Download presentation
Presentation is loading. Please wait.
Published byElijah Lane Modified over 9 years ago
1
Design of Advanced Erase Mechanism for NOR Flash EEPROM Amit Berman, June 2006 Intel Corporation
2
Agenda Flash Memory Introduction Project’s Targets Current Erase Solution Proposed Methodology for New Mechanism Lab Results and Evaluation Conclusions Questions Session AEM project
3
Flash Memory Introduction AEM project Substrate p- Drain n+ Source n+ Tox FG ONO Gate Bit-line (Vd) Control Gate Wordline (Vg) - - - - - - Source (Vs) - - What is Flash? - Non-Volatile Memory - Examples: Cell-phone OS, PC BIOS MOSFET with floating gate: Information is stored as electrons in the floating gate - Many Electrons – Logic ‘0’ - Few Electrons – Logic ‘1’ Best memory alternative: Low cost, Low power, High density and High speed.
4
Flash Memory Introduction (2) AEM project Id ( A) Vg (V) E 5 0 12 3 45 R P Program - CHEErase – FN tunnelingRead – Sensing Substrate p-p- Drain n+ Source n+ Tox FG ONO Gate Bit-line (Float) Control Gate Wordline Vg << 0 - - -- - - - Source (Float) - - - - - - -- - Vwell=positive Vg Substrate p- Drain n+ Source n+ Tox FG ONO Gate Vread Control Gate Vcc - - - - - - Source - - Sense Amp Output REF SEN Drain I I
5
Project ’ s Targets AEM project Speed-up Erase operation (bottleneck 0.9sec per block) Improve product reliability by reducing “Post Erase Repair” phenomena (DPM) Reduce product test time
6
Current Erase Solution Channel Erase Pulses Block basis (shared bulk) Increasing pulse series Erase Verify – Vt Measurement (read) until all block is erased Disadvantages: Since not all cells are similar,some will be over-erased : always logic ‘1’. Post erase repair – soft programming is needed. AEM project
7
Proposed Methodology Detect an “Optimum Point” on channel erase where fast and normal bits are erased but slow-erased bits are programmed. Handle Slow-erased bits with “special care” – erase via drain. Reduce over-erased cells to minimum. AEM project
8
Lab Results and Evaluation We can identify slow-erased bit by constant charge loss in each erase pulse. Flash Vt will be briefly reduced. Model for slow-erased bits: on erase pulse #10. Trade-off between reducing over-erased bits and erase speed. Erase operation speed is improved from 0.9sec per block to 0.7sec per block. PP project
9
Conclusions Potential speedup: future implementation on embedded flash microcode and on test program platforms. Quality and Reliability check is needed. Examine cost-effective trade-off with Q&R and device operation speed. PP project
10
Questions Session AEM project
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.