Why do some metals and metal alloys have Fatigue Limits, and others do not? I recently learned of Fatigue Limits, defined as "the stress level below which an infinite number of loading cycles can be applied to a material without causing fatigue failure."  Somewhat surprising to me, some metals and metal alloys have a fatigue limit, and others do not.
What properties of a pure metal or metal alloy determine if a fatigue limit exists, and if it does exist, what the limit is?
 A: Cyclic stressing of a piece of metal causes dislocations to walk back and forth within the lattice structure of each crystallite within the bulk. The grain boundaries where the crystallites meet confine the dislocations to the interior of each crystallite.
Those dislocations pile up against the grain boundaries on each successive stress cycle and while piled up, it is possible for them to get entangled with each other, depending on the exact details of the crystal lattice structure through which they move and the particulars of the grain boundaries. The entangled dislocations become pinned in place within the pileups and when pinned can no longer contribute flexibility to the crystallite, which then begins to become brittle.
The dislocation pileups represent voids within the lattice and allow the grain boundaries to separate from one another- forming microcracks, which then concentrate the stresses locally and hence begin to propagate and connect up. The part fails along the crack line, which represents a debonded surface that cannot transmit stresses between the two sides of the crack.
In some but not all metals there is a mechanism called dislocation climb which allows pinned dislocations to escape the pileups by shifting up and out of the crystallographic plane in which they are pinned, after which they are once again free to move and contribute elasticity to the grain.
If the dislocation climb mechanism is active at room temperature then the material has a fatigue limit. If not, then it has none.
A: There is a correlation between materials with strain hardening properties and the existence of a fatigue limit$^1$.
The puzzling of fatigue is that the stresses are always clearly below the yield limit, and are so supposed to have an elastic behaviour. Nevertheless, if they break after some number of cycles, it suggests some accumulative change, what is not a feature of elastic, but of plastic deformation.
The plastic deformation happens indeed, but only at microscopic level, without changing the object dimensions. When that micro-plastic deformation starts, it is a sign that the yield point was reached locally.
For a strain hardening material, the consequence of an initial plastic deformation is an increase of the required stress for further deformation. It stabilizes the situation, preventing a failure due to great localized plastic strain.
(1) F.C. Rally and G.M. Sinclair "Influence of strain aging on the shape of the S-N diagram" University of Illinois, 1955.
