Tools / Single events / DRAM stuck bits

How many stuck bits will this DRAM carry, and when?

Enter the bit errors counted after each exposure and how they changed with time and storage temperature. The tool turns them into a per-bit creation response and a persistence curve in demarcation energy, carries both over the particle spectra at the die, returns bit errors against mission duration at the flight temperature, and writes out the measurement plan that would cover the mission.

v1.0 · Oct 2026ISO 15390 cosmic rays, AP-8, ESP and JPL event modelsrecovery in demarcation energyruns in your browserMechanism in 3DHelp & Reference
Synthetic example loaded: proton data patterned on a measured set, not a measurement
Device and mission
Sets how far along the recovery curve the mission gets.
Where the reads are made, and the storage temperature of rows left blank.
A hold of t at temperature T releases every cell with activation energy below E* = kT ln(νt), and the flight age is converted the same way, so a hot hold stands in for years at the flight temperature. 1e12 to 1e14 s⁻¹ is the usual range; a decade moves E* by about 0.06 eV. The fit needs parts stored at two or more temperatures.
Environment
Galactic cosmic rays by ISO 15390 at DLR, with the orbit's cutoff and Earth shadow in Earth orbit, then one-dimensional aluminum. Albedo neutrons and trapped electrons are not included.
Solar particle events
Model spectra are free-space fluences at 1 AU; the page transports them through the shielding and the sky above.
Both models count only active years (about seven of each eleven); quiet years add nothing. Default equals the mission length, the conservative choice.
Fluence scaled by 1/r³ inside 1 AU and 1/r² beyond, a working convention rather than a model result.
Feeds the double-bit panel: the moment right after this event late in the mission.
Memory architecture
Optional, from your SEU rate. Adds the stuck-plus-upset term.
Used only for the upset-plus-upset baseline.
Measurements

One row per read: the part, the beam, the cumulative fluence the part had received, the hours since the end of the exposure, the temperature the part was stored at since its previous read (blank: the test room), and the number of distinct failing cells counted at that read. Rows of one part after one exposure form its time series. Protons: ion H with the energy in MeV; the LET column is then ignored.

Bit errors against mission duration

Double-bit errors against mission duration

What the measurements gave

Measurement plan

What to measure, how, and when, written for this device, mission and temperature. It is generated from the inputs above and changes with them; where it assumes something, it says so.

Conclusions: what the number rests on

"May be low" means the true number is likely above the prediction (the prediction is not a bound); "may be high" the reverse; "either way" the direction depends on the device or the environment.

The trades the tool makes, in plain language

Creation

  • One number per beam from the first count. σ = first count ÷ (bits × fluence). Everything the mission predicts scales with it, so its counting uncertainty (1/√count) is the band on the chart. The trade: it is only as good as the first read's timing and refresh setting, which is why the plan fixes both.
  • Linear in fluence. Exposures of one beam at several fluences are pooled as one straight line. Simple and right at low dose; it stops being right where accumulated dose makes the count rise steeply, so the page flags the mission dose and the test dose should stay below that point.
  • Heavy ions carried in LET, protons in energy. Between measured points the curve is a straight line in the logarithms; below the lowest heavy-ion LET it falls in proportion (a single-recoil tail), above the highest it is flat; protons are flat below the lowest energy down to a 15 MeV floor and flat above the highest. These extensions are the cheapest defensible choices, not physics; more beams replace them.
  • Omnidirectional flux times the normal-incidence cross section. Exact when creation is proportional to path length in the sensitive volume, which is the case for recoil production in a thin layer; no angular factor is applied.
  • Only protons and heavy ions. Electrons, neutrons and event heavy ions are not carried. The trade is simplicity against completeness at GEO, MEO, Jupiter and planetary surfaces, where the missing terms are not small.

Environment

  • ISO 15390 cosmic rays at a chosen solar level, one-dimensional aluminum. Solar minimum is the bound. The slab stands in for the real enclosure; a box with a thin face lets more through than the slab of its average thickness, a thick corner less.
  • Earth orbits by a sampled cutoff and the Earth shadow, checked against SPENVIS to a few percent; trapped protons from AP-8 presets or a pasted spectrum; event protons by the ESP or JPL model at a confidence that is the accepted risk, transported through the same shielding and, in orbit, through the SPENVIS transmission tables. The trade is a built-in environment against the mission's own environment document, which should win where they differ.

