betaFeedback

WorksPhysics and Astronomy

Wrinkling dynamics of thin crystalline sheets and its dependency on topological defects

Abstract

We investigate the wrinkling dynamics of thin crystalline sheets under uniaxial compression using a Surface Phase-Field-Crystal-Helfrich (SPFCH) model, which captures the coupling between crystalline order, out-of-plane deformations, and topological defects. In particular, the model allows us to track and characterize two distinct classes of defects, namely those forming in the underlying hexagonal crystal lattice, such as dislocations and grain boundaries, and defects in the wrinkling patterns, known as wrinklons. By resolving the evolution and interactions of these defects, we investigate how the interplay between in-plane and out-of-plane relaxation mechanisms influences wrinkling onset and the coarsening regime. Our results indicate that crystalline defects do not affect coarsening scaling, but they quantitatively affect the onset of wrinkling and the emerging wrinkling wavelength, promote more irregular wrinkle patterns with a higher density of wrinklons, and modify long-timescale morphological evolution.

Introduction

While thin sheets typically buckle under compression, they also develop wrinkles when attached to a viscous or viscoelastic substrate. For small quasistatic uniaxial compression, classical elasticity theory predicts regular sinusoidal wrinkles with a wavelength determined by the elastic properties of the film as well as the film thickness. Due to their predicted regularity, the resulting equilibrium patterns are appealing for various technical applications, e.g., in light harvesting [38] or flexible electronics [36, 37]. However, such wrinkles result from a stress-driven instability with an initial wavelength much shorter than its equilibrium value, and subsequent coarsening. As a consequence of such complex dynamics, small perturbations can critically affect the emergence of regular patterns. We investigate how topological defects in thin crystalline sheets affect the stress-driven instability under uniaxial compression, the coarsening of wrinkles, and, as a result, the regularity of the emerging pattern.

Understanding the morphological evolution of compressed thin films or sheets and controlling wrinkling broadens their applications in many fields [35]. However, even in classical elasticity theory, wrinkling dynamics remain poorly understood. Previous studies predict coarsening laws with a scaling of the wavelength that varies from t1/3t^{1/3} [31] to t1/4t^{1/4} [32, 25] and t1/6t^{1/6} [33, 34]. This diversity arises from different material properties, computational approaches, and treatments of the substrate’s viscous or viscoelastic properties. Because we focus on the impact of topological defects in the wrinkling of thin crystalline sheets, we will follow Ref. [25] and account for substrate viscosity only implicitly by using a finite mobility coefficient, e.g., an overdamped equation of motion for out-of-plane deformations. Rather than relying on classical elasticity theory, we use the Phase-Field-Crystal (PFC) model [11, 12], which resolves crystalline structures on atomic length and diffusive time scales and naturally incorporates crystalline defects and elasticity [15, 3, 2]. In particular, we model the thin crystalline sheet using a Surface Phase-Field-Crystal-Helfrich (SPFCH) model, building on [17, 20, 1, 29]. We study out-of-plane deformation of a perfect hexagonal crystal, hexagonal crystals hosting an increasing number of isolated dislocations, and a hexagonal polycrystal with small- and large-angle grain boundaries. We compare the onset of wrinkling in these cases with the theory following from classical elasticity and explore the coarsening of the wrinkle patterns. While averaged quantities for scaling and roughness essentially reproduce the results in [25], strong deviations appear in the emerging wrinkle patterns and morphological evolution when crystalline defects are considered. These deviations become more pronounced as the number of crystalline defects increases, affecting the regularity of the resulting patterns and also promoting topological defects within the ordered stripe-like patterns that emerge during wrinkling (i.e., wrinklons [26]). The interplay between crystalline defects, such as dislocations and grain boundaries, and wrinkling patterns under uniaxial compression reveals a complex coupling between atomic-scale processes and macroscopic deformation mechanisms, providing fundamental insights into the mechanical behavior of thin crystalline sheets and potentially enabling new routes to tailor applications.

Methods

Mathematical model

The Phase-Field-Crystal (PFC) model [11, 12] operates on atomic length and diffusive time scales and is grounded in dynamical density functional theory [13, 14]. In its simplest form, it reduces to conserved dynamics (a H−1H^{-1} - gradient flow) of a Swift-Hohenberg free energy F[ψ]{\mathcal{F}}[\psi] with order parameter ψ\psi mimicking the averaged in time atomic density and periodic fields as minimizers representing the crystal lattice. Under small deformations, Hooke’s law is recovered [15], and lattice defects such as dislocations and grain boundaries emerge spontaneously through energy minimization. Thanks to these features, the PFC model enables mesoscale investigation of a broad class of phenomena in crystalline systems [30, 28], including the interplay between crystalline order and mechanical deformation as well as defect dynamics [4, 7, 10, 6, 5].

Applying this approach to 2D crystals on curved surfaces S\mathcal{S} yields the Surface Phase-Field-Crystal (SPFC) model [17, 20, 18], which retains all features of the (planar) PFC model except that distances are now measured geodesically. The model becomes more complex once deformable surfaces are considered. Rather than a straightforward conserved gradient flow acting on ψ\psi alone over a fixed surface, one now needs a gradient flow that minimizes the free energy through the simultaneous evolution of ψ\psi and the surface S\mathcal{S}. We consider a (L2,H−1L^2, H^{-1}) - surface gradient flows: a L2L^2 - gradient flow with respect to the evolution of S\mathcal{S} and a H−1H^{-1} - gradient flow with respect to the evolution of ψ\psi. Besides energy dissipation by evolving S\mathcal{S} and ψ\psi simultaneously, we also want to ensure conservation of ψ\psi on the evolving surface, i.e., ddt∫SψdS=0\frac{d}{dt} \int_\mathcal{S} \psi d\mathcal{S} = 0. In order to deal with local expansion or contraction of the surface, the scalar Truesdell time derivative ψ˚=ψ˙+ψ DivCV\mathring{\psi} = \dot{\psi} + \psi\, {\rm Div}_C \mathbf{V}, where ∙˙\dot{\bullet} is the material time derivative, DivC{\rm Div}_C is the componentwise trace-divergence, and V\mathbf{V} is the material velocity, has been introduced [21]. In this way, the Reynolds transport theorem is elegantly incorporated into the mathematical formalism to build (L2L^2, H−1H^{-1}) - surface gradient flows that guarantee energy dissipation and conservation properties. Following this general approach, as in [1, 29], we consider the free energy F=FL+FP+Fb\mathcal{F} = \mathcal{F_L} + \mathcal{F_P} + \mathcal{F}_b with

FL=A2∫S(ψLSψ)dS,FP=∫S(B2ψ2+C3ψ3+D4ψ4)dS,Fb=β2∫SH2dS,\begin{align} &\mathcal{F_L} = \frac{A}{2} \int_S \left( \psi \mathcal{L}_S\psi\right)d\mathcal{S}, \tag*{(1)} \\ &\mathcal{F_P}=\int_S \left( \frac{B}{2} \psi^2+ \frac{C}{3} \psi^3+\frac{D}{4} \psi^4 \right) d\mathcal{S}, \tag*{(2)}\\ &\mathcal{F}_b= \frac{\beta}{2}\int _S \mathcal{H}^2 d\mathcal{S}, \tag*{(3)} \end{align}

with A,B,C,DA, B, C, D parameters of the Swift-Hohenberg energy, characterizing the phase space and material properties together with the global average density ψˉ=∫SψdS\bar{\psi} = \int_{\mathcal{S}} \psi d\mathcal{S}. The operator LS=(q2+ΔS)2\mathcal{L}_\mathcal{S}= (q^2+\Delta_\mathcal{S} )^2, with ΔS\Delta_\mathcal{S} the Laplace-Beltrami operator, approximates the two-point correlation function and encodes the crystal symmetry, which here corresponds to the hexagonal lattice, i.e., the equilibrium density is well approximated by

ψ=ψˉ+ϕ0∑qhexeiqhex⋅r,\begin{equation} \psi= \bar{\psi} + \phi_0 \sum_{\mathbf{q}_{\rm hex}} e^{i \mathbf{q}_{\rm {hex}}\mathbin{\mkern-2mu\cdot \mkern-2mu}\mathbf{r}}, \tag*{(4)} \end{equation}

with qhex\mathbf{q}_{\rm hex} the shortest wavevectors reproducing the hexagonal symmetry (see also Eq. (11)) [28]:

