•Norbert Huber
Forschungszentrum Karlsruhe, Institut für Materialforschung II, Postfach 3640, 76021 Karlsruhe

Using pyramidal indenters only the hardness and the Young's modulus of bulk materials can be investigated from indentation experiments. In the case of a film/substrate system it can be shown that the penetration depth to film thickness ratio represents a strain measure similar to the depth to radius ratio in conjunction with spherical indenters.

Neural networks are able to determine the stress-strain curve of metal films from the depth dependent hardness and stiffness of the film/substrate system. The neural networks are trained using Finite Element results. Four material parameters representing the true stress-strain curve are identified by the neural network: the Young's modulus, the yield stress, the initial work hardening rate and the ultimate tensile strength.

The stress-strain curves of aluminum films of different thickness on glass and silicon substrates are determined. Only small differences in the hardening behavior can be observed for the same thickness but different substrates while the yield stress increases significantly with decreasing film thickness.