Radiation environment
Single-Event-Effect Rate
How often a cosmic-ray ion flips a bit in your part — the single-event upset (SEU) rate, from the device's cross-section and the orbit's heavy-ion environment.
// CREME96 IRPP method: the device Weibull SEE cross-section convolved with the orbit galactic-cosmic-ray LET spectrum. A parametric order-of-magnitude screen for part selection + EDAC sizing — run CREME96 / CREME-MC in SPENVIS for the design value.
This single-event upset (SEU) rate calculator convolves the device's Weibull cross-section — the four parameters a JEDEC JESD57 heavy-ion test produces — with the orbit's galactic-cosmic-ray linear-energy-transfer (LET) spectrum, following the CREME96 integral rectangular-parallelepiped (IRPP) method (Tylka et al. 1997). It reports upsets per bit-day, upsets per device-day, and the mean time between upsets (MTBU), with altitude, inclination, and solar-cycle scaling of the environment built in. A parametric order-of-magnitude screen for part selection and error-correction sizing — the design value comes from CREME96 / CREME-MC run in SPENVIS.
How this model works & what it omits
A single-event effect is caused by one energetic particle: a heavy ion (or a proton, via a nuclear reaction) passes through a sensitive node and deposits enough charge to flip a memory bit (single-event upset, SEU), trip a latch-up (SEL), or inject a transient. Unlike total ionizing dose, which accumulates, an SEE is a rate — so what matters is the flux of ionizing particles, not how long the mission runs. This tool estimates that rate.
The device is characterised by its Weibull cross-section: the LET threshold L₀ below which no ion can upset it (the single most important hardness number), the saturation cross-section (the geometric sensitive area once every strike upsets), and the width and exponent that shape the turn-on. The environment is the orbit's integral galactic-cosmic-ray (GCR) LET spectrum — a high flux of low-LET ions falling steeply to an iron "knee" around 25–30 MeV·cm²/mg and a hard cutoff past relativistic iron. The IRPP method (Integral Rectangular Parallelepiped, as implemented in CREME96) convolves the two: the rate is the integral of the differential ion flux times the cross-section over LET.
Three environment facts drive the answer. The GCR flux is highest at GEO and interplanetary (full exposure), lower in MEO, and strongly suppressed in low-inclination LEO where Earth's magnetic field deflects the ions; polar orbits let GCR pour in at the poles. Counter-intuitively the rate peaks at solar minimum — at solar max the stronger heliospheric field modulates the incoming galactic ions (the opposite of the total-dose story). And shielding barely helps: the iron-group ions that cause most upsets are so penetrating that practical aluminum only trims the low-LET tail. The real mitigations are a higher LET threshold (rad-hard-by-design), EDAC/memory scrubbing, and redundancy.
What this tool does not do: it is a parametric analytic model of the IRPP convolution, tuned to CREME96-class magnitudes — an order-of-magnitude screen, not a physics-first sensitive- volume transport code. Real prediction uses CREME96 or CREME-MC (in SPENVIS) with the device's measured sensitive-volume geometry and the specific orbit-integrated spectrum. It models only heavy-ion direct ionization; proton-direct and proton-nuclear SEE — which dominate for very-low-threshold parts inside the proton belts — are not included, nor are SEL/SET/SEFI severity, angle-of-incidence path-length distributions, or single-event burnout in power devices.
// pick a device + orbit, or dial the Weibull cross-section directly.
Device SEE cross-section
// 4-parameter Weibull. LET threshold is the key hardness number.
Device size
// scales per-bit rate to the whole part.
Orbit & environment
// GCR peaks at GEO and at solar minimum.
Upset rate
// GEO / interplanetary — full GCR exposure
1.95 /day
device upset rate
12.3 h
mean time between upsets
2.33e-7
upsets / bit-day
17.8
LET₅₀ (MeV·cm²/mg)
// where the upsets come from (rate contribution by ion LET)
// screening only
Parametric IRPP-convolution model tuned to CREME96-class magnitudes — an order-of-magnitude screen for part selection and EDAC sizing. For a design value run CREME96 / CREME-MC in SPENVIS with the device's sensitive-volume geometry. Shielding barely helps against heavy-ion SEE — the real mitigations are a higher LET threshold, EDAC/scrubbing, and redundancy. Proton-induced SEE (dominant for very-low-threshold parts in the belts) is not modelled here.
// shareable URL encodes every input. no backend.
// device sensitivity — Weibull SEE cross-section σ(LET)
// ai-generated breakdown of what these numbers mean — with diagrams.
Common questions
How often will a commercial SRAM upset in orbit?
For a soft part, distressingly often. The default preset — an 8 Mbit commercial static RAM (SRAM) with a low ~1.5 MeV·cm²/mg threshold at geostationary orbit (GEO), solar minimum — upsets about twice per device-day: a mean time between upsets (MTBU) of roughly half a day. That is why unprotected commercial memory in a high orbit needs error correction and scrubbing from day one, not as a later hardening pass.
How much does the orbit change the SEU rate?
By an order of magnitude. The same SRAM that upsets ~2 times a day at GEO drops to ~0.19 upsets per day — one every five days or so — in an International Space Station (ISS)-like orbit (550 km, 51.6°), because Earth's magnetic field deflects most galactic cosmic rays before they reach low-inclination low Earth orbit (LEO). Tilt the same orbit polar (700 km, 98°) and the rate doubles to ~0.38 per day — the field lines funnel ions straight in at the poles. Altitude and inclination together set the heavy-ion environment.
Does shielding stop single-event upsets?
Barely — the iron-group ions that cause most upsets punch through any practical aluminum thickness, so extra shielding only trims the low-LET tail of the spectrum. The mitigations that actually work are a harder part (higher LET threshold), error detection and correction (EDAC) with regular memory scrubbing, and redundancy with voting. This is the opposite of total ionizing dose, which does respond to shielding — screen that side with the radiation dose & shielding tool.
What makes a part "rad-hard" against single-event effects?
Mostly its LET threshold — the minimum ionization density that can flip it. The galactic-cosmic-ray spectrum falls off steeply past the iron "knee" near 25–30 MeV·cm²/mg, so a part that needs more than that almost never sees a capable ion: the rad-hard preset (threshold 40, versus the soft SRAM's 1.5) upsets about 8×10⁻⁵ times per device-day at GEO — an MTBU of ~35 years against the soft part's half-day. Buying threshold is buying orders of magnitude.
Why is the upset rate worse at solar minimum?
Because the Sun's heliospheric magnetic field is what keeps galactic cosmic rays out of the inner solar system — at solar maximum the stronger field modulates them down, and at solar minimum they pour in (the opposite of the total-dose story). The tool's solar-condition toggle applies this modulation; the quietest Sun gives the harshest heavy-ion environment, so a conservative SEU screen uses solar minimum.
References
- // Tylka, A. J. et al. (1997). CREME96: A Revision of the Cosmic Ray Effects on Micro-Electronics Code. IEEE Trans. Nucl. Sci. 44(6).
- // Petersen, E. L. (2011). Single Event Effects in Aerospace. Wiley/IEEE — the IRPP method.
- // Weller, R. A. et al. (2010). Monte Carlo Simulation of Single Event Effects (CREME-MC). IEEE Trans. Nucl. Sci. 57.
- // ESA SPENVIS — Space Environment Information System (CREME96 module).
- // JEDEC JESD57 — test procedures that produce the Weibull SEE cross-section parameters used here.
