A Statistical Treatment of Non-Normal SEM Data and the Application to Designed Fiber/Filler/Polymer Structures

Abstract

One of the primary objectives of this thesis was to design fiber/filler/polymer structures for newsprint and in the process develop a greater understanding of fiber/filler/ polymer structures. Five different designed structures were created for study. The designed structures were composed of virgin, hydrosulfite bleached, TMP southern pine, Georgian kaolin clay and various polymers. Five filler levels from 0% to 20% were employed with each of these different structures. Numerous physical tests were used to gather data which would help to develop an understanding for the macroscopic properties of the structures.

Paper structures were created and data from bulk physical tests and particle based SEM image analyses were compared. Comparisons were made using a statistical method called Principal Component Analysis (PCA) where the data is grouped and reduced to find data correlations not readily apparent in the raw data.