# How Chips Use On-Chip Terahertz Sensors for Security Scanning

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 15 Jul 2026, 7:33.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0112.mp3

This episode of The Hardware Podcast dives into on-chip terahertz sensors — a technology that could replace bulky airport body scanners with a tiny silicon die. Lucas explains how researchers at the University of Texas and imec have integrated a 0.3 THz detector array into a standard CMOS process, achieving 1-millimeter resolution at a cost under $10 per chip. Luna asks about power consumption, privacy implications, and whether this will actually make it into phones. We also touch on the commercial timeline: a startup called TeraSense expects to ship evaluation kits by Q4 2026. If you've ever wondered how a chip can 'see' through clothes without X-rays, this is the episode. #TerahertzSensors #OnChipSensors #SecurityScanning #CMOS #Imec #UniversityOfTexas #TeraSense #MillimeterWave #ChipDesign #HardwareEngineering #NonIonizingRadiation #PrivacyTech #Semiconductor #Technology #FexingoBusiness #BusinessPodcast #HardwarePodcast #ElectronicsEngineering Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: Luna, imagine you're at airport security. You take off your shoes, your belt, your laptop — and you walk through a body scanner that uses millimeter-wave radiation. That scanner is the size of a phone booth and costs around a hundred and fifty thousand dollars.
Luna: Right, and it takes up a huge footprint. I've always wondered why we can't shrink that into something smaller.
Lucas: Well, that's exactly what a handful of research groups and startups have been working on. And the answer is: you can shrink it onto a single chip. I'm talking about on-chip terahertz sensors.
Luna: Terahertz — that's the frequency range between microwave and infrared, right? The one that people have been calling the 'next frontier' for like a decade?
Lucas: Exactly that. Terahertz waves, usually defined as 0.1 to 10 terahertz, have this neat property: they pass through clothing, paper, plastic, and drywall — but they reflect off metal and water. So they can 'see' concealed objects without using ionizing radiation like X-rays.
Luna: So the holy grail is a chip that can do that scanning in your pocket. How close are we?
Lucas: Closer than you'd think. The key breakthrough came from integrating terahertz detectors into standard CMOS — complementary metal oxide semiconductor — the same process used to make computer processors and phone chips. Historically, terahertz required exotic materials like indium phosphide or gallium arsenide. But around 2023, researchers at the University of Texas at Dallas and imec in Belgium showed you could build a 0.3 terahertz detector array using just standard 65-nanometer CMOS.
Luna: And what kind of performance did they get?
Lucas: The UT Dallas chip had a 32-by-32 pixel array — so 1,024 detectors — on a die about 3 millimeters by 3 millimeters. It could resolve objects down to about 1 millimeter at a distance of 10 centimeters. At the time, the noise-equivalent power was around 100 picowatts per square root hertz, which is competitive with much bulkier systems.
Luna: That's impressive for a tiny die. But 10 centimeters — that's not exactly through a full-body scanner. More like a handheld wand?
Lucas: Right. For a full-body scanner you'd need to stand about 1 to 2 meters away. That requires either a much larger array or a focusing lens. Some groups are using a silicon lens bonded to the chip — that increases the effective aperture. But the chip itself is still the core.
Luna: So the lens adds cost. But even with a lens, the total system might be a few hundred dollars instead of a hundred and fifty thousand?
Lucas: Exactly. A startup called TeraSense — spun out of imec — claims their evaluation kits, which include a chip and a lens, will be under $500 in volume. They're aiming for Q4 2026 to ship those kits. That's just a few months from now.
Luna: Wait, today is July 15, 2026. So that's... this quarter?
Lucas: Yes. If they hit that timeline, we could see the first commercial on-chip terahertz sensors before the end of the year.
Luna: What's the power consumption like? Because if this goes into a phone or a portable device, it can't be a power hog.
Lucas: The UT Dallas chip consumed about 50 milliwatts for the detector array. That's roughly what a Bluetooth Low Energy module draws during transmission. So it's definitely phone-feasible.
Luna: And the source? You need a terahertz source too, right? Not just a detector. Is that on-chip as well?
Lucas: That's the harder part. Generating terahertz waves efficiently is tough. Most CMOS terahertz sources use something called a 'harmonic oscillator' — they generate a lower frequency, say 100 gigahertz, and then use the nonlinearity of the transistor to produce the third harmonic at 300 gigahertz. The efficiency is low — typically under 1%. But even a few milliwatts of output power is enough for near-field sensing.
Luna: So the source might be a separate chip or a module. Still, having the detector on CMOS is the big leap.
Lucas: Right. And there's another interesting application beyond security. On-chip terahertz sensors can also be used for industrial quality control — checking the thickness of paint coatings, detecting moisture in paper, even inspecting pharmaceutical tablets for defects. The chip can be embedded in a production line.
Luna: That's a huge market. But let's go back to the privacy angle. A terahertz chip in a phone could effectively see through clothing — that raises obvious concerns about misuse.
Lucas: It's a real issue. The physics limits resolution at a distance — at 1 meter, the diffraction-limited spot size for 0.3 terahertz is about 1 centimeter. So you're not getting a detailed image of a person, just blobs of reflected energy. But still, the perception is problematic. Some researchers are already working on 'privacy filters' — pixelating the output or using AI to only detect metallic and liquid shapes, not human anatomy.
Luna: Interesting. So the industry is ahead of the backlash this time. Or at least trying to be.
Lucas: I think they learned from the Google Glass experience. You can't just drop a powerful sensor into the wild without thinking about the social implications.
Luna: Speaking of learning — this kind of deep dive into chip engineering is exactly why people listen to this show. And it's also why a small group of listeners chip in monthly through buy me a coffee dot com slash fexingo. That support literally funds the research that goes into episodes like this.
Lucas: Yeah, it's a model we really believe in. No ads, no sponsors — just listeners who find value in explaining how the hardware actually works. If that's you, we appreciate it more than we can say.
Luna: Right. And it keeps us free for everyone. Anyway — back to terahertz. So the timeline: evaluation kits by end of 2026. When would we see this in a consumer phone?
Lucas: I'd guess 2028 or 2029 at the earliest. The chip itself is ready, but integrating a terahertz source and lens into a phone form factor takes industrial design cycles. Plus regulatory approval — the FCC has to certify any new radiator. That takes time.
Luna: But the direction is clear. In a few years, the security scanner at the airport could be a chip in a kiosk, not a phone booth. And your phone might be able to detect studs in walls or find gas leaks.
Lucas: Exactly. The same chip that sees through clothes can also see through drywall — but for construction, not surveillance. I think the industrial applications will actually arrive first, because the regulatory path is simpler.
Luna: Makes sense. So watch for TeraSense's kits in Q4, and then expect to see on-chip terahertz in production lines before it hits your pocket.
Lucas: That's my bet. It's one of those rare technologies where the science has been solid for years, and the engineering is finally catching up.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.