THIN SECTION GRAIN SIZE ANALYSIS TECHNIQUE IN SILTY MUDROCKS: CALIBRATION OF BACKSCATTERED ELECTRON IMAGE ANALYSIS

Thin section grain size analysis of siltstone is possible using SEM-backscattered electron imaging and image analysis software, calibrated with standards of known grain size. This in-situ method of grain size analysis is superior and less time consuming than disaggregative techniques such as sieving and settling. Dissagregation may alter grain size due to acid dissolution or fragmentation of fractured grains. Thin section analysis is nondestructive, requires less sample and is repeatable. This method may be carried out at mm-scale and used to generate a high resolution grain size record in laminated silty mudrock. The thin section is scanned under backscattered electron (BSE) detection to produce an image that isolates the grains from the matrix. The image is refined by adjusting the contrast and brightness on the SEM to further isolate the peak of the mineral of interest. High-resolution images are then imported into freeware image analysis software Scion Image and/or Image Tool. After noise reduction and thresholding, the program produces an output file with measurements of the major and minor axes of the grains of interest. This data is put into an equation to convert the apparent mean ellipsoidal diameter to mean actual diameter. Thin section-sieve size conversion equations exist for sand size sediment, but none have been published for <63 micron sediment. Linear correlation of BSE images from thin sections against data collected by Sedigraph 5100 was used to determine the constants in the empirical conversion equations for this technique. A manufactured quartz silt standard was used as an ‘anchor’ point when creating the regression to reduce the possibility of biases resulting from disaggregation of rock samples. Sensitivity tests were run to determine the most effective resolution and image size. Problems such as non-spherical clasts, matrix, touching grains, overgrowths, and minerals with similar molecular densities are addressed.