LONGEVITY OF CD MEDIA RESEARCH AT THE LIBRARY OF CONGRESS

Investigations into the aging of CD-Audio media currently in progress at the Library of Congress are described. Aging trends under natural as well as accelerated conditions are under study. A natural aging study consisting of a pilot project designed to monitor the long-term effects of real-life storage and use conditions on the integrity of a limited sample population selected randomly from the Library’s CD collection is described. This sample population is retained within the collection and remains subject to normal wear and tear, and is withdrawn only briefly for periodic testing every few years. Results obtained over the past seven years from this pilot project are discussed with implications for the future course of this investigation. Also presented is the status of an accelerated aging study of CD-Audio media currently in progress. Visual observations of defects generated as a result of the accelerated aging process are presented. The goal of these studies is to gain an understanding of the scope of the challenge that may lie ahead for the longterm preservation of a contemporary CD collection that must inevitably comprise a wide range of compositional characteristics in every layer of its manufacture. It is hoped that this research will ultimately enable us to 1) provide the best long-term protection for our collections, 2) establish relevant parameters for monitoring CDs that would signal the need for creating backup copies or migration of data, and 3) identify the failure mechanisms that arise form deficiencies in the manufacture of CD’s. Introduction Ever since their advent, optical storage media have had a strong appeal for preservation-conscious institutions because of their presumed potential to last indefinitely since they are not subject to the same wear and tear as other recorded media. This is why the Library of Congress has long been committed to gaining an understanding of the lasting qualities of optical storage media. The earliest studies at the Library date back to accelerated aging studies with 12-inch optical WORM discs during the mid eighties. An accelerated aging study was undertaken to investigate the application of the Arrhenius aging methodology for the determination of an end-of-life point for the discs in question. Specific test procedures were developed for these analyses and formalized in ANSI standard X3.199 ­ 1991 (1). Later, with support from NIST, AIIM developed an error-monitoring scheme for the WORM, MO, and the CD optical media (2,3). These tests involved accelerated aging of the discs in an elevated temperature and humidity environment typically for repeated periods of 500 to 2000 hours to accelerate potential chemical changes that may affect the operational characteristics of the discs. At about that time, IT9, an ANSI standards committee, was developing life-expectancy procedures for digital optical media and approved three standards dealing with CD­ ROM media (4). A primary objective of the current research is to assess the preservation needs of the CD-Audio collections at the Library of Congress, and to devise strategies to minimize any data losses that may result as the collections continue to age. To this end, an essential first step is to gain an understanding of failure modes by monitoring errors that may accumulate upon aging of CDs. New insights might also be gained through a scientific examination of any visually perceptible defects as well. It is our hope that we will ultimately be able to link compositional characteristics of individual CDs to failure mechanisms in a way that would enable us to flag more vulnerable CDs for replication or backup so as to preserve the data before any of it is lost. The effects of the aging process on CDs have been studied in this work under accelerated as well as natural aging conditions. The accelerated aging study was modeled after the ANSI standard mentioned above (4). Although these experiments are still in progress and some of the data sets have yet to be completed, several interesting observations have been recorded about the development of physical defects that are clearly visible in some of the discs. The nature of these defects and their impact on the playability of the discs has been presented here. However, like most accelerated aging experiments, this series of experiments also begs the question whether the failure modes observed here are truly representative of phenomena that occur in real life, and if the differences that will soon emerge in the aging of CDs with different manufacturing characteristics would ultimately stand the test of time. In anticipation of these inevitable doubts, a natural aging study that monitors the aging of a representative sample of CDs was initiated in 1996. With the limited resources for testing of CDs available to us at that time, this initiative was restricted to a pilot study with a limited sample of 125 discs selected randomly from the total population of the Library’s CD collection, which at that time was estimated to consist of 60,000 items. While this was not a statistically significant sample, its size was deemed adequate enough to enable us to justify a full-fledged natural aging study if significant data errors were to develop in even a few discs. Indeed, that is exactly what has transpired as will be seen from the data presented below. In the eight years since this study was initiated, the size of the Library’s CD collection has mushroomed from 60,000 to 380,000. A statistically valid random sample is being assembled at present for a more definitive natural aging study. I. Natural aging study