130 Prakt. Met. Sonderband 38 (2006) 
(COD) [6,15], or the average void height to determine the separation energy for micro- 
ductile fracture which is the energy necessary to form the voids on the fracture surfaces 
[6,14,16,17]. 
The average void height and CTOA can be also derived semi-automatically, if the 
registration of the two surface halves is available: Then a difference height map (or misfit 
map) can be deduced which contains the differences between two surfaces. Such a map 
delivers small values in regions with good correspondence (such as the fatigue pre-crack 
regions), large positive values in regions with voids and large negative values in regions 
with material overlap. Additionally the misfit map allows calculating the total volume of all 
void areas and all areas with overlapping material. 
5. RESULTS 
The automatic registration procedure is demonstrated using two fracture mechanics 
specimens which show a different behaviour. In both cases the material is an MMC with 
the aluminium alloy Al-6061 as matrix material. The first specimen is the above mentioned 
cast MMC with 10% Al,O3 particles which have an average size of 10 ym. This material 
shows a micro-ductile fracture with broken particles at the bottom of the dimples and 
significant plastic deformation in the parts of final fracture. The second specimen is a 
powder-metallurgy MMC with SIC particles with an average size of 100 um. This specimen 
behaves much more brittle, showing very little plastic deformation in the final fracture 
region. 
The SEM images in Fig. 2 show corresponding regions S1 and S2 from both halves of the 
cast MMC specimen. The fatigue fracture surface is seen in the lower half of the images, 
the micro-ductile final fracture region in the upper half. Two corresponding regions near 
the fatigue pre-crack front are marked by a circle. From the two stereo-pairs, 3D models of 
the surfaces S1 and S2 have been reconstructed as depicted in Fig. 3. For an easier 
comparison, the surface S2 was rotated by 180°. The lower region of the models 
corresponds to the fatigue fracture areas and have very similar shape. The pre-crack front 
runs horizontally through the models from left to right. The overload fracture region in the 
upper part is different for both models and shows large plastic deformations that occurred 
during the fracture process. 
In order to measure local toughness parameters both models have been registered to 
each other using the registration procedure in Section 3. No other user input has been 
necessary apart from the information that for the registration only the lower parts of the 
images should be used. As results of the registration procedure, we get the two 3D models 
in the same world-coordinate frame (Fig. 3) with one model being rotated with respect to 
the reconstruction. Additionally the algorithm provides a map of the height differences 
between the two 3D models (misfit map), see Fig. 5a. For an ideally brittle material, no gap 
would appear between the two surfaces and the misfit would be zero. Grey values 
correspond to regions with similar height, white regions to void areas and dark regions 
mark areas with material overlap. The map clearly shows the pre-crack front as the border 
that separates the grey fatigue fracture region from the rest. 
The determination of COD; is demonstrated in Fig. 4. After marking the profile in Fig. 2, a 
second profile of the corresponding half has been extracted automatically. By vertical 
shifting of the profiles a COD; value of ~18um has been derived. From the misfit map the 
total volume of the void regions and the total volume of regions with material overlap has 
been calculated. Vig has been calculated as 54376um?. Voyeriap as 346429um?, thus being 
about 6.4 times as large as Vyoid.