Subthreshold voltage references can help reduce power consumption in space electronics, where energy is limited and stable reference signals are essential. But operating CMOS transistors below threshold also makes circuit behavior sensitive to threshold-voltage changes—one of several effects that radiation can cause. Testing two subthreshold reference designs under gamma rays, X-rays, protons and heavy ions illustrates both the potential and the engineering trade-offs of this approach.
Why reference circuits matter in space systems
Voltage references support analog and mixed-signal functions in systems such as data converters, amplifiers, memories and power-management circuits. If a reference shifts, downstream circuits may no longer meet their intended operating specifications. In a spacecraft, where radiation exposure and limited power are both design constraints, the reference must maintain useful performance while consuming little energy.
Conventional bandgap references use bipolar-transistor temperature behavior to help compensate for changes in temperature. Low-power CMOS alternatives can instead combine currents with proportional-to-absolute-temperature (PTAT) and complementary-to-absolute-temperature (CTAT) characteristics. The designs discussed here add curvature correction to improve temperature behavior, while operating their MOS devices in the subthreshold region.
Subthreshold voltage references and radiation sensitivity
In subthreshold operation, transistor current depends exponentially on the difference between gate-source voltage and threshold voltage. As a result, radiation-induced threshold shifts can have a pronounced effect on drain current. Radiation may also change carrier mobility, increase leakage, or affect device characteristics through damage to the silicon lattice.
Three broad effects are relevant to circuit evaluation:
- Total ionizing dose (TID): Radiation creates charge in insulating materials. Trapped charge can alter transistor characteristics, including threshold voltage and leakage current.
- Single-event effects: A particle strike can deposit charge at a sensitive circuit node, producing a temporary transient or, in some cases, a failure.
- Displacement damage: Energetic particles can displace atoms in the silicon lattice, potentially affecting carrier properties.
The relative importance of these effects depends on the circuit, its operating conditions and the radiation exposure. In particular, a threshold shift that has a different relationship to current in strong inversion can have a substantial impact in subthreshold circuits.
Design choices for radiation tolerance
The two custom reference circuits were fabricated in a 0.18 µm standard CMOS process. Their designs used radiation-hardening-by-design (RHBD) measures, including enclosed-layout geometries for key NMOS transistors, guard rings and layout choices intended to limit leakage associated with trapped charge in isolation oxides. The circuits used CMOS transistors and polysilicon resistors, without external compensation capacitors.
Both topologies generate and combine PTAT and CTAT-related currents, with resistor and transistor behavior contributing to higher-order temperature compensation. The first reference operated from a 0.75 V supply and consumed 4 µW; its reported temperature coefficient was 15 ppm/°C across a 190°C range. The second operated from 0.7 V, consumed 2.7 µW and achieved 12.9 ppm/°C over a 180°C range. These temperature results describe the characterized designs; radiation exposure can change electrical performance.
What the irradiation tests showed
Heavy-ion single-event transients
Testing with silicon, krypton and xenon ions evaluated analog single-event transients (ASETs). The second reference showed no measured sensitivity to silicon ions in the reported test and, overall, had lower measured sensitivity than the first design. It also produced shorter and smaller transients in the reported comparisons with krypton and xenon.
These results are specific to the tested circuits and conditions. The study notes that subthreshold designs can take longer to recover from a transient because of lower current drive and slower feedback. Their low bandwidth can also filter some transient effects before they appear at the output. The balance depends on circuit architecture and operating conditions.
Total-dose and proton exposure
Gamma-ray and X-ray tests showed changes in output voltage and current consumption, with variation between individual chips. In several tests, changes were more pronounced at lower accumulated doses and then became more stable at higher doses. Room-temperature annealing produced partial recovery in some measurements. The researchers attributed the observed behavior to competing effects from oxide-trapped and interface-trapped charge, alongside device and process variation.
The circuits were also tested with protons, including a sequence combining proton exposure and X-rays. In the reported 170.5 MeV proton test up to 1.4 Mrad(Si), the first reference showed a smaller output-voltage change than the second. In the combined 2 Mrad(Si) proton and subsequent X-ray exposure, the first topology also showed greater output-voltage resilience. Results differed across irradiation conditions, so these comparisons should not be treated as a universal ranking of the designs.
Practical implications for circuit selection
For engineers evaluating references for radiation environments, these findings point to several considerations:
- Assess the full radiation profile. TID, displacement damage and single-event transients are distinct effects and call for different tests.
- Track more than output voltage. Current consumption can change as transistor characteristics and leakage change.
- Consider layout alongside circuit topology. Enclosed NMOS geometries and other RHBD measures can address specific leakage mechanisms.
- Test across devices and conditions. Chip-to-chip variation and irradiation conditions influence measured results.
Subthreshold voltage references offer a route to low-power operation, but their radiation response must be evaluated in the context of the intended mission and system requirements. The tested designs maintained operation through a range of irradiation experiments, while showing measurable changes and differences between topologies. For design teams, the practical lesson is to combine low-power circuit techniques with targeted RHBD choices and radiation testing that reflects the expected environment.