space-rha.comShort courseChapter 6 · Evidence strategy and facility selection

Chapter 6 · Lesson 2 of 4 · about 14 minutes

Choosing TID and displacement-damage sources

Gamma rays, electrons and X-rays all deposit ionizing dose, but a dose reading means something only once the material and the place where the energy is deposited are named. A convenient source has to be shown equivalent for the device or board in question.

What you need to know

Dose is energy absorbed per unit mass in a named material. Cobalt-60 gamma rays (1.17 and 1.33 MeV), electron beams and X-ray sources all deposit ionizing dose, but the same reading on the dosimeter does not mean the same energy in the oxide of the part. Every dose number should name its material: rad(Si), rad(SiO₂) or Gy(water), with 1 Gy = 100 rad. Co-60 remains the conventional reference for total ionizing dose because its photons penetrate well enough that the dose is nearly uniform through a package, and because decades of device data were taken with it. Near high-atomic-number materials (gold plating, kovar lids), lower-energy photons produce photoelectrons that raise the local dose, and electron beams build up charge and produce bremsstrahlung. Both sources can be used, but the equivalence to Co-60 has to be established for the device rather than assumed.

The rate at which the dose is delivered matters as much as the dose. MIL-STD-883 Test Method 1019 defines a standard condition of 50 to 300 rad(Si)/s, and a low-dose-rate condition at or below 0.01 rad(Si)/s for parts that may show enhanced low-dose-rate sensitivity (ELDRS), which bipolar and BiCMOS linear circuits generally do. ESCC 22900 defines lower standard windows. For planning, the arithmetic is what matters: 20 minutes delivers 60 krad(Si) at 50 rad/s, while the same dose at 0.01 rad/s takes 69 days. A low-dose-rate test therefore has to be scheduled months ahead, so a project decides early whether its linear parts need one, whether an accelerated alternative (elevated-temperature irradiation, where it has been validated for the part) is acceptable, or whether the parts will be bounded by an enhancement factor and extra margin.

The bias and temperature during irradiation, and the timing of the measurements afterward, are part of the result. MOS oxides trap holes and build interface traps on different time scales, so a part measured an hour after irradiation and a part measured a week after can read differently. TM1019 therefore times the post-irradiation measurements and, for MOS parts, provides a room-temperature anneal and a 168-hour 100 °C rebound test that exposes interface-trap growth after the trapped holes anneal. Bipolar parts generally do not recover in the same way, and some continue to degrade. The experiment should follow the mission's worst case. For MOS parts the bias state that traps the most charge is generally the one with the largest field across the oxide, while some bipolar linear parts degrade most at zero bias because of their structure, so test the bias states the mission will use rather than assuming one.

Displacement damage is a different quantity. A proton, electron or neutron displaces atoms in proportion to its non-ionizing energy loss (NIEL), so fluences of different particles and energies can be compared only after conversion to a damage-equivalent fluence at a reference energy (10 MeV protons or 1 MeV electrons are common references). NIEL scaling holds reasonably well for many silicon devices. It holds less well where the defect type matters (clusters from heavy recoils against isolated defects from electrons) or where the material is not silicon, so optoelectronic parts, which U4 represents, should be tested with the particle and energy that reach the active region and checked at a second energy if the mission spectrum is broad. A proton exposure also delivers ionizing dose. At 50 MeV the dose is about 1.7 × 10⁻⁷ rad(Si) per proton per cm², so 10¹¹ protons per cm² brings about 17 krad(Si) along with the displacement damage. That dose either confounds the DD endpoint or serves as the TID exposure, and the plan has to say which.

