall Inc., Englewood Cliffs, N.J., 1967.
R. Schnaubelt: Stable and Unstable Manifolds for Quasilinear Parabolic Problems with Fully Nonlinear Dynamical Boundary Conditions. Advances in Differential Equations 22 (2017), 541–592.
O. Souček, V. Orava, J. Málek, D. Bothe: A continuum model of heterogeneous catalysis: thermodynamic framework for multicomponent bulk and surface phenomena coupled by sorption. Int. J. Eng. Sci. 138 (2019), 82–117.
B. Terreni: Hölder regularity results for nonhomogeneous parabolic initial-boundary value linear problems. Semigroup theory and applications (Trieste, 1987), 387–401, Lecture Notes in Pure and Appl. Math. 116, Dekker, New York 1989.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Corresponding author
Additional information
Dedicated to Matthias Hieber on the occasion of his 60 $$^{th}$$ th birthday.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors gratefully acknowledge financial support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), Project-ID 265191195, SFB 1194, sub-project B01.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Augner, B., Bothe, D. Analysis of some heterogeneous catalysis models with fast sorption and fast surface chemistry. J. Evol. Equ. 21, 3521–3552 (2021). https://doi.org/10.1007/s00028-021-00692-4
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00028-021-00692-4