Digital image processing in scanning electron microscopy analysis of protective coatings.

Davidescu, Arjana ; Savii, George ; Sticlaru, Carmen 等

1. INTRODUCTION

High temperature coatings are used to avoid surface degradation or to insulate the material against the hot environment. As the lifetime of structural components is frequently controlled by surface degradation, significant performance improvements are obtained by application of coatings. Coatings are applied to provide oxidation, corrosion or thermal protection depending on the nature of the operating environment and thermal loads to be endured. The coating should possess the required mechanical properties, adhesion and metallurgical stability in contact with the substrate.

MCrAlY's are a family of materials, which have a base metal (M) of cobalt, nickel and/or iron, plus chromium, aluminium, yttrium and sometimes other alloying elements. MCrAlY's are used as overlay coatings for turbine components improving their resistance and providing a longer lifetime for turbines, even under hard environmental conditions. High temperature oxidation resistance is achieved by one or more alloy components, which have to form a dense, stable, slow-growing, external oxide layer such as [Al.sub.2][O.sub.3] (Sloof & Jeurgens, 2004).

In the last few years, a new thermal-spraying process, high-velocity oxygen fuel (HVOF), was developed and used to produce MCrAlY coatings. The application of HVOF spraying has implications for the oxidation of powder particles, because of the free-oxygen content in the combustion gas and high temperatures required to melt the powder to assure a certain homogeneity level of the coating. There are two elements in the metallic powder, aluminium and yttrium, which have a high affinity for oxygen. These elements are oxidized during thermal spraying, so that the as-sprayed coatings contain oxides, such as [Al.sub.2][O.sub.3]. After annealing, the oxides are uniformly distributed in the material. The HVOF coatings showed good oxidation behaviour at high temperature in various atmospheres (synthetic air, He-[O.sub.2 ]mixture or He-synthetic air), which suggested that the HVOF process can be used as a technological alternative to the more expensive VPS technology (Toma et al, 1999).

Aluminium content in the coating alloy is important, because selective oxidation of the Al occurs only on the surface of alloys with adequate Al contents. However, a high Al content in coatings leads to coating brittleness and a strong tendency to crack (Tang et al, 2004). Such cracks can propagate into the substrate material and lead to premature failure of the coated component. It has been proved that a fine grain size has a positive effect on the oxidation behaviour of alumina forming alloys (Neidel & Riesenbeck, 2004).

2. GRAPHICAL USER INTERFACE

The objective of the work presented in this paper was to process SEM images of HVOF MCrAlY coatings in order to study the uniformity of the coating, the thicknesses of the oxide and diffusion layer and the percentage of the [beta]--phase from the substrate.

An instance of the GUI is presented in Figure 1.

[FIGURE 1 OMITTED]

The GUI allows loading a specific image file with tif extension (tagged image file). This goal is achieved by means of an edit box; where one has to type the name of the file, and a push button to update the image in the memory.

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

In order to depict the proper threshold level for the oxide layer, in the left side of the GUI a push button is presented which allows selection of a region from oxide layer. The threshold value is the mean of the pixel's grey levels extended by [+ or -] 10. Some morphological operations are applied on binaries image: erosion, filling holes and removing isolated pixels. Finally, the perimeter of the binaries image is overlaid on the original image so the user can observe if any error occurs. The result can be seen in Fig. 2.

Two corrections can be made: an automated one--by means of the slider element which establishes the threshold level, and a manual one--by means of a push button. It has been introduced for the case the automated correction is not efficient enough. When this correction is selected from the push button, the cursor transforms into a cross-hair and the user can select two pixels from the image. A white line is drawn between the two pixels in order to attach a region to the interested feature. The results (a detail) can be seen in figure 3. Pushing End button indicates that all corrections are finished and only the largest feature of the image (the oxide layer) is maintained.

In order to obtain quantitative results the image must be calibrated by pushing Calibration button. The cursor transforms into a cross-hair; the user has to select the extremities of the image scale and to type the length value indicated in the text box below. When this operation takes place the GUI shows the original image for avoiding the influence of the applied morphological operations upon the scale.

At this point the image of the oxide layer is ready for measuring. At the bottom of the GUI there is a pop-up menu which allows selecting the image, thickness and the thickness histogram. The results are presented in Fig. 4. A linear grid (Russ & Dehoff, 2002) is overlaid on the oxide layer image consisting of 100 equally spaced vertices. The lengths between the two intersections are calculated for each line generating a series of 100 values for the thickness which allows statistical calculations. The mean value is associated with the confidence limit. Also, the histogram of the data is presented.

A region from the [beta] phase must be selected in order to calculate the thickness of the diffusion layer. Due to the very small differences between grey levels of [beta] phase and the substrate, an intensity adjustment was applied to the initial image with oxide layer removed. Also, some morphological operations (closing, removing isolated pixels) were applied.

[FIGURE 4 OMITTED]

After cropping a region from p phase, the image is binaries by a threshold value equal to the mean of the pixel's grey levels extended by [+ or -] 5. The inner contour of the oxide layer was overlaid in order to depict the thickness of the diffusion layer and the picture can be seen in Fig. 6. The diffusion layer's thickness was calculated in the same manner as the thickness of the oxide layer. The two intersections of the vertices were depicted by Boolean operations. A confidence interval is attached to the mean value of the thickness considering a probability of 95%. The histogram of the layer's thickness is shown. The results presented in Fig. 5 can be selected by a pop-up menu.

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

The last objective was to depict some characteristics of the substrate. The pop-up menu has two options:

--Picture--showing the binary image of [beta]--phase;

--Percentage--indicating the percentage of [beta]--phase from the substrate.

For proper calculation the scale image is removed from the picture and the results can be seen in Fig. 6.

3. CONCLUSIONS

Digital image processing is a challenging field, considering that each application requires a specific algoritm, depending on the nature of the target parameters of the investigated process.

In this paper, we have presented the results in designing a GUI for objective, rapid, accurate analysis of SEM images for HVOF-sprayed MCrAlY coatings in order to obtain by statistic analysis the key parameters: the uniformity of the coating, the thicknesses of the oxide and diffusion layer and the percentage of the [beta]--phase from the substrate. An advantage of the GUI is the fact that no expertise in digital image processing techniques is needed for the user.

The authors will further improve and extend the application capabilities in order to meet the specific requirements of other types of protective coating layers analysis.

4. REFERENCES

Neidel, A. et al. (2004). Journal of Metalography, Structure e2, 14, Available from : http://www.struers.com/resources /elements/12/38856/e-Structure%202_EN.pdf. Accessed: 15.03.2008 .

Russ, J.C. & Dehoff, R.T. (2002). Practical Stereology, Plenum Publisher, ISBN 0-306-46476-4, New York.

Sloof, W.G. & Jeurgens, L.P.H, (2004). Microchimica Acta, vol.145, 1-4, p. 220, Wien, ISSN 0026-3672.

Tang, F.; Ajdelsztajn, L. & Schoenung, J.M. (2004). Oxidation of Metals, 61, 219-238, ISSN 0030-770X.

Toma, D.; Brandl, W. & Koster, U. (1999). Oxidation of Metals, 53, 125-137, ISSN 0030-770X.