Measures of CD-R Longevity

Contents of this Document

Introduction: Background of this study.

Objective: Outline of purpose and approach.

Degradation Modes: Longevity limitations.

Test Methods: Approach to destructive aging.

Test Results: Observed defects for parameters and errors.

End-of-Life Indicators: Measures of longevity.

Conclusions: Findings of this CD-R longevity study.

INTRODUCTION

Each media type has distinctive capabilities and limitations, including environmental instability. CD-R has become popular for mass storage because of its high capacity, low cost, reliable interchange, and a large installed base of CD-ROM and CD-R drives. Interchange is not assured by readability in one or a few drives. Interchange requires that every disc must be readable in every drive, and is achieved by conformance to ISO and Philips Standards. Comparable standards do not exist for longevity that is defined as satisfactory interchange over some unspecified period of time.

In the absence of an industry standard, CD-R longevity is defined by user requirements. Application dependent expectations and murky vendor claims lead to uncertainty, especially when media is used for long term archiving of valuable information. Longevity can be limited both by media deterioration and by technological obsolescence. Product support cycles are typically 5-10 years, while computer system use rarely exceeds 20 years. Information transfer to upgraded platforms may therefore occur every 10-20 years, requiring significantly longer media longevity.

CD-R discs and drives are technically sophisticated. Users may not fully understand complex requirements that must be satisfied in order to attain interchange and longevity. Because these products are sold as commodities, users may find that performance claims originated by marketing departments may not be supported by accurate test results.

OBJECTIVE

Standards specify methods and limits for a broad scope of tests that confirm interchange. The purpose of this independent study by Media Sciences was to distinguish those tests that can accurately predict media end-of-life. Electrical parameter tests are important because they directly evaluate media compatibility with read drive servos. Soft errors and unpredictable read failure can result from parameters that are outside of tolerance limits. Electrical error tests measure correctable error rates and detect uncorrectable errors caused by local defects. Visual and mechanical inspections discover flaws that may not be detected by high quality test drives.

Evaluations were conducted using destructive test methods that intentionally induced failures. This approach identified end-of-life tests that are important to longevity, and isolated other tests that might produce misleading results. This study did not attempt to evaluate the longevity of various dye types or brands, since these are changing issues where manufacturers must be responsible for claims based upon accurate, objective test results on their products.

DEGRADATION MODES

Exposure to radiation, inks, other chemicals, water, or pollutants can adversely affect CD-R quality. Although environmental degradation should be avoided, a major cause of deterioration is improper user handling based upon overconfidence in the robust construction and error correction of CD-R media. Handling or storage conditions may degrade good discs, especially if the thin, vulnerable protective coating on the label surface is flawed or damaged. In addition, high quality media that is recorded in poor writers may fail interchange standards. Such issues are outside the scope of this study that is focussed on the longevity of media that has been properly recorded and handled.

CD-R discs normally contain four layers consisting of a pre-grooved substrate, a dye layer, a metal layer, and a protective coating. All but the metal layer contain organic compounds that can degrade as a result of changes in their chemical structures or because of unstable defects that grow in size. Mechanical stress induced by rapid environmental changes may result in excessive differential expansion of the various layers and delamination or excessive birefringence. These and other ageing modes limit media longevity.

CD-R ageing processes can be accelerated by high temperature and humidity environments. Media manufacturers may estimate longevity by evaluating discs aged at elevated temperatures and humidities. Extrapolation to ambient conditions, typically 21 C-23 C and 15%-60% RH, then provide a lifetime for the product. Such methods are valid only when a proper end-of-life criteria is applied. BLER is commonly used for this purpose, although supporting evidence has not been published. Some vendors have used a maximum BLER of 50 per sec. to determine end-of-life. Standards ISO/IEC 10149, ANSI/NAPM IT9.21-1996, and others use a BLER limit of 220 per sec. Draft Standard ISO/DIS 12024 required zero E22 and E32 errors in addition to BLER limits.

