Microstructural Effects on Cleavage Fracture

When designing engineering structures, great emphasis is placed on brittle fracture of metals, where it is assumed that cleavage occurs when the stress intensity factor exceed a

“macroscopic” or “engineering” value of fracture toughness, K_IC. Techniques have been developed that collate metallurgical features to microscope applied fracture stress, such as the size of any defect within the material. Many researchers have investigated if a crystal cleavage event could be defined as an extension of cracked carbide, an inclusion, a M-A-C constituent, or pearlitic carbides.

Fracture of mild steels is believed to occur when a microcrack occurs at brittle grain-boundary carbide and propagates at the critical tensile stress into the ferrite matrix. Smith’s [190]

model predicts cleavage fracture stress is mainly affected by the carbide precipitates in contrast to experimental results that demonstrate that σF is dependent on grain size. However Curry and Knott showed [155, 191] that in normalized and annealed mild steels there is a

general relationship between grain size and the largest grain-boundary carbide width.

Confirming that lack of grain size dependence on Smiths model is overcome by an increase in the carbide thickness with increase in grain size. Okumura [192] measured σF values of steel where carbide and grain sizes variations are independent, thus concluded that regardless of increasing fracture resistance for fine-grain sizes, the effects of grain-boundary carbide size on cleavage fracture is the principal factor controlling the fracture stress.

Weld metals have different initiators for cleavage than those in wrought steels. A model was proposed by Tweed and Knott [7] for cleavage fracture in the AD microstructure of C-Mn weld metal after non-metallic inclusions were found to act as a cleavage microcrack initiator within this microstructure. McRobie [52], Novovic [13], Wenman [14] and do Patrocinio [15] showed that a large variety of inclusions are responsible for cleavage fracture initiation.

Research carried out on pearlitic steel [193-195] showed that the nuclei for cleavage fracture could be a microcrack or small oxide inclusion, nucleating across numerous pearlitic carbides.

Also different variations in toughness within the weld metal are known to occur due to inclusion population, which contains oxides formed during the deoxidation process. More recent studies have shown that the most important factor influencing toughness [7, 10, 13-15, 52] is the microstructure. Within weld metals inclusions are associated with initiation which is fibrous in nature through the coalescence of voids around them. Detailed examination of these inclusions could not find any clear differences in the chemical composition of the “void-initiating” and “crack-“void-initiating” inclusions. Knott [113, 196] rationalised that there could be subtle differences and these could be associated with a sulphide shell around the inclusion.

More recently, Miao [76] showed that C-Mn weld metal contained large “patches” of sulphide, resulting from the application of different cooling rates, these generally seem to have higher fracture toughness than the same material with fewer patches, it was also noted that there were differences in the cleavage initiation modes, these may have arisen from different inclusion surface features.

Weld metals are treated with warm prestressing (WPS), because after this operation the distribution of cleavage fracture initiators changes. Reed and Knott [113, 197, 198] used

A508 weld metal and showed that after WPS treatments the low-temperature fracture load/fracture toughness increase, this was primarily due to the compressive residual stress distribution generated during the WPS, also it is associated with the change in nature of the fracture initiation sites. When looking at specimens that were not treated with WPS, initiation sites corresponded to the largest inclusion in a “well-behaved distribution”, whereas for WPS specimens they are associated with inclusions with a size close to the mean of the size distribution, large inclusions were observed to have voids around them. It was concluded that during WPS treatments the inclusion matrix interface for large inclusions was decohered at low levels of strain, thus they could not act as cleavage initiation sites following low temperature fracture. Under these condition fracture need to be initiated by a smaller inclusion without interface decohesion so that statistically determined critical distance X₀ in the Ritchie, Knott and Rice (RKR) [199] model is also changed in comparison to the non WPS specimens.

As shown in Fig. 3.23 if cleavage fracture occurs after ductile crack growth, the inclusions in the weld material may not be the cleavage initiator. This is due to the fact that local conditions under which cleavage fracture begins, ahead of a blunted crack tip are not the same as those ahead of a sharp crack at low temperature because the plastic strain distribution ahead of the crack-tip is different in both cases. In the former case the plastic strain over a large distance is very high so that any likely inclusion microcrack nucleus produced near the yield-point strain would be blunted to become a non-virulent crack nucleus for cleavage fracture, see Fig. 3.23(c) and 3.23(d). Knott [113] therefore suggested that any microcrack nucleus would have to be newly formed in a conceptually different set of particles. Zhang and Knott [200-202] also demonstrated that ahead of a fibrous growing crack, voids formed around inclusions and cleavage is nucleated between inclusions, probably on MAC products.