4.09.1.2.1 Precipitation-induced cracking

Solid-state cracks in welds often occur near the time/ temperature regime of a phase transformation in which the local stress or strain produced from the phase transformation interacts with global stresses in the weldment and results in cracking. This basic phenomenon has several different names based on the alloy system it occurs in and includes ‘ductility

dip’ cracking in low-strength nickel-based alloys and stainless steels,10’18 ‘strain-age’ cracking in precipita­tion hardenable nickel — and iron-based alloys,1 — ‘reheat cracking’ in 2%Cr-1Mo-type steels,2 and ‘subsolidus cracking’ in titanium alloys.23

Ductility dip cracking has been studied in detail by Young and Capobianco, who provide a good example of how this phenomenon occurs.18 The cracking derives its name from the corresponding

loss of tensile ductility in the homologous tempera­ture range (~0.4—0.9 Tm) that corresponds to the time/temperature regime of the precipitation of a partially or fully coherent second phase. In low — strength nickel-chromium alloys, the ductility dip occurs during on-cooling from a peak temperature high enough to solutionize existing carbides and cause intergranular precipitation of the detrimental phase (M23C6 carbides, in this case).

The relationship between the precipitation kinet­ics of the detrimental phase and the macroscopic tensile ductility is shown in Figure 9, which com­pares a calculated TTT plot for M23C6 precipitation in a Ni-29Cr-9Fe-0.01C (wt%) alloy (i. e., an analog to EN52/Alloy 690), with experimental on-cooling tensile ductility data for the alloy.10 As shown, if very rapid cooling suppresses precipitation, there is no ductility loss (region 1). The ductility minimum occurs near the nose of the precipitation curve when the local strain contribution from intergranular carbide precipitation is maximized (region 2). Duc­tility recovery occurs as precipitation progresses because local misfit strains decrease as chromium depletion occurs and as misfit dislocations are gener­ated (region 3). Ductility is restored when precipita­tion is complete (region 4).

In Figure 10, the stages of ductility dip crack formation are outlined, in which (often in reheated weld metal of a multipass weld or in the base metal heat-affected zone) (Cr, Fe)23C6 carbides preferen­tially nucleate during on-cooling on grain boundaries with partial, cube-on-cube coherency (Figure 10(a)). Due to misfit strains, tension develops between the carbides, producing intermittent microscopic cracking (Figure 10(b)). Upon the development of global stres­ses (e. g., from thermal strains on-cooling or applied during hot ductility testing), these cracks often link up
and form the classic ‘ductility dip’ crack (Figure 10(c)), that is, an intergranular crack that typically extends < 1 grain in length. Compared to a solidification-type crack, the fracture surfaces of these solid-state cracks show less evidence of the underlying dendritic struc­ture and are littered with (Cr, Fe)23C6-type carbides.24 Figure 10(d) illustrates how the misfit strain between the carbide and matrix increases with increasing chro­mium concentration in the alloy. In part, this explains why 30 wt% alloys (A690 and EN52) are more sus­ceptible to this defect than their lower chromium counterparts (A600/E-182).

The transient nature of ductility loss with time and temperature, which are important dependencies cannot be explained by other proposed mechanisms for this solid-state cracking.2 -32 Specifically, in the Ni-Cr alloys of interest to nuclear systems, neither impurity segregation (at least at ‘typical’ levels of <50 wt ppm sulfur, <100 wt ppm P in the bulk alloy) nor grain boundary sliding plays a significant role in cracking. For example, if sulfur is migrating to grain boundaries at ^870 °C (1600 °F), ductility would not be expected to recover after short hold times (~10s as shown in Figure 9). Similarly, the temperature dependence of the ductility minimum must be explained, as the effect of embrittling agents

such as sulfur should persist to low homologous

18

temperatures.

Similarly, if grain boundary sliding were contri­buting to the intergranular fracture, the fastest quenched sample in Figure 9 should show the most embrittlement, that is, where sliding would be favored by a microstructure without carbides to pin the grain boundaries.29,30 A relative grain-by-grain map of the plastic strains from samples strained to 5 and 10% in the ductility dip temperature range (Figure 11) shows direct evidence against the

10mil	EB house 1c 1% nital etch
HAZ
Weld
S
1
4
10.0 keV 2.1kX
9/21/104
C
2.1 kX
9/20/104 Ф	10.0 keV
10.0 mm 9/20/104 10.0keV 2.1kX Ф

Figure 8 Illustration of intergranular liquation-type cracks in a low-alloy steel. The cracks occurred in both the partially melted zone and heat-affected zone along prior austenite grain boundaries (top) and subsequent analysis of crack surfaces via Auger electron microscopy identified bands of MnS-type sulfide inclusions (bottom).

notion that solid-state cracking is caused by sliding, that is, the uniformity of slip with some strain accu­mulation (yellow and red areas) near the ductility dip cracks. If grain boundary sliding played a role in ductility dip cracking, there would be less strain contrast near the cracks, not more, as this technique is sensitive only to the diffraction pattern rotation produced by dislocations.

The scenario of weld cracking occurring in the time/temperature regime of precipitation of a par­tially or fully coherent second phase is also well
recognized as controlling strain-age or ‘reheat’-type cracking in y’/y"-strengthened alloys.10,18,20,21,33 While susceptibility to reheat cracking is often plot­ted as a function of the aluminum and titanium content of the alloy, a more fundamental correlation based on the transformation kinetics and precipitate/ matrix mismatch of the alloy is possible.18 As shown in Figure 12, alloys with high susceptibility to strain — age cracking display fast transformation kinetics and a large negative precipitate/matrix mismatch (i. e., tension develops between precipitates).

As transformation stresses are displaced to longer times and precipitation-induced stresses become compressive, weldability is improved. However, while weldability is increased by displacing precipitation — induced stresses to longer times, hot workability is often degraded for the same reasons.

It is notable that some titanium alloys undergo a tensile ductility loss when tested near the p! a tra­nsformation.23,34,35 While the authors do not know of equivalent research on zirconium alloys, these alloys could also be susceptible to this form of precipitation — induced cracking (PIC). Mechanistically, this could be caused by the nucleation of a from the p-phase
if the (110)pk(0001)a is significant, or from some other phase with partial coherency (e. g., HCP Laves on HCP-a).