Crystal twinning

Spinel law contact twinning. A single crystal is shown at left with the composition plane in red. At right, the crystal has effectively been cut on the composition plane and the front half rotated by 180° to produce a contact twin. This creates reentrants at the top and lower left of the composition plane.[4]

In the isometric system, the most common types of twins are the Spinel Law (twin plane, parallel to an octahedron) <111>, where the twin axis is perpendicular to an octahedral face, and the Iron Cross <001>, which is the interpenetration of two pyritohedrons, a subtype of dodecahedron.[6]

In the hexagonal system, calcite shows the contact twin laws {0001} and {0112}. Quartz shows the Brazil Law {1120}, and Dauphiné Law <0001>, which are penetration twins caused by transformation, and Japan Law {1122}, which is often caused by accidents during growth.[6]

In the tetragonal system, cyclical contact twins are the most commonly observed type of twin, such as in rutile titanium dioxide and cassiterite tin oxide.[6]

In the orthorhombic system, crystals usually twin on planes parallel to the prism face, where the most common is a {110} twin, which produces cyclical twins, such as in aragonite, chrysoberyl, and cerussite.[6]

In the monoclinic system, twins occur most often on the planes {100} and {001} by the Manebach Law {001}, Carlsbad Law [001], Braveno Law {021} in orthoclase, and the Swallow Tail Twins (Manebach law) {001} in gypsum.[6]

In the triclinic system, the most commonly twinned crystals are the feldspar minerals plagioclase and microcline. These minerals show the Albite and Pericline Laws.[5][6]

The most common twin operations by crystal system are tabulated below. This list is not exhaustive, particularly for the crystal systems of lowest symmetry, such as the triclinic system.[7][2][6]

30° twin of staurolite

90° twins of staurolite

Iron pyrite cross twin

Fivefold twinning in a gold nanoparticle (electron micrograph).

Growth twinning likely begins early in crystal growth, since the contact surface usually passes through the center of the crystal. An atom joins a crystal face in a less than ideal position, forming a seed for growth of a twin. The original crystal and its twin then grow together and closely resemble each other. This is characteristic enough of certain minerals to suggest that it is thermodynamically or kinetically favored under conditions of rapid growth.[4][2] In certain materials cyclic twins such as that in the image can occur as they are lower in energy at small sizes.[13]

Illustration of transformation twinning in 2-D. The orthorhombic crystal at left transforms to a monoclinic crystal at right, with two parallel twinning planes (polysynthetic twinning)

Transformation and annealing twinning takes place when a cooling crystal experiences a displacive polymorphic transition. For example, leucite has an isometric crystal structure above about 665 °C (1,229 °F), but becomes tetragonal below this temperature. Any one of the three original axes of a crystal can become the long axis when this phase change takes place. Twinning results when different parts of the crystal break their isometric symmetry along a different choice of axis. This is typically polysynthetic twinning, which enables the crystal to maintain its isometric shape by averaging out the displacement in each direction. This produces a pseudomorphic crystal that appears to have isometric symmetry. Potassium feldspar likewise experiences polysynthetic twinning as it transforms from a monoclinic structure (orthoclase) to a triclinic structure (microcline) on slow cooling.[2]

Animation of deformation twinning of crystal

Deformation twinning is a response to shear stress. The crystal structure is displaced along successive planes of the crystal, a process also called glide. The twinning is always reflection twinning and the glide plane is also the mirror plane. Deformation twinning can be observed in a calcite cleavage fragment by applying gentle pressure with a knife blade near an edge. This particular glide twinning, {102}, is found almost universally in deformed rock beds containing calcite.[2]

Twinning and slip are competitive mechanisms for crystal deformation. Each mechanism is dominant in certain crystal systems and under certain conditions.[14] In fcc metals, slip is almost always dominant because the stress required is far less than twinning stress.[15]

Twinning can occur by cooperative displacement of atoms along the face of the twin boundary. This displacement of a large quantity of atoms simultaneously requires significant energy to perform. Therefore, the theoretical stress required to form a twin is quite high. It is believed that twinning is associated with dislocation motion on a coordinated scale, in contrast to slip, which is caused by independent glide at several locations in the crystal.

Compared to slip, twinning produces a deformation pattern that is more heterogeneous in nature. This deformation produces a local gradient across the material and near intersections between twins and grain boundaries. The deformation gradient can lead to fracture along the boundaries, particularly in bcc transition metals at low temperatures.

