space-rha.comShort courseChapter 1 · The natural space environment

Chapter 1 · Lesson 5 of 6 · about 14 minutes

What shielding changes, and what it cannot

The environment at the part differs from the environment outside the spacecraft. Structure, boxes, boards and the package itself absorb, slow and transform the particles before they reach the die. Aluminum stops trapped electrons in a few millimeters and barely touches a galactic iron ion. A shielding trade only makes sense once you know which population the shielding removes and which it leaves alone, and the same knowledge keeps a requirement from being written against the wrong spectrum.

What you need to know

Electrons. Trapped electrons of a few MeV have ranges of a few millimeters in aluminum. Their dose therefore falls steeply with the first millimeters of shielding, and what remains behind thick shielding is the bremsstrahlung they generate on the way, which is far smaller but penetrating. High-Z shielding stops electrons in less mass but makes more bremsstrahlung; a graded shield (low-Z outside, high-Z inside) is the usual compromise.

Protons. A 10 MeV proton stops in about 0.17 g/cm² of aluminum (0.6 mm), a 100 MeV proton in about 10 g/cm² (37 mm). Shielding of a few millimeters removes the low-energy part of the trapped and solar proton spectra and lowers the dose and displacement damage accordingly, but the proton spectrum above 100 MeV passes through any practical box, and nuclear reactions in the shield add secondaries.

Heavy ions. Galactic cosmic rays near 1 GeV per nucleon have ranges of tens of centimeters in aluminum, and the shielding that stops electrons does not change their flux appreciably. Shielding does shift the LET of the ions that pass (slowing an ion raises its LET until it stops), so the spectrum behind shielding differs somewhat from free space, but the integral flux above the iron knee is generally within a factor of a few of its free-space value for any realistic thickness. Heavy-ion SEE rates are therefore generally set by the device rather than by the box.

Geometry. Real shielding is not a sphere of aluminum. A part on a board sees different thickness in different directions: the chassis wall on one side, another board on the other, a connector cutout, the package lid. The standard approach is sector analysis, which traces rays from the part through the CAD model, converts the materials met along each ray to an equivalent aluminum thickness, and weights the dose-depth curve by the solid angle of each sector. Transport codes (Monte Carlo or deterministic) replace the curve with a calculation through the actual materials when the answer matters, for instance behind tantalum or near a cutout. Chapter 2 of this course covers these methods.

Figure 5.1. One year of dose at 800 km and 98.6° against aluminum shielding, by particle, from the Space-RHA hardware panel (AE-8 MAX electrons, AP-8 MAX protons, one August 1972 solar event through the orbit's cutoff, SHIELDOSE-2, center of a solid aluminum sphere). The vertical line marks 100 mil (0.69 g/cm²). The electron curve falls by two orders of magnitude between 1 and 3 g/cm²; the proton curve falls by less than a factor of two.
Polar LEO 800 km: one year of dose behind aluminum, by particle 0.1 1 10 1 10 100 1000 104 105 aluminum shielding, g/cm² (1 g/cm² = 3.7 mm = 146 mil) rad(Si) per year, center of a solid sphere 100 mil trapped electrons + bremsstrahlung trapped protons one August 1972 solar event bremsstrahlung alone

3D scene

Inside the box: what each particle does in a wall

The transport workbench: a teaching enclosure on the spacecraft with its board and target locations. Choose the particle to see the branches a shield opens (slowing, scattering, bremsstrahlung, nuclear secondaries), and change the wall material.

  1. Trapped electrons of a few MeV have ranges of a few millimeters in aluminum, so their dose falls steeply with the first millimeters of shielding. What remains behind thick shielding is the bremsstrahlung they generate on the way, far smaller but penetrating. High-Z material stops electrons in less mass but makes more bremsstrahlung, which is why a graded shield, low-Z outside and high-Z inside, is the usual compromise.

  2. A 10 MeV proton stops in about 0.17 g/cm² of aluminum (0.6 mm), a 100 MeV proton in about 10 g/cm² (37 mm). A few millimeters remove the low-energy part of the trapped and solar proton spectra and lower the dose and displacement damage accordingly, but the spectrum above 100 MeV passes through any practical box, and nuclear reactions in the shield add secondaries.

