Effect of Trace Water Vapor on Measurement of Fatigue Crack Growth Rates in Hydrogen Gas at Low ΔK

Abstract Anomalous and inconsistent results were observed while performing fatigue crack growth tests in hydrogen gas on 4130X Cr-Mo steel when the stress-intensity factor range (ΔK) was less than 10 MPa m1/2. Two particular abnormal characteristics were noted: i) sequential measurements of fatigue crack growth rate (da/dN) vs. ΔK from a single test specimen were not repeatable at lower ΔK, and ii) at constant ΔK ∼8 MPa m1/2, da/dN never reached a steady-state level, as crack growth rates continuously increased as a function of time. It was hypothesized that both of these characteristics could be related to water vapor in the hydrogen gas. This hypothesis was tested by varying the environment in the test chamber through modification of the pressure purging and vacuum evacuation procedures. The resulting water vapor levels in the hydrogen test gas were either higher or lower than the nominal value, and measurements of fatigue crack growth rates in these environments confirmed the hypothesis that water vapor governed da/dN at lower ΔK. One consequence of such water vapor-dominated fatigue crack growth rates is that the da/dN vs. ΔK relationship in dry hydrogen gas at lower ΔK cannot be reliably represented by the da/dN vs. ΔK relationship measured in humid air. As a result, when the lower-ΔK portion of the da/dN vs. ΔK relationship in hydrogen gas is constructed from the da/dN vs. ΔK relationship in air, fatigue crack growth rates can be overestimated by as much as an order of magnitude.