AN601 EEPROM Endurance Tutorial Author: BASIC TERMS David Wilkie Reliability Engineer INTRODUCTION The endurance of an EEPROM-based device will be quoted by a manufacturer in terms of the minimum number of erase/write cycles (write cycles) that the device is capable of sustaining before failure. A write cycle is generally considered to be the operation that changes data in a device from one value to the next. There are several EEPROM-based devices available on the market. Microchip Technology Incorporated makes three general types of EEPROM-based product: Serial EEPROMs, Parallel EEPROMs, and EEPROM-based Microcontrollers. As a manufacturer of many EEPROM products, Microchip is concerned with endurance and continues to try to educate its customers on the importance of this topic. There are many differences in the interpretation of "endurance" that can result in misleading or inaccurate information being used in design decisions. This paper hopes to clear up any questions that the customer may have in the subject of endurance, without becoming so technical that the information given is not helpful. There is no widely used standard for any type of endurance test. Each manufacturer will use their own endurance testing methodology. This report will describe all the testing options, and which tests Microchip performs on its EEPROM-based products. The MIL-STD cycling test (Method 1033) has not been updated since 1977 and is well out of date as applied to EEPROM non-volatile memories. The standard does not distinguish the difference between block cycling and byte cycling, and gives a very poor failure criteria. Microchip does not use this standard. The definition of "endurance" (as applied to EEPROMs) in the first part of this introduction contains various words and phrases that require clear definition and understanding. As shown in the following paragraphs, different manufacturers will use different standards. "Endurance cycling" is a test performed by all manufacturers (and some customers) to determine how many "write cycles" the product will achieve before failing. The "minimum number of write cycles" is the least number of times that you can expect to subject the product to a "write cycle" before it fails. "Failure" is a somewhat arbitrary definition, since a device only truly fails when it no longer meets the customer expectation, and does not operate in his system. A failure can be defined in this, the loosest of definitions, or the most stringent of definitions (whereby a device would fail if it did not meet any of the data sheet parameters), as well as a wide range in between. For example, if the device did not correctly store data into a particular address that the customer was not using, then the device would work correctly for the customer but would fail a functional test set by the manufacturer. Likewise, if the device drew more current than the data sheet specified after some time, but the customer application could supply the current needed, the device would work in the customer application but would fail a parametric test set by the manufacturer. Microchip uses the most stringent definition: A failure occurs when the device fails to meet any data sheet condition under any guaranteed operating condition of temperature and voltage. The number of devices that can fail before a particular endurance criteria is not met is also somewhat flexible. Even the most quality conscious manufacturer will occasionally have a failure, so a failure level is defined. The industry standard conditions for many types of reliability tests are set by JEDEC (the Joint Electronic Device Engineering Council). JEDEC defines that if 5% or less of a given sample fails at a given endurance goal, then that goal has been met. For example, if a sample of 100 units are endurance cycled to 1 million cycles and 3 have failed at 100,000 cycles and a further 7 have failed at 1 million cycles, then the sample would have an endurance of 100,000 cycles. 