This is actually two questions: Stress or strain? Stress, because it increases the energy of the atoms in the stressed lattice, increases the chemical potential of surface atoms. However, this effect is very small, and Pound et al. (1959) ,estimated it (for isothermal conditions) to be in the range of 1e-7 to 1e-8 volts per ksi of hydrostatic applied stress. This is far less than the noise typically observed during free corrosion conditions. On the other hand, stress, when it is greater than the Peierls stress, causes dislocation motion that then ruptures surface films exposing bare metal under conditions favorable for rapid oxidation, and transport from the surface. The stress required to accomplish this, the Peierls stress, is well below the bulk yield stress, and stress concentration at flaws and crack tips make the nominal applied stress to nucleate this phenomena quite small. An easy way to test this is to measure the potential of a sample as a load is applied. If the load is applied instantly, then the potential will shift in the active direction. Now, if the shift is more than a few mV, the calculations of Pound et al. (1959) would indicate that it is due to film rupture, and not stress increasing chemical potential.  However, if we ignore the calculations of Pound et al., the active shift in potential should not be recovered (by any amount) until the stress is removed. You could also use the same approach Faraday used, that is, you could scratch an unloaded sample to see if the resulting transient is similar to that caused by a step increase in the load. Finally, tensile and compressive stresses should have essentially identical effects on the chemical potential of surface atoms while compression should expose considerably less bare surface than tensile.

Fatigue is probably not the argument to look at. For example if you have a surface treatment on your material, it will generate residual stresses which in turn have an effect on corrosion cracking. Also corrosion tends to drive material properties down which could explain why being under the fatigue threshold still has an impact.

Regards,

It could happen enbritlement of material on the crack tip. The passivation layer is more brittle, therefore, easier to be broken. This is just one of the mechanisms that would explain that.

It could happen even if there is not cyclic loading. The corrosion along grain boundary will happen under stress corrosion due to presence of segregation and preciptation (pure metal are resistant to this phenomena). Also can happen slip of planes causing rupture of passive layer (transgranular crack).

Hi,

 Presence of any stress in the material will accelerate corrosion phenomena with the environment.

When the operating stress is less than your endurance limit, you do not have the problem of corrosion fatigue. Whereas Stress corrosion cracking (SCC) can happen in the presence of a tensile stress. This stress may be a residual stress from the heat treatment / thermo mechanical processing or due to the operating load conditions. 

SCC:

https://www.nace.org/Corrosion-Central/Corrosion-101/Stress-Corrosion-Cracking-(SCC)/

The presence of stress leads to form zones with high energy, i.e., more active and less stable than unstressed zone and then corrosion cells can be created. The  stressed zones act as anodic sites and the others are cathodic sites.

i agree with Daniel George

This is actually two questions: Stress or strain? Stress, because it increases the energy of the atoms in the stressed lattice, increases the chemical potential of surface atoms. However, this effect is very small, and Pound et al. (1959) ,estimated it (for isothermal conditions) to be in the range of 1e-7 to 1e-8 volts per ksi of hydrostatic applied stress. This is far less than the noise typically observed during free corrosion conditions. On the other hand, stress, when it is greater than the Peierls stress, causes dislocation motion that then ruptures surface films exposing bare metal under conditions favorable for rapid oxidation, and transport from the surface. The stress required to accomplish this, the Peierls stress, is well below the bulk yield stress, and stress concentration at flaws and crack tips make the nominal applied stress to nucleate this phenomena quite small. An easy way to test this is to measure the potential of a sample as a load is applied. If the load is applied instantly, then the potential will shift in the active direction. Now, if the shift is more than a few mV, the calculations of Pound et al. (1959) would indicate that it is due to film rupture, and not stress increasing chemical potential.  However, if we ignore the calculations of Pound et al., the active shift in potential should not be recovered (by any amount) until the stress is removed. You could also use the same approach Faraday used, that is, you could scratch an unloaded sample to see if the resulting transient is similar to that caused by a step increase in the load. Finally, tensile and compressive stresses should have essentially identical effects on the chemical potential of surface atoms while compression should expose considerably less bare surface than tensile.

In this situation, there are three driving forces on crack tip.  They are mechanical driving force, chemical driving force and corroded material property.  One of them weaken orginal material property. The effects can be combined together, and explain why being under the fatigue threshold still has an impact. In aqueous environments, hydrogen will play the main role of fatigue phenomenon (chemical driving force).  

Almost all crystals contain dislocations unless they are grown under special conditions. In addition to these pre-existing dislocations, new dislocations are generated during plastic deformation. The dislocations move under the influence of the externally applied stress to maintain material continuity during plastic deformation. Their motion is interfered by grain boundaries and other dislocations in their path during this process, which results in dislocation pile-ups and strain localization. The strain localization can accelerate dissolution of metal that occur as a result of a breakdown of the protective passive film on a metal surface lead to accelerating the corrosion.

Any stress can change the metal properties in the stress position. stress increase the corrosion of metals.