  3. Galactic cosmic rays near 1 GeV per nucleon have ranges of tens of centimeters in aluminum, and the shielding that stops electrons does not change their flux appreciably. Slowing an ion raises its LET until it stops, so the spectrum behind shielding differs somewhat from free space, but the integral flux above the iron knee generally stays within a factor of a few of its free-space value for any realistic thickness.

  4. Bremsstrahlung photons and the neutrons made by nuclear reactions are the secondary radiations a shield creates. Photons deposit their energy through the secondary electrons they set in motion; neutrons scatter, produce recoils and are captured with the emission of photons. The branches drawn here are schematic; the quantities come from a transport calculation.

  5. The same electrons in tantalum: they are stopped in far less thickness and scattered more strongly, and more bremsstrahlung is produced. Which material is the better choice depends on the incident spectrum, the volume available, the order of the layers and the quantity scored at the target, so compare candidates at equal thickness and at equal mass as two separate comparisons.

The original space-rha.com environment and mission renderer (NASA imagery and spacecraft models, credits in the view), with the course scenes drawn by the same renderer. Trails, clocks and counts are illustrative; the scene computes no dose or rate.

Worked example: what thicker aluminum changes, and what it leaves alone

The question. A board on the 800 km polar mission sits behind 1 g/cm² of aluminum (3.7 mm). The designer proposes 3 g/cm² (11 mm). What changes in the dose, and what happens to the heavy-ion single-event rate?

Inputs. The dose-depth curves of figure 5.1 (one year, rad(Si), solid sphere) and the CREME96 integral LET spectrum behind 100 mil of aluminum in free space (solar minimum) from the Space-RHA SEE tool.

AluminumElectrons + brems.Trapped protonsOne 1972 eventElectron share
0.5 g/cm² (1.9 mm)3,99052459888 %
1.0 g/cm² (3.7 mm)77940741566 %
2.0 g/cm² (7.4 mm)30.33062339 %
3.0 g/cm² (11.1 mm)5.82621562 %
5.0 g/cm² (18.5 mm)3.8217792 %
10 g/cm² (37 mm)2.1141211 %

Steps.

  1. Read the electron column. From 1 to 3 g/cm² the electron-plus-bremsstrahlung dose falls by a factor of 130, from 779 to 5.8 rad(Si) per year. The extra 2 g/cm² removes the electrons almost completely, and what is left is bremsstrahlung.
  2. Read the proton column. Over the same step the trapped-proton dose falls by a factor of 1.6, from 407 to 262 rad(Si) per year. The protons that reach this depth are above roughly 30 to 50 MeV, and the extra aluminum slows them without stopping them.
  3. Read the shares. At 0.5 g/cm² electrons are 88 % of the trapped dose; at 3 g/cm² they are 2 %. The proposed shielding changes which population sets the dose. Beyond 3 g/cm² further aluminum does little, since the proton dose falls only to 141 rad(Si) per year at 10 g/cm².
  4. Check the solar event. The one-event dose falls from 415 to 156 rad(Si), a factor of 2.7, because the solar spectrum at this orbit is softer than the trapped-proton spectrum behind the cutoff. The event is a larger share of the total at 3 g/cm² than at 1 g/cm².
  5. Now the heavy ions. Behind 100 mil in free space a 1 cm² die is crossed by an ion above LET 10 about 0.9 times per day, above LET 20 about 0.13 times per day (once a week), above LET 30 about 3.6 × 10⁻⁴ times per day (once in 8 years), and above LET 40 once in about 66 years. Going from 1 to 3 g/cm² does not change these figures appreciably. If the part on the board has a measured LET threshold of 30 MeV·cm²/mg, its galactic heavy-ion exposure is of order one crossing per cm² per decade whatever the box thickness. If its threshold is 10, it is crossed daily, and only the device cross section (chapter 3 and the SEE rate tool) decides how many of those crossings become events.

Result. The extra 2 g/cm² takes the yearly dose on this orbit from about 1,186 to about 268 rad(Si) in the sphere model, which may well be worth its mass if the parts are dose limited. It changes the heavy-ion single-event rate very little and does little for displacement damage from protons above 50 MeV. Whether it is worth the mass depends on which effect limits the parts, so the effect has to be named before the level (lesson 1).

