Readying radiation-hardened ICs for space flight
StoryJune 21, 2016
Josh Broline
Renesas
Nick van Vonno
Renesas
The radiation environment of space presents several challenges for satellites and deep-space flight systems. Acceptance testing of integrated circuits (ICs) ensures predictable performance and prevents system failure while in flight through the various radiation environments encountered in nearly all mission profiles.
To understand the effects of radiation on electronic systems and integrated circuits, one must first understand the source of radiation. Radiation in space consists almost entirely of particles, including electrons, protons, and energetic heavy ions. Most of these particles originate from the solar wind or solar flares. Superimposed on this particle flux are very high-energy protons and heavy ions, which are isotropic. The moderate energy particles – such as electrons and protons – are trapped by the Earth’s magnetic field in the Van Allen belts; depending on a satellite’s orbit, the belts may cause most of the ionizing radiation damage. High-energy protons and heavy ions also are affected by the magnetosphere but are much more difficult to trap.
All of these particles interact with the materials used in electronic devices. This interaction occurs because solar electrons and protons are abundant and cause ionization in materials. In a simplified model, low- and moderate-energy charged particles generate hole-electron pairs in the thermal oxides used in integrated circuits. The electron mobility in these oxides is very high and any applied electric field sweeps the electrons out of the oxide in picoseconds. The hole mobility is much lower, so a much greater proportion of holes get trapped. The result of these asymmetrical trapping dynamics is the positive volume charging of dielectric layers and degradation of both bipolar and metal-oxide semiconductor (MOS) circuit devices.
Figure 1: Galactic cosmic-ray energy spectrum – particle flux plotted as a function of particle energy.
(Click graphic to zoom)
|
|
Figure 1 shows that when the particle energy is increased, the particle abundance decreases, with the abundance-versus-energy curve spanning 25 orders of magnitude and culminating in very low fluxes of relativistic TeV (teraelectronvolt) heavy ions. The effects of the abundant lower-energy particles are uniform through the volume of the IC, but high-energy heavy ions cause single-event effects, defined as the interaction of a single energetic ion with a silicon device. Energy is lost by these high-energy particles as they pass through the semiconductor lattice, generating a track of hole-electron pairs. The resulting charge is collected and can change the voltage on sensitive nodes, which can affect circuit operation.
Single-event effects can be divided into destructive and nondestructive phenomena. Nondestructive effects include bit flips, functional interrupts in digital applications, and transients on the outputs of analog functions. Destructive effects include latchup, burnout, and MOS gate oxide rupture, which can lead to permanent damage. The most commonly used unit in single-event-effects work is the linear energy transfer (LET) of the incoming ion, which equates to the energy loss (dE/dx) per unit track length for a given material density; this unit is expressed in MeV.cm2/mg. The key takeaway: Shielding is ineffective as it only filters out the low-energy end of the spectrum. The high-energy ion flux is unaffected, and mitigation must be performed at the IC and system levels.
Impact of commercial space
Since the end of the space shuttle program, the U.S. has seen a definite trend towards increased commercial participation in launches and missions. Today’s launch-vehicle market has changed from government-run programs carried out by private contractors to programs run by private industry. Also seen has been the growth of space tourism and commercial spaceports. These trends, in addition to cost-reduction pressures in government-funded programs, have resulted in increased market penetration by low-cost alternatives to high-reliability ICs. Some of the alternatives include military-grade Class Q parts, straight commercial off-the-shelf (COTS) parts, and even automotive parts. The leading edge of this trend is in commercial launch vehicles, which did not use high-grade ICs to begin with, due to short mission time and emphasis on cost control.
The objectives of private industry are to reduce the cost and increase the reliability of access to space – and to make a profit. This model demands simplicity, reliability, and sharply reduced cost throughout the entire launch vehicle, payload, and launch infrastructure system. Commercial space products are designed for compatibility with multiple and limited capability infrastructure facilities used for production, launch, and mission support.
NASA’s focus
NASA’s key initiative in this space is the Orion program, which is developing a manned spacecraft to be used as an exploration vehicle for a range of missions including a manned mission to Mars. Orion provides launch, launch abort, transit to the specific objective, and reentry upon its return to Earth. The Orion EFT-1 flight launched from Cape Canaveral on December 5, 2014, for a two-orbit, four-hour demonstration mission. Orion EFT-1 evaluated launch and re-entry systems such as avionics, attitude control, parachutes, and the heat shield required for high-speed re-entry. Orion is a steppingstone to later manned exploration of Mars. Until now, Mars has been explored through an intensive program of rovers, orbiters, and imagers, supplemented by extensive research into the human and health aspects of a long-duration mission in a difficult radiation environment.
NASA has historically categorized parts for space applications by reliability assurance levels. In this system, a Grade-1 part is considered the most reliable and is suitable for mission-critical and manned flight applications. Grade-2 parts are intended for general-purpose applications, while Grade-3 parts are suitable for non-mission, higher-risk applications. These grades correlate with MIL-PRF-38535 QML-V or QML-S compliance, Class B, Q, or H compliance, and MIL-STD-883 compliance, respectively, and the part-procurement cost follows this sequence. The main differences in grades are the types and extent of screening and the product assurance procedures used in production.
