D. Raabe, Z. Zhao, F. Roters — Scripta Materialia 50 (2004) 1085–1090

1. Background and Scientific Motivation

Cube-oriented face-centred cubic (FCC) metals — particularly aluminium and copper alloys — develop the {001}〈100〉 cube texture component prominently after recrystallisation. During subsequent cold rolling under plane strain compression, the cube orientation is known to be kinematically metastable: it sits at a saddle point of the orientation space energy landscape, such that small perturbations in either the initial crystallographic orientation or the mechanical boundary conditions can drive substantially different lattice rotation paths. Depending on these perturbations, the crystal can re-orient by rotating about the normal direction (ND), the rolling direction (RD), or the transverse direction (TD).

This sensitivity has been documented experimentally by numerous groups over several decades, yet no single model had succeeded in reconciling the full range of observations. Some early models predicted ND and RD rotations; later studies by Hansen, Driver, and co-workers favoured TD rotations. The discrepancy pointed to an intrinsic sensitivity of the cube orientation to fine details of the deformation state — specifically to the initial orientation spread within a grain and to the friction conditions imposed by the tooling. A model capable of capturing this sensitivity needed to resolve both the spatial heterogeneity of orientation within a grain and the effect of surface friction on the local stress state.

The present study addresses precisely these two factors using a newly formulated texture component crystal plasticity finite element method (TCCP-FEM). The goal is to explain, from first principles, under which starting and boundary conditions cube crystals rotate about ND, RD, or TD, and to quantify how sensitively the re-orientation path depends on each factor.

2. Theoretical Framework: Texture Component Crystal Plasticity FEM

2.1 Orientation Component Representation

Classical crystal plasticity finite element (CPFE) methods assign a single discrete orientation to each integration point. This works well for coarse polycrystal simulations but cannot efficiently represent the continuous, smooth orientation distribution that characterises a real single crystal or a textured grain. The authors introduce an approach in which the initial crystallographic orientation distribution of a grain is mapped onto the Gauss points of a finite element mesh in a compact, mathematically consistent form.

The key construct is the orientation component method: the orientation distribution function (ODF) of each crystallographic component is approximated as a spherical Gaussian — a central standard function — defined by a centre orientation gcgc and a full width at half maximum (FWHM) that quantifies the orientation scatter. For the cube component, the centre is gc=(φ1=0°,Φ=0°,φ2=0°)gc=(φ1=0°,Φ=0°,φ2=0°) in Bunge–Euler notation, and the FWHM used in this study is 2.5°. This compact Gauss function is then discretised: a set of single orientations is sampled from the distribution and assigned — in random lateral order — to the individual integration points of the mesh, such that the ensemble of integration point orientations reproduces the prescribed spherical Gaussian.

This two-step mapping procedure allows the initial texture scatter to be represented both in real space (spatially distributed across the mesh) and in orientation space (as a statistical distribution), simultaneously. Critically, once the simulation begins, each integration point evolves its orientation independently according to the local stress and strain-rate state — the orientation component concept is used only to initialise the simulation, not to constrain it during deformation.

2.2 Crystal Plasticity Constitutive Model

The constitutive framework follows the implicit viscoplastic crystal plasticity formulation of Kalidindi, Bronkhorst, and Anand (1992). Plastic deformation is carried by slip on the twelve {111}〈110〉 slip systems of the FCC lattice. The resolved shear stress on each slip system is compared to the current slip resistance via a power-law flow rule, and the slip resistances evolve according to a hardening law. The lattice rotation at each integration point is updated incrementally from the plastic velocity gradient, providing a continuous tracking of the local crystallographic re-orientation throughout the deformation history.

2.3 Finite Element Model Assembly

The finite element model consists of 500 three-dimensional linear hexahedral elements, each with eight integration points (4,000 integration points in total). The geometry represents a rectangular specimen deformed in plane strain compression in a channel die configuration. Four distinct rigid surfaces define the boundary conditions: surfaces 1 and 2 are the top and bottom compression surfaces (contact with the die platens), and surfaces 3 and 4 are the two lateral channel walls. Friction between the specimen and each surface is prescribed independently using a linear Coulomb friction law, allowing systematic investigation of the effect of friction on orientation evolution. All simulations are carried to 50% engineering thickness reduction, which corresponds to a true strain of approximately 0.69 — sufficient to develop well-defined re-orientation patterns.

