Chapter 9 · Lesson 2 of 4 · about 14 minutes
Protons, heavy ions and the time cases
Low Earth orbit changes which particle causes the events. Trapped protons, mostly in the anomaly passes, can dominate the rate of a soft part, while galactic and solar heavy ions still matter for others. A proton test and a heavy-ion test answer different questions, so one response curve does not generally replace the other. When the events arrive matters as much as how many there are.
What you need to know
Low Earth orbit does not remove the single-event concern. It changes which particle causes the events. The geomagnetic cutoff removes most galactic ions below a few GeV per nucleon, so the heavy-ion rate in a 400 km orbit is a fraction of the interplanetary value. The inner belt, however, supplies trapped protons along the South Atlantic Anomaly. The outer-zone horns crossed at high inclination are electron-dominated, and the protons above the 15 MeV floor are confined to low L-shells, so the proton exposure of a sun-synchronous orbit is essentially the anomaly passes. A proton upsets a soft part through the recoils of nuclear reactions. The cross section per proton is small and the flux is large. For U2 the proton cross section per bit is 1.0 × 10⁻¹⁴ cm², four orders of magnitude below the heavy-ion saturation per bit, and the AP-8 MIN orbit-averaged flux above 15 MeV in a sun-synchronous orbit is 150 protons per cm² per second, which is 1.3 × 10⁷ per cm² per day. The product, 139 upsets per day, is 134 times the heavy-ion rate of lesson 9.1. In the ISS orbit the ratio is 39. For a soft part in low Earth orbit, the proton term is the rate.
The full proton calculation folds an energy-dependent cross section with the differential spectrum, which is what the CREME96 proton module does. A widely used first estimate multiplies the saturated high-energy cross section (measured near 200 MeV) by the integral flux above a floor energy. This works because the proton cross section is near its plateau over most of the trapped spectrum, and the flux below the floor is both uncertain after transport and less effective. The floor is set in the range 10 to 20 MeV in common practice; the calculator uses 15 MeV. A part without proton data is estimated from its heavy-ion curve through a proxy model. The proxy is generally conservative for parts with onsets of a few MeV·cm²/mg and useless for parts whose onset is above what proton recoils can deposit (about 15 MeV·cm²/mg in silicon). For that reason a heavy-ion onset above that value is generally read as immunity to proton-induced upset through recoils. Direct ionization by low-energy protons is a separate question for scaled technologies.
When the events arrive matters as much as how many there are. The galactic flux varies by a factor of about two to three over the solar cycle at low LET, and by more at high LET. The trapped flux arrives in passes. A sun-synchronous spacecraft collects most of its proton upsets in the 12 % of the time it spends in flux, with an instantaneous peak about 50 times the orbit average (the pass-averaged rate is lower; chapter 1 lesson 4). Solar particle events add days of elevated flux a few times per cycle. The same mission total can be spread evenly or concentrated into hours, and a recovery mechanism that keeps up with the average can be overrun by the peak. The rate is therefore computed for several cases: the quiet background (galactic ions plus trapped protons), the worst week and worst day of a reference event, and the peak five minutes. Each case raises its own question about what the events do to the function. The event cases are scenarios with no probability of occurrence attached. The cumulative models of chapter 1 (ESP, PSYCHIC) supply the mission fluence at a confidence level, and a common convention turns the worst-week flux into a per-event count by treating one event as 7.5 days at that flux.
Worked example 9.2. The U2 proton rate, and what a solar event adds
| Orbit | AP-8 MIN flux above 15 MeV, per cm² per s | Proton upsets per day (σ(200 MeV) × flux) | Heavy-ion upsets per day (IRPP X/5) | Proton share of the total |
|---|---|---|---|---|
| ISS | 15.5 | 14.3 | 0.37 | 97.5 % |
| SSO | 150.2 | 139.3 | 1.04 | 99.3 % |
In the sun-synchronous orbit U2 sees 140 upsets per day, almost all from trapped protons and concentrated in the anomaly passes. A solar event adds a burst. The ESP-PSYCHIC worst event at 95 % confidence delivers 1.6 × 10¹⁰ protons per cm² above 15 MeV in free space. At geosynchronous orbit (transmission near one) U2 would collect about 1.8 × 10⁵ upsets over the event, roughly 135 years of its galactic background in a week. In the sun-synchronous orbit the geomagnetic transmission reduces the event fluence to about a fifth (chapter 1 lesson 3). That still leaves about 3.5 × 10⁴ upsets in a few days, about 8 months of the quiet-time rate, and the scrubber of chapter 10 is sized for that week rather than for the average day. The event is a scenario. The probability that it occurs during the mission comes from the cumulative models, not from the scenario itself.
Exercise
Before you sum any rates, make a table of the particle species and mechanisms the mission can supply (galactic ions, trapped protons through recoils, solar protons, direct ionization where the technology makes it credible) and mark which of them the data cover. Resolve each credible contribution that has no entry.
Record the quiet-time result, the event-integrated result and the peak-window result separately, each with the scenario and the operating assumptions behind it.
Going deeper
The same LET does not mean the same track
Nuclear reactions produce a spread of secondary species, energies and track lengths. In a structure sensitive enough to respond to direct ionization by protons, there is a second pathway. Whether a proxy from the heavy-ion curve stays useful depends on the overburden above the sensitive volume, on how the charge is collected and on the angular response. A transfer model needs to be checked against the mechanism and the construction of the part; agreement at one energy is limited evidence.
Record in the rate calculation which mechanism was used, which secondaries were included and what the proxy assumed.