Figure 5.2. The free-space galactic LET spectrum behind 100 mil of aluminum at solar minimum, read as a crossing rate for a 1 cm² die: ions above the LET on the axis cross the die at the rate on the vertical axis. Horizontal lines mark one crossing per day, week, year and decade. This is a count of crossings, not a rate of events; the device cross section decides how many become events.
Free space behind 100 mil of aluminum: how often an ion above a given LET crosses 1 cm² 1 10 100 10-8 10-7 10-6 10-5 10-4 0.001 0.01 0.1 1 10 100 1000 LET in silicon, MeV·cm²/mg ions per cm² per day above the LET one crossing per day one per week one per year one per decade GCR, solar minimum, Z = 2 to 92 LET 30: one crossing in about 8 years LET 10: about one a day LET 20: about one a week
Try it: dose behind a shield

Choose the orbit, the shield material and its areal density. The table gives one year of trapped dose by particle, the dose of one August 1972 event, and the mission total for the years and events you enter.

Going deeper

rad(Si), rad(Al) and the detector material

Dose is deposited energy per unit mass of a named material. SHIELDOSE-2 reports dose in silicon, in silicon dioxide, in aluminum or in water, and the choice matters at the percent to tens-of-percent level because the stopping powers and photon cross sections differ. Historical curves are often in rad(Al); modern requirements are generally in rad(Si), with rad(SiO₂) for oxide-dominated effects. State the material with the unit every time, and convert explicitly when combining sources. The figures in this lesson are rad(Si) at the center of a solid aluminum sphere, which is the SHIELDOSE-2 geometry closest to a part surrounded by structure on all sides. The finite-slab geometry gives lower values and the semi-infinite medium higher.

LET, range and why the knee is where it is

Linear energy transfer is the energy lost per unit path length, normalized by density, in MeV·cm²/mg; for a given material it depends on the ion's charge and velocity through the Bethe formula, roughly as Z²/β². An iron nucleus has its minimum LET of about 1.2 MeV·cm²/mg in silicon at several GeV per nucleon and its maximum of about 28 near 1 MeV per nucleon, where it is close to stopping. The galactic spectrum peaks near 1 GeV per nucleon, where iron's LET is about 1.5; the ions above LET 20 are the slower iron nuclei, and nothing in the galactic spectrum with an appreciable flux has a higher LET than the iron maximum. That produces the knee at 27 to 30 in figure 5.2 and explains why the flux above LET 40 is more than four orders of magnitude below the flux above LET 10. Shielding moves individual ions along their own LET-versus-energy curves without creating new high-LET ions, apart from rare fragments and recoils.

Equal mass is not equal shielding

Two shields of the same areal density (g/cm²) do not give the same dose behind them. For electrons, higher-Z materials scatter more and stop in less mass but produce more bremsstrahlung. For protons, hydrogen-rich materials stop more per unit mass. For neutrons produced in the shield, the composition decides the secondary field. The site's hardware panel compares aluminum, polyethylene, tantalum and a graded stack for the same three orbits, with Monte Carlo electron kernels; chapter 2 of this course takes up the material question in detail.

The common mistake

The common mistake is to carry "100 mil of aluminum" from the environment specification into the part requirement as if it were the real shielding. It is a convention for comparing environments, chosen because it is close to a typical box wall, and it is the depth at which most published LET spectra and many dose figures are quoted. The dose at a real part can be several times lower behind a thick chassis and several times higher next to a cutout. Sector analysis of the actual configuration replaces the convention before the part-level requirement is written.

Takeaway

Shielding of a few millimeters removes electrons and the softest protons and leaves the heavy ions and the hard protons alone, so its value depends on which population limits the part.

Check yourself

A box wall is thickened from 2 to 4 mm of aluminum to protect a commercial processor from latchup. Is that likely to help?

Generally not. Latchup in a commercial CMOS processor is caused by heavy ions and by proton-induced recoils, and 2 mm of aluminum changes neither flux appreciably: the galactic ions above the latchup threshold pass through, and the protons above a few tens of MeV that produce the recoils pass through as well. The mass would be better spent on current limiting and power cycling, or on a part with a demonstrated latchup threshold above the iron knee.