{qhex}≡[±(0,1),±(32,12),±(−32,12)].\begin{equation} \{\mathbf{q}_{\rm hex}\}\equiv \left[\pm(0,1),\pm\bigg(\frac{\sqrt{3}}{2},\frac{1}{2}\bigg),\pm\bigg(-\frac{\sqrt{3}}{2},\frac{1}{2}\bigg)\right]. \tag*{(5)} \end{equation}

FP\mathcal{F_P} addresses the thermotropic part. Fb\mathcal{F}_b is a macroscopic Helfrich energy with β\beta the bending rigidity and H\mathcal{H} the mean curvature of S\mathcal{S}. The (L2L^2, H−1H^{-1}) - surface gradient flow, which we term Surface Phase-Field-Crystal-Helfrich (SPFCH) model reads

MXV=−DXF,Mψψ˚=ΔSDψF,\begin{align} M_{\mathbf{X}} \mathbf{V} &= - \mathcal{D}_{\mathbf{X}} {\mathcal{F}} , \tag*{(6)}\\ M_\psi \mathring{\psi} &= \Delta_{\mathcal{S}} \mathcal{D}_{\psi} {\mathcal{F}}, \tag*{(7)} \end{align}

where X\mathbf{X} denotes a paramaterization of S\mathcal{S}, MX,Mψ>0M_\mathbf{X}, M_\psi > 0 are mobility coefficients and DXF\mathcal{D}_{\mathbf{X}} {\mathcal{F}} and DψF\mathcal{D}_{\psi} {\mathcal{F}} denote functional derivatives leading to R3\mathbb{R}^3 vector fields and scalar fields defined on S\mathcal{S}. In this form, the surface gradient flow guarantees the conservation property ddt∫SψdS=0\frac{d}{dt} \int_\mathcal{S} \psi d\mathcal{S} = 0 and energy dissipation ddtF≤0\frac{d}{dt} \mathcal{F} \leq 0.

In the following, we restrict the surface evolution to the normal direction, e.g., MXv⊥=νD ⁣XFM_{\mathbf{X}} v_{\bot}= \boldsymbol{\nu}D_{\!\boldsymbol{X}}{\mathcal{F}}, with ν\boldsymbol{\nu} the outward-pointing normal to S\mathcal{S} and v⊥=V⋅νv_{\bot}= \mathbf{V} \mathbin{\mkern-2mu\cdot \mkern-2mu}\boldsymbol{\nu} the normal component of V\mathbf{V}, and consider a height observer given by the parameterization

Xo: (t,x,y)↦Xo(t,x,y):=[x,y,h(t,x,y)]T∈S⊂R3 ,\begin{align} \boldsymbol{X}_{\mathfrak{o}}:\ (t,x,y)\mapsto \boldsymbol{X}_{\mathfrak{o}}(t,x,y) &:= \left[ x, y, h(t,x,y) \right]^T \in\mathcal{S}\subset\mathbb{R}^3\,\text{,} \tag*{(8)} \end{align}

where h(t,x,y)h(t,x,y) is the scalar height field and (x,y)∈U⊆R2(x,y)\in\mathcal{U}\subseteq\mathbb{R}^2 are local coordinates. The resulting problem corresponding to (6) and (7) reads:

∂tψ=(∂th)(⟨∇h,∇ψ⟩∣g∣+ψΔSh)+Mψ−1ΔS(A(q4ψ+2q2ΔSψ+ΔS2ψ)+Bψ+Cψ2+Dψ3),∂th=−MX−1∣g∣(HA2((ΔSψ)2−q4ψ2)−2A(∇SψΠQSII∇S(ΔSψ+q2ψ))+⟨ ⁣⟨A(q4ψ+2q2ΔSψ+ΔS2ψ)⟩ ⁣⟩Hψ+β(ΔSH+H(H22−2K))),\begin{align} \partial_t \psi = &(\partial_t h) (\frac{\langle\nabla h, \nabla \psi\rangle}{|\boldsymbol{g}|} + \psi \Delta_S h ) \nonumber \\ & + M_\psi^{-1} \Delta_S ( A ( q^4 \psi + 2q^2 \Delta_S \psi + \Delta_S^2 \psi )+ B\psi + C \psi^2 + D\psi^3 ), \tag*{(9)} \\ \partial_t h = & -M_{\mathbf{X}}^{-1} \sqrt{|\boldsymbol{g}|} (\mathcal{H}\frac{A}{2}((\Delta_S \psi)^2 - q^4\psi^2) - 2 A (\nabla_S \psi \Pi_{QS} \textit{II} \nabla_S (\Delta_S \psi + q^2\psi) ) \nonumber \\ & + \langle\!\langle A ( q^4 \psi + 2q^2 \Delta_S \psi + \Delta_S^2 \psi ) \rangle\!\rangle \mathcal{H}\psi+\beta ( \Delta_S \mathcal{H} + \mathcal{H} ( \frac{\mathcal{H}^2}{2} - 2\mathcal{K} ) ) ), \tag*{(10)} \end{align}

with g\boldsymbol{g} the metric tensor, ΠQSII\Pi_{QS} \textit{II} the trace-free part of the shape operator, K\mathcal{K} the Gaussian curvature (see Appendix § 5.1), and ⟨ ⁣⟨⋅⟩ ⁣⟩\langle\!\langle\mathbin{\mkern-2mu\cdot \mkern-2mu}\rangle\!\rangle denoting an average over the unit cell via a smoothing kernel in reciprocal space (see Appendix § 5.3). Following previous work [1, 29, 16], we only considered the elastic parts (neglecting FP\mathcal{F_P}) of F\mathcal{F} in (10). This allows for comparison with classical elasticity theories. However, one central difference from such theories, e.g., the Föppl–von Kármán (FvK) equations for thin sheets, remains. The Surface Phase-Field-Crystal-Helfrich (SPFCH) model has dissipative structure. This is expressed by the finite mobility coefficients Mψ,MX>0M_\psi, M_{\mathbf{X}} > 0. While MψM_\psi follows from the PFC modeling approach, a finite MXM_{\mathbf{X}} essentially realizes viscous dissipation in the surrounding medium, e.g., an underlying viscous or viscoelastic substrate [25]. As a consequence, the model is expected to develop wrinkles under compression without explicitly modeling an underlying dissipative substrate.

Stress in the Surface Phase-Field-Crystal (SPFC) model

The elastic stress field is computed within the Surface Phase-Field-Crystal (SPFC) framework by extending the flat-space formulation of [9, 8] to non-planar surfaces. Elastic deformations of the density entering the PFC model can be characterized by considering the (macroscopic) displacement field u\mathbf{u},

ψ(r,t)=ψref(r−u,t)=ψˉ+∑qjϕ0e−iqj ⋅ ueiqj ⋅ r,\begin{equation} \psi(\mathbf{r}, t) = \psi^{\mathrm{ref}}(\mathbf{r} - \mathbf{u}, t) = \bar\psi+\sum_{\mathbf{q}_j}\phi_0 e^{-i\mathbf{q}_j\,\mathbin{\mkern-2mu\cdot \mkern-2mu}\, \mathbf{u}}e^{i\mathbf{q}_j\,\mathbin{\mkern-2mu\cdot \mkern-2mu}\, \mathbf{r}}, \tag*{(11)} \end{equation}

where ψref\psi^{\mathrm{ref}} denotes the reference (undistorted) density field (u=0\mathbf{u}=\mathbf{0}) and qj\mathbf{q}_j the shortest wavevectors reproducing a given lattice symmetry [28]. The macroscopic stress tensor is obtained by spatially coarse-graining the microscopic stress, which is determined by the variation of the free energy functional with respect to mass displacements [9, 8]. We generalize this expression to a curved surface S\mathcal{S} by following a similar approach to [23, 24], i.e., defining a displacement u∈TS\mathbf{u} \in T \mathcal{S}, with TST \mathcal{S} the tangent space of S\mathcal{S}, replacing flat differential operators with their surface counterparts and thus introducing additional terms through the surface derivatives. We consider only the stress contributions arising from the elastic part of the energy FL\mathcal{F}_{\mathcal{L}}, so excluding isotropic pressure terms. The microscopic stress tensor for a hexagonal lattice, generalized to a curved surface, reads