1995 Microchip Technology Inc. DS00601A-page 1 Thi d t t d ith F M k 404 AN601 Microchip uses a more stringent criteria for endurance: no more than 2.5% of devices can have failed for the given endurance goal to have been met. A "write cycle" is also a somewhat flexible definition since almost every customer will write the device in a different way. For example, if the customer application uses only the first three bytes of the array to store variable data, and the remainder of the array is used as a lookup table, then a write cycle will be complete when the three data bytes have been re-written to their new data state. A write cycle is often described as an erase/write cycle, since almost all technologies employ an "auto-erase" before the data is actually written to the array. This is also used by Microchip, but we will use the term "write cycle" since the auto-erase is invisible to, and cannot be suppressed by, the customer. The term "data changes" is occasionally used in place of "write cycle" or "erase/write cycle." A data change will occur when an auto-erase cycle is initiated, and a second data change will occur upon the write cycle, therefore, one "erase/write cycle" is equivalent to two "data changes." The term "data change" also implies that a different type of cycling is being used than "erase/write cycle." This will be described later. System Design Considerations There are a number of design considerations that the system designer can use to maximize the endurance of an EEPROM-based device, if endurance is the application's limiting factor. As will be described in more detail later, if the designer has any control over certain environmental or operation conditions he should observe the following basic guidelines: * Keep the application temperature as low as possible * Keep the application voltage (or the VCC voltage on the EEPROM-based device) as low as possible * Write as few bytes as possible * Use page write features wherever possible * Write data as infrequently as possible With these basic guidelines applied to the fullest extent, the endurance of EEPROM-based devices can be extended well beyond the guaranteed minimum endurance. Under certain very specific conditions, Microchip Serial EEPROMs have been shown to last for well over 100 million cycles. The term "write cycle" does not define under what conditions the cycling was done (unless explicitly stated) nor does it define the type of cycling that was done. The endurance cycling can be done at any number of conditions of voltage and temperature (e.g., 85C and 5.5V, or 25C and 5.0V) that may or may not meet with a customer's application. The cycling mode used in endurance cycling can affect the endurance of the product. All these effects will be described later. Microchip uses the most stringent conditions that are reasonable for endurance cycling. We use byte or page mode cycling at a temperature of 85C at 5.5V. All data not explicitly defined at other conditions is taken at these conditions. A "read cycle" is completed when any number of bytes of data have been read from the device. For the FLOTOX-design EEPROM-based devices made by Microchip a read cycle does not affect endurance, since the data in the EEPROM is not changed. Other technologies, such as Ferroelectric technology, may have a limited number of read cycles since data is corrupted during a read. DS00601A-page 2 1995 Microchip Technology Inc. AN601 WRITE MODES IN EEPROMS There are three ways that EEPROM-based devices can have the entire array data contents changed. These are: byte mode, page mode, and block (or bulk) mode. Some types of devices support all three modes, others may only support one or two modes. The mode that you use to write an EEPROM-based device will affect the long term endurance of the product. Byte mode writing is used when the contents of the array are changed one byte at a time. For many devices this is the only user-accessible write mode available. To change the entire contents of a Serial EEPROM in this way would take up to 10 seconds (using 10 ms per page on a 64K Serial EEPROM). Parallel EEPROMs such as a 28C64, which have a faster byte write time (1 ms rather than 10 ms), but no page mode would also take up to 10 seconds. Page mode writing is a popular feature on many new designs of EEPROM memory products. This feature allows up to 8 bytes of data to be written to the memory in the same time that one byte would normally take. In this mode, the write time for a 64K Serial EEPROM can be cut from eight seconds to one second. Block cycle is generally a test mode used by EEPROM manufacturers to make it easier to test the products. Some types of EEPROM-based products have these modes as user options (such as the ERAL and WRAL mode in 93CXX products, or the Chip Clear mode in 28CXXX products) but generally this mode is not user accessible. A block write can be done in as little as 1 ms, allowing millions of write cycles to be completed in a few hours. A general rule to follow in choosing write modes is that the larger the number of bytes being written in a single instruction, the longer the device will last. For example, in byte mode a device might start to fail after 300,000 cycles under a particular set of conditions, but the device may last 600,000 cycles in page mode under the same conditions. In block mode the device might last 1 million cycles, under the same conditions. 