Polarisation behaviour and corrosion initiation mechanisms of Mo coated with amorphous hydrogenated silicon alloy thin ceramic films L. H. Hihara, A. S. Iwane and R. E. Rocheleau Amorphous hydrogenated silicon, silicon nitride, and silicon–carbon films on molybdenum substrates were anodically and cathodically polarised in NaCl and Na 2 SO 4 solutions at 30uC. The silicon films showed signs of extensive deterioration, whereas the silicon nitride and silicon–carbon films were inert, but led to pitting of the Mo substrate predominantly at pre-existing coating breaches. Analysis of the corrosion morphology, and comparisons of actual polarisation diagrams to those generated by electrochemical polarisation models indicated that submicron cracks accounted for most of the breaches, while larger breaches, such as those caused by coating spallation were less frequent. To a lesser degree, pits also initiated from regions lacking pre-existing breaches during the later stages of anodic polarisation. CEST/2139 Keywords: polarisation, corrosion initiation, thin ceramic films L. H. Hihara is in the Department of Mechanical Engineering, University of Hawaii at Manoa, Honolulu, HI 96822, USA (hihara@wiliki.eng.hawaii.edu). A. S. Iwane is currently at Tesoro Hawaii Corporation, Honolulu, HI 96813, USA. R. E. Rocheleau is in the Hawaii Natural Energy Institute, University of Hawaii at Manoa, Honolulu, HI 96822, USA. Manuscript received 17 February 2004; accepted 20 July 2004. # 2004 Institute of Materials, Minerals and Mining. Published by Maney on behalf of the Institute. INTRODUCTION Silicon alloy thin films deposited using plasma-enhanced chemical vapour deposition (PECVD) are generally con- sidered to be inert chemically resistant materials. The effectiveness of silicon alloy thin films as corrosion barriers, however, has been compromised by film defects that have led to localised substrate corrosion. Silicon alloy films have been previously studied as corrosion barriers for integrated circuits 1–3 and medical implants. 4 The studies demonstrated that silicon nitride 1,3 and silicon carbide 4 films generally improved corrosion resistance, but could not prevent localised corrosion of the substrate at film breaches. Corrosion initiation sites on substrates coated with thin silicon alloys have been identified as regions of spallation, microcracks, and localised areas where no visible film breaches could be detected using scanning electron micro- scopy. 5 Since the utility of using PECVD silicon alloy thin films as effective corrosion barriers hinges on their ability to prevent or substantially reduce corrosion, further research is needed to develop a comprehensive understanding of the localised corrosion mechanisms. In this study, polarisation experiments were conducted on pure Mo, and Mo substrates coated with amorphous, hydrogenated a-Si : H, a-SiN x : H, and a-SiC x : H thin films. Comparisons were made between bare Mo, the coated Mo substrates, and electrochemical polarisation models. Anodic polarisation experiments were conducted to study the degradation mechanisms of the coated substrates, and cathodic experiments were conducted to study whether additional information on pre-existing coating breaches could be obtained. The experiments were conducted in deaerated and oxygenated 0 . 5M Na 2 SO 4 and 3 . 15 wt-%NaCl solutions. Chlorides are particularly aggressive towards metals with naturally occurring passive films and, hence, it was of interest to determine if chlorides had a similar effect on metals coated with the synthetic films. Molybdenum was chosen as the substrate material since its polarisation behaviour is very similar when exposed to either 0 . 5M Na 2 SO 4 or 3 . 15 wt-%NaCl; hence, differences which may appear in corrosion behaviour in the sulphate and chloride solutions are likely to be attributable to the effect of Cl - and SO 22 4 on the films rather than their effect on Mo. The corrosion morphology and anodic polarisation behaviour was compared to that predicted using polarisa- tion models to identify processes by which localised substrate corrosion occurs. The anodic polarisation model 6 that was used in this study was based on the hypothesis that localised corrosion predominantly initiates from pre- existing coating breaches during the initial stages of polarisation, and not from the nucleation of new breaches. Pre-existing breaches would serve as nucleation sites for localised corrosion because neither the silicon alloy films nor the substrate Mo have the ability to heal coating breaches. New breaches were not assumed to nucleate during anodic polarisation because the electric field through the silicon alloy films was assumed to be less than that necessary to cause breakdown. Anodic polarisation curves generated from the polarisation models were generally consistent with those observed in the actual experiments. The results indicated that the majority of coating defects appeared to be pre-existing breaches, and their initial size and density strongly affected the shape of the polarisation curves. Relatively large breaches (e.g. 15 mm) such as those caused by coating spallation were significantly less common than smaller coating cracks. There was also evidence that a relatively small number of pits and micron sized, localised regions of film–substrate detachment nucleated during the later stages of anodic polarisation (i.e. at higher potentials). The comparison of the cathodic polarisation data and polarisation models was able to provide some information on pre-existing coating breaches. The cathodic polarisation model 6 that was used in this study was based on the assumption that cathodic currents exclusively emanate from the substrate exposed at coating breaches. Results from oxygenated solutions corroborated findings from the anodic polarisation studies on initial coating breach sizes and distributions. MATERIALS Thin films The deposition parameters and composition of the silicon alloy films have been described elsewhere. 5 The a-Si : H, a-SiN x : H, and a-SiC x :H films were deposited to a thickness of 0 . 7 mm on 1 mm thick, 5 cm by 5 cm pure Mo (99 . 98%) plates. The samples are denoted as a-Si : H/Mo, a-SiN x : H/Mo, and a-SiC x : H/Mo, respectively. DOI 10.1179/147842204X2862 Corrosion Engineering, Science and Technology 2004 Vol. 39 No. 4 277