σ~ij=−(q2ψ+ΔSψ)∇S2ψ∣ij+∇i ⁣(q2ψ+ΔSψ) ∇jψ,\begin{equation} \tilde{\sigma}_{ij} = -\left(q^2\psi + \Delta_S\psi\right)\nabla^2_S\psi\big|_{ij} + \nabla_i\!\left(q^2\psi + \Delta_S\psi\right) \, \nabla_j\psi, \tag*{(12)} \end{equation}

where i,j∈{x,y}i,j \in \{x,y\}, ∇S2\nabla^2_S is the Hessian and ∇\nabla is the Euclidean gradient. We work with the contravariant components by σ~ij=gikgjlσ~kl\tilde\sigma^{ij}=g^{ik}g^{jl}\tilde\sigma_{kl}. The stress tensor is intrinsically tangential, σ~∈TS⊗TS\tilde{\boldsymbol\sigma}\in T\mathcal{S}\otimes T\mathcal{S}, i.e. σ~ ⋅ n=0\tilde{\boldsymbol\sigma}\,\mathbin{\mkern-2mu\cdot \mkern-2mu}\,\mathbf{n}=0, since u∈TS\mathbf{u}\in T\mathcal{S} by construction. In the Monge parametrization X=(x,y,h(x,y))\mathbf{X}=(x,y,h(x,y)), the tangent vectors are ex=(1,0,hx)T\mathbf{e}_x=(1,0,h_x)^T and ey=(0,1,hy)T\mathbf{e}_y=(0,1,h_y)^T. From this, the stress in the Cartesian components (the stress embedded in R3\mathbb{R}^3) is computed:

T~=σ~ij ei⊗ej=J σ~ JT,J=[ex  ey].\begin{equation} \tilde{T} = \tilde{\sigma}^{ij}\,\mathbf{e}_i\otimes\mathbf{e}_j = J\,\tilde{\sigma}\,J^T, \qquad J=[\mathbf{e}_x\;\mathbf{e}_y]. \tag*{(13)} \end{equation}

T~\tilde{T} is a microscopic stress. In order to get a macroscopic (hydrodynamic) quantity as in classical continuum mechanics, we average T~\tilde{T} over the unit cell via a smoothing kernel in reciprocal space (see Appendix § 5.3),

T=⟨⟨T~⟩⟩.\begin{equation} T = \langle \langle \tilde{T} \rangle \rangle . \tag*{(14)} \end{equation}

Note that since the Cartesian basis is fixed, the composition of the coarse-graining of each component yields the average of the tensor itself. The coarse-grained tensor is then projected back onto the tangent plane, σij=eiTT  ej\sigma_{ij}=\mathbf{e}_i^T T\;\mathbf{e}_j, and normalized by the lengths of the basis vectors, σi^j^=σij/(∣ei∣∣ej∣)\sigma_{\hat i\hat j}=\sigma_{ij}/(|\mathbf{e}_i||\mathbf{e}_j|), so that the plotted quantity is a physical stress along the unit vectors e^i\hat{\mathbf{e}}_i and is comparable across regions of different slope. For small out-of-plane deformation, averaging does not significantly alter tangentiality. To discuss the results, we also compute the von Mises stress, a scalar quantity that is independent of the basis in which the stress is expressed. It is defined as σVM=3/2τijτij\sigma_{\rm VM} = \sqrt{3/2 \tau_{ij}\tau_{ij}} where τij\tau_{ij} are the components of the deviatoric stress tensor Tdev=T−tr(T)3IdT^{\rm dev}= T - \frac{\text{tr}(T)}{3}Id with IdId the identity in R3\mathbb{R}^3. Since τijτij=tr  (T2)−13(tr  T)2\tau_{ij}\tau_{ij}= \text{tr}\;(T ^2) -\frac{1}{3}(\text{tr}\; T)^2, this yields:

σVM=3  tr  (T2)−(tr  T)22.\sigma_{\rm VM} = \sqrt{\frac{3\; \text{tr}\; (T ^2) -(\text{tr}\; T)^2}{2}} .

Numerics

We consider a rectangular domain (0,Lx)×(0,Ly)(0,L_x)\times(0,L_y) with periodic boundary conditions and appropriate initial conditions for ψ\psi and hh. We solve the model’s equation numerically using a Fourier pseudo-spectral method. The periodic rectangular cell containing two atoms has sides ax=4π/3=aa_x=4\pi/\sqrt{3}=a, with aa the lattice constant, and ay=3 aa_y=\sqrt{3}\,a. The system size is Lx=240 axL_x=240\,a_x and Ly=160 ayL_y=160\,a_y. For discretization, we use a uniform 2048×20482048 \times 2048 grid. The unstrained configuration corresponds to Lx=256 axL_x=256\,a_x, so that the domain is subject to a compressive strain of 6.25%6.25\% along xx (εij=−δixδjx0.0625\varepsilon_{ij}=-\delta_{ix}\delta_{jx}0.0625). We evaluate the nonlinear terms in physical space and transform to and from Fourier space using the fast Fourier transform (FFT), while we treat the linear terms fully in Fourier space, enabling a time-stepping scheme based on an exponential integrator. For details, see [1]. The method is implemented on GPU using CuPy, exploiting GPU-accelerated FFT to efficiently handle the large system sizes and long integration times required to resolve both crystalline defect dynamics and wrinkling pattern coarsening. We use a constant time step Δt=tn−tn−1=1\Delta t = t^n - t^{n-1} = 1 and mobilities Mψ=1M_\psi = 1 and MX=0.1M_{\mathbf{X}} = 0.1. ψ\psi is initialized with ψˉ=0\bar\psi=0 and ϕ0=(C+C2−15  B  D)/(15  D)\phi_0=(C+\sqrt{C^2-15\;B \;D})/(15\;D). We consider A=1,B=−0.2,C=32  0.98,D=1A=1, B=-0.2, C=\frac{ \sqrt{3}}{2\; \sqrt{0.98}}, D=1, corresponding to previous studies [1, 29]. The bending rigidity is β=1\beta=1. In the following, lengths, such as the height hh, are reported in units of aa.

Simulation setup and postprocessing

Starting from a flat film, we add a small random perturbation to the initial height field to allow for out-of-plane deformations, breaking the symmetry in which no out-of-plane deformation occurs for a perfectly flat domain (h(r)=const.h(\mathbf{r})=\text{const.}). We generate the perturbation as Gaussian noise in Fourier space with random amplitudes and phases, imposing zero mean and a fixed rms amplitude of 10−5a10^{-5}a. We choose a small amplitude to avoid biasing the selected wrinkling pattern. We verified with different random seeds that the overall coarsening dynamics are independent of the noise realization, while the positions of the topological defects in the wrinkle arrangement, namely the wrinklons [26], and the coarsening events differ from run to run. An example of a configuration obtained by solving the model’s equations is reported in Figure 1, illustrating the wrinkled thin sheet and its underlying crystal lattice, along with the field ψ\psi and the reconstructed atomic positions. All simulations use the same initial random height perturbation for direct comparison, but we consider different initial conditions for the crystal lattice, including isolated dislocations and grain boundaries. Figure 2 explicitly illustrates some of these configurations.

Figure 1. Modeling framework and sample wrinkling morphology. a) Wrinkling of a thin sheet under uniaxial compression (εxx=−6.25%\varepsilon_{xx} = -6.25 \%, see black arrows). b) Hexagonal lattice in the region marked by the black square in panel a), illustrated by atomic arrangement obtained by reconstructing the peak positions of the density field ψ\psi, see inset. Illustration via OVITO [22].

Figure 2. Stress field projected on U\mathcal{U} (first row) and zoomed-in crystalline structure reconstructed as in Figure 1b) (second row). a) Perfect hexagonal crystal. b) Crystal with Nψ=16N_\psi=16 dislocations. The 5−75-7 dislocation core is highlighted in black. c) Polycrystal. Different grain orientations are color-coded.

