Predicting Corrosion Damage in the Human Body Using Artificial Intelligence: In Vitro Progress and Future Applications

Artificial intelligence (AI) is used in the clinic to improve patient care. While the successes illustrate AI's impact, few studies have led to improved clinical outcomes. In this review, we focus on how AI models implemented in nonorthopedic fields of corrosion science may apply to the study of orthopedic alloys. We first define and introduce fundamental AI concepts and models, as well as physiologically relevant corrosion damage modes. We then systematically review the corrosion/AI literature. Finally, we identify several AI models that may be implemented to study fretting, crevice, and pitting corrosion of titanium and cobalt chrome alloys.

An Atomistic Study of the Stress Corrosion Cracking in Graphene

Using molecular dynamics (MD) simulations, we study the mechanism of stress corrosion cracking in graphene. Two sets of modelings are conducted. In the first one, large graphene sheets with cracks in the armchair and zigzag directions are exposed to oxygen molecules. The crack growth as a result of chemical reactions between carbon radicals and oxygen molecules at different mechanical tensile stress levels is studied. In the second set of simulations, MD simulations are combined with the density functional-based tight bonding method to enhance the accuracy. This set of modelings focuses on a smaller zone in the vicinity of the crack tip. The impact of initial crack orientation on corrosion is studied by investigating corrosion of cracks in both armchair and zigzag directions. We investigate the subcritical crack propagation occurring as a result of the combined effects of both mechanical loading and chemical reactions. Our results show that cracks in graphene can grow due to chemical reactions with the environmental molecules. The MD modelings also predict that reaction of carbon atoms with oxygen molecules might lead to a stress relaxation at the crack tip, hence preventing further crack propagation. The results show that subcritical crack growth can happen by two mechanisms, which include the failure of C-C bonds or by removing the carbon atoms from graphene sheets in the form of CO or CO₂ molecules.

Phononic thermal transport properties of C3N nanotubes

Reverse nonequilibrium molecular dynamics (RNEMD) is employed to study the phononic thermal transport properties of C₃N nanotubes. We study the effect of nanotube length and diameter on the thermal conductivity and investigate the phonon transport transition from ballistic to diffusive regime in C₃N nanotubes. It is found that the thermal conductivity of C₃N nanotubes is significantly lower than those of carbon nanotubes across the entire ballistic-diffusive range. In addition, significantly lower ballistic to diffusive transition length (72-80 nm) is observed in C₃N nanotubes compared to carbon nanotubes. The inspection of phonon dispersion curves shows that carbon nanotubes have stiffer acoustic modes than C₃N nanotubes which results in lower group velocities for C₃N nanotubes. Due to the presence of nitrogen atoms, the phonon mean free paths and relaxation times of C₃N nanotubes are shorter than those of the carbon nanotubes. The combined effect of lower group velocities and relaxation times leads to the lower thermal conductivity of C₃N nanotubes.