Industrial challenges and trends in terrestrial singleevent effects

Industrial challenges and trends in terrestrial singleevent effects (SEE) Dr. Robert Baumann TI/IEEE Fellow, Technology Office Aerospace & Defense (MHRS Group) High Performance Analog Products Texas Instruments, Dallas, Texas, USA © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 1/42 1/24 TI Information–– Selective Disclosure TI Information Disclosure

Radiation Environments Natural • Terrestrial Background (neutrons and a particles) • Avionics (neutrons) • Space (protons, heavy ions, electrons) Man-made • Accelerators/Nuclear reactors (x-ray, gamma, proton, neutron, etc. ) • Weapons (x-ray, gamma, neutrons) • Industrial/Security (x-ray, gamma, e-beam) • Medical (x-ray, gamma, protons, neutrons, e-beam) © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 2/42 2/24 TI Information–– Selective Disclosure TI Information Disclosure

Flux, Total Ionizing Dose, and Neutron/Proton Dose Comparison Adapted from M. Brugger (CERN) © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 3/42 3/24 TI Information–– Selective Disclosure TI Information Disclosure

Energetic Ions in Matter Silicon surface Generated with SRIM 2008 27 um 5 Me. V He in Silicon © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 4/42 4/24 TI Information–– Selective Disclosure TI Information Disclosure

Physical Manifestations of Radiation • Transient Charge Generation • Charge Trapping/Interface Damage • Nuclear Reactions • Structural (Lattice) Damage Single Event Effects Dose Effects chronic stochastic Dose Rate Effects © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 5/42 5/24 TI Information–– Selective Disclosure TI Information Disclosure

Basic Reliability Definitions Hard Failure (GOI, EM, NBTI, ESD, …TID, ND, SEE) An error induced by faulty device operation. DATA is lost AND function is lost and can no longer operate at that location. Soft Failure (glitch, noise, SEE) An event corrupting only the DATA stored in a device. The device itself is not damaged and functionality is restored when new data is written. 1 FIT is 1 failure in 114, 155 years! or 1, 000 FIT is ~ 1 failure/month © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 6/42 6/24 TI Information–– Selective Disclosure TI Information Disclosure

Who cares about SEE (SEU, SEL)? Don’t Care Really Care • Consumer Goods • High Reliability • Single-chip • Multi-chip systems • Non-critical • Cell phones • MP 3 Players Catalog DSP, MCUs etc. 1 MFIT/chip ok (~1 fail/month) © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 7/42 7/24 • Life support • Safety systems • Medical electronics • Automotive, Avionics < 1 k. FIT/Chip (~ 1 fail/114 yrs) TI Information–– Selective Disclosure TI Information Disclosure

BIG Business Impact Sun Screen Daniel Lyons, Forbes Global, 11. 13. 00 mysterious glitch has been popping up since late last year… for America Online, Ebay and dozens of other major corporate accounts…The SUN (server) has caused crashes at dozens of customer sites. An odd problem involving stray cosmic rays and memory chips in the flagship Enterprise server line… A dotcom company bought a Sun 6500 server to run…the core of its business. The server crashed and rebooted four times over a few months. "It's ridiculous. I've got a $300, 000 server that doesn't work. The thing should be bulletproof, " says the company's president. Loss of customer confidence © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 8/42 8/24 = Loss of revenue TI Information–– Selective Disclosure TI Information Disclosure

Safety Impact: QANTAS Flight 72 “In-flight upset, 154 km west of Learmonth, WA, 7 Oct. 2008, VHQPA Airbus A 330 -303, ” ATSB Transp. Safety Report - Aviation Occurrence Invest. , AO-2008 -070, pp. 1 – 313, Dec. 2011. Single subatomic event has human-scale impact! © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 9/42 9/24 TI Information–– Selective Disclosure TI Information Disclosure

Terrestrial + Avionics Environments Alpha particles and neutrons © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 10/42 10/24 TI Information–– Selective Disclosure TI Information Disclosure

Actual & Simulated alpha spectra 238 U 232 Th Measured Thick 232 Th © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 11/42 11/24 TI Information–– Selective Disclosure TI Information Disclosure Packagesimulation U: Th (50: 50)