The dislocations are inserted by initializing the density via Eq. (11) and imposing a singularity in the phase φi=−qi⋅u\varphi_i=-\mathbf{q}_i \mathbin{\mkern-2mu\cdot \mkern-2mu}\mathbf{u} [27], such that

∮dφi=∮du⋅qi=b⋅qi=2πs,\begin{equation} \oint {\rm d}\varphi_i = \oint {\rm d}\mathbf{u}\mathbin{\mkern-2mu\cdot \mkern-2mu}\mathbf{q}_i=\mathbf{b}\mathbin{\mkern-2mu\cdot \mkern-2mu}\mathbf{q}_i=2\pi s, \tag*{(15)} \end{equation}

for a qi∈{qhex}\mathbf{q}_i\in \{\mathbf{q}_{\rm hex}\}, see Eq. (5), with ss the winding number and b\mathbf{b} the dislocation Burgers vector, thus linked to the lattice symmetry via the scalar product with qi\mathbf{q}_i. This can be realized by imposing φj(r)=s(2π)−1θ(r−r0)\varphi_j(\mathbf{r})=s (2 \pi)^{-1} \theta (\mathbf{r}-\mathbf{r}_0) for a given jj, with r0\mathbf{r}_0 the nominal dislocation core location and s=1s=1, i.e., introducing a perfect dislocation. We always introduce dislocations as dipoles, i.e., pairs with opposite Burgers vectors, to ensure zero total topological charge while still imposing random positions. We impose a minimum distance between all cores to avoid overlaps in the initial condition. The lattice deformation induced by dislocations is then obtained after a first relaxation of the initial condition. We recall that the resulting elastic field agrees with continuum-mechanics predictions [27, 6, 1, 2].

We initialize polycrystalline films by placing Nseeds=40N_{\rm seeds}=40 hexagonal crystal seeds with randomly assigned orientations in the range [0,π/3)[0,\pi/3). Seed positions are randomly sampled within the computational domain, subject to a minimum separation of (dmin⁡=0.8min⁡(Lx,Ly)/Nseedsd_{\min}=0.8\min(L_x,L_y)/\sqrt{N_{\rm seeds}}) from all previously accepted seeds. Each grain is represented by a rotated hexagonal lattice modulated by a Gaussian envelope with width (min⁡(Lx,Ly)/(2Nseeds)\min(L_x,L_y)/(2\sqrt{N_{\rm seeds}})). During subsequent relaxation, the grains grow and eventually impinge on one another, forming grain boundaries.

The uniaxial compression induces out-of-plane deformations, creating a corrugated profile. Wrinkles orient to relax the deformation (decrease the film’s elastic energy), forming aligned stripes perpendicular to the compression axis. Because of the initial Gaussian noise, in the early stages of pattern formation the stripe alignment is not yet fully established and the pattern can display local irregularities including topological defects, i.e., wrinklons [26]. We locate them by demodulating the height field around the dominant wavevector k0\mathbf{k}_0 (see Appendix § 5.2), namely computing z=∣z∣eiζz=|z|e^{i\zeta} such that h−hˉ=zeik0 ⋅ r+c.c.h-\bar{h}=ze^{i\mathbf{k}_0\,\mathbin{\mkern-2mu\cdot \mkern-2mu}\, \mathbf{r}}+{\rm c.c.} with hˉ\bar{h} the average height, and counting its phase windings [40].

w=12π∮dζ  ∈  Z.w=\frac{1}{2\pi}\oint d\zeta\;\in\;\mathbb Z.

The sign of ww gives the wrinklon’s topological charge.

The analysis of the wrinkling dynamics is performed by computing a few representative quantities and studying their evolution over time. They are: the mean wavelength of the wrinkled height profile λˉ(t)\bar{\lambda}(t), the root-mean-square amplitude of the surface profile WRMS(t)W_{\rm RMS}(t) (the roughness), the root-mean-square Gaussian curvature of the surface KRMSK_{\rm RMS}, the mean von Mises stress σˉVM\bar\sigma_{\rm VM}, the number of crystalline defects Nψ(t)N_\psi(t) and the number of wrinklons Nh(t)N_{h}(t). Details concerning their evaluation are reported in the Appendix § 5.2.

Results

We consider uniaxial compression of a thin sheet, as illustrated in Figure 1. We first derive an analytic result, instrumental for discussing the numerical results of the full Surface Phase-Field-Crystal-Helfrich (SPFCH) model. Then, we illustrate and analyze the morphological evolution and deformation of thin crystalline sheets hosting different defect configurations. By comparing them, we derive general trends on the impact of crystalline defects on wrinkling dynamics.

Derivation of the onset wrinkling wavelength

To describe the out-of-plane deformation instability within the one-dimensional continuum framework, we consider the 1D Föppl–von Kármán (FvK) equations for a thin film, which we couple to an overdamped dynamics for the height field hh:

0=ddy[dudy+12(dhdy)2],1Γ∂h∂t=−β∂4h∂y4+(λ+2μ)ddy[(dudy+12(dhdy)2)dhdy],\begin{align} 0 &= \frac{\mathrm{d}}{\mathrm{d}y} \left[ \frac{\mathrm{d}u}{\mathrm{d}y} + \frac{1}{2} \left( \frac{\mathrm{d}h}{\mathrm{d}y} \right)^2 \right], \\ \frac{1}{\Gamma}\frac{\partial h}{\partial t} &= -\beta \frac{\partial^4 h}{\partial y^4} + (\lambda + 2\mu) \frac{\mathrm{d}}{\mathrm{d}y} \left[ \left( \frac{\mathrm{d}u}{\mathrm{d}y} + \frac{1}{2} \left( \frac{\mathrm{d}h}{\mathrm{d}y} \right)^2 \right) \frac{\mathrm{d}h}{\mathrm{d}y} \right], \end{align}

where u(y)u(y) is the in-plane displacement, h(y,t)h(y,t) represents the out-of-plane deflection (height), β\beta is the bending rigidity, 1/Γ1/\Gamma is the mobility coefficient, equivalent to MXM_{\mathbf{X}} and containing the viscosity of the substrate, and λ,μ\lambda, \mu are the Lamé elastic parameters. Following the same computation as [25], under a uniform in-plane compressive (negative) strain ε0<0\varepsilon_0 < 0, we have dudy+12(dhdy)2=−∣ε0∣\frac{\mathrm{d}u}{\mathrm{d}y} + \frac{1}{2} ( \frac{\mathrm{d}h}{\mathrm{d}y} )^2 = -|\varepsilon_0| and the onset wavelength can be determined. Substituting this relation into Eq. eq:fvk_outplane yields the linearized overdamped equation of motion for small out-of-plane perturbations:

∂h∂t=−Γ[β∂4h∂y4+(λ+2μ)∣ε0∣∂2h∂y2].\begin{equation} \frac{\partial h}{\partial t} = -\Gamma \left[ \beta \frac{\partial^4 h}{\partial y^4} + (\lambda + 2\mu) |\varepsilon_0| \frac{\partial^2 h}{\partial y^2} \right]. \tag*{(16)} \end{equation}

By assuming a plane-wave perturbation of the form h(y,t)=h^ eγ(k)teikyh(y, t) = \hat{h} \, e^{\gamma(k) t} e^{i k y}, with γ(k)\gamma(k) growth rate, and inserting in (16) we obtain:

γ(k)=Γ[(λ+2μ)∣ε0∣k2−βk4].\begin{equation} \gamma(k) = \Gamma \left[ (\lambda + 2\mu) |\varepsilon_0| k^2 - \beta k^4 \right]. \tag*{(17)} \end{equation}

The dominant mode k∗k^* corresponding to the maximum growth rate (dγ/dk=0\mathrm{d}\gamma / \mathrm{d}k = 0) is given by:

k∗=(λ+2μ)∣ε0∣2β,\begin{equation} k^* = \sqrt{\frac{(\lambda + 2\mu) |\varepsilon_0|}{2\beta}}, \tag*{(18)} \end{equation}

which defines the characteristic onset wavelength:

λonset=2πk∗=2π2β(λ+2μ)∣ε0∣.\begin{equation} \lambda_{\text{onset}} = \frac{2\pi}{k^*} = 2\pi \sqrt{\frac{2\beta}{(\lambda + 2\mu) |\varepsilon_0|}}. \tag*{(19)} \end{equation}

Without energy dissipation in the substrate, accounted for by the viscosity in the mobility coefficient 1/Γ1/\Gamma, e.g. for Γ→0\Gamma \to 0, the result drastically changes and we obtain the typical buckling of thin sheets under compression [19].

