Avalanches and plastic flow in crystal plasticity: an overview

Abstract

Crystal plasticity is mediated through dislocations, which form knotted configurations in a complex energy landscape. Once they disentangle and move, they may also be impeded by permanent obstacles with finite energy barriers or frustrating long-range interactions. The outcome of such complexity is the emergence of dislocation avalanches as the basic mechanism of plastic flow in solids at the nanoscale. While the deformation behavior of bulk materials appears smooth, a predictive model should clearly be based upon the character of these dislocation avalanches and their associated strain bursts. We provide here a comprehensive overview of experimental observations, theoretical models and computational approaches that have been developed to unravel the multiple aspects of dislocation avalanche physics and the phenomena leading to strain bursts in crystal plasticity.

Learning local, quenched disorder in plasticity and other crackling noise phenomena

“…2 The evidence for quenched disorder in initial defect microstructures has been accumulated through observations of abrupt plastic events or material-crackling noise in a large variety of materials, such as FCC and BCC crystals, [3][4][5][6] amorphous solids 7 and also earthquake geological faults. 8,9 This crackling noise 10 has been commonly explained by random field models 11,12 or interface depinning ones, [13][14][15][16][17][18] where the major component is homogeneous solid elasticity, but also a spatially inhomogeneous and random distribution of local, quenched disorder (typically entering local flow stress information) 4,17,[19][20][21] and the allowed microstates are characterized by its stress and strain and minimize the elastic energy functional. The evidence of crackling noise has led to an extensive study of the local, statistical properties of abrupt events and their properties, such as their sizes, durations, average shapes, and their critical exponents.…”

“…However, the major concern has been the fact that while homogeneous elastic properties are relatively straightforward to measure and test at virtually any scale, 22 the model distribution of local, quenched disorder is elusive, despite its commonly observed signature response of stochastic plastic bursts. [3][4][5]7 In this paper, we propose a feasible approach to "learn" the quenched disorder distributions directly from load-response timeseries: We argue that the full characteristics of the timeseries may unveil the information on the form of the quenched disorder distribution which is not available through typical temporally local observables (such as abrupt event size/duration). 23 While the major motivation of this work stems from plastic deformation, this method is generally applicable across crackling noise phenomena, defined through timeseries of an applied field (magnetic field, force, stress) and the associated response variable (magnetization, displacement, and strain).…”

“…31 However, its mechanism has been debated. 4,26 A plausible phenomenological scenario that explains the overall behavior is that the stochastic yield distribution at a representative volume has non-trivially large tails 32 at the nanoscale (that could either be a fat-tailed/large-kurtosis distribution or special distributions such as Weibull or Gumbel) and it gradually self-averages into the central limit behavior as the volume increases (cf. Fig.…”

“…While still a phenomenologically-driven hypothesis, the validity of this scenario would explain the self-consistent emergence of size effects and stochasticity in small-scale plastic deformation, in a material and sample independent manner. 4,5,33 In contrast to this picture, it is common in multiscale modeling of material deformation 22,34 to assume that the statistics of yield parameters in a representative material volume are dominated solely by their averages. Such a severe assumption is clearly not valid in a scenario where the yielding distribution has a large variance, or it is non-Gaussian in nature.…”

“…We use typical methods of unsupervised machine learning that naturally depend on physical modeling's completeness and the possible universality class. 4,23 The method does not aim to identify novel physical mechanisms of crackling noise; instead, it may provide a classification of parent quenched disorder distributions despite the existence of coexisting universal behavior. In the following, we perform simulations in two characteristic models of crackling noise, the T = 0 mean-field Random Field Ising model 24 (RFIM) and the elastic long-range dipolar interface depinning model(LRDIDM), 14,35 designed for crystalline or amorphous solids.…”