Of the three common crystalline structures bcc, fcc, and hcp, the hcp structure is the most likely to form deformation twins when strained, because they rarely have a sufficient number of slip systems for an arbitrary shape change. High strain rates, low stacking-fault energy and low temperatures facilitate deformation twinning.[1]

If a metal with face-centered cubic (fcc) structure, like Al, Cu, Ag, Au, etc., is subjected to stress, it will experience twinning. The formation and migration of twin boundaries is partly responsible for ductility and malleability of fcc metals.[16]

Twin boundaries are partly responsible for shock hardening and for many of the changes that occur in cold work of metals with limited slip systems or at very low temperatures. They also occur due to martensitic transformations: the motion of twin boundaries is responsible for the pseudoelastic and shape-memory behavior of nitinol, and their presence is partly responsible for the hardness due to quenching of steel. In certain types of high strength steels, very fine deformation twins act as primary obstacles against dislocation motion. These steels are referred to as 'TWIP' steels, where TWIP stands for twinning-induced plasticity.[17]

Deformation twinning crystallographic planes

Twinning is crystallographically defined by its twin plane 𝑲_𝟏, the mirror plane in the twin and parent material, and 𝜼_𝟏, which is the twinning shear direction. Deformation twins in Zr are generally lenticular in shape, lengthening in the 𝜼_𝟏 direction and thickening along the 𝑲_𝟏 plane normal.[18]

The twin plane, shear direction, and shear plane form the basis vectors of an orthogonal set. The axis-angle misorientation relationship between the parent and twin is a rotation of angle 𝜉 about the shear plane's normal direction 𝑷.

More generally, twinning can be described as a 180° rotation about an axis (𝑲_𝟏 for type I twins or 𝜼_𝟏 for type II twins normal direction) , or a mirror reflection in a plane (𝑲_𝟏 or 𝜼_𝟏 normal plane).[19]

In addition to a homogeneous shear, atomic shuffles are sometimes required to reform the correct crystal structure in the twinned lattice. For each twin variant, a reciprocal twin with swapped 𝑲_𝟏 and 𝑲_2, 𝜼_𝟏 and 𝜼₂ is possible, but one variant may appear more frequently in reality due to complexities with the required shuffles.[20]

there are only two crystallographic planes in a shearing action that do not change their shape and size as a consequence of the shear. The first 𝑲_𝟏 is the plane defining the upper and lower surfaces of the sheared volume. This plane contains the shear direction. The other plane, designated C. The shear direction is shown with an arrow and labelled with its customary designation 𝜼_𝟏. It follows from the above that there are three ways that a crystal lattice can be sheared while still retaining its crystal structure and symmetry:

1. When 𝑲_𝟏 is a rational plane and 𝜼₂ a rational direction, a twin of the first kind
2. When 𝑲₂ is a rational plane and 𝜼_𝟏 a rational direction, a twin of the second kind, rare
3. When all four elements 𝑲_𝟏, 𝑲₂, 𝜼_𝟏, and 𝜼₂ are rational, a compound twin

Regular twinning planes in calcite crystal. Crossed nicols image, magnification 10× (Field of view = 2 mm)

A deformation twin embryo forms in BCC metal by accumulating stacking faults, with a variant selection governed by the local stress state.[21][22][23] Variation of the stress field close to twins inferred from HR-EBSD experimental[24][25] and crystal plasticity finite element (CPFE) simulation data indicated that twins nucleate on sites with maximum strain energy density and twin resolved shear stress; thus, reducing the total elastic energy after formation. This relaxation depends on the twin thickness and is a deciding factor in the spacing between twins.[26] Experimental[27] and three-dimensional[28] analysis has focussed on the (stored) strain energy density measured along a path. This highly localised stress field can provide a sufficient driving force for concurrent twin nucleation[29] and inter/intra-granular crack nucleation.

Deformation twin growth can be perceived as a two-step process of i) thickening that is mediated by the interaction between the residual and mobile twin partials at the coherent twin-parent interface,[30] and ii) dislocation mobility along the twin shear direction.[31] The twin propagates when the homogeneous shear stress reaches a critical value, and a twin-parent interface advances inside the parent grain [240]. The propagating deformation twin generates a stress field due to its confinement by the surrounding parent crystal, and deformation twins develop a 3D oblate spheroid shape (which appears in 2D sections as a bi-convex lens) with a mixed coherent and non-coherent interface (Figure b).[31]

Kannan et al.[32] found, using in-situ ultra-high-speed optical imaging, that twin nucleation in single-crystal magnesium is stress-driven accompanied by instantaneous propagation at a speed of 1 km/s (initially) that prioritises volume lateral thickening over forward propagation, past a critical width where growth is then become faster along the shear direction. Barnett[33] also indicated that growth is due to twin tip extension. Furthermore, elastic simulations of the local stress field surrounding the ellipsoidal twin tip find that the field can be described using its lens angle (${\displaystyle \beta }$) and that the stress field magnitude increases with twin thickness.[34]

