Andrew Wilson, Editor, [email protected]
Material grain size is an important measurement that allows metal fabricators to estimate mechanical properties such as strength, ductility, and hardness. Because of this, it is vital that material grain characteristics are quantified in an accurate manner. In the past, this was done manually using optical microscopes. However, such methods are by nature highly subjective and require extensive operator training. Today, semiautomated image analysis systems can be used to analyze grain sizes automatically.
“It is difficult to correctly segment grains from a micrographic image,” says Gervais Gauthier, European sales manager of ADCIS (Hérouville Saint-Clair, France; www.adcis.net), “because chemical etching, which should reveal the boundaries, can quickly erode the grains if applied for too long. Conversely, grain boundaries may not become present if too little chemical etching is performed.”
To automate this process, the American Society for Testing and Materials (ASTM; West Conshohocken, PA, USA; www.astm.org) has developed a number of methods to estimate and express the average grain size of all single-phase metals. Using the ASTM E112-96 standard, metallurgists can now automatically evaluate the average grain sizes of metals. “The advantage of this method against a method that segments the grains from images is its robustness,” says Gauthier. Such test methods can also be applied to other materials that have similar appearances to metallic structures.
Two main methods exist within the ASTM E112 standard to provide a grain size number. The first, a planimetric method, involves counting the number of grains within a known area. The numbers of grains per unit area then determine the grain size number. Rather than count the number of grains per unit area, the ASTM intercept method counts the number of grains intercepted by a test line overlayed on an image of the specimen. The number of grain boundary intersections per unit length of this test line then determines the grain size. A European standard, EN ISO 643:2003, uses similar methods to compute grain sizes based upon the ASTM standard.
“The precision of the method,” says Gauthier, “is a function of the number of intercepts or intersections counted. And because an accurate count can be made without the need to mark intercepts or intersections, the intercept method is faster than the planimetric method for the same level of precision. However, the drawback of the ASTM E112 is that it provides only an average grain size while the planimetric method is based on grain segmentation and provides a size distribution.”
Mark E. Cavaleri and his colleagues at 3M’s Microscopy Laboratories (St. Paul, MN, USA; www.3m.com), are routinely computing grain sizes using tools based on the company’s Aphelion image-processing software. Based on the ASTM standard, the software provides Calvaleri and his colleagues with a means to automate the procedure of materials grain size counting.
“Before any material can be analyzed,” says Gauthier, “images of the specimen are first placed under a microscope, usually with 10X or 20X magnification. The operator then defines a set of test lines by specifying a number of equally-spaced horizontal and vertical lines across the image.
After images are digitized into a host computer, it is important to remove any etched parts of the sample which are not grain boundaries and are not connected (see figure). This is done by thresholding, skeletonizing, and filtering the image to extract region-based information.”
An estimate of the number of grains intersected by each of the test lines is then determined by counting the number of intercepts along the lines. The total length of the test lines and the estimated average size of the grains are then computed and statistics displayed in a Microsoft Excel template.
