1 / 23

SEU limits 130nm technology

SEU limits 130nm technology. F.Faccio CERN – PH/ESE. Outline. ITAR limit for SEU Error rate projection for “standard” SRAM and Flip-Flops in CMOS8RF (130nm) Error rate projection for “hardened” cells in CMOS8RF (130nm) Conclusion. ITAR limit for SEU.

cheri
Download Presentation

SEU limits 130nm technology

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. SEU limits 130nm technology F.Faccio CERN – PH/ESE

  2. Outline • ITAR limit for SEU • Error rate projection for “standard” SRAM and Flip-Flops in CMOS8RF (130nm) • Error rate projection for “hardened” cells in CMOS8RF (130nm) • Conclusion F. Faccio – CERN/PH

  3. ITAR limit for SEU • Listed in the Code of Federal Regulation (CFR), Title 22 (Foreign Regulation), Part 121 (The US Munitions List), in Category XV (Spacecraft Systems and Associated Equipment) • The line referring to SEU has been recently changed • The limit error rate has been decreased by 3 orders of magnitude, and the environment has been specified • The present formulation for the specific characteristic to be met or exceeded (one of the 5 for microelectronics circuits) is: • (4) A single event upset rate of 1×10−10errors/bit-day or less, for the CREME96 geosynchronous orbit, Solar Minimum Environment; F. Faccio – CERN/PH

  4. Interpretation of the limit (1) • Despite the recent addition of the environment, still there is large margin for different interpretations • The limit is “per bit” • Is the underlying assumption that all storage cells are identical (like SRAM or DRAM circuits)? What if different storage cells are embedded? • What about EDAC or TMR, in particular if these techniques are used for only a part of the circuit? • The limit only seems to concern digital functions • An analog circuit should hence be always considered as failing the combination of the 5 characteristics? Or always passing characteristic 4 (regardless the SET rate of the circuit)? F. Faccio – CERN/PH

  5. Interpretation of the limit (2) • The environment for calculating the error rate • Integral Spectra for Heavy Ions in Geostationary orbit as provided from CREME96 (left, source: NSREC1997 short course notes) • Integral spectra for protons at solar minimum in Geostationary orbit according to the IDS (Integrated Data Set) approach proposed by NASA in 2004 (IEEE TNS Dec2004, right image) F. Faccio – CERN/PH

  6. Outline • ITAR limit for SEU • Error rate projection for “standard” SRAM and Flip-Flops • Compared to ITAR limit • In the LHC/SLHC environment • Error rate projection for “hardened” cells • Conclusion F. Faccio – CERN/PH

  7. Error rate projection from available data and environment • A (very) approximate projection of the error rate for logic circuits can be made using available radiation data (Heavy Ions and proton irradiation tests performed on different circuits) • For Geosynchronous orbit, error rates for protons and Heavy Ions are calculated separately, then summed to yield the projected rate for the cell (integral flux from CREME96 or IDS) F. Faccio – CERN/PH

  8. Available radiation data for “standard” SRAM cells in IBM 130nm • SRAM cell Irradiation done by CERN at LNL (It) in 2005 on custom samples designed with Artisan SRAM generator Irradiation done by CERN at PSI (Ch) in 2002/2003 on IBM-provided SRAM samples F. Faccio – CERN/PH

  9. Error rate projection for SRAM (ITAR) - Protons WARNING: this is a very approximate estimate! Cross-section Particle flux Error rate Integral flux above 10MeV is taken. Different models for solar minimum (we take the average solar min year) Flux ≈ 5 cm-2s-1 ≈ 4.3∙105 cm-2day-1 Experimental cross-section measured by CERN. Above 10MeV, the cross-section can be approximated as flat: s≈ 6∙10-14 cm2/bit Calculated as product of cross-section and flux: Error rate ≈ 3∙10-8 bit-1 day-1 F. Faccio – CERN/PH

