Surface and Coatings Technology, 50 (1991) 25—32 25 Failure modes in scratch adhesion testing* S. J. Bull AEA Industrial Technology, Ha,well Laboratory, Oxfordshire, OX11 ORA (UK) (Received March 12, 1991; accepted July 10, 1991) Abstract The scratch test has been used to assess coating adhesion for some time. In this test, a diamond indenter is drawn across the coated surface under an increasing load (either stepwise or continuous) until at some load, termed the critical load L~, a well-defined failure event occurs; if this failure event represents the loss of coating—substrate adhesion then the critical load can be used as a qualitative measure of coating—substrate adhesion. However, it is well known that a range of possible failure modes can occur and only some of these are dependent on adhesion; other failure modes which depend on plastic deformation and fracture within the coating, rather than any adhesive failure at the coating—substrate interface, may be just as useful in the assessment of coating quality for tribological applications. In this study, titanium nitride coatings have been deposited onto a range of different substrates from soft nickel to a hard cemented carbide and the failure modes which occur during scratch testing have been identified. Failures fall into two general groups, depending on whether the substrate behaves in a brittle or ductile manner during the scratch test. Similar failure modes are observed for titanium nitride coatings produced by several deposition technologies, although there is some variation in the appearance of specific failures which is process dependent. Many of the same failure modes are also observed for other coating materials, such as ZrN or hard carbon, but several different types of failure are also observed. The generation of acoustic emission during the test is related to the occurrence of these different failure modes. The origin of the failure modes and the use of the scratch test to assess coating—substrate adhesion are discussed in the light of these observations. 1. Introduction factors which still need to be understood if the test is to be used more reliably. A qualitative measure of the adhesion of a thin hard A number of workers have identified failure modes coating to its substrate can generally be achieved using associated with the scratch test bypostfacto observation the scratch adhesion test [1—3].The test is very easy of the scratch track using reflected light microscopy to perform; a loaded diamond stylus (generally a Rock- (RLM) or scanning electron microscopy (SEM) [4-9]. well C diamond with a tip of 200 ~im radius) is drawn Some of the failure mechanisms have recently been across the surface under increasing load (either stepwise confirmed by in situ SEM scratch testing [10] and the or continuous) until some well-defined failure occurs viewing of the scratch testing of coatings on transparent at a load which is called the critical load L~. If the substrates from beneath by transmitted light microscopy test is to be used as a measure of adhesion, this failure during the test [11]. Additional failure modes have must occur as a result of detachment at the coat- been identified in these latter studies which may not ing—substrate interface which is not always very easy be directly relevant to the scratch adhesion test owing to identify. Thus a recognizable event along a track is to the geometry and size of the stylus used [10] and generally used, such as the complete stripping of the the properties of the coating material [11]. Although coating to uncover the substrate or regular chipping only a few of the failure modes depend on detachment along the edge of the track, which is then related to at the coating—substrate interface, and are thus of direct coating—substrate adhesion but not in any simple man- relevance to the scratch test as a measure of adhesion, ner. If performed with care, the technique can be other types of coating failure (such as cracking through reproducible and enables the adhesion of a number of similar coatings to be ranked; however, there are several the coating thickness) may determine the behaviour of a coated component in a particular application and are *Paper presented at the 18th International Conference on thus equally important. In this paper, some of the Metallurgical Coatings and Thin Films, San Diego, CA, April coating failure modes which are observed for sputtered 22—26, 1991. and arc-evaporated titanium nitride coatings on a range