Alpha Particles from Materials • Distributed throughout materials • Flux depends on types of materials & purity • Most of the alphas are from packaging • Ultra low alpha materials < 0. 002 a/cm 2 -hr © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 12/42 12/24 TI Information–– Selective Disclosure TI Information Disclosure

Single Incoming Cosmic Particle Cosmic Cascade p N o J. F. Ziegler, “Terrestrial Cosmic Ray Intensities, ” IBM J. Res. Develop. , Vol. 42(1), p. 125, Jan. 1998. p ± g p g + m e e - e ± + p - - g g me ± m+ m- © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 13/42 13/24 P m+ m+ TI Information–– Selective Disclosure TI Information Disclosure

Effect of Altitude/Latitude Adapted from Eugene Normand, “Single Event Effects in Avionics”, IEEE Trans. Nucl. Sci. , 43(2), April 1996, pp. 463. . 350 G. A. Glatzmaier and P. H. Roberts, "Rotation and magnetism of Earth's inner core, " Science, 274, 1887 -1891 (1996). 300 250 6. 0 Terrestrial Altitudes 200 150 100 Flight Altitudes 50 0 0 5 10 15 20 25 Altitude (km) at 3, 000 meters relative neutron flux ~ 11 x higher than sea-level Relative Neutron Flux (sea-level=1) 400 North South Flight altitudes 5. 0 4. 0 3. 0 Sea-level 2. 0 1. 0 0. 0 equatorial 0 10 20 30 40 50 60 70 80 90 Latitude (degrees) © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 14/42 14/24 polar TI Information–– Selective Disclosure TI Information Disclosure

10 B Thermal neutron and Thermal Neutrons Generated with SRIM 7 Li 10 B 0. 84 Me. V 4 He 1. 47 Me. V 104 103 snth (barns) 102 101 Slide 15/42 15/24 Boron-11 Oxygen Phosphorus Silicon Nitrogen Copper Arsenic © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Aluminum 10 -3 Titanium 10 -2 Tungsten 10 -1 BORON 10 100 R. Baumann, T. Hossain, E. Smith, S. Murata, H. Kitagawa, “Boron as a primary source of radiation in high density DRAMs”, IEEE Symp. VLSI Tech. , June 1995, pp. 81 - 82 TI Information–– Selective Disclosure TI Information Disclosure

Single Event Effects (SEEs) A single event (nuclear reaction or energetic ion) creates transient charge that induces a disruption in circuit operation or data state. Typically SEE are very rare events (1 per month, etc. ) Of all possible nuclear events only a fraction cause a SEE Of all possible SEE only a few will cause machine state failures (derating effects) © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 16/42 16/24 TI Information–– Selective Disclosure TI Information Disclosure

SEU and SEL – Most Common SEE +V Ion Track n+ diffusion on ctr n Ele ectio l col p- epi Potential Contour Deformation Drift Collection Diffusion Collection Electron-Hole Pairs Recombination Parasitic bipolar action Reverse-biased N+/P junction Single Event Upset © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 17/42 17/24 Single Event Latch Up TI Information–– Selective Disclosure TI Information Disclosure

SRAM/DRAM Bit SEU Scaling Trend Planar DRAM sensitivity has been decreasing with scaling R. H. Edwards, C. S. Dyer, E. Normand, “Technical standard for atmospheric radiation single event effects, (SEE) on avionics electronics”, IEEE Rad. Effects Data Workshop, 2004, pp. 1 - 5 Magnitudes from two different curves cannot be compared as these curves were individually normalized! FINFET SRAM sensitivity has been decreasing with scaling (since the 130 nm node) © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 18/42 18/24 TI Information–– Selective Disclosure TI Information Disclosure

