Chapter 3 showed that single-event effects (SEEs) hardness assurance is a complex methodology involving many different processes and systems, all of which must work smoothly if the radiation testing system is to function properly. With the previously small radiation testing community, the process has been adequate, with most conventional satellites using few state-of-the-art electronic components. Now that the community is growing to include entities outside that small community, and as the technologies that are planned to be incorporated advance, there is a need to improve the coordination and capabilities of the radiation testing infrastructure. There are 27 active proton-treatment facilities in the United States (with another nine being built). It is at least conceivable that future demand could be met through relationships with these facilities. Options for heavy ions are much more limited. The K150 project at Texas A&M University (TAMU) could add a significant amount of moderate-energy heavy-ion beam time. Resolution of funding issues at Lawrence Berkeley National Laboratory (LBNL) could increase beam time there by about 50 percent. Taken together, these actions could perhaps double available heavy-ion time. In this rapidly evolving field, however, future demand is difficult to quantify. The actual number of hours required depends on the evolution of the commercial space sector and electronics technology as well as future developments in testing methodology and the level of reliability required for future missions. Greater demands for reliability will result in greater demand for heavy-ion testing, while proton testing demand may increase if there is a large increase in the number of risk-tolerant missions.
Planning for the future of testing infrastructure needs to take into account both the changing mix of users—from primarily government-funded or large commercial ventures to including an increasing number of smaller missions and short lifetime satellites. This planning also needs to consider the evolution of electronics, pushing the bounds of complementary metal–oxide–semiconductor (CMOS) integration to the end of Moore’s Law and beyond with new device structures and materials.
This chapter specifically focuses on the changes that are likely to drive the space electronics testing infrastructure needed in fiscal year (FY) 2018 and beyond (nominally through 2030, particularly in the case of particle accelerators) to adequately provide the required capabilities for new and emerging electronic technologies. It also identifies the principal gaps that exist between existing and needed infrastructure.
The numbers of spacecraft designated for low Earth orbit (LEO), geostationary orbit (GEO), and beyond are expected to increase significantly out to 2030. Large government missions will continue, and there will be an increasing number of smaller satellite and CubeSat missions, both government and commercial. In addition, with a larger number of operating satellites, failures could have system-wide effects, as failed satellites could pose collision risks to other satellites. With the expansion of the previously close-knit satellite community comes expansion of demands on the radiation testing infrastructure, both additional capacity and additional capability.
There will be continued emphasis on expanding the commercial sector in contributing to U.S. leadership in space, which implies continued attention paid to improving spacecraft safety; fault detection, isolation, and recovery (FDIR); and diagnosing and potentially solving system failures. This places paramount importance on meeting the radiation-hardening needs of an expanding space community requiring testing and modeling services. The movement to smaller satellites for both commercial and government use will increase the pressure to develop less expensive radiation-hardened devices and systems. Commercial operators, especially those proposing smaller, less expensive satellites, may be less inclined to conduct high-cost radiation tests for several reasons, including lack of financial resources, testing expertise, and shorter planned lifetimes compared with traditional spacecraft. Therefore, spacecraft failures after a few months to a few years can be expected to increase and need to be accounted for in business planning and financial risk analysis.
Plans for large constellations of commercial satellites in LEO (e.g., OneWeb, SpaceX), taken together with evolving commercial test and evaluation practices, will increase the demand for proton testing. These and other commercial mission developers are increasingly using protons up to ~200 MeV to test their electronic subsystems in lieu of heavy-ion tests on individual parts. Often the tests are considered pass-fail for total system performance, in contrast to conventional methods of characterizing individual parts for SEE. A pass-fail methodology can require additional test time if the test fails in contrast to traditional SEE characterization.
Looking out to 2030, many missions will involve the processing of massive data stores and the use of high-bandwidth communications (e.g., optical, hybrid optical/radio frequency, and quantum) for faster data streaming. Optical communications and sensor integration will assume an increasingly important role in the future. These goals impose demanding requirements for high-volume data collection and management and greater autonomous precision assist in guidance, navigation, and control (GNC); approach, proximity, and landing (APL); and grappling maneuvers. The addition of such demanding and safety-critical capabilities will increase demand for radiation-hardened solutions and for radiation testing of commercial technologies. This will require new methodologies for testing and modeling.
