A study of the fatigue limit of copper

This paper shows that, in fatigue tests on copper, a great number of cycles beyond expectation is necessary for determining the fatigue limit. An ultrahigh frequency fatigue testing machine was used andS-N curves covering 10¹⁰ cycles were obtained for annealed specimens and for stretched ones. In annealed copper, the fatigue limit appeared at about 9.8 × 10⁹ cycles. Observations were made on the development of slip bands and substructures in the course of a test at the strain level (3.8 × 10^-4) of the fatigue limit. The slip bands continued to develop up to about 9 × 10⁹ cycles, but remained unchanged if further cycled. On a searching examination of the behavior of microcracks, one end of which stayed in a grain without propagation, it was proved that fatigue hardening was responsible for the existence of the fatigue limit. Contrary to expectation, cell structures were found after 10¹⁰ cycles in such a low strain fatigue. In stretched copper, however, microcracks continued to grow even in the stage beyond 10¹⁰ cycles when fatigued at the strain level of the fatigue limit inferred from itsS-N curve.