Within the PFC model framework, and assuming small deformation (linear elasticity), the effective in-plane elastic modulus for a defect-free hexagonal crystal is (λ+2μ)=9Aϕ02(\lambda + 2\mu) = 9 A \phi_0^2. AA is the coefficient in (1) and ϕ0\phi_0 is the (real) equilibrium amplitude of the lattice in (4). Nonlinearities lead to a (small) correction to the elastic modulus of (1−54ε0)\left(1- \frac{5}{4}\varepsilon_0\right) as discussed in [1]. Substituting this elasticity relation into Eq. (19) yields the explicit onset wavelength as a function of the simulation parameters:

λonset=2π2β9Aϕ02(1−54ε0) ∣ε0∣≈8.36.\begin{equation} \lambda_{\text{onset}} = 2\pi \sqrt{\frac{2\beta}{9A \phi_0^2 \left(1-\frac{5}{4}\varepsilon_0\right) \, |\varepsilon_0|} } \approx 8.36. \tag*{(20)} \end{equation}

This provides a simple validation criterion for the numerical simulations based on the Surface Phase-Field-Crystal-Helfrich (SPFCH) model.

Numerical Simulations

Besides the examples in Figure 2, namely the sheets with Nψ=0N_\psi = 0 and Nψ=16N_\psi = 16 dislocations, as well as a polycrystalline sheet, we also consider Nψ=8,60,100N_\psi= 8, 60, 100 to inspect various distributions and degrees of plastic relaxation. The polycrystalline case serves as a representative case, featuring both small- and large-angle grain boundaries.

Figure 3. Height profile hh at representative stages: t=5×104,1×105,5×105,5×106,1×107t= 5 \times 10 ^4, 1 \times10^5, 5 \times 10^5, 5 \times 10^6, 1 \times 10^7. a) Perfect crystal, b)-e) Crystal with isolated dislocations with Nψ=8,16,60,100N_\psi = 8, 16, 60, 100, respectively. f) Polycrystal.

Figure 4. Stress field σxx\sigma_{xx} at representative stages: t=5×104,1×105,5×105,5×106,1×107t= 5 \times 10 ^4, 1 \times10^5, 5 \times 10^5, 5 \times 10^6, 1 \times 10^7. a) Perfect crystal, b)-e) Crystal with isolated dislocations with Nψ=8,16,60,100N_\psi = 8, 16, 60, 100, respectively. f) Polycrystal.

The considered examples show that different distributions and types of crystalline defects (dislocations and grain boundaries) introduce different deformation fields and affect wrinkling patterns and their dynamics differently. Figure 3 illustrates the morphological evolution of thin sheets hosting such distributions over long times through representative stages. The wrinkling instability is associated with the development of a pattern featuring wrinklons both in the presence and absence of initial crystalline defects. Without dislocations or grain boundaries, a regular pattern develops, eventually leading after coarsening to the buckled state with a wavelength equal to the domain size LxL_x (not shown). As defect density increases, we observe the nucleation of more wrinklons, while the emerging height-profile pattern becomes more irregular. With high crystalline dislocation density, the wrinkle crests also strongly deviate from directions perpendicular to the uniaxial deformation, with persistent irregularities. Figure 4 illustrates the evolution of thin-sheet stress field σxx\sigma_{xx} at the same stages shown in Figure 3. In all cases, the initial compression relaxes. Higher defect density leads to larger stress variations in the crystal between dislocations, which in turn increases the wrinklon density.

While Figure 3 and Figure 4 clearly indicate the increased irregularity in the height field and its relation to crystalline defects, they do not reveal the microscopic mechanisms mediating complex morphological evolution. To better understand them, Figure 5 reports a small intermediate time sequence of the case with initially Nψ=16N_\psi=16 dislocations, corresponding to Figure 3 b) and Figure 4 b), focusing on the annihilation of wrinklons. Wrinklons with positive and negative charges (w=+1w=+1 red circles w=−1w=-1 white triangles) are illustrated. The total topological charge ∑iwi=0\sum_i w_i=0 vanishes at each time instance, and the number of wrinklons decreases over time due to these annihilation events. The evolution of the height profile reveals a progressive coarsening of the wrinkling pattern, driven by the migration and subsequent annihilation of wrinklon pairs, as illustrated in Figure 5. The wrinklons are also identifiable in the corresponding stress profiles, where they give rise to pronounced localized stress variations (Figure 5b). These regions constrain out-of-plane deformation, hindering stress relaxation by forming an ideal wrinkling pattern.

Figure 5. Wrinkling dynamics for the case with Nψ=16N_\psi = 16 initial dislocations. Representative stages are selected. a) Height profile hh, wrinklons with w=+1w=+1 are marked with red circles and with w=−1w=-1 with white triangles. b) Stress field of configurations in a) projected on U\mathcal{U} together with winklons positions.

Figure 6. Example of dislocation reaction from the case with Nψ=16N_\psi = 16 initial dislocations. Representative stages are selected. a) Stress field projected on U\mathcal{U}. b) Hexagonal lattice in the white squares of panel a) and dislocation cores illustrated as in Figure 2b).

As evident from Figure 5, crystalline defects such as dislocations also migrate and interact as the system evolves. They also cause localized perturbations in the height profiles [1, 29, 41]. In Figure 6, an early intermediate time sequence of the case with initially Nψ=16N_\psi=16 dislocations, corresponding to Figure 4 b), is reported. We concentrate on the interaction of two dislocations. In the case illustrated there, they approach and react, forming a new dislocation with a different orientation. Annihilation also occurs when two dislocations with equal and opposite Burgers vectors react. We do not report the height field at the dislocation cores, as the out-of-plane deformation is negligible compared to the wrinkling. Detailed investigations of out-of-plane deformations of isolated dislocations using the Surface Phase-Field-Crystal-Helfrich (SPFCH) model are available in [29].

The dynamics illustrated above show that the interplay of these processes, namely the migration and annihilation of wrinklons as well as the migration and annihilation/reaction of crystalline defects, affects the wrinkling dynamics. In order to quantify and characterize related effects, we compute a few representative quantities (detailed in the Appendix § 5.2) and study their evolution over time, see Figure 7.

Figure 7. Analysis of the coarsening dynamics. a) Average wavelength λˉ(t)\bar{\lambda}(t), together with theoretical wavelength from Eq. (20) and t1/4t^{1/4} power law. b) Root-mean-square amplitude WRMS(t)W_{\rm RMS}(t), together with t1/4t^{1/4} power law. c) Root mean square Gaussian curvature KRMS(t)K_{\rm RMS}(t). d) Mean von Mises stress σˉVM(t)\bar{\sigma}_{\rm VM}(t). e) Number of crystalline defects N(t)N(t). f) Number of wrinklons, detected with winding number ±1\pm 1, Nh(t)N_{h}(t). The cases correspond to the wrinkling dynamics reported in Figure 3 and Figure 4.

For the defect-free sheet, the onset wavelength for wrinkling agrees with the theoretical prediction in Eq. (20). With crystalline defects, the same process occurs, but we observe minor deviations from this wavelength. Furthermore, after an initial transient, the wavelength coarsening dynamics follow the theoretically predicted t1/4t^{1/4} scaling [25] across all considered systems, regardless of the density or nature of the crystalline defects. This behavior is illustrated by the black dashed line in Figure 7 a). These findings indicate that crystalline defects influence the initial wrinkling wavelength but do not significantly alter its power-law scaling over time.

The surface roughness WRMSW_{\rm RMS} likewise exhibits t1/4t^{1/4} power-law scaling across all considered systems, consistent with [25]. This behavior is illustrated in Figure 7 b), where the black dashed line represents this scaling. Notably, although the scaling exponent remains unchanged, the onset of this regime is progressively delayed with increasing defect density, with the most pronounced delay observed in the polycrystalline system. These results suggest that crystallographic defects partially hinder wrinkle development by delaying amplitude growth, without altering the underlying scaling behavior observed above.

The root-mean-square Gaussian curvature KRMSK_{\rm RMS} in Figure 7 c) exhibits a qualitatively similar evolution across all considered systems during the initial growth of the wrinkle amplitudes, with only minor differences. At later stages, however, this quantity decreases markedly in the defect-free crystal, whereas it decreases much less in the presence of crystalline defects. In the latter cases, the decrease is at least one order of magnitude smaller than in the defect-free system and becomes progressively more pronounced as the defect density decreases.

Stress and topological defects evolve differently across all considered cases. The mean von Mises stress, σˉVM\bar{\sigma}_{\rm VM}, is shown in Figure 7 d). It initially decreases, with a more pronounced relaxation observed at higher defect densities. This behavior can be attributed to defect rearrangements and the associated plastic flow, which are absent in the perfect crystal.

