Technical Brief: Fatigue Dimples

By Arthur H. Griebel, III, Metallurgical and Electrical Engineer, Stork Climax Research Services

Introduction
This technical brief is intended to warn failure analysts against assuming that dimples indicate overload fracture.  Dimples simply mean that microvoid processes were active; these processes can occur in cyclic as well as monotonic fractures.  The failure analyst must examine the entire fracture, the macroscopic as well as the microscopic features, to be confident of the cracking mode.

Fatigue Dimples
Failure analysis often relies (heavily) upon fracture analysis, which entails observing a crack surface and making conclusions about the mode and cause of cracking. Interpreting a crack surface can be complicated by rub, post fracture damage, and corrosion, but in the absence of these obscuring effects the analyst hopes to establish the mode of cracking based upon the crack appearance.  Once the mode is known, the analyst tries to identify the conditions (material or environmental) that caused the crack.

To successfully follow this chain of deduction from crack appearance backwards to cause, the analyst must recognize a correlation between crack appearance and the cracking mode. The correlation is preferably one-to-one, so that an observed crack surface represents only one mode. 

Figure 1: Fatigue cracking initiated in the thread root at the bottom of this image. The crack propagated from the bottom towards the top of the image. All subsequent images maintain the same orientation as this image.		Figure 2: Crack appearance 0.1 mm from the crack initiation site.		Figure 3: Crack appearance 0.2 mm from the crack initiation site.

At the very least, the analyst wants to distinguish between cracking caused by cyclic loading (fatigue) in which a crack of length a extends incrementally (da/dN) with each stress cycle N, and cracking caused by overload, in which the crack traverses a large distance in a single stress event. Distinguishing between fatigue and overload in ductile materials is frequently based upon the presence of striations or dimples, which often are associated with fatigue and overload, respectively. Actually, fatigue does not always produce striations, and dimples are not exclusively associated with overload. The first fact is well recognized, but the second is not, and fatigue dimples are seldom mentioned in the literature.

Figure 4: Crack appearance 0.4 mm from the crack initiation site.		Figure 5: Crack appearance 0.6 mm from the crack initiation site.		Figure 6: Crack appearance 0.8 mm from the crack initiation site.

An experienced analyst recognizes that fatigue crack morphology changes considerably with propagation rate, da/dN. Striations form only in a limited range of crack propagation rates, and only in some materials. At lower propagation rates the appearance can range from crystallographic to “ropey” and at higher propagation rates fatigue dimples can form.      

The following sequence of fractographs demonstrates the range of fatigue crack appearances observed on the fracture surface of a single small screw.  Fatigue cracking began in the thread root and the crack propagated through about 1.6 millimeters of the 3 mm cross section before the remaining material fractured due to overload. Figure 1 is an overview of the crack initiation site.  Figures 2 through 11 were obtained at increasing distances from the crack initiation site. 

Figure 7: Crack appearance 1.0 mm from the crack initiation site.		Figure 8: Crack appearance 1.2 mm from the crack initiation site.		Figure 9: Crack appearance 1.4 mm from the crack initiation site.

The crack propagation rate, da/dN, increased as the crack lengthened and, consequently, the crack surface appearance changes continuously between the crack initiation site and the region of overload.  Striation cracks observed in Figures 2 through 7 leave no doubt that fatigue is the cracking mode in this area.  However, there is an increasing amount of “dimple-like” character to the surfaces shown in Figures 5 through 7.  The regions shown in Figures 8 and 9 consist mostly of poorly formed dimples with blunt edges; these might be interpreted as dimple rupture due to overload if they were not part of the sequence of images that clearly shows the crack began as fatigue.  

Figure 10 shows dimples with slight blunting of the edges that might be argued as fatigue or overload, with proper classification dependent upon macroscopic appearance of the fracture or other evidence.  Figure 11 shows well formed dimples with sharp edges that were observed in the region of final fracture.  This is what classic tensile dimples look like. 

Figure 10: Crack appearance 1.6 mm from the crack initiation site.		Figure 11 Crack appearance 1.8 mm from the crack initiation site.

This screw exhibited a rather wide range of fatigue appearances; not all fatigue fractures display such a wide variation in morphology.  Sometimes a crack is comprised mostly one type of surface, such as the poorly formed dimples shown in Figure 8.  It is important for a fracture analyst to examine the entire fracture surface, and consider the macroscopic as well as the microscopic evidence of fracture mode.   Otherwise, the microscopic fracture morphology can lead to incorrect conclusions about the failure mode.