  10. Error rate projection for SRAM (ITAR) – Heavy Ions WARNING: this is a very approximate estimate! Cross-section Particle flux Error rate Integration of cross-section times the flux (with discrete binning) gives: Error rate ≈ 3∙10-7 bit-1day-1 Daily integral flux calculated with CREME96 for Heavy Ions with Z 2-92 Experimental cross-section measured by CERN F. Faccio – CERN/PH

  11. Error rate projection for SRAM (ITAR) – Combined protons and HI • The combined error rate is the sum of the error rates induced by the two types of particles. Therefore the resulting estimate for the error rate is • Error rate ≈ 3.3∙10-7 bit-1day-1 • Given the large approximation, this is uncomfortably close to the previous ITAR limit (1∙10-7 bit-1day-1), but very safely far from the present limit of 10-10 bit-1day-1 F. Faccio – CERN/PH

  12. Available radiation data for “standard” FF cells in IBM 130nm • Standard FF • Results from the FNAL group presented by J.Hoff in May 2006. “Standard” FF from Artisan library irradiated with monoenergetic protons (200MeV) at the Indiana University Cyclotron Facility, with devices operating as storage cells (no continuous clock). Measured cross section: 4.86∙10-14 cm-2bit-1 • Results from the CERN ESE group presented at TWEPP07 and published in JINST. “Standard” FF from Artisan library irradiated with Heavy Ions at UCL-CRC (Be), in static (no clock) or dinamic (clocked) conditions. Extrapolation of results to a mono-energetic 200MeV environment yields a cross section of 2∙10-14 cm-2bit-1 F. Faccio – CERN/PH

  13. Error rate projection for FF (ITAR) - Protons WARNING: this is a very approximate estimate! Cross-section Particle flux Error rate Experimental cross-section measured by FNAL. Above 10MeV, the cross-section can be approximated as flat: s≈ 4.86∙10-14 cm2/bit Integral flux above 10MeV is taken. Different models for solar minimum (we take the average solar min year) Flux ≈ 5 cm-2s-1 ≈ 4.3∙105 cm-2day-1 Calculated as product of cross-section and flux: Error rate ≈ 2∙10-8 bit-1 day-1 F. Faccio – CERN/PH

  14. Error rate projection for FF (ITAR) – Heavy Ions WARNING: this is a very approximate estimate! Cross-section Particle flux Error rate Integration of cross-section times the flux (with discrete binning) gives: Error rate ≈ 1∙10-6 bit-1day-1 Daily integral flux calculated with CREME96 for Heavy Ions with Z 2-92 Experimental cross-section measured by CERN F. Faccio – CERN/PH

  15. Error rate projection for FF (ITAR) – Combined protons and HI • The combined error rate is the sum of the error rates induced by the two types of particles. Therefore the resulting estimate for the error rate is • Error rate ≈ 1∙10-6 bit-1day-1 • As for the SRAM, this figure is very safely far from the present limit of 10-10 bit-1day-1 F. Faccio – CERN/PH

  16. Probability curve from the simulation of the environment Estimate cross-section in LHC environment (cm2/bit) Weibull curve (from HI test) FF SRAM ) 2 7.0x10-14 2.9x10-14 Pixel cross section (cm 6.0x10-14 Outer trk 2.5x10-14 6.3x10-14 Endcap ECAL 2.6x10-14 5.2x10-14 Exp hall 2.2x10-14 40 60 80 Deposited energy 0 20 100 120 Error rate projection for SRAM and FF (LHC/SLHC) (1) It is possible to estimate the error rate in LHC by using the cross-section data measured using the Heavy Ion beam. This gives the information on the sensitivity of the circuit. With this information and the estimate (from simulation) of the probability for energy deposition in LHC, it is possible to compute the “cross-section” of the SRAM and FF in the LHC environment. Detailed explanation of the procedure can be found in: M.Huhtinen and F.Faccio, NIM A 450 (2000) 155-172 F. Faccio – CERN/PH