Advanced microelectronic and optoelectronic devices will continue to be required to support space science and human and robotic explorations both within the solar system and in Earth orbits and deep space. Some missions will extend to several years and have limited connections with Earth control centers, making the following goals even more important than they are today: (1) achieving improved spacecraft systems reliability and performance; (2) enabling advanced in situ measurements, autonomous operations, and remote manipulation and sampling; and (3) employing advanced sensors of various types to better understand Earth and the universe. All space-based activities would continue to address the imperative of withstanding the space environment, to include space radiation in all of its forms as well as impacts from micrometeoroids and orbital debris (MMOD).
Complementary metal–oxide–semiconductor (CMOS) is the primary enabling technology of integrated electronics, now and for the foreseeable future. Traditional scaling (geometrical scaling during the 1970s, 1980s, and 1990s, followed by “equivalent scaling” since then) has enabled device integration to keep pace with Moore’s Law that predicts the number of transistors on cutting-edge integrated circuits doubling approximately every 2 years without a corresponding increase in manufacturing costs. Eventually, expected performance gains from simply making the individual transistors smaller will stall, due to the combined effects of the need to reduce the supply voltages to control device leakage and reduce power consumption, and the increasing wiring load of densely
compacted transistors.1 The addition of such demanding and safety-critical capabilities will increase demand for radiation-hardened solutions and for radiation testing of commercial technologies.
Figure 4.1 illustrates the progression of commercial electronic parts, which are not designed for use in radiation environments, and radiation-hardened electronic parts, which are designed, manufactured, and tested to ensure performance in radiation environments. During this period, scaling of both commercial and radiation-hardened devices has progressed exponentially; however, the slope of the respective scaling curves have diverged, indicating an increasing performance gap between commercially available electronic parts and radiation-hardened parts. Figure 4.1 also illustrates a time lag in radiation-hardened parts compared with commercial-part scaling that increases from a 5-year lag in 2000 to an 8-year lag in 2015. With decreasing feature size often comes increased performance speed and reduced power consumption, and it is clear that spacecraft designers face pressure to use the more advanced commercial parts to take advantage of their capabilities. It seems inevitable that the radiation-hardened electronics will not keep up with those designed for a more benign Earth environment, increasing the pressure to use commercial electronic parts, especially for the smaller commercial satellites.
Finding: Most estimates show commercial CMOS/flash devices stretching Moore’s Law for at least a few more generations (8 to 10 years). Radiation-hardened devices will likely reach their scaling limit sooner because of increasing uncertainties in predicting and mitigating SEE rates. Thus, in a decade, it is likely that SEE testing will have to be done on novel devices using a broad range of technologies that will change with each new generation of devices.
Current trends in CMOS technology include smaller feature sizes and novel device structures to maintain Moore’s Law, which also drives more integration on individual chips, and stacking chips to increase integration within a package.
The 2015 International Technology Roadmap for Semiconductors (ITRS) report (“ITRS 2.0”), released in July 2016, forecasts the end of traditional scaling by 2021—the shrinking of chip features to achieve greater functional
1 G.E. Moore, Cramming more components onto integrated circuits, Electronics 38(8):114-117, 1965.
density and performance. The end will be driven primarily by the unfavorable economics of further scaling, as determined by the cost per transistor metric. That is, the increased costs of lithography steps and wafer processing at advanced technology nodes are pushing the industry toward alternatives to improve the performance and functionality of electronic devices at lower costs. Also there is the desire to integrate disparate technologies (e.g., memory, logic, sensors, radio frequency) in small form factors. Chip manufacturers will instead pivot toward vertical (3D) integration, so-called “3D power scaling,” building multiple layers of circuitry atop one another and connecting them with a dense wiring network. Specifically, the ITRS predicts that chip manufacturers will most likely move from device structures currently used in leading-edge chip production (such as the Fin Field effect transistor [FinFET], a device with a thin 3D channel with its gate draped on three sides, and the fully depleted silicon on insulator [FDSOI] conventional device with an ultrathin channel built above a buried insulator layer) to a lateral, gate-all-around device that extends that horizontal channel concept of a FinFET by having a wraparound gate on all four sides (Lateral Nanowire), as depicted in Figure 4.2. Next, their channels will be built vertically so that they are standing on end, leading ultimately to monolithic 3D (vertical) integration. Devices with channels made using alternative materials, such as silicon germanium and III-V compound semiconductor materials (e.g., indium gallium arsenide, or InGaAs), are also being developed for potential future device options targeted at the 7 nm technology node. However, engineers must find a way to reliably build these alternative transistor channels on industry-standard silicon wafers.2
CMOS technology scaling has been a primary driver for the higher levels of device integration. As device channel lengths have shortened, the supply voltages have been reduced to control device leakage, and so at a circuit level, the power per switching event has decreased, increasing the noise sensitivity of the circuit. Consequently, circuits have become more vulnerable to ion-induced SEEs, with some circuits being susceptible to direct, ionization-driven SEEs from abundantly plentiful protons. For advanced technologies, space systems may be subject to SEE threats from protons through direct ionization, as well as indirect ionization. Assessing these potential proton effects will drive more demand for low- to medium-energy (direct) and high-energy (indirect) proton testing. Increased integration on a single chip also increases the number of test states that should be interrogated for SEE, driving a need for more proton testing time and more proton facilities for testing electronics.