Persistence (annealing)

  • Time and temperature on one axis. A hold of t at T releases every cell with activation energy below E* = kT ln(νt), so a hot hold stands in for a long cold one. This rests on first-order release with a single attempt frequency ν; it is what lets 1000 h in an oven speak for years in flight, and the arms at two temperatures are the test of it.
  • One curve per particle class, chain-linked across arms, weighted by first counts. Protons and heavy ions each get their own curve where they have a time series; a class without one borrows the other's. Cohorts with more cells count for more, which is right for counting noise and wrong only if a part is unrepresentative.
  • The first read defines the start. Cells released before the first read are in neither σ nor the curve. In flight they would also be gone within the hour at the same or a higher temperature, so nothing is lost there; a colder flight keeps them longer, and the tool says so.
  • Held flat beyond the last read. Past the highest E* measured, the curve keeps its last value: an upper bound when recovery is the only process and the flight is not colder than the reads, and not a bound otherwise. The plan's hot arm is what removes the extrapolation.
  • Visibility fixed at the read. The count is what the read's refresh period and temperature made visible; the plan sets the read's refresh period from the flight temperature (retention time taken to halve per 10 °C) so that the count is the flight count.

Double bits

  • Stuck bits placed at random over the codewords, the same population in every device; uncorrectable words follow from the binomial tail, so they grow as the square of the stuck population and follow the persistence curve twice over. Clusters from one ion, row and column faults and the bit mapping are not represented. The stuck-plus-upset rate needs the upset rate; the largest event adds its cells at once late in the mission as the worst moment.

What is not traded

No device parameter is built in: every number on the page comes from the measurements entered, the environment chosen and the temperatures given. The Conclusions card above lists, for this run, which of these trades is measured, assumed or missing, and which way each one pushes the number.

References

Stuck bits and their recovery

  1. G. M. Swift, D. J. Padgett and A. H. Johnston, "A new class of single event hard errors," IEEE Transactions on Nuclear Science, vol. 41, no. 6, pp. 2043 to 2048, 1994.
  2. L. D. Edmonds, S. M. Guertin, L. Z. Scheick, D. Nguyen and G. M. Swift, "Ion-induced stuck bits in 1T/1C SDRAM cells," IEEE Transactions on Nuclear Science, vol. 48, no. 6, pp. 1925 to 1930, 2001.
  3. H. Shindou, S. Kuboyama, N. Ikeda, T. Hirao and S. Matsuda, "Bulk damage caused by single protons in SDRAMs," IEEE Transactions on Nuclear Science, vol. 50, no. 6, pp. 1839 to 1845, 2003.
  4. L. D. Edmonds and L. Z. Scheick, "Physical mechanisms of ion-induced stuck bits in the Hyundai 16M x 4 SDRAM," IEEE Transactions on Nuclear Science, vol. 55, no. 6, 2008.
  5. L. Scheick, D. Nguyen and S. Guertin, "Investigation of the mechanism of stuck bits in high capacity SDRAMs," IEEE Radiation Effects Data Workshop, 2008.
  6. A. M. Chugg, A. J. Burnell, P. H. Duncan, S. Parker and J. J. Ward, "The random telegraph signal behavior of intermittently stuck bits in SDRAMs," IEEE Transactions on Nuclear Science, vol. 56, no. 6, pp. 3057 to 3064, 2009 (variable retention of stuck cells).
  7. S. M. Guertin, A. C. Daniel, W. P. Parker, D. N. Nguyen and S. Delaney, "Radiation test results for ISSI 1 Gb DDR2," IEEE Radiation Effects Data Workshop, 2020 (refresh period as the analog of operating temperature).
  8. J. R. Srour and J. W. Palko, "Displacement damage effects in irradiated semiconductor devices," IEEE Transactions on Nuclear Science, vol. 60, no. 3, pp. 1740 to 1766, 2013 (annealing of displacement damage, review).
  9. W. Primak, "Kinetics of processes distributed in activation energy," Physical Review, vol. 100, no. 6, pp. 1677 to 1689, 1955 (the demarcation energy E* = kT ln(νt) used to put time and temperature on one axis).
  10. I. Jun, M. A. Xapsos, S. R. Messenger, E. A. Burke, R. J. Walters, G. P. Summers and T. Jordan, "Proton nonionizing energy loss (NIEL) for device applications," IEEE Transactions on Nuclear Science, vol. 50, no. 6, pp. 1924 to 1928, 2003.

