Sample preparation 
First, polished samples were submitted to a 7A etching (1) which attacks carbides and lets appear 
them in white without affecting the martensitic matrix which appears in black, Fig. 5. Three 
different types of carbides are identified from the textural criteria (shape, position along or inside 
grains, ). The MC carbides have globular shape and are along or inside grains. The M,C carbides 
form fibrous structures and the MgC carbides form fishbone-like structures. These two types are 
present only along the grains boundaries, Fig. 2-4. 
(a 
Quantitat 
Three par 
commuted | 
be charachi 
- ratio betwe 
Fig. 2 : MC carbides Fig. 3: M,C carbides Fig. 4: McC carbides 
The inter 
The human brain can differentiate phases thanks to textural criteria. Unfortunately, in image principle | 
analysis, the textural segmentation is often not very reliable. In other words, for the separation of ym; 
phases, image analysis techniques are more efficient when a spectral contrast (a colour difference) yp, i 
exists between the phases. So, we found an etching which gives a different spectral response to each pc 
carbide type. The Groesbeck’s etching (2) attacks the MC and the MC carbides and does not affect yn, 6 
the bright matrix and the pink MC carbides. After the etching, M,C carbides appear in dark brown; character 
M;C carbides in yellow or blue according to the etching time. To improve the contrast between the mean inter 
phases, images are acquired with an interference filter centred on 489 nm. the MICc 
The carbit 
between € 
from the 
Wiring 
nim 
Fig. 5 : High speed steel. Groesbeck’s etching. (a) without an interference filter 
(b) with an interference filter centred on 489 nm. 
Image analysis 
Observations 
The region near the surface cools more rapidly than any deeper region. The effect of this cooling 
difference is observed in the microstructure. The most the depth increases, the most the martensite >. 
and carbides increase in size. As carbides are mainly situated along the grain boundaries, their |. 
interdistance is directly related to the grain size, increasing with depth. The goal is to quantify these a Ny 
differences by image analysis. a. 
10 the ons 
140