Technologists have developed 3D integration approaches as the future paths to achieve higher performance, increased functionality, lower power consumption, and a smaller footprint. Monolithic 3D integrated circuits
2 R. Stevenson, “Changing the Transistor Channel,” IEEE Spectrum, 2013; International Technology Roadmap for Semiconductors, International Technology Roadmap for Semiconductors 2.0, 2015, http://www.itrs2.net/itrs-reports.html.
have explicitly been predicted in the 2015 ITRS report as the integration approach moving forward, identified as “3D power scaling.” Electronic systems built with 3D circuits that are intended for space applications will need to undergo comprehensive SEE testing. 3D integration technologies include 3D packaging schemes, like 3D wafer-level packaging and 3D stacked packaged die, and 3D integrated circuit (IC) schemes, like 3D stacked ICs and monolithic 3D ICs. In comparison to the drive for increased integration and drive to low power on a single chip driving needs for increased proton testing time, increased integration within packages will drive the need for higher-energy heavy-ion testing to penetrate through to the sensitive device areas.3 This 3D integration places demand on heavy-ion facilities with accelerators that can produce a variety of ions with high energies, sufficiently high enough to penetrate these intensely packed, vertically stacked modules to reach the sensitive areas of the devices.
Non-conventional device designs such as FinFET and FDSOI are already in commercial production, and new alternative materials to replace silicon are being investigated to extend the capabilities of CMOS technology into the future. The 2015 ITRS report predicts that traditional silicon channels may ultimately be replaced by channels made with alternative materials, such as silicon germanium, germanium, and III-V compound semiconductors like InGaAs. These alternative materials have been found to transport charge faster, and so using them will allow the production of transistors that can deliver the same amount of current as their silicon counterparts but at a lower voltage, saving power. Understanding the radiation response characteristics of devices and circuits made with these new materials will be critical to being ready to insert them in next-generation space systems. As new materials and new device designs begin to prove their potential for use in space electronics, they will continue to be integrated with standard silicon CMOS in slow evolution, replacing CMOS with new technologies, and as such, traditional silicon CMOS testing will continue to dominate radiation testing volumes at least until 2030. Such fast technologies will require new test hardware capable of handling their speed. In spacecraft avionics and instrumentation, they will require processors, memories, and other supporting electronics (likely commercial) capable of handling such rapid data streams. All of these advanced electronics will require SEE testing.
With the need for space electronics with lower power consumption and lighter weight, there is constant pressure on research and development to innovate with new materials and device designs for space electronics. Many of these requirements dovetail with the evolution of commercial electronics, and space electronics can take advantage of the tremendous amount of research funded by companies and government agencies. However, the additional requirement for these electronics to function for long periods of time in harsh space radiation environments imposes additional constraints on the materials and devices and additional testing requirements to ensure adequate performance in radiation environments. For example, while the responses of digital electronics in radiation environments have been thoroughly investigated and their radiation effects and hardening approaches are generally well understood, a significant challenge remains in achieving that same level of clarity in understanding and being able to predict (and mitigate) the radiation responses of analog (e.g., radio frequency, mixed signal) electronics.
Historically, space system electronics have benefited greatly from commercial technology development. The commercial industry, driven largely by its economic interests, has been the leader in driving technologies into new frontiers of enhanced capabilities. Space systems manufacturers have then taken commercial electronics, assessed their radiation effects, and devised approaches to mitigate these effects, to ultimately use them in space applications. However, now with advances in technologies driving circuit nodal switching powers to new low levels, making them vulnerable to ground-level atomic elements, a nexus has formed in that now commercial manufacturers need to be knowledgeable and concerned about terrestrial effects on their commercial systems. This in effect broadens the community of interest in radiation effects and ultimately drives more future demand for radiation test time.4
3 3DICs may incorporate dense materials, such as copper heat sinks. A heavy-ion beam (e.g., gold, or at least xenon) with the ability to penetrate at least 3 mm of silicon would likely be adequate. This equates to an energy per nucleon ~150–200 MeV/nucleon, or about 35–50 GeV for gold.