1995 Microchip Technology Inc. The reason for this is related to the internal design of any FLOTOX EEPROM-based product. In these devices, an internal "charge pump" takes the applied VCC voltage (typically 2.5 to 5.5V) and increases it to 15 to 25V. This voltage is required to induce the "Fowler-Nordhiem Tunneling" or "Enhanced Emission" effects that are used to program and erase EEPROM-based devices. The specific effect that is used is manufacturer-dependent. Microchip EEPROMs program by Fouler-Nordheim Tunneling. The charge pump voltage is used to program however many EEPROM-cells are being programmed. For example, in byte mode, all the cells in a byte (8 to 16) are biased with the charge pump voltage. In block mode, all the cells in the array (up to 100,000, depending on the device) are biased with the charge pump voltage. The charge pump is like a current source during conditions of high load, so the voltage put out by the charge pump will be reduced slightly if more bytes are being written. If the whole array is being programmed then the charge pump voltage will be significantly reduced, but the programming current Ipp will be very high. Generally, the lower the charge pump voltage the better the endurance (there is a limit since the charge pump voltage needs to be high enough to program the cell) and so the best endurance is generally achieved by using block mode cycling. Page mode is worse than block mode, but better than byte mode. Block mode is generally not a very useful cycling mode to the end user, since the data contents in the whole array will be changed to the same value (generally 00 or FF). When Microchip tests EEPROM-based products we use byte mode cycling on devices which do not have a page mode, and page mode cycling for those that do. We encourage our customers to use page mode writing on all products which have page mode, to ensure high endurance. DS00601A-page 3 AN601 ENDURANCE TESTING METHODOLOGIES Different manufacturers use different ways to both endurance cycle and test EEPROM-based products. There is no standard for endurance cycling, or testing devices after cycling. There are two groups of testing that Microchip performs on all products: qualification and production. Qualification testing is done for all new products, and major changes to a product or manufacturing process. Production testing is done on all devices shipped to customers. Qualification testing at Microchip is used to test the reliability of new products, and to guarantee that the device is reliable. A great deal of testing is done, including endurance testing on all EEPROM-based products. Endurance cycling is done at the maximum rated data book value, generally 85C or 125C. After the rated number of cycles (10,000, 100,000, or 1,000,000) the sample (around 300 from multiple wafer lots) is tested to a full production test program. After endurance, the units are subject to "data retention" to guarantee that the required 40 years of data retention will be achieved, after the maximum number of cycles has been completed. Endurance cycling is done under the conditions previously described, and the data retention test is performed after this. After the data retention stress is completed (which takes six weeks) the devices are tested again, to confirm the functionality of the device to all data sheet parameters. No more than 2.5% failures are allowed after endurance, and a failure rate higher than 100 FITs after 1000 hours of data retention stress (equivalent to more than 10 years at 55C) is unacceptable. DS00601A-page 4 Production testing is done by Microchip on all devices shipped to a customer. Production testing begins immediately after a wafer lot is finished being processed, continuing in various stages until the devices are shipped to a customer. The first tests that are done on EEPROM-based products at Microchip are called wafersort. They are done before the wafer is cut up into dice for assembly. There is a series of tests which include large numbers of write cycles (up to 