A constant rate is an assumption
A constant-rate Poisson process is a fair model for a defined quiet interval. A hazard that changes with time needs a model whose rate changes with it, or one conditioned on the state of the environment. Storms, a shared environment and changes in the detection can all make events appear clustered. A percentile from a model is not the probability that any given mission follows that trajectory.
A worst-day rate multiplied by the full mission lifetime is not a mission total. The worst day is a scenario, not a long-term average.
The common mistake
A proton screen and a single high-LET test each cover some modes and leave out others. Neither one generally bounds every destructive and nondestructive mode of a part.
Takeaway
- A rate built from heavy-ion data alone can miss the proton term, which for a soft part in low Earth orbit is most of the rate.
- A mitigation that keeps up with the average count can still be overrun during an anomaly pass or a solar event.
Check yourself
Why does the heavy-ion curve of U2 not determine its proton rate?
Protons upset the cell through nuclear recoils, a different physical pathway, with a cross section four or five orders of magnitude below the heavy-ion saturation. The proton cross section is measured directly, or estimated from the heavy-ion onset through a proxy model whose conservatism is stated.
A memory controller is sized for 140 upsets per day. What does the anomaly pass do to that figure?
The 140 per day arrive during the 12 % of the time spent in flux, at an instantaneous peak up to 50 times the orbit-average rate. The controller has to keep up with the peak, and the scrub interval has to be short compared with a pass.
What is the probability that the worst-week scenario occurs during the mission?
The scenario has no probability attached; it is a reference flux. The probability that the mission sees a given fluence comes from the cumulative models (ESP, PSYCHIC) at a stated confidence. The two are reported as different quantities.
U1 has a heavy-ion onset near LET 21 (lesson 9.1) and no proton data. Why is it generally read as immune to proton-induced upset, and what does that reading not cover?
Proton recoils in silicon can deposit at most about 15 MeV·cm²/mg, so a heavy-ion onset above that value is generally read as immunity to proton-induced upset through recoils, and a proxy model from the heavy-ion curve is useless there rather than conservative. The reading covers the recoil pathway only. Direct ionization by low-energy protons is a separate question for scaled technologies, and a proton screen and a single high-LET test each cover some modes and leave out others, so neither one generally bounds every destructive and nondestructive mode of a part.
A reviewer multiplies the worst-day proton rate of the reference event by the three-year mission length and calls the result a conservative mission total. Why is that not a mission total?
The worst day is a scenario, not a long-term average, and a worst-day rate multiplied by the full mission lifetime is not a mission total. The mission total is the quiet background (galactic ions plus trapped protons) over the mission plus the event fluence that the cumulative models (ESP, PSYCHIC) give at a stated confidence; the worst week, the worst day and the peak five minutes are separate cases, each asking what the events do to the function during that interval. Where a per-event count is wanted, a common convention treats one event as 7.5 days at the worst-week flux.
Why does multiplying the proton cross section measured near 200 MeV by the integral flux above 15 MeV give a usable first estimate of the rate, and when would you not rely on it?
The proton cross section is near its plateau over most of the trapped spectrum, and the flux below the floor is both uncertain after transport and less effective, so the product is close to what the full fold of an energy-dependent cross section with the differential spectrum (the CREME96 proton module) gives; for U2 in the sun-synchronous orbit it is 139 upsets per day. The floor is a convention set in the range 10 to 20 MeV, the calculator uses 15 MeV, and the result is quoted with that choice. The estimate is generally not relied on where the cross section has not reached its plateau over the energies that carry the trapped flux, or where direct ionization by low-energy protons is credible, and the full calculation is used instead.
Further reading
- E. L. Petersen, "The SEU figure of merit and proton upset rate calculations," IEEE Trans. Nucl. Sci., vol. 45, no. 6, pp. 2550-2562, 1998.
- D. M. Sawyer and J. I. Vette, "AP-8 trapped proton environment for solar maximum and solar minimum," NSSDC/WDC-A-R&S 76-06, 1976.
- M. A. Xapsos, G. P. Summers, J. L. Barth, E. G. Stassinopoulos and E. A. Burke, "Probability model for cumulative solar proton event fluences," IEEE Trans. Nucl. Sci., vol. 47, no. 3, pp. 486-490, 2000.
- M. A. Xapsos, C. Stauffer, T. Jordan, J. L. Barth and R. A. Mewaldt, "Model for cumulative solar heavy ion energy and linear energy transfer spectra," IEEE Trans. Nucl. Sci., vol. 54, no. 6, pp. 1985-1989, 2007.
- P. Jiggens et al., "Long-term destructive SEE risk and calculations using multiple worst-case events versus modelling," IEEE Trans. Nucl. Sci., vol. 61, no. 4, pp. 1695-1702, 2014. doi:10.1109/TNS.2014.2302994.
- M. A. Xapsos et al., "Inclusion of radiation environment variability in total dose hardness assurance methodology," IEEE Trans. Nucl. Sci., vol. 64, no. 1, pp. 325-331, 2017. doi:10.1109/TNS.2016.2607021.
- R. Ladbury, J.-M. Lauenstein and K. P. Hayes, "Use of proton SEE data as a proxy for bounding heavy-ion SEE susceptibility," IEEE Trans. Nucl. Sci., vol. 62, no. 6, pp. 2505-2510, 2015.
- A. J. Tylka et al., "CREME96: a revision of the Cosmic Ray Effects on Micro-Electronics code," IEEE Trans. Nucl. Sci., vol. 44, no. 6, pp. 2150-2160, 1997.