The initial relaxation is also accompanied by defect annihilation, as shown in Figure 7 e). As wrinkling enters the coarsening regime, out-of-plane deformation becomes the dominant relaxation mechanism, leading to a marked change in the rate of stress reduction. The stress change in this phase is more pronounced at lower defect densities, indicating more efficient out-of-plane deformation in those cases. At later stages, when the wrinkles reach larger wavelengths, the remaining crystalline defects increasingly determine the residual stress. Consequently, σˉVM\bar{\sigma}_{\rm VM} increases with defect density, consistent with the dislocation populations shown in Figure 7 e).

A particularly notable difference appears in the evolution of the number of wrinklons, shown in Figure 7 f). This number increases with dislocation density and reaches its highest value in the polycrystalline system. Crystallographic defects are then found to trigger these defects in the height profile. This also explains the mechanism behind the delays in the growth of the wrinkled profile amplitude and related associated stress release: see the number of wrinklons and the change in the slope of σˉVM(t)\bar{\sigma}_{\rm VM}(t) and WRMS(t)W_{\rm RMS}(t). Note also that, at late times, the drop of σˉVM(t)\bar{\sigma}_{\rm VM}(t) for the case of a perfect crystal without dislocations can be associated with the formation of perfectly aligned wrinkles (no wrinklons), see Figure 7 f).

Conclusions

Wrinkling dynamics of thin sheets under uniaxial compression has previously been explored by classical elasticity theory, identifying two regimes: a stress-driven instability with an initial wrinkle wavelength much shorter than the equilibrium value, corresponding to buckling of the entire film, and wrinkle coarsening until they reach an equilibrium wavelength. Averaged quantities like the initial wavelength and the scaling exponent for the coarsening process depend on the considered material, but seem to be universal with respect to the strength of uniaxial compression upon rescaling. Here, we explored how emerging wrinkling patterns and their coarsening depend on the material’s crystalline microstructure, particularly on the presence of topological defects of different kinds.

Thanks to the Surface Phase-Field-Crystal-Helfrich (SPFCH model, we can capture the coupling between in-plane and out-of-plane relaxation mechanisms. This coupling is central to the physics at play, as it allows the system to redistribute imposed compressive stress between two competing channels: in-plane relaxation, mediated by defect migration/annihilation and grain boundary motion within the crystalline lattice, and out-of-plane relaxation, mediated by bending and the emergence of a wrinkling pattern. A key strength of the model is its ability to simultaneously investigate two distinct classes of topological defects and their interplay: dislocations and grain boundaries in the crystalline order, described by ψ\psi, and wrinklons in the out-of-plane wrinkling pattern, described by hh.

Our results show that the emergence and evolution of the wrinkling pattern are influenced by the underlying microstructure of the crystal. Crystalline defects are found to slightly affect the wrinkling onset wavelength and, more significantly, the growth of the emerging wrinkles. Still, they play a negligible role in the scaling law for the increase in wrinkle wavelength during coarsening over time. Higher crystalline defect density leads to the nucleation of more wrinklons. This is also associated with more irregular and persistent wrinkling patterns and less efficient stress relaxation at later stages. These findings highlight that crystalline defects primarily control the morphology and relaxation efficiency of wrinkled thin sheets rather than their underlying coarsening laws.

We remark that our results are consistent with, and to some extent reproduce, predictions of classical elasticity theory for wrinkling and coarsening [32, 25]. However, for various applications such as strain sensors, functional surfaces, flexible electronics, and energy storage [35], the observed differences in wrinkling patterns, the presence of wrinklons, and the overall long-term morphological evolution may play a crucial role in determining the mechanical and functional properties of thin films. These aspects are found to be crucially coupled with crystalline microstructure, with additional phenomenology that can be relevant for optimization and novel design concepts. Extending these findings to specific materials and such applications would require material-specific parametrizations and calibration against experimental conditions. Nevertheless, the general mechanisms and trends identified in this work provide a foundation for in-depth understanding and potentially controlling the morphological evolution of crystalline thin films.

Acknowledgments

E.R. and A.V. acknowledge support from DurAMat under the European Union - Marie Skłodowska-Curie Actions Doctoral Networks (ITN) Call H2020-MSCA-DN-2022, grant no. 101119767. M.S. acknowledges support from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), project numbers 417223351 (FOR3013). The authors gratefully acknowledge the computing time made available to them on the high-performance computer at the NHR Center of TU Dresden. This center is jointly supported by the Federal Ministry of Research, Technology and Space of Germany and the state governments participating in the NHR.

Appendix

Height formalism

In order to formulate the equations in local coordinates (x,y)∈U⊆R2(x,y)\in\mathcal{U}\subseteq\mathbb{R}^2 use:

gij=δij+∂ih  ∂jh,∣g∣=det⁡(gij)=1+∣∇h∣2,gij=δij−∂ih  ∂jh∣g∣,g_{ij}=\delta_{ij}+\partial_i h \; \partial_j h,\qquad |\mathbf{g}|=\det(g_{ij})=1+|\nabla h|^2,\qquad g^{ij}=\delta^{ij}-\frac{\partial_i h \; \partial_j h}{|\mathbf{g}|},
H=1∣g∣(Δh−⟨∇h,(∇2h) ∇h⟩∣g∣),K=det⁡(∇2h)∣g∣2,\mathcal{H} = \frac{1}{\sqrt{|\mathbf{g}|}}\left( \Delta h - \frac{\langle \nabla h, (\nabla^2 h)\,\nabla h \rangle}{|\mathbf{g}|} \right), \qquad K = \frac{\det(\nabla^2 h)}{|\mathbf{g}|^2},
⟨∇Sf,∇Sℓ⟩S=⟨∇f,∇ℓ⟩−⟨∇h,∇f⟩⟨∇h,∇ℓ⟩∣g∣,\langle \nabla_S f, \nabla_S \ell \rangle_S = \langle \nabla f , \nabla \ell \rangle - \frac{\langle \nabla h, \nabla f \rangle \langle \nabla h, \nabla \ell \rangle}{|\mathbf{g}|},
ΔSf=Δf−⟨∇h,∇f⟩∣g∣Δh−⟨∇h,(∇2f) ∇h⟩∣g∣+⟨∇h,(∇2h) ∇h⟩⟨∇h,∇f⟩∣g∣2.\Delta_S f = \Delta f - \frac{\langle \nabla h, \nabla f \rangle}{|\mathbf{g}|} \Delta h - \frac{\langle \nabla h, (\nabla^2 f)\,\nabla h \rangle}{|\mathbf{g}|} + \frac{\langle \nabla h, (\nabla^2 h)\,\nabla h \rangle \langle \nabla h, \nabla f \rangle}{|\mathbf{g}|^2}.
∇Sf=∇f−⟨∇h,∇f⟩∣g∣ ∇h,∇S2f=∇2f−⟨∇h,∇f⟩∣g∣ ∇2h.\nabla_S f = \nabla f-\frac{\langle \nabla h,\nabla f\rangle}{|\mathbf{g}|}\,\nabla h ,\qquad \nabla_S^2 f = \nabla^2 f-\frac{\langle \nabla h,\nabla f\rangle}{|\mathbf{g}|}\, \nabla^2 h .
II=∇2h∣g∣,(ΠQSII)=II−H2 IdTS\textit{II} = \frac{ \nabla^2 h}{\sqrt{|\mathbf{g}|}}, \qquad (\Pi_{QS}\textit{II}) = \textit{II} - \frac{\mathcal{H}}{2}\,\mathrm{Id}_{TS}

For f,ℓ:U→Rf, \ell : \mathcal{U} \to \mathbb{R} and IdTS\mathrm{Id}_{TS} is the identity operator in the tangential space, ∣g∣\sqrt{|\mathbf{g}|} is the area element. Here ∇\nabla, ∇2\nabla^2, Δ\Delta and ⟨⋅,⋅⟩\langle\mathbin{\mkern-2mu\cdot \mkern-2mu},\mathbin{\mkern-2mu\cdot \mkern-2mu}\rangle (no subscript SS) are the Euclidean operators in (x,y)(x,y).

Quantitative measures of wrinkling evolution

Let hijh_{ij} denote hh on the Nx×NyN_x\times N_y grid, hˉ=1NxNy∑i,jhij\bar h=\frac{1}{N_xN_y}\sum_{i,j}h_{ij}, and h^(k)\hat h(\mathbf k) the discrete Fourier transform of h−hˉh-\bar h. In Figure 7 are reported the roughness, WRMS(t)W_{\rm RMS}(t), the power-weighted RMS wavelength λˉ(t)\bar\lambda(t) in agreement with [25].