TEST METHODS

All test methods of Media Sciences were based upon ISO/IEC 10149, and employed test equipments that had been directly correlated to multiple Philips test discs. Discs of various dye types from different manufacturers were tested to ensure that conclusions were representative of current CD-R technology. Ageing was accelerated by storage at elevated temperature and humidity. One set of twenty samples was recorded in a high quality drive prior to storage and testing. Another set of four samples was stored in their unrecorded condition, after which the unrecorded discs were tested, recorded in a high quality drive, and then tested for their recorded properties.

Both pre-recorded and unrecorded sets were tested initially, were loaded into special fixtures, and were then destructively aged by storage for 100 hours at 85 C and 85% relative humidity. Non-condensing ramp-up and ramp-down conditions were maintained in the environmental chamber. After interim testing, the samples were subjected to an additional 100 hours at 85 C and 85% RH, after which final testing was conducted.

TEST RESULTS

Longevity results were not the same for all discs. Significant differences were observed between manufacturers and also between samples from the same manufacturer. No clear differences were observed between dye types within the limited sample. Degradation was more severe for discs of very poor initial quality than for high quality samples, indicating that initial recorded quality was important to longevity for multiple reasons.

Identification of specific degradation mechanisms were not studied, and is the responsibility of each manufacturer. Mechanical tests did not disclose any warping or other physical deformation. Visual examination disclosed penetration of label ink into the dye layer of two samples. This caused severe defects and unreadable discs. Further testing was discontinued on these two samples. The protective coating delaminated on two different samples, resulting in uncorrectable errors where the metal layer was exposed. Comprehensive testing was conducted on these samples. The following tables summarize test results of the two sets of samples.

Minor defects indicate significant, but marginal flaws. Major defects reflect significant interchange risks. Critical defects are expected to cause interchange failures. Parameters are time averages. Errors evaluate local defect density and size.

The following changes, bold designating serious changes, were observed along with attribute results tabulated above:

Reflectance: 12% of the samples changed less than 1%, 12% decreased by 3-6%, and 76% increased by 3% to 5%.

I3/Itop: 76% changed less than 0.01, 18% decreased by 0.03 to 0.09, 6% decreased by 0.22.

I11/Itop: 76% changed less than 0.01, 12% decreased by 0.03 to 0.09, 12% decreased by 0.17 to 0.27.

Asymmetry: 82% changed less than 1%, 6% decreased by 4%, 6% increased by 4%, 6% increased by 19%.

Radial Tracking: 59% changed less than 0.02, 12% increased by 0.03, 29% decreased by 0.06.

Radial Noise: 35% changed less than 1 nm rms, 65% increased by 1 nm rms to 5 nm rms.

Cross Talk: 88% changed less than 0.02, 6% decreased by 0.07, 6% decreased by 0.21.

Radial Contrast After: 82% changed less than 0.02, 18% decreased by 0.05 to 0.12.

Jitter: 53% changed less than 2 ns. 35% increased by 4 ns to 7 ns, 6% increased by 18 ns, 6% increased by 48 ns.

Minor defects indicate significant, but marginal flaws. Major defects reflect significant interchange risks. Critical defects are expected to cause interchange failures. Parameters are time averages. Errors evaluate local defect density and size.

The following changes, bold designating serious changes, were observed along with attribute results tabulated above:

Normalized Wobble Amplitude: 75% changed less than 0.002, 25% decreased by 0.003 to 0.005.

Wobble Carrier-to-Noise Ratio: 50% changed less than 1 dB, 50% decreased by 1 dB to 2 dB.

ILAND: 50% changed less than 1%, 50% increased by 1% to 2%.

IGROOVE: 25% changed less than 1%, 25% increased by 1-2%, 25% decreased by 1-2%, 25% decreased by 2% to 4%.

Radial Contrast Before: 75% changed less than 0.01, 25% increased by 0.06 to 0.08.

ATIP Error Rate: 100% changed less than 1%.