3. Results

3.1 Ideal Cube Single Crystal: No Initial Scatter

In the first set of simulations, every element in the mesh is assigned the exact cube orientation (φ1=0°,Φ=0°,φ2=0°φ1=0°,Φ=0°,φ2=0°) with zero orientation scatter. Friction coefficients on the top and bottom surfaces are set to μ=0.1μ=0.1; the lateral surfaces are frictionless.

After 50% reduction, the specimen retains a highly regular, brick-like shape and the von Mises stress distribution is nearly homogeneous throughout. The {111} pole figure shows that orientation splitting occurs exclusively about ND. The scatter is symmetric and confined to a narrow cone around the cube orientation. This result confirms that an ideal, perfectly homogeneous cube crystal subjected to moderate friction deforms in a well-ordered fashion with predictable ND rotations — the textbook case of cube metastability under plane strain.

3.2 Cube Crystal with Gaussian Orientation Scatter (Low Friction)

As-grown single crystals are never perfectly oriented; they contain subgrain misorientations and lattice curvature. To model this, a spherical Gaussian orientation scatter of FWHM = 2.5° is superimposed on the cube orientation and discretised onto the 500 mesh elements. Under the same low-friction conditions (μ=0.1μ=0.1 on top and bottom surfaces only), the deformation response changes qualitatively.

The {111} pole figure now shows orientation splitting about both ND and RD. Quantitative analysis of the φ2=0°φ2=0° section of the ODF reveals that after deformation the orientation spread about RD (~22° for f(g)=10f(g)=10) is considerably larger than the spread about ND (~14° for the same ODF intensity level). The finite element mesh exhibits slight shear distortion in most elements, indicating that even the small 2.5° initial scatter is sufficient to break the local plane strain symmetry and activate RD rotation paths alongside the dominant ND rotations. This finding directly demonstrates that a tiny pre-existing orientation spread within a grain is sufficient to qualitatively alter the re-orientation trajectory of a nominally cube-oriented crystal.

3.3 Cube Crystal with Gaussian Scatter under Higher Friction

Increasing the friction coefficient to μ=0.3μ=0.3 on the top and bottom surfaces — while keeping all other parameters identical — amplifies the deformation heterogeneity substantially. The {111} pole figure now displays orientation splitting about ND, RD, and TD simultaneously. The additional TD scatter arises because the higher friction induces significant forward shear strains (longitudinal shear) near the contact surfaces. TD rotations in FCC metals under plane strain are physically linked to the activation of shear-driven lattice rotations; the enhanced friction provides the forward shear component that promotes this rotation mode.

The overall orientation scatter in the ODF is more diffuse and more smeared out than in the low-friction case. The element distortion is more pronounced and the stress distribution is more inhomogeneous, particularly near the contact surfaces. Notably, the peak of the ODF at the exact cube position is weaker under high friction, indicating that a larger fraction of the crystal volume has migrated away from the cube orientation into neighbouring orientation space.

3.4 Cube Crystal with Surround Friction

In a fourth configuration, friction (μ=0.1μ=0.1) is applied to all four longitudinal surfaces simultaneously, representing a more constrained die geometry. This boundary condition suppresses free lateral flow and increases the internal shear field across the full cross-section of the specimen.

The resulting orientation distribution shows a pronounced TD scatter component alongside weaker ND and RD components. The RD scatter is smaller than in the two-surface friction case with comparable μμ. The four-surface friction condition creates a more uniform shear field across the specimen, which shifts the preferred re-orientation axis toward TD. The stress distribution is more inhomogeneous than in the two-surface cases, confirming that the friction geometry itself — not only the friction magnitude — is a controlling variable in the re-orientation behaviour of cube crystals.

4. Interpretation and Physical Mechanisms

4.1 Metastability and Divergence of Re-orientation Paths

The cube orientation under plane strain sits at a point of positive divergence in the re-orientation rate vector field of orientation space. This means that infinitesimally neighbouring orientations have non-zero re-orientation rates that point away from the cube — the cube is an unstable fixed point of the deformation-induced rotation field. The divergence operator applied to the re-orientation rate field provides a scalar measure of orientational instability: a positive value characterises orientations that are kinematically unstable and prone to grain fragmentation and build-up of orientation gradients, even under nominally gradient-free macroscopic loads.

A crucial consequence is that the specific direction of re-orientation — whether toward ND, RD, or TD rotations — is not uniquely determined by the macroscopic strain state alone. It depends on the fine structure of the local stress field, which is in turn controlled by (i) the initial orientation spread within the grain, and (ii) the friction conditions at the die–specimen interface. The simulations demonstrate this dependence quantitatively and systematically.

