Effect of substrate surface roughness on mechanical properties of diamond-like carbon coatings

Abstract

A plasma enhanced chemical vapour deposition (PECVD) amorphous carbon coating was deposited onto 100Cr6 steel substrates having varying degrees of surface roughness. The samples were subsequently evaluated to determine the correlation between substrate roughness and coating performance. The steel substrates were prepared before coating deposition to attain five different levels of roughness: (a) ground; (b) superfinished (SF); (c) polished to 1000 grit; (d) polished to 220 grit and (e) polished to a 1 μm diamond finish. The aim of the investigation was to determine the degree of finish required for good tribological performance and coating adhesion. The mechanical and tribological properties of the samples were assessed by nanoindentation, ramped load scratch testing, and pin on disk wear testing. Nanoindentation testing was used to determine the hardness of the samples and the relative contributions to the system hardness from the substrate and coating were separated using the model of Korsunsky et al. Nanoindentation testing showed that the coating hardness (when separated from the system hardness) was lower for the samples with the SF substrate than the others: the reasons for this are discussed in the light of Raman measurements on the fractions of diamond-like and graphite-like bonding in the coatings. Ramped load scratch testing was used to determine coating adhesion and the scratch test failure mode. With the exception of the samples with the ground substrate finish, studies of the friction coefficient plots during scratch testing showed little variation between the samples, and SEM imaging revealed a common failure mode of severe spallation at the scratch track border. The samples with the ground substrate showed differences in response between scratches parallel and perpendicular to the grinding direction, with scratches parallel to the grinding direction showing more severe spallation. The average critical load to failure, as determined by the point of first failure in the scanning electron microscope, was lower for the coatings on the SF substrate than the coatings on the 220 grit, 1000 grit and 1 micron finished substrates. The critical load to failure for the samples with ground substrates was lower than the other substrate surface finishes. Pin on disk wear testing of the samples against a steel ball revealed that the major effect of the varying substrate roughness was on the wear of the counterface, with rougher substrate finishes generally resulting in higher wear rates of the counterface, although the smoothest substrate finish, the micrometre finish, also resulted in higher wear. The sample whose substrate was superfinished gave least wear of the counterface and this was therefore the optimum finish for the samples when considering their performance in a tribological couple.