Test Simulation of Neutron Damage to Electronic Components using Accelerator Facilities Donald King, Patrick Griffin, Ed Bielejec, William Wampler, Chuck Hembree, Kyle McDonald, Tim Sheridan, George Vizkelethy, Victor Harper-Slaboszewicz Sandia National Laboratories International Topical Meeting on Nuclear Research Applications and Utilization of Accelerators May 4-9, 29 Sandia is a Multiprogram Laboratory Operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy Under Contract DE-ACO4-94AL85.
Motivation for the use of Accelerator Facilities Fast burst neutron facilities have been used to study the response of electronics to displacement damage and ionization The availability of fast burst neutron facilities is decreasing in the United States A new test methodology is being developed - high-fidelity computational models combined with testing of devices and circuits at alternative accelerator experimental facilities The computational models are initially validated at the fast neutron facilities and then applied to the test results at alternative facilities In the future, we will test and model at an alternate facility and then predict a neutron response
The Sandia Pulse Reactor SPR-III provided fast burst neutrons Facility statistics maximum pulse 4E14 n/cm 2 1 MeV Si equivalent 12 krad (Si) 1 µsec pulse @ FWHM
SNL - Ion Beam Laboratory IBL 6.5 MV Tandem van de Graaff with a nuclear microprobe Magnet slots 2ns to seconds Deflection Plates Quadrupole Lens Device Under Test V slits V slits Sample Chamber Ions: H to Au Si: 4.5, 1, 19, 28, 36 MeV Focused beam (mm µm) Currents: na (mm beam) fa (subµ beam) Pulse length > 2 ns seconds with a 9 ns rise & fall time High energy ions are focused into a 1x1 mm^2 area to simulate neutron displacement damage conditions.
Little Mountain Test Facility (LMTF) is used to decouple displacement damage and photocurrent effects LINAC operated in electron beam mode Electron energy tuned from 5 to 3 MeV Pulse widths: 5 nsec to 5 µsec Beam currents:.1 to 2 Amps Dose rate: 5E6 to 4E13 rad(si)/sec achieved with a variety of diffusers, target positions, and pulse widths Beam diameters (with ~ 8% uniformity) range from 1 cm for high dose rates to 4 cm for low dose rates The facility can operate with a maximum repetition rate of 2 pulses per second
The transistor gain is a traditional metric Transistor current response to radiation is measured collector Maximum SPR pulse on a BJT 2N2222 base emitter Test circuit uses ASTM Standard F 98M-961 techniques Gain = IC/IB Inverse Gain = IB/IC
Current (ma) Current (ma) 1. 5 4.5 3 2 1. -1 -.5-2 -3 5 4 3 2 1-1 -2-3 -4-5 Si ions create a response in transistors similar to neutrons -1. -4 1 1 2 2 3 3 4 5 56 6 7 Time (µsec) 1 2 3 Time (µsec) IB IC IB IC IB IC SPR fast neutron 3E14 n/cm 2 1 MeV Si Eqv 1E9 rad(si)/sec 9 µsec pulse width FWHM Displacement damage increases base current and decreases collector current IBL 1 MeV Si 3E14 n/cm 2 1 MeV Si Eqv 1 µsec pulse width FWHM
Deep Level Transient Spectroscopy (DLTS) relates ion to neutron damage at late times Defect density ( C) C (pf).25.2.15.1.5 4.5 MeV Si 3E14 n/cm 2 SPR-III 3E14 n/cm 2 DLTS spectra of the base region of a PNP device V 2 =/-.36 ev VP + V 2 -/ DLTS peak amplitude is proportional to number of traps. The type and number of defects are the same for a given 1 MeV Si eqv neutron fluence (matched late-time gain degradation).. 5 1 15 2 25 3 Temperature (K) The spectra of defects as measured by DLTS in the base of pnp transistors that are responsible for the gain degradation are the same after neutron (including SPR-III and ACRR) and ion irradiations.
The transient annealing factors between SPR neutron and ion beam irradiation are similar SPR maximum pulse Annealing Factor Annealing Factor 12 12 T3219 G =22, G =1.1 11 T3297 G =1.35 =175, G G =29, =1.6 1 113245 G13245 G =175, G =1.6 9 SPR 8 8 4.5 MeV Si SPR 7 1 MeV Si 6 6 5 4 4 3 2 2 1 1E-41E-4 1E-31E-3.1.1.1.1 1 1 1 1 11 Time (s) Time (s) astm astm astm astm Ie =.22 ma, 2N2222 Annealing Factor 1 1 G (t ) G AF (t ) = 1 1 G G AF uncertainty 5% Agreement between the annealing factors indicates that the annealing kinetics are similar for ion and neutron irradiations critical for early-time predictive capabilities SPR-III gamma environment delays gain measurement compared to IBL The SPR-III early-time annealing factor can be matched using ion irradiations (simulating a wide range of SPR-III fluence values).