Future Terrestrial Mechanisms? From B. Sierawski et al. , “Impact of Low-Energy Proton Induced Upsets on Test Methods and Rate Predictions”, IEEE Trans. NS, 56 (6), Part 1, Dec. 2009, pp. 3085 - 3092 (with TI) Adapted from B. Sierawski et al. , “Effects of Scaling on Muon-Induced Soft Errors”, 2011 IEEE IRPS, pp. 3 C. 3. 1 3 C. 3. 6. (with TI) Proton SEU s for direct ionization Muon SEU s for prior technologies Terrestrial Protons may also dominate as Qcrit is reduced due to the much higher cross-section for direct ionization. © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 19/42 19/24 n + a SER When Qcrit reaches 0. 2 f. C SER will double from muons alone and at 0. 1 f. C SER will be 5 -10 x higher. Note: Qcrit for 40 nm is ~ 0. 5 f. C TI Information–– Selective Disclosure TI Information Disclosure

Accelerated Testing & Facilities © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 20/42 20/24 TI Information–– Selective Disclosure TI Information Disclosure

Extrapolating Neutron Results Actual neutron particle flux (≥ 10 Me. V n/hr/cm 2) reaching the Si as defined by JEDEC JESD 89 A NYC Sea-level = 13 n/hr/cm 2 Failure rate due to neutrons (errors/hr) Neutron beam always induce > 100 upsets per test so that s ≤ 10%) Conversion of TTL Neutron Sensitivity ULC from U 238 foil (errors/neutron/cm 2) to actual n/hrcm 2 sources are inexpensive and in-house BUT (≥ 10 Me. V) extrapolating the alpha-particle SEE is much more difficult and requires simulation n. ASER Test © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 21/42 21/24 TI Information–– Selective Disclosure TI Information Disclosure

Accelerated Neutron Testing neutron test procedure in JESD 89 and JESD 89 A Neutrons/cm 2 -sec-Me. V 1 E-02 Proton beam r te as er m utt sh Tungsten spallation target shut t 15ºR © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 0º Slide 22/42 22/24 1 E-05 1 E-06 DUT er 15ºL WNR Beam / 1. 38 E 08 Atmosphere 1 30ºL 30ºR 1 E-04 1 E-07 Fission Foil Neutron beams 1 E-03 10 1000 Neutron Energy (Me. V) Close match between terrestrial background spectrum and Los Alamos means extrapolation is based on a simple multiplication TI Information–– Selective Disclosure TI Information Disclosure

“Atmospheric” Neutron Test Facilities From Charlie Slayman, “Theoretical Correlation of Broad Spectrum Neutron Sources for Accelerated Soft Error Testing”, IEEE Nuclear and Space Radiation Effects Conference (NSREC), Denver, July 22, 2010 (to be published Trans. on Nuc. Sci. December 2010) Svedberg Laboratory, Uppsala University, Sweden (TSL, ANITA) Los Alamos Neutron Science Center Ice House (LANSCE), New Mexico, USA x Tri-University Meson Facility - Univ. of British Columbia (TRIUMF) + Research Center for Nuclear Physics Osaka University (RCNP) Vesuvio Beamline - Rutherford Appleton Lab, Oxfordshire, UK (ISIS) JEDEC JESD 89 A “standard” flux x (3 x 108) Extend En > 800 Me. V Need for muon testing will grow © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 23/42 23/24 TI Information–– Selective Disclosure TI Information Disclosure

Summary • SEE sensitivity is decreasing with new generations due to Vdd saturation, HOWEVER increased bit density leads to similar or increasing system SER. • Nano-devices may offer improved resilience against SEE (bulk/SOI Fin. FETs, etc. ) but will NOT eliminate them. Protons and muons are a growing concern and 10 B reactions with thermal neutrons can still be a risk. • SER is application-specific so one failure rate specification for all products is NOT viable (e. g. for catalog products). Extensive support for radiation effects engineers needed to extrapolate reliability in a wide variety of environments. • Vendors that ignore the soft error problem will end up paying for it in loss of customer confidence – leading to significant revenue and market share loss. • Control and detection electronics in accelerator facilities share many of the problems induced by terrestrial and avionics radiation environments (typically at much higher equivalent fluxes). • Use of space-grade or enhanced COTS may be required for many accelerator applications and/or fault-tolerant system design. © © Robert Baumann 4/22/2014 6/10/2014 9/18/2013 Slide 24/42 24/24 TI Information–– Selective Disclosure TI Information Disclosure