Tag Archives: Dislocation density

Nucleation of Recrystallization

As a metallic material is plastically deformed, energy will be accumulated in the material through generation and rearrangement of defects, mainly dislocations. As dislocation networks develop they will tend to form entanglements that subsequently form dislocation cells which, in turn, eventually leads to the formation of subgrains in the grain interiors. These subgrains will be the birthplaces of the recrystallization nuclei. In order for the subgrains to become viable nuclei, however, two criteria need to be met locally in the microstructure. The first is a kinematic criterion requiring mobile high-angle grain boundaries to be formed by the nucleation event. The second is a thermodynamic criterion requiring a stored energy gradient to be present across the interface that is sufficient to provide enough positive driving pressure for grain growth to occur. These criteria are usually found to be met at grain boundary triple junctions and along grain boundaries. Such sites provide enough lattice curvature and sufficient stored energy differences for providing possible nucleation sites. Nucleation may also, but less common, take place in the grain interiors, for example along shear bands and near particle inclusions. The latter is often referred to as particle-stimulated nucleation. As nucleation preferentially takes place along pre-existing grain boundaries, it is common to observe so-called necklace patterns of recrystallized material along the boundaries.

The process of subgrain formation is in mesoscale models usually captured by considering an incubation time t_{\mathrm{c}} or a threshold dislocation density \rho_{\mathrm{c}}, i.e. a threshold stored energy, that needs to be reached before nucleation is initiated. In macroscale models, the same process is usually formulated in terms of a critical strain \varepsilon_{\mathrm{c}} that has to be achieved in order for nucleation to take place. Also the notion of a critical stress \sigma_{\mathrm{c}} has been proposed to identify the onset of recrystallization.

During the formation of the subgrains, recovery processes very rapidly reduce the dislocation content in the subgrain interiors by annihilation of dislocations and by accumulation of dislocations at the subgrain boundaries. This increases the subgrain misorientation with respect to neighboring regions and eventually provide high-angle, mobile, boundaries. The same processes also provide subgrains with very low internal stored energy.

The crystallographic orientation of the nuclei will influence both the progression of recrystallization and also the evolution of any recrystallization texture in the material. Recrystallization nuclei are products of the cold worked parent material but they do not necessarily inherit their orientation from the parent grains although it is not uncommon that clear evidence of the texture present in the initial, cold worked, microstructure can be observed also in the recrystallized material. High-angle boundaries with respect to the surrounding material are also required in order to permit growth of the nuclei, causing differences to develop between the initial and recrystallized textures. It can be noted that the development of recrystallization texture is usually attributed to either orientated nucleation or oriented growth or possibly a combination of these processes.

Having defined the initiation of recrystallization, the next question is at what frequency new nucleation events take place. This issue has been approached in a number of ways in different studies. The two main trends is to consider either of  site-saturated nucleation where all nuclei are assumed to be present at the start of the simulation and no new nuclei are added over time or, alternatively, continuous nucleation where new nuclei are continuously added according to some expression of the rate of nucleation. In the latter scenario, both constant and non-constant rates of nucleation have been considered.

Although constant nucleation rates are often assumed, there are experimental studies indicating nucleation to be non-constant and continuous process. However, it is inherently difficult to experimentally determine the rate of nucleation as it is not easy to define through experimental observations if a region in the microstructure is a subgrain, a nucleus or an expanding grain and the exact time at which one changes into the other.

Production of very fine grained materials

Since recrystallization is driven by reduction of stored energy, the amount of cold work influences the grain size evolution. Materials with grain sizes down to the nanometer scale can be obtained by exposing the material to severe plastic deformation (SPD). Such SPD processes include Equal Channel Angular Pressing/Extrusion (ECAP/ECAE), Asymmetric Rolling (ASR), Multi-Directional Forging (MDF), High-Pressure Torsion (HPT) and Twist Extrusion (TE). Some of these processes are discussed further below.

During ECAP, studied in Hallberg et al. (2010), the material specimen is pressed through a channel die as illustrated below.

Schematic Equal Channel Angle Pressing (ECAP) setup.
Schematic Equal Channel Angle Pressing (ECAP) setup.

The amount of effective plastic deformation, denoted by \displaystyle \varepsilon_{\mathrm{eff}}^{\mathrm{p}}, that is imposed onto the specimen in each pass can be estimated from knowledge of the die geometry according to

\displaystyle \varepsilon_{\mathrm{eff}}^{\mathrm{p}}=\frac{N_{\mathrm{pass}}}{\sqrt{3}}\left[2\mathrm{cot}\left(\frac{\Phi}{2}+\frac{\Psi}{2}\right)+\Psi\mathrm{cosec}\left(\frac{\Phi}{2}+\frac{\Psi}{2}\right)\right]

Varying the channel geometry thus allows control of the amount of plastic deformation that is exerted onto the specimen in each pass through the die. If the specimen is rotated between each pass through the die, different standardized processing routes can be obtained.