Even though most unrecorded discs passed appropriate quality tests after storage, E22, E32, and BURST errors were more severe in samples that were recorded after storage than in discs recorded prior to storage. It was clear that severe changes in properties of unrecorded discs had degraded media quality so as to seriously affect recording quality.

Further interpretations of these test results are inappropriate because of the variety of manufacturers and dye types that were present in this study. Instead, it should be realized that both attribute and variables data are important in evaluating both the interchange and longevity capabilities of CD-R media, and that one, common degradation mode and end-of-life indicator could not be identified.

END-OF-LIFE INDICATORS

Test data did not identify one universal end-of-life indicator, probably because of the diverse media sources included in this study. BLER was a poor indicator, showing acceptable results for samples that failed other important quality requirements. This is not surprising, since BLER does not distinguish between easily correctable errors and severe uncorrectable errors, and fails to evaluate important parameters that can affect read drive servos. Only 47% of the failed discs had unacceptable BLER values in excess of 220 per sec., 14% of the failed discs had BLER values between 100 and 220, and 14% had BLER values between 50 and 100. BLER values below 50 per sec., usually characteristic of high quality media, were present for 25% of the failed discs.

Increases in mark length as measured by beta, asymmetry, or effect length, have been proposed as an indicator. Only 6% of the samples used in this study showed such changes, but indicated a decrease in mark length. The onset of E22, E32, or BURST errors clearly indicate end-of-life, but does not allow extrapolation over multiple storage time intervals to yield a failure time. Both total and peak E12 error rates as well as jitter may be appropriate for extrapolation purposes, but each manufacturer should determine the end-of-life measures appropriate to their processes.

CONCLUSIONS

CD-R discs are capable of excellent longevity, but achieving that potential requires diligence by both manufacturers and users. Manufacturers claims may be valid, or may be based upon flawed or non-existent data. Proper end-of-life indicators must be used to estimate longevity. This study has shown that BLER is not a universal indicator of media life, although most published longevity estimates have utilized BLER as the sole end-of-life indicator.

E22, E32, and BURST errors are valid end-of-life indicators. When present, they indicate a need for immediate duplication if a disc containing archival information is still readable. Such errors are not useful for estimating media life by extrapolating test results of discs that have been subjected to accelerated ageing. All quality indicators must be considered when selecting end-of-life indicators. This study suggests that total and peak E12 error rates as well as jitter may be useful indicators, provided that all other quality requirements are met.

High initial quality for each disc can only be achieved by managing variations in media and recording drive quality. Individual CD-R lot qualification should be employed where possible to confirm that manufacturing quality was high and was not degraded by subsequent packing, shipping, and storage events. Media handling and storage is very important. Both unrecorded and recorded disks should be archived in clean jewel cases in a stable storage environment of 10 C-15 C and 20%-50% RH, and protected from sunlight and other radiation sources.

CD-R media and drive manufacturers are responsible for product quality levels that support interchange and longevity. Not all manufacturers achieve this goal. Increasing demand may require new facilities or production lines that inevitably undergo growth pains. Technical advances can lead to new manufacturing processes that must be debugged. Price pressures may force compromises in quality that adversely affect baseline quality or increase fluctuations about that baseline. Identification of CD-R media and drives that support consistent, acceptable levels of interchange and longevity is the responsibility of the archivist. Proper vendor qualification and monitoring enables the user to confidently utilize the rich capabilities of CD-R, and rewards the manufacture with recognition of their diligent efforts to attain and maintain product quality.

Reliance upon brand names or upon readability of discs in one or a few drives cannot verify longevity. Confidence in longevity can only be achieved by initial testing of drives and media, through proper handling and storage, and by periodic resampling to confirm longevity or to identify a need for duplication while the original disc is still readable. Short cuts do not exist. The level of confidence will always be proportional to the amount of effort and expense incurred by the archivist in establishing and maintaining a high level of CD-R quality. Such methods are appropriate for all media types, and their proper application to CD-R information storage will satisfy the most critical requirements for interchange and longevity.