4.2 Role of Initial Orientation Scatter

The symmetry of an ideal cube crystal under plane strain is such that ND and RD rotation modes are degenerate — both are equally accessible from the exact cube position. A real crystal with a 2.5° Gaussian scatter breaks this local symmetry: those integration points that are slightly displaced toward RD-favourable orientations preferentially rotate about RD, while those displaced toward ND-favourable orientations rotate about ND. The spatial distribution of these orientations across the mesh creates a locally heterogeneous stress field, which in turn reinforces the differentiation of rotation paths. This self-amplifying mechanism explains why even sub-degree orientation fluctuations in a real crystal can generate macroscopic orientation gradients and ultimately grain fragmentation into differently oriented cell blocks.

4.3 Role of Friction

Friction at the die–specimen interface introduces a forward shear stress component that is superimposed on the nominally plane strain stress state. In FCC crystals, forward longitudinal shear favours TD lattice rotations over ND and RD rotations. With increasing friction coefficient (μ=0.1→0.3μ=0.1→0.3), the balance between ND/RD rotations and TD rotations shifts progressively toward TD. When friction acts on all four surfaces, the shear field is more uniform across the cross-section, which promotes TD rotations more uniformly and suppresses the spatial heterogeneity that otherwise promotes RD rotations in the corner regions of the specimen.

These observations are consistent with prior experimental results from several groups and with recent theoretical analyses based on homogenisation and CPFE methods, which have shown that orientation gradients can develop in initially uniform cube crystals even under macroscopically homogeneous loads — driven entirely by the intrinsic instability of the cube orientation and amplified by local stress heterogeneity.

5. Principal Conclusions

The study establishes a clear, quantitative picture of how two physical variables — initial orientation scatter and friction coefficient — govern the re-orientation behaviour of cube-oriented FCC crystals under plane strain compression. The main conclusions are:

• An exactly cube-oriented crystal with no initial scatter and low friction (μ=0.1μ=0.1) re-orients exclusively about ND after 50% plane strain reduction.

• A cube-oriented crystal with a 2.5° Gaussian scatter and low friction (μ=0.1μ=0.1) re-orients about both ND and RD; the RD spread (~22°) exceeds the ND spread (~14°) at equivalent ODF intensity levels.

• A cube-oriented crystal with 2.5° scatter and high friction (μ=0.3μ=0.3) re-orients about ND, RD, and TD; the additional TD scatter is attributed to forward shear induced by higher friction.

• Applying friction to all four surfaces at μ=0.1μ=0.1 produces strong TD scatter with weaker ND and RD components, confirming that the friction geometry — not only its magnitude — is a controlling variable.

• The sensitivity of the re-orientation path to these conditions originates from the kinematic metastability of the cube orientation, which sits at a positive-divergence point in the re-orientation rate vector field. Small starting perturbations are amplified by the deformation into macroscopic orientation gradients.

These results reconcile previously contradictory experimental observations of ND, RD, and TD rotations in cube-textured FCC metals, and establish a physically transparent framework for predicting grain fragmentation and in-grain microtexture development as a function of processing conditions. The texture component CPFE approach is shown to be sufficiently sensitive to initial texture scatter at the sub-2.5° level while remaining computationally tractable and scalable to larger microstructural domains.

6. Methodological Significance

The texture component CPFE method introduced in this work offers a direct bridge between orientation distribution functions — the standard tool of quantitative texture analysis — and spatially resolved crystal plasticity simulations. By encoding the full Gaussian orientation spread of a grain onto the integration points of a finite element mesh, the method captures the interplay between microstructural heterogeneity, local stress state, and lattice rotation in a manner that neither Taylor-type homogenisation models nor classical single-orientation CPFE approaches can achieve. The method is formulated on a strictly scaleable and quantitative basis: the prescribed Gauss function parameters (centre orientation, FWHM) map directly and reproducibly onto the mesh, and the subsequent evolution of each orientation point is physically unconstrained. This makes the approach well suited for systematic parameter studies — as demonstrated here for friction and scatter — and for the investigation of orientational stability across a broad range of FCC alloy compositions and thermomechanical processing routes.

Reference: Raabe D, Zhao Z, Roters F. Study on the orientational stability of cube-oriented FCC crystals under plane strain by use of a texture component crystal plasticity finite element method. Scripta Materialia 50 (2004) 1085–1090. doi:10.1016/j.scriptamat.2003.11.061