Date of Award

Spring 2021

Document Type

Open Access Dissertation


Chemistry and Biochemistry

First Advisor

Stephen L. Morgan


This dissertation explores and discusses conservation of magnetic audio tapes and discrimination of fiber dyes employing spectroscopic techniques and chemometrics.

Conservation of polyester urethane magnetic audio tapes has become a major challenge since they are susceptible to degradation via hydrolysis. The world’s modern cultural history is vastly recorded in this media, and now conservationists need a non-destructive technique to determine the playability status of these tapes. Once conservationists identify degraded tapes, they can be subjected to the baking process. This process temporarily reverses hydrolysis and provides enough time to digitize tapes, preserving information for future generations. The first three chapters of this dissertation investigates use of attenuated total reflectance Fourier transform infrared spectroscopy (ATR FT-IR) and chemometrics as a nondestructive technique to determine the degradation status of tapes under different circumstances.

The first chapter of this study focuses on determining how reliable it is to collect spectra at the beginning of the magnetic audio tapes. This is a main concern since tapes used in this study are lengthy, and it is crucial to find the best sample location to represent the entire tape to minimize misclassifications. Using two different test sets and six different classification techniques, it was found with above 90% prediction accuracy that taking spectra from the beginning of the tape is probably representative of the degradation behavior of most tapes. The second chapter focuses on building a more robust model to determine the degradation status of problematic tapes. This was successfully achieved by using neural networks and least absolute shrinkage and selection operator (Lasso). The third part of this study explores the degradation taking place along the back-coat and compares the results with the magnetic layer of the same tape identities. Results obtained show poor predictability in the back layer, indicating that the magnetic layer is the preferred side to obtain spectra to determine the degradation status.

The final part of this dissertation discusses the applications of chemometrics and microspectrophotometry to differentiate fiber dyes. This was demonstrated using the blue nylon 6 subgroup with seven different fiber identities. Applying three different machining learning approaches, it was determined that Lasso is probably the best technique to differentiate fiber dyes.


© 2021, Nilmini Hemanthika Ratnasena

Included in

Chemistry Commons