WRMS(t)=1NxNy∑i,j(hij−hˉ)2,kˉ(t)=∑k∣h^(k)∣2 ∣k∣2∑k∣h^(k)∣2,λˉ(t)=2πkˉ(t),\begin{align} W_{\rm RMS}(t) &= \sqrt{\frac{1}{N_xN_y}\sum_{i,j}\left(h_{ij}-\bar h\right)^2}, \tag*{(21)} \\ \bar k(t) &= \sqrt{\frac{\sum_{\mathbf k}|\hat h(\mathbf k)|^2\,|\mathbf k|^2} {\sum_{\mathbf k}|\hat h(\mathbf k)|^2}}, \qquad \bar\lambda(t)=\frac{2\pi}{\bar k(t)}, \tag*{(22)} \end{align}

Gaussian curvature and stress are also reported. Since they are scalar fields on S\mathcal S, they are averaged with respect to the surface measure dA=∣g∣ dx dy=1+∣∇h∣2 dx dyd\mathcal{A}=\sqrt{|\mathbf g|}\,dx\,dy=\sqrt{1+|\nabla h|^2}\,dx\,dy, with A=∫SdA\mathcal{A}=\int_{\mathcal S}d\mathcal{A}:

KRMS=1A∫SK2 dA,σˉVM=1A∫SσVM dA.\begin{equation} K_{\rm RMS}=\sqrt{\frac{1}{\mathcal{A}}\int_{\mathcal S}K^2\,d\mathcal{A}}, \qquad \bar\sigma_{\rm VM}=\frac{1}{\mathcal{A}}\int_{\mathcal S}\sigma_{\rm VM}\,d\mathcal{A} . \tag*{(23)} \end{equation}

To compute the number of wrinklons, we start by writing h−hˉh- \bar h as a slowly modulated plane wave, (h−hˉ)(r)≃z(r) eik0⋅r+c.c.(h- \bar h) (\mathbf{r})\simeq z (\mathbf{r})\,e^{i\mathbf k_0\mathbin{\mkern-2mu\cdot \mkern-2mu}\mathbf r}+\mathrm{c.c.}, where k0\mathbf{k_0} is the dominant peak of the pattern spectrum. k0\mathbf{k_0} is shifted to the origin of Fourier space, and a Gaussian filter of width ∣k0∣|\mathbf{k_0}| suppresses the conjugate band at −2k0-2\mathbf{k_0}.

z=F−1 ⁣[F[(h−hˉ) e−ik0⋅r] e−∣k∣2/2(∣k0∣)2],z=\mathcal F^{-1}\!\Big[\mathcal F\big[(h-\bar h)\,e^{-i\mathbf k_0\mathbin{\mkern-2mu\cdot \mkern-2mu}\mathbf r}\big]\,e^{-|\mathbf k|^2/2(|\mathbf k_0|)^2}\Big],

with F[f(r)]=f^(k)F\big[f(\mathbf{r})\big]=\hat f\mathbf{(k)} and F−1 ⁣[f^(k)]=f(r)F^{-1}\!\big[ \hat f\mathbf{(k)} \big]=f(\mathbf{r)}. The modulus of zz, ∣z∣|z|, is the local pattern amplitude and the argument, ζ=arg⁡z\zeta=\arg z, is the local phase. At a dislocation core, ∣z∣→0|z|\to0 and the phase winds by a multiple of 2π2\pi. Following a similar approach as [39], we sum the wrapped phase differences wrap(Δζ)∈(−π,π]\mathrm{wrap}(\Delta\zeta)\in(-\pi,\pi] around each elementary 2×22\times2 pixels plaquette of the grid (with periodic boundaries),

w=12π∮dζ∈Z,w=\frac{1}{2\pi}\oint d\zeta\in\mathbb Z,

and flag every plaquette with w≠0w\neq0 as a defect, with the sign of ww giving its topological charge. Defects in the density field ψ\psi are identified through a Delaunay triangulation of the local minima of ψ\psi, each of which corresponds to a lattice cell (the maxima being the lattice sites). In a perfect lattice, every minimum has coordination number 66 (six nearest neighbours), while dislocations are detected as bound 5−75 - 7 pairs. Periodic boundary conditions are taken into account.

Coarse-graining

The coarse-graining procedure is obtained by averaging a microscopic quantity f~\tilde{f} over the unit cell with a Gaussian smoothing kernel in reciprocal space to get its macroscopic counterpart ff

f=⟨⟨f~⟩⟩=F−1 ⁣[G^(k) F[f~](k)],G^(k)=exp⁡ ⁣[−12(ax2kx2+ay2ky2)],\begin{equation} f = \langle\langle \tilde{f} \rangle\rangle = \mathcal{F}^{-1}\!\left[\hat{G}(\mathbf{k})\,\mathcal{F}[\tilde{f}](\mathbf{k})\right], \qquad \hat{G}(\mathbf{k}) = \exp\!\left[-\tfrac{1}{2}\left(a_x^2 k_x^2 + a_y^2 k_y^2\right)\right], \tag*{(24)} \end{equation}

where ax=4π/3a_x=4\pi/\sqrt{3} and ay=4πa_y=4\pi.

