RADECS 2026 · Paper · accepted
Total Ionizing Dose Influence on Trapping Activities in Nanoscale CMOS
Yurii Chubenko, Semih Ramazanoglu, Alicja Michałowska-Forsyth · Institute of Electronics, TU Graz
Radiation and its Effects on Components and Systems · Prague
Most studies of semiconductor reliability in a ionising radiation environment measure a transistor before and after exposure, which skips the part we find most interesting — what the device does while it is actually being hit. So we set out to watch a single minimum-size 40 nm NMOS device with the X-rays still on: kept biased and monitored continuously, at high sampling rate, through the whole radiation stress and the 100 degree heat annealing that follows.
What radiation does inside
A transistor this small has not so much to it. It has a conducting channel a few hundrets atoms wide, with a thin layer of insulator on top. That insulator is where the trouble starts: it holds a handful of atomic-scale flaws (defects / charge traps), and each one can accumulate a charge and hold onto it. In a device this tiny a single trapped charge is enough to slightly change the current through the whole transistor.
Radiation may make more of these traps, and the 3D model above demonstrates that process. As we begin the irradiation the X-rays seed fresh defects in the insulator; each one that catches a charge changes the current up or down by its own small amount. The damage does not just grow though as charges slowly leak back out and traps heal on their own
Why you have to catch it live
The usual way to measure radiation damage in a seconductor device is to pause the exposure now and then and take a careful reading, then repeat at a few dose levels along the way. It is precise, but it only gives you snapshots of the damage — and anything that flickers or settles between those snapshots is gone before you can see it.
So we never pause the measurements during the experiment run. The device stays powered and we read its current without a break. The two lanes below make the difference plain: only the continuous record can accuratly capture the sharp events and the growing restlessness in the transistor.
What we saw
The current does not rise/drop smoothly as dose builds, it rather moves in sharp steps, caused by single defects catching or releasing a charge
Those steps make out a low-frequency noise that appears when the irradiation begins, and returns again when the device is heated for annealing, yet stays quiet at room temperature with the irradiation stopped. That recorded noise is the result the paper is built on.
Experiment at a glance
- 40 nm bulk CMOS
- Technology
- min-size n-MOSFET
- Devices
- ~42 h, continuous
- Under the beam
- 100 deg C, ~1 day
- Then annealed
- almost gap-free, µs-scale
- Sampling
As far as we can tell this effect has been underreported, and it matters wherever nanoscale transistors have to stay quiet under heavy radiation — the giant detectors such as CERN's Large Hadron Collider, but also medical imaging and electronics bound for space.