(a) forescatter electron diode (FSD) image for deformation twins at grain boundary in age-hardened ferrite at I) 18 mm working distance and II) 38 mm working distance. (b) Schematic of a lenticular twin with interface dislocations and (c) Twin band.[35]

In practice, plastic accommodation occurs in the parent crystal; thus, it also depends on the material’s yield stress, the anisotropic elastic stiffness of the parent crystal lattice, and the deformation twinning shear magnitude.[30] This can also be accompanied by long-range diffusion of elements and elemental segregation (e.g., Cr and Co in single crystal Ni-based superalloy MD2), which occurs at the twin boundary to facilitate twin growth by lowering the critical stacking fault energy.[36] A linear variation has been observed between twin thickness, stacking fault energy and grain size,[37] and to a lesser degree, the stress state of the twinning grain (Schmid Factor).[38] The twin thickness saturated once a critical residual dislocations’ density reached the coherent twin-parent crystal boundary.[23][39]

Significant attention has been paid to the crystallography,[40] morphology[41] and macro mechanical effects[42] of deformation twinning. Although the criterion for deformation twin growth is not entirely understood, it is a tip-controlled phenomenon linked to the interaction between the residual and mobile twin partials at the twin interface; thermodynamically, this involves the elastic energy of the strained lattice, the interface and volume free-energy of the twin, and the dissipated energy of the growth mechanism.[43] To fully understand the interactions between microstructure (i.e., grain size, texture), temperature and strain rate on deformation twinning, it is crucial to characterise the (high) local stress and strain field associated with twin thickening and propagation. This is especially important for materials where cleavage fracture can be initiated by twinning (e.g., iron-silicon, the ferrite phase of age-hardened duplex stainless-steel, and single-crystal magnesium) as a stress-relieving mechanism.

Early studies of deformation twins arrested within grains of niobium[44] and iron[45] visualised the highly local strain concentration at the twin tip using an etch-pit procedure. More recently, high-resolution electron backscatter diffraction (HR-EBSD) has been used to investigate the strain 'singularity' ahead of a twin tip in hexagonal close-packed (HCP) zirconium alloy. A deformation twin in commercial purity titanium was characterised similarly and then quantified using a local Schmid factor (LSF) at the twin tip,[46] as described in equation below.

$${\displaystyle \mathrm {LSF} ={\frac {{\boldsymbol {\sigma }}:{\boldsymbol {P}}^{i}}{\|{\boldsymbol {\sigma }}\|}},\quad {\boldsymbol {S}}^{i}={\boldsymbol {d}}^{i}\otimes \mathbf {n} ^{i}}$$

where σ is the stress tensor, Sⁱ is the Schmid tensor, Pⁱ is its symmetric part, dⁱ is the shear direction and nⁱ is the shear plane normal for ith slip system. The authors concluded that conditions at the twin tip control thickening and propagation in a manner analogous to the operation of dislocation sources ahead of a crack-tip.[47] In the analysis, a broad region of high LSF ahead of the twin tip favoured propagation, whereas a narrow region of high LSF promoted thickening. Since then, it has been argued[48] that the LSF firmly controls the twin variant selection, as twinning has strong polarity.

The LSF novelty – compared to other criteria to describe conditions at the twin[37] – lies in combining a geometrical criterion with the deformation field in the parent grain to provide an approximate indication of the local twin mode (i.e., thickening or propagation). However, the LSF analysis does not take advantage of the available full-field data, relies on global information on the applied stress, and does not consider the energy balance that drives twin growth. There have been few in-situ experiments to quantify the strain field ahead of a propagating deformation twin.[35][25] Such observations might validate geometrical or hybrid geometrical-energy-based criteria[48] for growth. Nanoscale testing (i.e., transmission electron microscopy) may not represent the behaviour in bulk samples due to plasticity starvation, i.e., large surface area to volume ratio,[49] so a suitable analysis method is needed.

Lloyd[39] described the stress concentration field ahead of the twin tip using a two-dimensional dislocation-based model within a single magnesium grain. Wang and Li,[50] who considered microscopic phase-field (MPF) models of cracks, noted that the stress fields were similar for dislocations, deformation twinning and martensitic transformations, with differences only in the traction of the created surface, i.e., there is 100% traction recovery for dislocations and a traction-free surface for a crack. They highlighted that the stress field singularity regulates the advancement of the crack-tip and dislocations. This stress concentration can be characterised using a path-independent line integral, as shown by Eshelby for dislocations considering the contribution from the surface traction and ellipsoidal inclusions,[51] and Rice[52] for cracks and stress concentrations with traction-free surfaces. Furthermore, Venables[53] noted that the oblate spheroid shape of the twin tip is the ideal example of an ellipsoid inclusion or a notch.