5000) to ensure reliability by weeding out weak devices so they never get shipped. After assembly full testing is done which includes further write cycles across the guaranteed temperature range, to ensure device functionality at the temperature extremes. After the normal production testing a sample of 128 units is taken from every wafer lot, and cycled to 10,000 cycles (Parallel EEPROMs) or 1,000,000 cycles (Serial EEPROMs and EEPROM microcontrollers) at 85C using the conditions already described. After endurance testing the devices are tested for functionality at 85C. Any sample of Parallel EEPROMs which shows significant failures at the end of cycling causes the entire wafer lot to be pre-cycled prior to production testing. Serial EEPROMs and EEPROM-based microcontrollers do not receive the same disposition. Lots with significant failures come under scrutiny and full failure analysis with corrective actions is done. This is, however, a rare occurrence. The testing that Microchip does is unique. Manufacturers will generally do different testing from each other. Microchip firmly believes that our testing ensures excellent quality and reliability. 1995 Microchip Technology Inc. AN601 THE EFFECT OF TEMPERATURE ON EEPROM ENDURANCE THE EFFECT OF VOLTAGE ON EEPROM ENDURANCE The temperature at which cycling is done will affect the number of write cycles that can be executed before the device fails. The higher the temperature, the worse the endurance will be. Generally, and approximately, a device which fails at 10 million cycles at 25C will fail at 2 million cycles at 85C and 1 million cycles at 125C. The reasons for this are not conclusive (although there is much technical literature supporting one theory or another) but it is apparent that the failure mode of EEPROM cells (generally considered to be electron trapping in the tunnel dielectric causing shielding and dielectric breakdown) is strongly dependent on temperature. The voltage at which a device is written can also affect the endurance. This is simply because the charge pump (used to program and erase EEPROM cells) is more powerful at higher voltages. As has already been described, a higher programming voltage will reduce the endurance of an EEPROM cell, and a stronger charge pump will produce a higher voltage. Data taken by Microchip suggests that if the typical failure of an EEPROM-based device is 10 million cycles at 25C, the mean failure will occur at other temperatures according to the following table: TABLE 1: TEMPERATURE MEAN FAILURE Write Cycle Temperature Mean Failure (Cycles) -40C 37.1 million 0C 16.7 million 25C 10.0 million 40C 7.4 million 55C 5.4 million 70C 4.0 million 85C 2.9 million 100C 2.2 million 125C 1.3 million Data taken by Microchip suggests that if typical failure of an EEPROM-based device is 1 million cycles when endurance cycling is done at 5.5V, mean failure occurs at other temperatures according to the following table: TABLE 2: VOLTAGE MEAN FAILURE Endurance Cycling Voltage Mean Failure (Cycles) 5.5V 1.0 Million 5.0V 1.2 Million 4.5V 1.4 Million 4.0V 1.7 Million 3.5V 2.0 Million 3.0V 2.4 Million 2.5V 2.8 Million 2.0V 3.3 Million This data was taken on Microchip FLOTOX Fowler-Nordhiem Tunneling EEPROMs and formed a part of the data set used to create the Total Endurance disk. Other technologies (such as FLOTOX Enhanced Emission or Ferroelectric technologies) may have different characteristics. This data was taken on Microchip FLOTOX Fowler-Nordhiem Tunneling EEPROMs and formed a part of the data set used to create the Total Endurance disk. Other technologies (such as FLOTOX Enhanced Emission or Ferroelectric technologies) may have different characteristics. As is clearly seen, any cycling done at 25C can be misleading in the extreme if the application requires a device that can be cycled 10 million times at, say, 55C. 1995 Microchip Technology Inc. DS00601A-page 5 AN601 THE EFFECT OF WRITE MODE ON EEPROM ENDURANCE THE TOTAL ENDURANCE PREDICTIVE SOFTWARE As has been discussed there are three basic ways of writing data to an EEPROM-based device: Microchip has a Windows-based software model called Total Endurance. This program, based on all of the customers endurance parameters, will predict the failure level at the expected end of application life. This tool is invaluable for system designers who would like to fine-tune their application in favor of endurance. It is available now from your local Microchip distributor. * Byte mode * Page mode * Block mode This is related to the strength of the charge pump in applying the required programming voltages to the EEPROM cells. Data taken by Microchip suggests that if the typical failure of an EEPROM-based device is 1 million cycles when the endurance cycling is done in byte mode, the mean failure will occur in other modes according to the following table: TABLE 3: ENDURANCE MEAN FAILURE Endurance Cycling Mode Mean Failure (Cycles) Byte 1.0 Million Page 4.6 Million Block 13.2 Million This data was taken on Microchip FLOTOX Fowler-Nordhiem Tunneling EEPROMs and formed a part of the data set used to create the Total Endurance disk. Other technologies (such as FLOTOX Enhanced Emission or Ferroelectric technologies) may have different characteristics. This data was taken from a Microchip 24LC16. As you can see, the use of the block cycle data to guarantee endurance can be misleading. DS00601A-page 6 1995 Microchip Technology Inc. AN601 NOTES: 1995 Microchip Technology Inc. DS00601A-page 7 WORLDWIDE SALES & SERVICE AMERICAS ASIA/PACIFIC EUROPE Corporate Office Microchip Technology Inc. 2355 West Chandler Blvd. Chandler, AZ 85224-6199 Tel: 602 786-7200 Fax: 602 786-7277 Technical Support: 602 786-7627 Web: http://www.mchip.com/microchip Atlanta Microchip Technology Inc. 500 Sugar Mill Road, Suite 200B Atlanta, GA 30350 Tel: 770 640-0034 Fax: 770 640-0307 Boston Microchip Technology Inc. 5 Mount Royal Avenue Marlborough, MA 01752 Tel: 508 480-9990 Fax: 508 480-8575 Chicago Microchip Technology Inc. 333 Pierce Road, Suite 180 Itasca, IL 60143 Tel: 708 285-0071 Fax: 708 285-0075 Dallas Microchip Technology Inc. 14651 Dallas Parkway, Suite 816 Dallas, TX 75240-8809 Tel: 214 991-7177 Fax: 214 991-8588 Dayton Microchip Technology Inc. 35 Rockridge Road Englewood, OH 45322 Tel: 513 832-2543 Fax: 513 832-2841 Los Angeles Microchip Technology Inc. 18201 Von Karman, Suite 455 Irvine, CA 92715 Tel: 714 263-1888 Fax: 714 263-1338 New York Microchip Technology Inc. 150 Motor Parkway, Suite 416 Hauppauge, NY 11788 Tel: 516 273-5305 Fax: 516 273-5335 San Jose Microchip Technology Inc. 2107 North First Street, Suite 590 San Jose, CA 95131 Tel: 408 436-7950 Fax: 408 436-7955 Hong Kong Microchip Technology Unit No. 3002-3004, Tower 1 Metroplaza 223 Hing Fong Road Kwai Fong, N.T. Hong Kong Tel: 852 2 401 1200 Fax: 852 2 401 3431 Korea Microchip Technology 168-1, Youngbo Bldg. 3 Floor Samsung-Dong, Kangnam-Ku, Seoul, Korea Tel: 82 2 554 7200 Fax: 82 2 558 5934 Singapore Microchip Technology 200 Middle Road #10-03 Prime Centre Singapore 188980 Tel: 65 334 8870 Fax: 65 334 8850 Taiwan Microchip Technology 10F-1C 207 Tung Hua North Road Taipei, Taiwan, ROC Tel: 886 2 717 7175 Fax: 886 2 545 0139 United Kingdom Arizona Microchip Technology Ltd. Unit 6, The Courtyard Meadow Bank, Furlong Road Bourne End, Buckinghamshire SL8 5AJ Tel: 44 0 1628 851077 Fax: 44 0 1628 850259 France Arizona Microchip Technology SARL 2 Rue du Buisson aux Fraises 91300 Massy - France Tel: 33 1 69 53 63 20 Fax: 33 1 69 30 90 79 Germany Arizona Microchip Technology GmbH Gustav-Heinemann-Ring 125 D-81739 Muenchen, Germany Tel: 49 89 627 144 0 Fax: 49 89 627 144 44 Italy Arizona Microchip Technology SRL Centro Direzionale Colleoni Palazzo Pegaso Ingresso No. 2 Via Paracelso 23, 20041 Agrate Brianza (MI) Italy Tel: 39 039 689 9939 Fax: 39 039 689 9883 JAPAN Microchip Technology Intl. Inc. Benex S-1 6F 3-18-20, Shin Yokohama Kohoku-Ku, Yokohama Kanagawa 222 Japan Tel: 81 45 471 6166 Fax: 81 45 471 6122 9/22/95 All rights reserved. 1995, Microchip Technology Incorporated, USA. Information contained in this publication regarding device applications and the like is intended through suggestion only and may be superseded by updates. No representation or warranty is given and no liability is assumed by Microchip Technology Incorporated with respect to the accuracy or use of such information, or infringement of patents or other intellectual property rights arising from such use or otherwise. Use of Microchip's products as critical components in life support systems is not authorized except with express written approval by Microchip. No licenses are conveyed, implicitly or otherwise, under any intellectual property rights. The Microchip logo and name are registered trademarks of Microchip Technology Inc. All rights reserved. All other trademarks mentioned herein are the property of their respective companies. DS00601A-page 8 1995 Microchip Technology Inc.