Figure 6.2. Irradiation time against dose rate for 30 and 60 krad(Si). The low-dose-rate condition turns a half-hour exposure into weeks; a project that needs the result must start it early.
Irradiation time against dose rate for two total doses 0.001 0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000 104 dose rate, rad(Si)/s irradiation time, hours TM1019 low dose rate: 0.01 rad/s or less one day one week 60 krad(Si) 30 krad(Si) TM1019 condition A: 50 to 300 rad/s

Worked example 6.2. Planning the U3 and U4 cumulative tests

U3 is a bipolar precision amplifier with a mission dose of 15 krad(Si) behind its shielding and a required test dose of 30 krad(Si) with the design margin of 2 (chapter 3). Because bipolar linear parts generally show ELDRS, the plan has two legs.

LegDose rateDoseIrradiation timePurpose
High dose rate, five parts50 rad(Si)/s75 krad in steps of 10 krad to 30 krad, then 5 krad25 min of beam, two days with the twelve measurementsparameter trajectories to failure for the tolerance bound (chapter 8 lesson 3)
Low dose rate, five parts0.01 rad(Si)/s30 krad35 daysthe ELDRS factor at the required dose, which is the result the application claim depends on

The low-dose-rate leg is 35 days of chamber time and must start at least two months before the result is needed. U4 is an optocoupler whose current transfer ratio degrades with displacement damage. The plan irradiates it with 50 MeV protons to 3 × 10¹¹ per cm² in four steps and measures CTR at the flight drive current at each step. The 50 MeV exposure deposits about 51 krad(Si) by the last step, above the 15 krad mission dose, so the test is recorded as a combined TID and DD exposure and no separate Co-60 exposure is planned. The mission fluence that this test must cover is computed in chapter 10 lesson 4 from the sun-synchronous trapped-proton spectrum of chapter 1.

A question to carry through the course

When, if ever, can the dose applied to a part during a proton SEE test be used as a proxy for a TID test?

Things to weigh here:

  • The dose each proton deposits at the energies used: about 1.7 × 10⁻⁷ rad(Si) per proton per cm² at 50 MeV, and about a quarter of that at 200 MeV.
  • The dose rate the run delivers, set against the TM1019 windows: condition A is 50 to 300 rad(Si)/s, and the low-dose-rate condition is 0.01 rad(Si)/s or less.
  • For a bipolar part, the ELDRS question, which a fast proton run cannot answer.
  • The displacement damage that arrives with the dose, which matters for bipolar and optical parts.

Exercise

Before you choose a source, define the dose material, the coverage of the active region, the dose rate and the parametric measurements the analysis needs.

For a displacement-damage test, record the material, the reference particle, the evidence for the NIEL scaling, the temperature history and the ionizing dose delivered with the fluence.

Going deeper

Interfaces and competing time scales

With low-energy photons, attenuation and dose enhancement near high-Z materials and interfaces change the dose the oxide receives. Electron beams can produce dose gradients and bremsstrahlung. Oxide trapping, interface-trap growth and annealing run on different time scales, so matching the total dose alone does not make one source equivalent to another, or to the mission.

A high-dose-rate result should not be extrapolated to a low-dose-rate bipolar application without a validated method and evidence for that technology.

Which defects control the endpoint

NIEL describes the energy that goes into displacing atoms. It does not describe the electrically active defects that result, and isolated defects, clusters, material composition, injection conditions and annealing can all change the relation between NIEL and a measured endpoint. The scaling should be checked against evidence from the device in question rather than taken from one curve as a law of the material.

Where the response model is in doubt, a second energy or a second particle type helps to tell competing models apart and to separate ionization from displacement effects.

The common mistake

A TID exposure says nothing about single-event response, and a board-level exposure does not qualify every device on the board.

The silicon NIEL curve does not carry over unchanged to other materials or to every device endpoint.

Takeaway

  • The choice of source can change both the dose deposited in the part and the response per unit dose.
  • A displacement-damage claim needs a defensible link from the transported spectrum to the measured degradation of the function.

Check yourself

A test house offers a 10 keV X-ray irradiator for a quick TID screen of U3. What must be shown before the result is used in place of Co-60?

