Textures (sizes, shapes, distributions) of glass inclusions in crystals can provide important insight into magmatic processes. Three-dimensional X-ray tomography is an excellent nondestructive method for assessing igneous rock textures in general; however, standard absorption-contrast tomography is not useful for imaging glass inclusions because it typically does not provide enough contrast between the inclusion and crystal host to allow quantitative separation of these phases during image processing. Propagation phase-contrast tomography enhances contacts between different phases, allowing the use of edge-detection algorithms to separate glass inclusions from their host crystal. The sample-to-detector distance is increased to achieve edge enhancement at the cost of some loss in image resolution. We qualitatively assessed images acquired at a series of distances (30, 100, 150, 200 mm) to determine the best distance for imaging given this trade-off.

Typical image processing approaches used for absorption-contrast tomography are not useful for edge-detection problems, so we developed an IDL-based graphical user interface (GUI) to process propagation phase-contrast tomograms. A combination of grayscale filters, distance thresholding, erosion processes, and individual picking of outlier voxels is used to fit a convex hull to an individual inclusion, and the output provides information on its size and shape. The processed inclusions can be placed in the context of the original host, such that size, shape, and location of multiple inclusions can be assessed individually or in combination.

We apply this method to five glass inclusions in quartz crystals from early erupted Bishop Tuff. Results indicate that inclusions range in shape from nearly spherical to nearly ellipsoidal and from round to faceted. The relationship between degree of faceting, size, and location within the host is consistent with melt inclusion faceting under magmatic conditions over centennial to millennial time scales. We also show application of this method to other kinds of studies, such as glass inclusion size distributions.

You do not currently have access to this article.