Commercial parts are routinely “up-screened” through implementation of some or all of the screens for a given grade, but the resulting costs are typically equal to or even greater than parts screened by the manufacturer. Radiation hardness is a key issue here – one that COTS may have problems with – including radiation effects that range from total dose degradation of performance parameters, which can cause system degradation; to destructive single-event effects, which can be mission-ending.
COTS systems are predominantly one-off systems, typically built with stringent cost limitations and a mixture of COTS and high-reliability (hi-rel) parts: The hi-rel parts go into mission-critical sockets, while the COTS parts are used in the rest of the system.
In this model, COTS parts are procured on the commercial market, usually with several known date codes or other traceability indicator and in quantities sufficient for the entire mission requirements. The parts are then qualified on a sampling basis, with some lots passing and the failing lots discarded. Overall, this may be a cost-effective method, but its risk and unpredictability are not consistent with high-stakes missions such as communications satellites and national-security payloads. Commercial communications missions have stringent time-to-market and insurance constraints and so have historically used the highest-quality parts available. They are expected to continue doing so, as are national-security assets.
Low-dose-rate testing and qualification
The effects of space radiation are caused by particles interacting with the materials used in electronic devices, so truly rigorous radiation tests would replicate these particle environments. Electron and proton testing is inconvenient and expensive, though, so the space community has historically used gamma rays for ground testing. While it is in fact a simulation, behind it is 50 years of correlation and high confidence using high-energy photons to predict part response to charged particles.
One inexpensive gamma-ray source – 60Co – is a synthetic isotope of cobalt that has two photon energies at 1.17 million electron volts (MeV) and 1.33 MeV. It is widely used for “total dose” testing of electronic components, in which the actual dose rate during irradiation was found to play a major role. Note that the unit for total ionizing dose is the radiation-absorbed dose (rad), which is equivalent to 100 ergs per gram. The energy absorption is specific to the material being irradiated, so the common unit in silicon technology becomes the rad(Si).
Total dose testing was originally performed at high dose rates in the 50 to 300 rad(Si)/s range, which is a convenient approach as the test takes a few minutes for a 100 krad(Si) exposure. Research performed in 1992 showed enhanced vulnerability of bipolar analog parts at dose rates of as low as 0.01 rad(Si)/s. However, the problem is that the dose rate in space is even lower than that. High-dose-rate testing is now recognized as an excessively accelerated test in many technologies. The downside is that a low-dose-rate test to 50 krad(Si) at the 0.01 rad(Si)/s dose rate required by MIL-STD-883 Test Method 1019 takes 10 weeks, which increases the cost of acceptance testing. Part hardness at the low dose rates found in space has turned into an important hardness-assurance issue, with many users insisting on low-dose-rate testing on a characterization or acceptance basis and deemphasizing high-dose-rate testing altogether.
In response to customer demand, Intersil Corporation in 2012 introduced an industry-leading low-dose-rate hardness-assurance program, which performs wafer-by-wafer acceptance testing at both low and high dose rate using on-site production irradiators. The distinct part number suffix (EH) defines a part type as having been acceptance tested at both dose rates; most Intersil radiation-hardened ICs are now available with this option.
In the development stages of the low-dose-rate acceptance-testing program, Intersil considered using accelerated low-dose-rate testing, in which irradiations are carried out at elevated temperature or in a modified, hydrogen-rich package atmosphere. While these methods produced encouraging results in a research and development environment, they were inconsistent with the limited production of QML-V-compliant parts. A detailed correlation study would be required for each part number, with periodic verification of correlation accuracy, and it was found that simply performing the tests was the most cost-effective approach.
The Intersil EH parts (see Figure 2) are low-dose-rate tested (biased and unbiased) to 50 krad(Si), with a parallel high-dose-rate test (biased only) to the applicable data sheet level. At 0.01 rad(Si)/s, the 50 krad(Si) irradiation test takes 10 weeks to perform.
Figure 2: The ISL71840SEH 16-channel multiplexer is a drop-in replacement for Intersil’s HS9-1840ARH, which has been aboard nearly every satellite and space exploration mission, including NASA’s Orion spacecraft flight test.
(Click graphic to zoom)
|
|
Future missions
While commercial space will continue using a mix of COTS and high-reliability parts, high-stakes mission profiles will continue to rely on low-dose-rate testing to meet the mission-assurance needs of military and communications satellites as well as manned spacecraft for deep-space exploration. Successful spaceflight system performance requires radiation hardening at the part level as well as thorough characterization and acceptance testing.
Josh Broline is a product marketing manager for Intersil’s mil/aero products group. He is responsible for developing and driving strategic plans and product roadmaps for radiation-hardened products used in spaceflight applications. Josh holds a BSEE from the University of Central Florida and an MBA from Florida Institute of Technology. Readers may reach him at jbroline@intersil.com.
Nick van Vonno is a principal engineer in Intersil’s mil/aero products group. He has 42 years of service with Intersil and its predecessors in a number of technical and management posts. He is currently responsible for radiation-effects research, customer support, and product technology development. Nick holds a BSEE from the University of Florida; he is also a senior member of IEEE and won the 2009 IEEE Radiation Effects Award. His email address is nvanvonn@intersil.com.
Intersil www.intersil.com