  17. Error rate projection for SRAM and FF (LHC/SLHC) (2) Error rate in different locations of LHC experiments (ATLAS, CMS) Approximate at max luminosity, with cross-section = 2.8x10-14 cm2/bit for the SRAM and 7x10-14 cm2/bit for the FF Flux (all hadrons E>20MeV) particles/cm2s Error rate (SEU/bit s) Barrel; radius = SRAM FF 1.4x10-6 3.5x10-6 4 cm 5x107 1.4x10-7 12 cm 3.5x10-7 5x106 5.6x10-8 1.4x10-7 2x106 37 cm 2.8x10-8 7.0x10-8 1x106 52 cm 1.4x10-8 3.5x10-8 5x105 100 cm For SLHC, multiply the error rate by 5-10 depending on luminosity increase. For example: for a circuit with 100 bits, at 12 cm, we estimate about 1-2 errors/hour in SLHC F. Faccio – CERN/PH

  18. Outline • ITAR limit for SEU • Error rate projection for “standard” SRAM and Flip-Flops • Error rate projection for “hardened” cells • Conclusion F. Faccio – CERN/PH

  19. Available radiation data for “hardened” cells (1) • Large range of “hardened” cells custom designed and tested (200MeV protons) by FNAL. Results presented by J.Hoff in 2006. • Some of them have cross-section 3 orders of magnitude below the one measured for the Artisan cell F. Faccio – CERN/PH

  20. Available radiation data for “hardened” cells (2) • “Modified” DICE latches custom designed and tested in 2007 by the CERN ESE group. They have been tested with Heavy Ions (UCL-CRC, Be) in both static and dynamic (clocked) mode. Results presented at TWEPP07 and published in JINST • Dynamic: • large dependence on incidence angle for low LET • 1-2 orders of magnitude better than standard Artisan cells at high LET • Static: errors found only at 1 LET and 1 cell (very low statistic). Limit s around 4∙10-12 cm-2bit-1(but fluence too low!) • DICE cells custom designed and tested in 2008 by the ATLAS Pixel detector collaboration (results presented by M.Menouni at TWEPP 08). 3 different layouts integrated. Tests done with the CERN 24 GeV/c proton beam. Cross-section varies with layout but mainly around 2-3∙10-16 cm-2bit-1. This is 10 times larger than what measured by FNAL on the same design in 2006 (but different layout and proton energy). F. Faccio – CERN/PH

  21. Error rate projection for “hardened” cells (ITAR) • From FNAL tests with 200MeV protons, it appears that some hardened cells have a cross-section 3 orders of magnitude below the one measured for the Artisan cell • The projected error rate should therefore fall uncomfortably close to the present ITAR limit (uncertainty on estimate is too large here to judge on pass/fail) • Comparison with more recent data (Menouni) reveals large uncertainty on the actual rate for the DICE cells (is this due to layout, proton energy, systematic difference in experiments?) • From CERN tests with heavy ions (modified DICE, probably different than the others): • Dynamic test • Very difficult to project rate due to dependency on angle of incidence • Static test • Upper bound of error rate determined by total fluence during test (not sufficient, since no error detected). This yields an upper limit of about 10-9 errors/bit day, but certainly the rate will be considerably lower – ITAR limit will be exceeded • Summary: • Error rate depends on the detailed implementation of the cell (architecture, layout). Some cells can surely exceed the ITAR limit – at least when used as storage cells (not clocked) F. Faccio – CERN/PH

  22. Outline • ITAR limit for SEU • Error rate projection for “standard” SRAM and Flip-Flops • Error rate projection for “hardened” cells • Conclusion F. Faccio – CERN/PH

  23. Conclusion • Standard memory and FF cells in 130nm • Their error rates are comfortably below the ITAR limit. The use of only these cells leads to circuits failing to meet characteristic 4 • Hardened cells in 130nm • They improve the error rate by a very variable amount, depending on cell architecture, layout and whether the cell is used clocked or not (storage) • Some of the cells have been measured to lower the error rate by 3 orders of magnitude or more (comparison done ONLY at one proton energy) • We can conclude that the use of these cells will possibly (and in some cases certainly) lead to cells exceeding the ITAR limit F. Faccio – CERN/PH

More Related