Nonpolar solid continua under quasistatic conditions are considered in the following. Employing analytical notation as a mathematical basis, whereat summation is indicated by repeated indices, the balance equations read:
$$\begin{aligned}&\sigma _{ij,j}+b_{i}=0, \end{aligned}$$
(1)
$$\begin{aligned}&D_{i,i}=0. \end{aligned}$$
(2)
Equation (
1) describes the balance of momentum, containing Cauchy stresses
\(\sigma _{ij}\) and volume forces
\(b_{i}\), whereas Eq. (
2) implies electrostatic equilibrium with electric displacement
\(D_{i}\), disregarding volume charges. Here, as well as in the following, a comma in a subscript denotes partial spatial derivation. Moreover, infinitesimal mechanical deformations are assumed, leading to a linear relation between the strain
\(\varepsilon _{ij}\) and the displacement gradient:
$$\begin{aligned}&\varepsilon _{ij}=\frac{1}{2}(u_{i,j}+u_{j,i}). \end{aligned}$$
(3)
On the other hand, the electric field
\(E_{i}\) is related to the gradient of the electric potential
\(\phi \):
$$\begin{aligned}&E_{i}=-\phi _{,i}. \end{aligned}$$
(4)
Due to dissipation as a result of domain wall motions, thermodynamic consistency is a major issue in ferroelectric constitutive laws. In e.g. [
37], the electric enthalpy density
h as a thermodynamic potential of ferroelectric continua reads:
$$\begin{aligned} h(\varepsilon _{ij},E_{i})=&\frac{1}{2}C_{ijkl}\varepsilon _{ij}\varepsilon _{kl}-\frac{1}{2}\kappa _{ij}E_{i}E_{j}-e_{lij}E_{l}\varepsilon _{ij} \\&-\left( C_{ijkl}\varepsilon _{kl}-e_{lij}E_{l}\right) \varepsilon _{ij}^{\textrm{irr}}-\frac{1}{2}C_{ijkl}\varepsilon _{ij}^{\textrm{irr}}\varepsilon _{kl}^{\textrm{irr}}\nonumber \\&-E_{i}P_{i}^{\textrm{irr}}, \nonumber \end{aligned}$$
(5)
with the electric field and the total strain as independent variables. Internal variables, accounting for domain wall motion, are involved in the elastic tensor
\(C_{ijkl}\), the dielectric tensor
\(\kappa _{ij}\) and the piezoelectric coupling tensor
\(e_{ikl}\) as well as the irreversible strain
\(\varepsilon _{ij}^{\textrm{irr}}\) and polarization
\(P_{i}^{\textrm{irr}}\). Differentiating Eq. (
5) with respect to the independent variables, the associated variables stress and electric displacement are obtained as
$$\begin{aligned} \frac{\partial h}{\partial \varepsilon _{ij}}\bigg |_{E}&=\sigma _{ij}=C_{ijkl}\left( \varepsilon _{kl}-\varepsilon _{kl}^{\textrm{irr}}\right) -e_{lij}E_{l}, \end{aligned}$$
(6)
$$\begin{aligned} -\frac{\partial h}{\partial E_{i}}\bigg |_{\varepsilon }&=D_{i}=e_{ikl}\left( \varepsilon _{kl}-\varepsilon _{kl}^{\textrm{irr}}\right) +\kappa _{ij}E_{j}+P_{i}^{\textrm{irr}}, \end{aligned}$$
(7)
representing the ferroelectric constitutive law. Equations (
6) and (
7) are valid within sufficiently small changes of state, at which material tensors act as linear tangent moduli. Inelastic nonlinear ferroelectric behavior is achieved adapting material tangents and irreversible quantities iteratively, see Sect.
3.1. Considering thermodynamic consistency, a mathematical description of ferroelectricity has to satisfy the balance of entropy. Starting at the Clausius inequality according to
$$\begin{aligned} \Theta \dot{s}+q_{i,i}- \displaystyle {\frac{q_i}{\Theta }}\Theta _{,i}-\rho r \ge 0, \end{aligned}$$
(8)
where
\(q_i\),
\(\rho \),
r and
\(\Theta \) denote specific heat flux, mass density, volume heat source and temperature, respectively, the generalized Clausius–Duhem inequality
$$\begin{aligned} \Theta \dot{s}-\dot{u}+\sigma _{ij}\left( \dot{\varepsilon }^\textrm{rev}_{ij}+\dot{\varepsilon }^\textrm{irr}_{ij}\right) +E_i \left( \dot{D}^\textrm{rev}_i+\dot{P}^\textrm{irr}_i \right) -\displaystyle {\frac{q_i}{\Theta }}\Theta _{,i}\ge 0 \end{aligned}$$
(9)
is obtained by inserting the local energy balance into Eq. (
8). Here,
s denotes the specific entropy,
u the specific internal energy and the strain and electric displacement are decomposed into reversible (
\(\varepsilon _{ij}^{\textrm{rev}}\),
\(D_{i}^{\textrm{rev}}\)) and irreversible (
\(\varepsilon _{ij}^{\textrm{irr}}\),
\(P_{i}^{\textrm{irr}}\)) parts. While the statement
\(-\displaystyle {{q_i}{\Theta _{,i}}}/\Theta \ge 0\) is always fulfilled due the oppositional directions of temperature gradients and heat fluxes
\(q_i\), the remaining terms hold the inequality in case of reversible processes, i.e.,
\(\dot{\varepsilon }^\textrm{irr}_{ij}=0\) and
\(\dot{P}^\textrm{irr}_i=0\) [
45]. Consequently, an irreversible change of state due to domain wall motion has to satisfy
$$\begin{aligned} \sigma _{ij}\textrm{d}{\varepsilon }_{ij}^{\textrm{irr}}+E_{i}\textrm{d}{P}_{i}^{\textrm{irr}} \ge 0 \end{aligned}$$
(10)
for the sake of thermodynamic consistency.