Results from simulations of ECAP-processing. The top figure shows the distribution of average grain size after and two ECAP-passes, respectively. The bottom figure shows the distribution of normalized dislocation density, also after one and two ECAP-passes, respectively. Note that after two passes, both grain size and dislocation density remain at relatively constant levels all through the specimen along the indicated lines.
Results from simulations of ECAP-processing. The top figure shows the distribution of average grain size after and two ECAP-passes, respectively. The bottom figure shows the distribution of normalized dislocation density, also after one and two ECAP-passes, respectively. Note that after two passes, both grain size and dislocation density remain at relatively constant levels all through the specimen along the indicated lines.

Another SPD process is asymmetric rolling, discussed in Hallberg (2013), where a conventional rolling process can be made asymmetric by different methods.

asymmetric_rolling_setup
Schematic illustration of a rolling process.

The asymmetry of the process can be induced by having different radii r of the rolls, by different roller velocities, i.e. \omega_{1}\ne\omega_{2} or by different friction/lubrication conditions at each side of the sheet. The asymmetry increases the shear deformation of the rolled sheet and hence the total amount of effective plastic deformation.

Distribution of dislocation density obtained from a symmetric rolling operation with 40% thickness reduction per pass and a rolling friction of 0.25. The top figure shows the results after the first pass and the bottom figure after the second pass.
Distribution of dislocation density obtained from a symmetric rolling operation with 40% thickness reduction per pass and a rolling friction of 0.25. The top figure shows the results after the first pass and the bottom figure after the second pass.

 

Dislocation and grain boundary interaction

Mesoscale models of microstructure evolution allow studying of heterogeneous dislocation density distributions and related gradient effects. This is closely connected to the well-renowned Hall-Petch effect, stating a proportionality between the yield stress \sigma_{\mathrm{y}} and the inverse of the square-root of the average grain size d according to

\sigma_{\mathrm{y}}\propto\frac{1}{\sqrt{d}}

The interaction between dislocation motion and grain boundaries can be modeled by different approaches. Some examples are given below.

In Hallberg and Ristinmaa (2013) (also discussed in this conference presentation), a hybrid finite difference/cellular automaton model is established where dislocation density gradient are modeled in a reaction-diffusion system. This approach results in the expected Hall-Petch behavior of the macroscopic flow stress – without including an explicit dependence on the grain size – in addition to providing a physically sound dislocation density distribution, indicated in the figure below.

Simulated distribution of total dislocation density in an artificial polycrystal. Due to the reaction-diffusion modeling of dislocation density evolution and the presence of grain boundaries, dislocation pile-ups are formed along the grain boundaries and particularly at triple junctions.
Simulated distribution of total dislocation density in an artificial polycrystal. Due to the reaction-diffusion modeling of dislocation density evolution and the presence of grain boundaries, dislocation pile-ups are formed along the grain boundaries and particularly at triple junctions.

By this modeling approach, the dislocation density will be concentrated at grain boundaries and particularly at triple junctions, directly providing the sites for nucleation of recrystallization. This is in contrast to the common modeling approach where nuclei are placed manually at appropriate sites in the microstructure.

Another approach to modeling of dislocation and grain boundary interaction is taken in Hallberg (2013) where a level set formulation is used to model polycrystal grain structures.

Dislocation density distribution influenced by the presence of grain boundaries. The microstructure model is based on level sets, representing the individual grains. The extent of dislocation accumulation at grain boundaries can be controlled by the parameter w, as illustrated in images a-d.
Dislocation density distribution influenced by the presence of grain boundaries. The microstructure model is based on level sets, representing the individual grains. The extent of dislocation accumulation at grain boundaries can be controlled by the parameter w, as illustrated in images a-d.

Continuum scale modeling of recrystallization

Adopting a continuum mechanical approach, recrystallization can be modeled using an internal variable representation of the pertaining quantities, such as the average grain size and the dislocation density. The macro-scale material behavior will in this way be based on parameters related to the evolving microstructure.

An example of continuum-scale modeling and simulation of ECAP-processing of Aluminum is given in Hallberg et al. (2010). Some results on the distributions of grain size and dislocation density in the work piece are shown in the figures below.

Results from simulations of ECAP-processing. The top figure shows the distribution of average grain size after and two ECAP-passes, respectively. The bottom figure shows the distribution of normalized dislocation density, also after one and two ECAP-passes, respectively. Note that after two passes, both grain size and dislocation density remain at relatively constant levels all through the specimen along the indicated lines.

Results from simulations of ECAP-processing. The top figure shows the distribution of average grain size after and two ECAP-passes, respectively. The bottom figure shows the distribution of normalized dislocation density, also after one and two ECAP-passes, respectively. Note that after two passes, both grain size and dislocation density remain at relatively constant levels all through the specimen along the indicated lines.
Results from simulations of ECAP-processing. The top figure shows the distribution of average grain size after and two ECAP-passes, respectively. The bottom figure shows the distribution of normalized dislocation density, also after one and two ECAP-passes, respectively. Note that after two passes, both grain size and dislocation density remain at relatively constant levels all through the specimen along the indicated lines.