Retention and refresh

  1. T. Hamamoto, S. Sugiura and S. Sawada, "On the retention time distribution of dynamic random access memory (DRAM)," IEEE Transactions on Electron Devices, vol. 45, no. 6, pp. 1300 to 1309, 1998 (temperature dependence of retention time; the 10 °C rule is the engineering form of it).
  2. JEDEC Standard JESD79-4, DDR4 SDRAM (refresh interval 7.8 µs, 64 ms, at or below 85 °C; halved above).

Environment

  1. ISO 15390:2004, Space environment (natural and artificial): Galactic cosmic ray model.
  2. ECSS-E-ST-10-04C Rev.1, Space engineering: Space environment, ESA-ESTEC, 2020 (solar particle event model parameters, Annexes B and I).
  3. M. A. Xapsos, J. L. Barth, E. G. Stassinopoulos, E. A. Burke and G. B. Gee, "Space environment effects: model for emission of solar protons (ESP), cumulative and worst-case event fluences," NASA/TP-1999-209763, 1999.
  4. 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 Transactions on Nuclear Science, vol. 47, no. 3, pp. 486 to 490, 2000.
  5. M. A. Xapsos, J. L. Barth, E. G. Stassinopoulos, S. R. Messenger, R. J. Walters, G. P. Summers and E. A. Burke, "Characterizing solar proton energy spectra for radiation effects applications," IEEE Transactions on Nuclear Science, vol. 47, no. 6, pp. 2218 to 2223, 2000.
  6. J. Feynman, G. Spitale, J. Wang and S. Gabriel, "Interplanetary proton fluence model: JPL 1991," Journal of Geophysical Research, vol. 98, no. A8, pp. 13281 to 13294, 1993.
  7. L. Rosenqvist, A. Hilgers, H. Evans, E. Daly, M. Hapgood, R. Stamper, R. Zwickl, S. Bourdarie and D. Boscher, "Toolkit for updating interplanetary proton-cumulated fluence models," Journal of Spacecraft and Rockets, vol. 42, no. 6, pp. 1077 to 1090, 2005 (the updated JPL model parameters).
  8. C. Tranquille and E. J. Daly, "An evaluation of solar-proton event models for ESA missions," ESA Journal, vol. 16, pp. 275 to 297, 1992 (confidence levels by mission length).
  9. 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, NASA Goddard Space Flight Center, 1976.
  10. D. Heynderickx, B. Quaghebeur, J. Wera, E. J. Daly and H. D. R. Evans, "New radiation environment and effects models in the European Space Agency's Space Environment Information System (SPENVIS)," Space Weather, vol. 2, S10S03, 2004 (the AP-8 presets, the event transmission tables and the model checks on this page were made with SPENVIS 4.6.14 in October 2026).
  11. R. A. Leske, R. A. Mewaldt, E. C. Stone and T. T. von Rosenvinge, "Observations of geomagnetic cutoff variations during solar energetic particle events and implications for the radiation environment at the Space Station," Journal of Geophysical Research, vol. 106, no. A12, pp. 30011 to 30022, 2001 (the cutoff law sampled along the orbits).
  12. D. F. Smart and M. A. Shea, "A review of geomagnetic cutoff rigidities for earth-orbiting spacecraft," Advances in Space Research, vol. 36, pp. 2012 to 2020, 2005.

The constants of the models (ESP parameters, JPL parameters, AP-8 presets, transmission tables) and the checks against SPENVIS are listed in the source files, which carry the same references.

This page is a planning aid. Its numbers come from the measurements entered and the environment chosen, carry the assumptions listed under Conclusions, and are not a qualification result for any part or mission. Nothing on it has been reviewed by any organization.

What one particle does to a cell, in three dimensions

The mechanism behind the counts entered above. A single proton or heavy ion leaves a recoil cascade in the silicon; the vacancy complexes that survive it become generation centers, and when they sit in the depletion region of the storage node junction, next to the gate edge where the field is highest, trap-assisted tunneling multiplies their emission until the node is drained before the next refresh. Drag to turn the cell and step through the strike, the cascade, the defect chemistry, the leakage and the anneal; the full page carries the explanation, the numbers behind each step and the references.