1 / 11

Background Material

Background Material. Top Metal Layer (Metal 3 in RT54SX32S, Metal 4 in RT54SX72S). Barrier Metal Layer. Amorphous Silicon. Barrier Metal Layer. Second Metal Layer (Metal 2 in RT54SX32S, Metal 3 in RT54SX72S). What is an Antifuse?. Antifuses start life as capacitors

nevin
Download Presentation

Background Material

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Background Material

  2. Top Metal Layer (Metal 3 in RT54SX32S, Metal 4 in RT54SX72S) Barrier Metal Layer Amorphous Silicon Barrier Metal Layer Second Metal Layer (Metal 2 in RT54SX32S, Metal 3 in RT54SX72S) What is an Antifuse? • Antifuses start life as capacitors • Layer of amorphous silicon between layers of titanium nitride barrier metal

  3. Conducting Filament Programming an Antifuse • Programming • High voltage causes the amorphous silicon to break down • As conduction begins to occur, localized heating causes the amorphous silicon and barrier metal to fuse, forming a conducting filament • Careful control of voltage and current during programming pulses ensures tight distribution of programmed antifuse resistance

  4. About Programmed Antifuse Failures • 11 Confirmed failures. Other cases are suspected. • "Cluster of failures" - small number of users reporting a small percentage of failures on 0.25µm MEC SX-A/RTSX-S devices • Failures are occurring in a “dirty” or "out-of-spec" environment • Data strongly suggests early failures under "dirty" electrical environment • Experiments ongoing: • Determine reliability under “in-spec” conditions with respect to I/O and VCCA noise • Determine sensitivity to “out-of-spec” conditions, with respect to I/O and VCCA noise.

  5. Usage History • More than 1.2 Million commercial 0.25 µm MEC SX-A units shipped • Four aerospace customers with confirmed failures report an approximate failure rate of 2%-7% • These are from a small sample of shipped units • Actel does not see an overall 2% functional failure return rate • Actel Failure Analysis group reports 58 units of SX-A products submitted for analysis in the last 2 years • One of these 58 units were traced to antifuse damage • Breakdown of other failure causes • Design Timing Fault • EOS of the non-antifuse nature (e.g., clock pin EOS damage, ref IAT) • ESD

  6. Actel Published MEC 0.25µm Reliability Data • High Temperature Operating Life(HTOL): • 1 antifuse failure out of 805 devices • Equivalent 601,440 device hours • Failure occurred after 184 hrs (1st pull point) • The burnin system induced out of specification spikes of 9V on VCCA. This sequence has been since corrected. • Low Temperature Operating Life(LTOL): • 0 failures out of 334 devices • Equivalent 309,000 device hours

  7. EOS Model Reliability Data • Clean Burn-in (no VCCA spikes observed >3V) • 0/77 failures for 168 hrs HTOL at 150°C • No I/Os switching • Special burn-in stress: antifuses are stressed with operating current levels using VPP • 0/100 failures at 168 hrs HTOL. Subsequent additional 168 hours in LTOL had zero failures. • Contractor X application • Contractor X found zero failures in their applications. Typical designs were 33MHz, one at 120MHz. • Contractor X measured I/O undershoot of –1.8V. But only 1 or 2 pins switching. • On the PCI bus with 33 Simultaneously Switching Outputs, care was taken to limit the undershoot to less than 500 mV. • The above results support the Simultaneously Switching Undershoot model

  8. Early failure under out of specification conditions • Actel data and userdata indicates an early failure problem under out of specification conditions: • Actel experiments: • 10/246 units failed at 168 hours HTOL, LTOL “dirty” burn-in (out of spec -1.1V undershoots on I/Os) • First pull point for both HTOL and LTOL (first observation point) was at 168 hours • 0/207 fail at 500 hours cumulative HTOL, LTOL (same “dirty” burn-in) • 29 units held back due to burnin board capacity issues. • Single failure reported in Actel published reliability data was at first observation (184 hrs) • Extended burn-in • 0/8 RT54SX32S failures with 5000 hrs burn-in at 125°C • 0/8 RT54SX32S failures with 2000 hrs burn-in at 150°C • 0/22 RT54SX32S failures with 2000 hrs burn-in at 125°C

  9. Early failure under out of specification conditions • Sanitized Actel Customer data • Contractor A • 3/101 failed at first observation on ATE (unconfirmed failures) • Contractor A has no flight box failures. More than 3500 device hours on system. Some units >400hrs. (reported on 2/13/04) • Contractor B • Measured I/O undershoots exceed Actel maximum operating conditions. • 8 failures reported out of 100+ devices. Four parts confirmed antifuse failures. One failure could not be duplicated at Actel. One part was not returned to Actel. FA ongoing on other two units. • 1 failure was immediate. 6 failures less than 168 hrs. 1 failures at 0-400hrs (first observation) • No more failures seen. Contractor B has approx. 63,000 failure free hours (reported Dec 16, 2003). 18 parts >2000 hrs. 18 parts >1000 hrs. 18 parts >500 hrs. • Contractor C • Programmer with expired calibration, intermittent fan found. 1/5 device failure rate, of every 5 units programmed one failed • 10 failures reported out of 100+ devices. All in one design. All at first observation (< 1hr). Antifuse failure confirmed on 1 device. • No more failures seen after new programmer (replace bad fan). More than 100,000 device hrs (reported 2/13/04). Some parts more than 1year testing.

  10. Early Failure under out of specification conditions • Sanitized Actel Customer data (cont'd) • Contractor D • Awaiting information relating to device operating conditions • 2/6 failed at device level testing at time zero. Remaining 4 units pass on board level. • Contractor E • Industry review of post-programming burn-in setup identified critical design deficiencies • no bypass capacitors • clock signal termination not employed • 1/59 failed from 96 hours in post-programming burn-in test. No failures seen since.

  11. Field failures Hundreds of units zero failures

More Related