References

  1. [1]Benoit–Maréchal, L., Nitschke, I., Voigt, A., & Salvalaglio, M. (2024). Mesoscale modeling of deformations and defects in thin crystalline sheets. Mechanics of Materials, 198, 105114. https://doi.org/10.1016/j.mechmat.2024.105114
  2. [2]Benoit-Maréchal, L., & Salvalaglio, M. (2024). Gradient elasticity in Swift–Hohenberg and phase-field crystal models. Model. Simul. Mater. Sci. Eng., 32(5), 055005. https://doi.org/10.1088/1361-651X/ad42bb
  3. [3]Hüter, C., Friák, M., Weikamp, M., Neugebauer, J., Goldenfeld, N., Svendsen, B., & Spatschek, R. (2016). Nonlinear elastic effects in phase field crystal and amplitude equations: Comparison to ab initio simulations of bcc metals and graphene. Phys. Rev. B, 93, 214105. https://doi.org/10.1103/PhysRevB.93.214105
  4. [4]Trautt, Z., Adland, A., Karma, A., & Mishin, Y. (2012). Coupled motion of asymmetrical tilt grain boundaries: Molecular dynamics and phase field crystal simulations. Acta Mater., 60(19), 6528–6546. https://doi.org/10.1016/j.actamat.2012.08.018
  5. [5]Qiu, C., Punke, M., Tian, Y., Han, Y., Wang, S., Su, Y., Salvalaglio, M., Pan, X., Srolovitz, D. J., & Han, J. (2024). Grain boundaries are Brownian ratchets. Science, 385(6712), 980–985. https://doi.org/10.1126/science.adp1516
  6. [6]Skogvoll, V., Angheluta, L., Skaugen, A., Salvalaglio, M., & Viñals, J. (2022). A phase field crystal theory of the kinematics of dislocation lines. J. Mech. Phys. Solids, 166, 104932. https://doi.org/10.1016/j.jmps.2022.104932
  7. [7]Berry, J., Provatas, N., Rottler, J., & Sinclair, C. W. (2014). Phase field crystal modeling as a unified atomistic approach to defect dynamics. Phys. Rev. B, 89, 214117. https://doi.org/10.1103/PhysRevB.89.214117
  8. [8]Skogvoll, V., Skaugen, A., & Angheluta, L. (2021). Stress in ordered systems: Ginzburg-Landau-type density field theory. Physical Review B, 103, 224107. https://doi.org/10.1103/PhysRevB.103.224107
  9. [9]Salvalaglio, M., Voigt, A., & Elder, K. R. (2019). Closing the gap between atomic-scale lattice deformations and continuum elasticity. Npj Computational Materials, 5(1), 48.
  10. [10]Salvalaglio, M., Angheluta, L., Huang, Z.-F., Voigt, A., Elder, K. R., & Viñals, J. (2020). A coarse-grained phase-field crystal model of plastic motion. Journal of the Mechanics and Physics of Solids, 137, 103856.
  11. [11]Elder, K. R., Katakowski, M., Haataja, M., & Grant, M. (2002). Modeling Elasticity in Crystal Growth. Physical Review Letters, 88(24), 245701. https://doi.org/10.1103/PhysRevLett.88.245701
  12. [12]Elder, K. R., & Grant, M. (2004). Modeling Elastic and Plastic Deformations in Nonequilibrium Processing Using Phase Field Crystals. Physical Review E, 70(5), 051605. https://doi.org/10.1103/PhysRevE.70.051605
  13. [13]Elder, K. R., Provatas, N., Berry, J., Stefanovic, P., & Grant, M. (2007). Phase-Field Crystal Modeling and Classical Density Functional Theory of Freezing. Physical Review B, 75(6), 064107. https://doi.org/10.1103/PhysRevB.75.064107
  14. [14]van Teeffelen, S., Backofen, R., Voigt, A., & Löwen, H. (2009). Derivation of the phase-field-crystal model for colloidal solidification. Physical Review E, 79(5), 051404. https://doi.org/10.1103/PhysRevE.79.051404
  15. [15]Heinonen, V., Achim, C. V., Elder, K. R., Buyukdagli, S., & Ala-Nissila, T. (2014). Phase-Field-Crystal Models and Mechanical Equilibrium. Physical Review E, 89(3), 032411. https://doi.org/10.1103/PhysRevE.89.032411
  16. [16]Elder, K. R., Achim, C. V., Heinonen, V., Granato, E., Ying, S. C., & Ala-Nissila, T. (2021). Modeling buckling and topological defects in stacked two-dimensional layers of graphene and hexagonal boron nitride. Physical Review Materials, 5, 034004. https://doi.org/10.1103/PhysRevMaterials.5.034004
  17. [17]Backofen, R., Voigt, A., & Witkowski, T. (2010). Particles on curved surfaces: A dynamic approach by a phase-field-crystal model. Physical Review E, 81, 025701. https://doi.org/10.1103/PhysRevE.81.025701
  18. [18]Köhler, C., Backofen, R., & Voigt, A. (2016). Stress Induced Branching of Growing Crystals on Curved Surfaces. Physical Review Letters, 116(13), 135502. https://doi.org/10.1103/PhysRevLett.116.135502
  19. [19]Seung, H. S., & Nelson, D. R. (1988). Defects in Flexible Membranes with Crystalline Order. Physical Review A, 38(2), 1005–1018. https://doi.org/10.1103/PhysRevA.38.1005
  20. [20]Backofen, R., Gräf, M., Potts, D., Praetorius, S., Voigt, A., & Witkowski, T. (2011). A Continuous Approach to Discrete Ordering on $ S^2$. Multiscale Modeling & Simulation, 9(1), 314–334. https://doi.org/10.1137/100787532
  21. [21]Nitschke, I., & Voigt, A. (2026). Scalar Truesdell time derivative and $(L^2$,$H^-1)$-surface gradient flows. arXiv:2604.08186. https://arxiv.org/abs/2604.08186
  22. [22]Stukowski, A. (2009). Visualization and analysis of atomistic simulation data with OVITO–the Open Visualization Tool. Modelling and Simulation in Materials Science and Engineering, 18(1), 015012. https://doi.org/10.1088/0965-0393/18/1/015012
  23. [23]Seung, H. S., & Nelson, D. R. (1988). Defects in flexible membranes with crystalline order. Phys. Rev. A, 38, 1005–1018. https://doi.org/10.1103/PhysRevA.38.1005
  24. [24]Fei, C., Mao, S., Yan, J., Alert, R., Stone, H. A., Bassler, B. L., Wingreen, N. S., & Košmrlj, A. (2020). Nonuniform growth and surface friction determine bacterial biofilm morphology on soft substrates. Proceedings of the National Academy of Sciences, 117(14), 7622–7632. https://doi.org/10.1073/pnas.1919607117
  25. [25]Huang, R., & Im, S. H. (2006). Dynamics of wrinkle growth and coarsening in stressed thin films. Physical Review E, 74(2), 026214. https://doi.org/10.1103/PhysRevE.74.026214
  26. [26]Vandeparre, H., Piñeirua, M., Brau, F., Roman, B., Bico, J., Gay, C., Bao, W., Lau, C. N., Reis, P. M., & Damman, P. (2011). Wrinkling Hierarchy in Constrained Thin Sheets from Suspended Graphene to Curtains. Physical Review Letters, 106(22), 224301. https://doi.org/10.1103/PhysRevLett.106.224301
  27. [27]Skaugen, A., Angheluta, L., & Viñals, J. (2018). Dislocation dynamics and crystal plasticity in the phase-field crystal model. Phys. Rev. B, 97, 054113. https://doi.org/10.1103/PhysRevB.97.054113
  28. [28]Salvalaglio, M., & Elder, K. R. (2022). Coarse-grained modeling of crystals by the amplitude expansion of the phase-field crystal model: an overview. Modelling and Simulation in Materials Science and Engineering, 30(5), 053001. https://doi.org/10.1088/1361-651X/ac681e
  29. [29]Radice, E., Nitschke, I., Salvalaglio, M., & Voigt, A. (2026). Surface Phase-Field-Crystal-Helfrich model for out-of-plane deformations in thin crystalline sheets with lattice mismatch. arXiv:2607.12997. https://arxiv.org/abs/2607.12997
  30. [30]Emmerich, H., Löwen, H., Wittkowski, R., Gruhn, T., Tóth, G. I., Tegze, G., & Gránásy, L. (2012). Phase-field-crystal models for condensed matter dynamics on atomic length and diffusive time scales: an overview. Advances in Physics, 61(6), 665–743.
  31. [31]Draux, A., Rizzo, M., Vella, D., Démery, V., Brau, F., & Damman, P. (2026). Universal wrinkling dynamics of a sheet on viscous liquid. arXiv:2601.11682. https://arxiv.org/abs/2601.11682
  32. [32]Im, S. H., & Huang, R. (2005). Evolution of Wrinkles in Elastic-Viscoelastic Bilayer Thin Films. Journal of Applied Mechanics, 72(6), 955–961. https://doi.org/10.1115/1.2043191
  33. [33]Vandeparre, H., Gabriele, S., Brau, F., Gay, C., Parker, K. K., & Damman, P. (2010). Hierarchical wrinkling patterns. Soft Matter, 6(22), 5751–5756. https://doi.org/10.1039/c0sm00394h
  34. [34]Kodio, O., Griffiths, I. M., & Vella, D. (2017). Lubricated wrinkles: Imposed constraints affect the dynamics of wrinkle coarsening. Phys. Rev. Fluids, 2, 014202. https://doi.org/10.1103/PhysRevFluids.2.014202
  35. [35]Chen, W., Gui, X., Yang, L., Zhu, H., & Tang, Z. (2019). Wrinkling of two-dimensional materials: methods, properties and applications. Nanoscale Horizons, 4(2), 291–320. https://doi.org/10.1039/c8nh00112j
  36. [36]Khang, D.-Y., Jiang, H., Huang, Y., & Rogers, J. A. (2006). A Stretchable Form of Single-Crystal Silicon for High-Performance Electronics on Rubber Substrates. Science, 311(5758), 208–212. https://doi.org/10.1126/science.1121401
  37. [37]Rogers, J. A., Someya, T., & Huang, Y. (2010). Materials and Mechanics for Stretchable Electronics. Science, 327(5973), 1603–1607. https://doi.org/10.1126/science.1182383
  38. [38]Kim, J. B., Kim, P., Pégard, N. C., Oh, S. J., Kagan, C. R., Fleischer, J. W., Stone, H. A., & Loo, Y.-L. (2012). Wrinkles and deep folds as photonic structures in photovoltaics. Nature Photonics, 6(5), 327–332. https://doi.org/10.1038/nphoton.2012.70
  39. [39]Happel, L., Oberschelp, G., Richter, A., Rode, G. R., Grudtsyna, V., Doostmohammadi, A., & Voigt, A. (2026). A spectrum of p-atic symmetries and defects in confluent epithelia. Soft Matter, 22(5), 1109–1129. https://doi.org/10.1039/d5sm01010a
  40. [40]Skogvoll, V., Rønning, J., Salvalaglio, M., & Angheluta, L. (2023). A unified field theory of topological defects and non-linear local excitations. Npj Computational Materials, 9.
  41. [41]De Donno, M., Angheluta, L., & Salvalaglio, M. (2026). Dislocation dynamics on deformable surfaces. Proc. Royal Soc. A, 482(2347), 20260141. https://doi.org/10.1098/rspa.2026.0141

Paper details

vlnc:W_s2fc2wjt
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 14

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents