No abstract

No abstract

No abstract

Extended abstract of a paper presented at Microscopy and Microanalysis 2012 in Phoenix, Arizona, USA, July 29 – August 2, 2012.

Two fundamental assumptions in surface analysis are that the surface is representative of the whole and the analysis area is homogeneous at sampling depth [1]. As surface analysis techniques become more common for analysis of concrete specimens, it is important to test the validity of these assumptions. These assumptions were tested using a method developed to track and summarize the change in composition with depth through analyses with the micro X-ray fluorescence ( XRF) instrument. Typically, the XRF beam size is 30 µm laterally (for laboratory based instruments), and the depth of production for the fluoresced x-rays is approximately 30 µm in concrete [2]. After an initial surface analysis in the µXRF, a concrete specimen surface was ground using three grades of silicon carbide to remove 30 µm of material. Then, another X-ray spectrum image was taken of the same area. The grinding and re-scanning process was repeated 26 times until a total of 780 m of the material were analyzed.The data from each scan were used to develop compositional similarity models for cement and concrete analysis. The model categorizes the surface of the concrete into areas of similar composition, based on the intensity of each element embedded in the pixels of the µXRF image (see Figure 2). A total of seven phases were identified, three of which defined the cement paste based on the varying levels of sulfur in the cement, and four defined the coarse aggregate based on the varying levels of potassium, calcium, and silicon. For each X-ray spectrum image, these seven categories were color coded into an image, as shown in Figure 3. An analysis of relative phase abundances indicates that changes in the aggregate phases occur gradually with depth, whereas alteration in the cement phases can occur immediately. A three dimensional (depth into the material along the Z axis) image of the concrete sample was produced from the derived compositional images. Analysis of the images indicates that random surface selection (where the sample is exposed via successive grinding and polishing routines) and analysis of a concrete sample will show a generally representative phase distribution. However, researchers should interrogate samples at other depths in order to ensure a truly representative sample.