Connecting atomic-scale phenomena to macroscopic engineering performance in the general area of the mechanical behavior of materials remains an immense challenge.  This is particularly true in metals, where properties such as flow stress, fracture toughness, and fatigue life are controlled by the behavior of dislocations over a wide range of length and time scales.  Dislocation interactions with other defects (other dislocations, solutes, precipitates, crack tips, and grain boundaries) determine the overall evolution of the macroscopic stress, plastic strain, and strain rate, as well as the limiting failure phenomena (brittle or ductile fracture, ductility, crack growth), and all as a function of temperature.  This complex interplay between processes at many scales means that purely phenomenological approaches to material models are limited, as experimental results defy easy description without insights from more theoretical underpinnings.  As such, there has been an intense effort to address these problems “from the bottom up,” using atomistic models.  On the other hand, the challenges to predictive atomistic modeling of mechanical behaviour are many.  For example, the complex compositions of engineering alloys necessitates sophisticated reactive interatomic potentials.  Equally challenging is the massive gap between the atomistic nanoscale and the engineering macroscale (both length and time scales) ,which means that many of the processes we seek to understand are beyond the computational capacity of purely atomistic approaches.  As such, any hope for accurate, atomistically-informed modeling of mechanical behaviour requires innovative multiscale methods and algorithms that can take advantage of rapidly increasing computational power.With the above as background, we propose a CECAM workshop on one of the most challenging topics within the broader field, fundamental failure mechanisms in metals. The title of the proposed workshop is “Modeling Metal Failure Across Multiple Scales”.

State of the art in each of the 5 sub-topic areas is as follows. The references favor those of the organizers, but are not meant to be the definitive references on the topic but rather merely reflect the nature of the work being carried out in the field at the present time.

Dislocation-crack tip interactions: Understanding of thermally-activated nucleation for geometries with dislocation line parallel to crack front has been established reasonably well in pure metals; very limited work on 3d dislocations at atomic scales, limited work on emission on oblique planes and role of solutes and other defects, and challenges in solving appropriate boundary value problems at higher scales (e.g. discrete dislocation level).

Fatigue-crack growth: Studies using 2d Discrete Dislocation Dynamics reveal some origins of fatigue crack growth, but using ad-hoc cohesive zone models for failure, and with difficulties in realistic treatment of crack/dislocation interactions at the nanoscale; very limited multiscale approaches to bridge this gap, and limited ability to include larger-scale plasticity. In general, no real “multitime scale” methods that would enable modeling of many cycles of fatigue. Many open questions and computational challenges. Some new work on cyclic behavior without cracks using discrete dislocation modeling exists, demonstrating emergence of complex dislocation structures, but insufficient to connect to any failure modes to date.

Chemical-embrittlement: Multiscale models with quantum mechanics at the crack tip have just emerged to address crack-tip/single-embrittlement-atom effects, but computationally very challenging. MD-level methods require appropriate chemistry in the interaction potentials, and successes here are limited (recent work in Ni-H, Fe-H, and Ni-S). New mechanisms through nanoscale MD modeling are also starting, but bridging time scale issues and capturing both crack tip, near-tip, and far-field plasticity phenomena all together to predict real embrittlement has not been accomplished.

Ab initio prediction of environmental embrittlement at a crack tip in aluminum 

Work-hardening rate and rate sensitivity: The application of 3d Discrete Dislocation Dynamics models is now well-established and widely used, mainly in pure materials or in systems with defects unique to radiation damage. Extension to alloys, introduction of proper nanoscale dislocation core mechanics, and thermally-activated rate sensitivity are almost non-existent yet pose both fundamental and computational challenges.

Multiscale Methodologies: A number of new approaches that enable the treatment of quantum mechanics at a crack tip or dislocation core, with correct far-field elasticity response, have been developed recently. These methods have some limitations, and remain computationally intensive, and so it remains necessary to understand what can and cannot be accomplished with such methods, and how to develop improved methods. Bridging time scales is less developed in this context. Although accelerated MD methods are established and widely used now, such approaches are not immediately extendable about atomic scales and so new ideas or coupling methods are still needed. Coupling of atoms to dislocations and dislocations to continuum plasticity has seen some progress but again existing methods have limitations and are not suitable for all classes of problems.