From 1 to 3 g/cm² of aluminum the electron-plus-bremsstrahlung dose on the polar orbit falls by a factor of 130, while the galactic heavy-ion crossing rate above LET 30 stays near 3.6 × 10⁻⁴ per cm² per day. What makes the two populations respond so differently to the same 2 g/cm²?

Range. Trapped electrons of a few MeV have ranges of a few millimeters in aluminum, so going from 3.7 to 11.1 mm stops nearly all of them, and the 5.8 rad(Si) per year that remains is mostly the bremsstrahlung they produced on the way. Galactic cosmic rays near 1 GeV per nucleon have ranges of tens of centimeters, so the same aluminum slows them slightly, shifting their LET, without stopping them, and the integral flux above the iron knee generally stays within a factor of a few of its free-space value for any realistic thickness. Heavy-ion SEE rates are therefore generally set by the device rather than by the box.

A part-level TID requirement for the five-year polar mission is written as 12,900 rad(Si) because that is the figure behind 100 mil in the specification (lesson 4). The part sits behind the 11 mm chassis wall on one side and faces a connector cutout on the other. What has gone wrong?

The 100 mil solid-sphere figure is a convention for comparing environments, chosen because it is close to a typical box wall, and it is not the shielding of this part. The dose at a real part can be several times lower behind a thick chassis and several times higher next to a cutout, and this part has both. Sector analysis of the actual configuration (rays traced through the CAD model, materials converted to equivalent aluminum, the dose-depth curve weighted by solid angle) replaces the convention before the part-level requirement is written, and the project margin policy is added on top (chapter 3). As written the number could be too strict or too loose, and without the geometry there is no way to tell which.

A designer proposes replacing an aluminum lid with tantalum of the same areal density to cut the electron dose further. Why is equal g/cm² not a sufficient basis for the comparison?

Two shields of the same areal density do not give the same dose behind them. A higher-Z material such as tantalum stops electrons in less mass and scatters them more strongly, but it produces more bremsstrahlung, which is far more penetrating than the electrons that made it, and a graded shield with low-Z outside and high-Z inside is the usual compromise. For protons, hydrogen-rich materials stop more per unit mass, and for the neutrons made in the shield the composition decides the secondary field. Candidates are generally compared at equal thickness and at equal mass as two separate comparisons, with a transport calculation through the actual materials when the answer matters (chapter 2).

A 10 MeV proton stops in 0.17 g/cm² of aluminum (0.6 mm) and a 100 MeV proton in 10 g/cm² (37 mm). A reviewer concludes that the proposed 11 mm wall makes the board safe from protons. What does the dose table say, and what does the rule leave out?

The trapped-proton dose falls only from 407 to 262 rad(Si) per year between 1 and 3 g/cm², a factor of 1.6, and is still 141 at 10 g/cm², because the protons that reach these depths are above roughly 30 to 50 MeV and the extra aluminum slows them without stopping them. The spectrum above 100 MeV passes through any practical box, nuclear reactions in the shield add secondaries, and displacement damage from protons above 50 MeV changes little. The solar-event dose falls faster (415 to 156 rad(Si), a factor of 2.7) because that spectrum is softer at this orbit, so the event becomes a larger share of the remaining dose rather than a smaller one.

Further reading

  • S. M. Seltzer, "Updated calculations for routine space-shielding radiation dose estimates: SHIELDOSE-2," NISTIR 5477, 1994.
  • S. Bourdarie and M. Xapsos, "The near-Earth space radiation environment," IEEE Trans. Nucl. Sci., vol. 55, no. 4, pp. 1810-1832, 2008. NTRS 20080039622.
  • NASA, "Avionics Radiation Hardness Assurance (RHA) Guidelines," NASA/TM-20210018053, 2021. NTRS 20210018053.
  • 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.
  • Space-RHA, "Space radiation environment in 3D," space-rha.com/space-radiation-environment (the hardware panel supplies the dose figures used here).
  • Space-RHA, "SEE rate tool," space-rha.com/see-tool.html (the CREME96 LET spectra used here).