Stress Corrosion Cracking of Silicon

Cracks in silicon can initiate and propagate even if the energy supplied by the mechanical load is not sufficient to create new fracture surfaces, according to research recently published in Physical Review Letters. Overturning the conventional view of environmentally assisted fracture in silicon, the paper shows that “subcritical” crack propagation can be driven by dissociative chemisorption of oxygen molecules, as predicted by quantum mechanics-based molecular simulations and elegantly confirmed by experiments showing no cracking in an inert, oxygen free environment.

It has long been known that so-called “stress corrosion” cracking - where fracture proceeds under the concerted action of stress and chemistry - takes place in polar ionic materials such as silica, typically in the presence of water, although the exact mechanism remains elusive. However, covalent materials such as silicon were not thought to be vulnerable until now.

TYC researchers Giovanni Peralta, Dr James Kermode and Prof. Alessandro De Vita (KCL) used the `Learn on the Fly’ hybrid scheme to carry out QM-accurate calculations, revealing that oxygen molecules placed inside a crack under a subcritical applied load spontaneously dissociate and then chemisorb at the crack tip, releasing enough heat to cleave one Si-Si bond and advance the crack. One oxygen molecule is needed for each crack advance step, suggesting that the speed of the crack is controlled by the diffusion of oxygen to the tip. 

The simulations were complemented by experiments conducted by Anna Gleizer and Dov Sherman (Technion Institute, Israel) which confirmed that cracks can advance under subcritical loads, but only in the presence of oxygen, with experiments carried out in an inert argon environment requiring the full energetic cost of separating the two fracture surfaces to be paid.

Speaking about the research, Dr Kermode said “this work is exciting as it has possible implications for the manufacture of improved semiconductor devices resistant to fracture, as well as being a step along the path to an improved understanding of stress corrosion cracking in general, which is a problem of huge technological and scientific relevance, affecting sectors as diverse as mining, biomedicine, and even civil engineering”.

The work benefited from the recently awarded INCITE computer time allocation at Argonne National Laboratory, US, running on the MIRA Leadership-class Blue Gene/Q facility, connected with the ongoing TYC/ANL partnership. The theoretical modelling work was carried out at King’s College London within the Fragmentation project of the Rio Tinto Centre for Advanced Mineral Recovery, based at Imperial College.

Journal link: Dissociative Chemisorption of O2 Inducing Stress Corrosion Cracking in Silicon Crystals. A. Gleizer, G. Peralta, J. R. Kermode, A. De Vita, and D. Sherman, Phys. Rev. Lett. 112,115501 (2014).

Figure: Stress corrosion cracking on the silicon (110) cleavage plane can be mediated  by dissociative chemisorption of oxygen molecules. A large model system (above) is required to accurately model stress concentratrion, but QM accuracy is only needed in a  small part near the crack tip (below).


De Vita Highlight - March 2014 - small.jpg

Follow @tyc_london for updates from the Thomas Young Centre.