Physics based modeling codes used in this study 1D Numerically solves diffusion-drift equations similar to commercial device simulation codes One dimensional Single transistor Includes defect production, migration & reactions, and charge change reactions by carrier capture & emission. Includes photocurrent generation with synergistic displacement damage and annealing effects Xyce a high performance Sandia developed SPICE compatible tool zero dimensional to half dimensional Single transistor to integrated circuits Includes defect migration & reactions and charge change reactions by carrier capture & emission defect production is an input Includes photocurrent generation without synergistic displacement damage and annealing effects
Xyce and 1D are used to model transistor displacement damage at SPR and IBL Xyce experiment SPR IE=9 ma Xyce experiment Time from center of pulse (s) Excellent agreement between the shape and the fluence dependence IE=9 ma IBL
Xyce and 1D are used to model transistor photocurrent at SPR and IBL base current (A) Base current (ma) 8 6 4 2 1.3E-3 1.1E-3 9.E-4 7.E-4 5.E-4 3.E-4 SPR..1.2.3.4.5 SPR IB2 Xyce Time (ms) SPR data Simulations (no free parameters) no damage with damage 1D Xyce Photocurrent [A] IBL 36 MeV Si IB IBL 1.4E-1 IB 1D damage with annealing IB 1D damage, no annealing 1.2E-1 IB Xyce 1.E-1 8.E-2 6.E-2 4.E-2 2.E-2.E+.E+ 5.E-5 1.E-4 1.5E-4 Time [s] 1.E-4-1.E-4 1.E-4 2.E-4 2.E-4 3.E-4 3.E-4 4.E-4 4.E-4 time (s)
LMTF is used to calibrate Xyce transistor photocurrent models Photocurrent data (red) and simulation (blue) for short pulses which are dominated by contributions from prompt photocurrent. Photocurrent data (red) shown with simulated pulses where parameters are calibrated against all data simultaneously. These 5 µsec pulses have contributions from prompt and delayed photocurrents and are thus considered long pulses. Final calibration
For future tests, we must calibrate IBL photocurrent vs LMTF photocurrent Photocurrent (A) Photocurrent [A].18.16.14.12.1.8.6.4.2.. 4.x1 9 8.x1 9 1.2x1 1.16.14.12.1.8.6.4.2 LMTF IBL Dose rate [rads(si)/s].. 2.5x1 13 5.x1 13 7.5x1 13 1.x1 1 Photocurrent [A] 1.4E-1 1.2E-1 Dose Rate [rad(si)/s] 1.E-1 8.E-2 6.E-2 4.E-2 2.E-2 1.4x1 1 1.2x1 1 1.x1 1 8.x1 9 6.x1 9 4.x1 9 2.x1 9. IB 1D IB LMTF IB Xyce IB IBL.E+. 4.x1 13 8.x1 13.E+ 5.E-5 1.E-4 Ion Flux [ions/cm 2 /s] 1.5E-4 Time [s] 1.27E-4 rad(si) cm 2 /ion Ion Flux [ions/cm 2 /s]
IBL photocurrent as a function of ion energy is being studied 1.E+3 Ion strike 4.5, 1, 36 MeV Photocurrent (ma) 1.E+2 1.E+1 1.E+ IC 36 MeV IC 1 MeV IC 4.5 MeV 1.E-1 2 4 6 8 1 12 Time (µsec) 2n2222 cross section Ionization (1 1 eh/cm/ion) 3 2 1 eh 36 MeV Si into 2N2222 IV 8 6 4 2 Damage (1 6 IV/cm/ion) 5 1 15 Distance from emitter contact (micron)
Summary and Conclusions SNL will use accelerator facilities to simulate displacement damage and ionization effects observed in electronics at fast burst neutron facilities We will use high fidelity computational models to confirm our ability to predict displacement damage and ionization effects at accelerator and neutron facilities Ultimately, when fast burst neutron facilities are not available, tests at alternate facilities, combined with computational models, will be used to simulate fast burst neutrons We have presented preliminary LMTF and IBL test and computational results We observed excellent agreement between the experimental results and preliminary calculations at LMTF and IBL