4 See B. Schroeder, E. Pinheiro, and W.-D. Weber, “DRAM Errors in the Wild: A Large-Scale Field Study” in SIGMETRICS/Performance’09, https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/35162.pdf, and P. McLellan, “Soft Error Rates in Satellites and Cars,” Cadence Breakfast Bytes Blogs, May 9, 2017, https://community.cadence.com/cadence_blogs_8/b/breakfast-bytes/posts/ser-in-auto-and-space; both accessed on December 19, 2017.
Given the commercial electronics drive for innovation in new materials and devices at a rapid pace, with funding from a wide variety of sources, NASA and the Department of Energy (DOE) will have to identify and articulate specific areas where the requirements for space electronics diverge from those for commercial electronics—for example, performance in radiation environments and performance lifetimes in decades. Additionally, NASA and DOE will have to continue to follow innovations in commercial electronics, and as new technologies show promise for reduced cost, size, weight, and power (CSWaP), will be wise to support efforts to investigate failure mechanisms of the new technologies with the aim of identifying the need for new testing technologies. Commercial microelectronics that reduce power and weight are important considerations for spacecraft systems design and performance. Forecasting trends to 2030 are considered critical to determining requirements for a radiation hardness testing infrastructure. Additionally, both the commercial and space sectors will have to consider radiation effects (especially SEE) for new technologies. However, the space and accelerator environments are qualitatively and quantitatively much more severe than the terrestrial environment. NASA and DOE, as primary users for these environments, will need to make decisions as to what technologies they will need in what is likely to be an increasingly fragmented electronics market.
Chapter 3 found that the current proton and heavy-ion SEE testing infrastructure is showing signs of fragility and strain, and the list of facilities willing to participate in such testing, even as a secondary mission, is in a continual state of flux. Moving forward, the pressure on the testing infrastructure will be exacerbated by a number of issues, some of which have been discussed in the sections above.
- The growing commercial space industry and numbers and variety of spacecraft that increase testing requests.
- The pressure for spacecraft designers to incorporate commercial electronic parts for their performance and power advantages.
- The increasing integration and complexity of microelectronic and optoelectronic circuits and front-end electronics for sensors that increase the amount of testing configurations and time required to evaluate space radiation vulnerabilities (this need places a priority on high-energy proton and broad-spectrum heavy-ion accelerators).
- The increasing integration of functions performed by the spacecraft, including sensors generating large amounts of data, data compression, synthesis, storage, and communications over extended durations in hostile environments.
- The potential commercial testing needs for assuring the reliable and safe operation of advanced electronic systems exposed to terrestrial sources of radiation.
- The lack of dedicated accelerators for space radiation testing, leaving electronics testing reliant on unrelated external factors.
Finding: The combination of the three factors, (1) fragility and overloading of current beam-line facilities for space radiation testing, (2) increasingly complex commercial microelectronic parts and systems that will require longer test times to characterize, and (3) increasing requirement for accelerator testing by commercial interests, project a growing shortage of available testing facilities to support future space missions among space agencies and industry.
The committee was also briefed on radiation sources other than proton and heavy ion, such as X-ray and laser sources, to simulate space radiation–induced SEE failures. These techniques, including X-ray testing and laser testing, are useful as diagnostic methods to explore failure mechanisms. The committee believes that methods other than proton or heavy-ion irradiation only explore a single dimension of the radiation environment and are not suitable to replace proton and heavy-ion irradiation techniques. However, these alternate techniques are useful to investigate
specific failure locations, and to allow for exercising test equipment for test-related problems—for example, system software—before taking parts to be tested in expensive proton or heavy-ion beams.
Increasing emphasis is also being given to modeling and simulation (M&S), especially Monte Carlo estimation of SEE failure rates, to reduce the number of heavy-ion tests required to assure device reliability and lifetime. However, although improved M&S may reduce the need for electronic parts testing in the long term, in the short term, validating and verifying new models and simulations may in fact increase the need for radiation testing.