That the dose in the sensitive oxide matches, rather than the dose at the dosimeter. Low-energy photons are attenuated and enhanced by the package and metallization, so an equivalence for this device must be established, generally by comparison with Co-60 on the same part type.

Why does the dose-rate decision have to be made at the start of the program?

Because the low-dose-rate exposure to 30 krad at 0.01 rad(Si)/s takes about 35 days and cannot be inserted late. The alternatives (validated accelerated methods or an enhancement factor with extra margin) need evidence of their own.

A 1 MeV electron fluence and a 50 MeV proton fluence are proposed as interchangeable DD exposures for U4. What is required for that to hold?

That the device response scales with NIEL between the two particles, which has to be checked for the device and endpoint. Electrons produce mostly isolated defects while protons also produce clusters, and optoelectronic materials do not always follow the silicon NIEL curve.

The U4 proton run delivers about 51 krad(Si), above the 15 krad mission dose, and the plan drops the separate Co-60 exposure for U4. A colleague proposes the same shortcut for U3. Why does it not carry over?

U4 is an optocoupler whose endpoint is the current transfer ratio under displacement damage, so the ionizing dose that arrives with the fluence is recorded as part of a combined TID and DD exposure. U3 is a bipolar linear part whose application claim rests on the ELDRS factor at 30 krad(Si), and the dose rate a proton run delivers has to be set against the TM1019 windows; the ELDRS question is one a fast proton run cannot answer. The 35-day leg at 0.01 rad(Si)/s therefore stays in the plan.

Sixty krad(Si) at 50 rad(Si)/s is 20 minutes of beam, and the U3 high-dose-rate leg needs 25 minutes. A colleague books the chamber for an afternoon. What does the beam time leave out?

The twelve measurements. The leg delivers 75 krad(Si) in steps (10 krad to 30 krad, then 5 krad), and each step is followed by a parametric measurement within the window TM1019 sets, so the plan budgets two days for the leg. The timing of those measurements is part of the result, since a part measured an hour after irradiation and one measured a week after can read differently, so the schedule generally cannot be shortened by batching the measurements or letting them slide.

For MOS parts the bias that traps the most charge is generally the one with the largest field across the oxide. Why does the lesson still ask for the bias states the mission will use to be tested, rather than that one alone?

Because the rule describes MOS oxides and does not carry to every part. Some bipolar linear parts degrade most at zero bias because of their structure, and a mission may hold a part in a state (unpowered, standby) that the largest-field rule would not select. The experiment should follow the mission's worst case, which is found by testing the states the mission uses rather than by assuming the one the rule names.

Further reading

  • MIL-STD-883, Test Method 1019, "Ionizing radiation (total dose) test procedure."
  • ESCC Basic Specification No. 22900, "Total dose steady-state irradiation test method."
  • ASTM F1892, "Standard guide for ionizing radiation (total dose) effects testing of semiconductor devices."
  • R. L. Pease, R. D. Schrimpf and D. M. Fleetwood, "ELDRS in bipolar linear circuits: a review," IEEE Trans. Nucl. Sci., vol. 56, no. 4, pp. 1894-1908, 2009.
  • J. R. Schwank et al., "Radiation effects in MOS oxides," IEEE Trans. Nucl. Sci., vol. 55, no. 4, pp. 1833-1853, 2008.
  • J. R. Srour, C. J. Marshall and P. W. Marshall, "Review of displacement damage effects in silicon devices," IEEE Trans. Nucl. Sci., vol. 50, no. 3, pp. 653-670, 2003.
  • G. P. Summers, E. A. Burke, P. Shapiro, S. R. Messenger and R. J. Walters, "Damage correlations in semiconductors exposed to gamma, electron and proton radiations," IEEE Trans. Nucl. Sci., vol. 40, no. 6, pp. 1372-1379, 1993.
  • Space-RHA, "NIEL tool," space-rha.com/niel-tool.