# How Diamond NV Centers Are Shrinking Quantum Magnetometers onto a Chip

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 26 Jul 2026, 9:23.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0132.mp3

On-chip diamond nitrogen-vacancy centers are bringing quantum-grade magnetic field sensitivity to millimeter-scale sensors. Lucas and Luna explore a recent MIT breakthrough that integrates NV centers directly into a CMOS-compatible chip, achieving single-neuron-level magnetic detection. They discuss how nanofabrication techniques overcome diamond's challenges, applications from brain imaging to navigation, and the road to commercial devices. A concrete look at how quantum sensing is going mainstream — one chip at a time. #HardwarePodcast #Technology #Chips #Devices #ElectronicsEngineering #DiamondNV #QuantumSensors #Magnetometers #OnChip #Semiconductors #NVCenter #MagneticFieldSensing #QuantumTechnology #MIT #Nanofabrication #SensorTechnology #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: Alright Luna, I want to talk about a sensor that can detect the magnetic field of a single firing neuron — and it's built on a diamond chip smaller than your fingernail.
Luna: Diamond chips? You have my attention. Is this the nitrogen-vacancy center thing I keep seeing in physics journals?
Lucas: Exactly. Nitrogen-vacancy centers — NV centers for short — are atomic-scale defects in diamond's crystal lattice. A nitrogen atom sits next to a missing carbon atom, and that defect behaves like a tiny quantum magnet. When you shine green laser light on it, it fluoresces red, and the brightness depends on the local magnetic field.
Luna: So you can literally read out magnetic fields by measuring how much red light comes out. That's elegant. But why haven't we seen these in smartphones yet?
Lucas: Traditionally, NV sensors use bulk diamond crystals — expensive, hard to integrate with electronics. But a team at MIT just published a paper in Nature Electronics detailing a method to grow thin diamond films directly on a silicon wafer, then pattern them with standard CMOS lithography.
Luna: Growing diamond on silicon? That sounds like a materials science nightmare. Diamond's lattice constant is way different from silicon's.
Lucas: It is. The trick is to use an intermediate layer of iridium. They deposit iridium on the silicon, then grow diamond via chemical vapor deposition. The iridium acts as a template — diamond crystals form with the same orientation, and after etching away the iridium, you get a smooth, single-crystal diamond film just a few hundred nanometers thick.
Luna: And then they introduce the NV centers? How do you get the nitrogen in there?
Lucas: During the diamond growth, they add a trace of nitrogen gas. Then they bombard the film with high-energy electrons to knock out carbon atoms, creating the vacancies. A quick anneal at 800 degrees Celsius lets the vacancies migrate and pair with nitrogen atoms. You end up with billions of NV centers per square centimeter.
Luna: That's a lot of atomic defects. But how do you read the signal from a specific spot? You need to collect the fluorescence efficiently.
Lucas: Right. They integrated a solid-immersion lens — basically a tiny dome etched into the diamond itself — that focuses the emitted light into an on-chip photodiode. The whole sensor, including the laser driver and readout electronics, fits on a 2-millimeter-square chip. In their tests, it achieved a sensitivity of about one picotesla per square-root hertz.
Luna: One picotesla per root hertz — that's roughly a thousand times better than commercial Hall sensors. But magnetic fields from a single neuron are even smaller, right?
Lucas: Neuronal magnetic fields are around a few picotesla at the scalp surface. So with this sensitivity, you could theoretically detect the firing of a single neuron if the sensor is close enough. They demonstrated detecting the magnetic field from a small permanent magnet at a distance of one millimeter — that's about the same magnitude.
Luna: Okay, so the sensitivity is there. But what about real-world noise? A chip sitting in a lab is one thing — in a phone, you've got power supplies, Wi-Fi, and other interference.
Lucas: Great point. The MIT chip includes on-chip magnetic field modulation. They apply a small oscillating magnetic field to the NV centers, which shifts the resonance frequency. Then they lock in on that frequency — it's a classic lock-in amplifier technique, but done entirely on-chip. That rejects low-frequency noise from the environment.
Luna: So the chip is essentially doing its own noise cancellation. That's clever. What are the killer apps?
Lucas: One obvious one is magnetoencephalography — mapping brain activity. Current MEG machines are room-sized, require liquid helium, and cost millions. A chip-scale sensor could enable a wearable helmet that tracks neural activity in real time. Beyond that, there's non-destructive testing: detecting cracks in metal structures, monitoring battery currents, even navigation in gps denied environments by sensing the Earth's magnetic field.
Luna: Navigation without GPS — that's a big one for defense and autonomous vehicles. But I imagine the dynamic range is limited. How does it handle strong magnetic fields, like the Earth's field?
Lucas: The Earth's field is about 50 microtesla, which is five orders of magnitude larger than the sensor's noise floor. The chip uses a technique called 'pulsed ODMR' — optically detected magnetic resonance. They apply microwave pulses that flip the NV spin states, and the fluorescence response depends on the magnetic field. By varying the microwave frequency and looking at the resonance dip, they can measure fields from DC up to several kilohertz with a range of at least 100 microtesla.
Luna: So it's not just a single-point sensor — they can map the field vector? Because NV centers are sensitive to the projection of the field along the axis of the defect.
Lucas: Exactly. The diamond film has NV centers with four different orientations, each sensitive to a different axis. By combining signals from all four, you can reconstruct the full vector magnetic field. The MIT chip actually has an array of 16 sensors on a single chip, allowing for imaging with sub-millimeter spatial resolution.
Luna: An array of vector magnetometers on a chip — that's basically a solid-state magnetic camera. Could this replace SQUID sensors in scientific instruments?
Lucas: SQUIDs — superconducting quantum interference devices — are still more sensitive, down to femtotesla levels. But they require cryogenic cooling. The NV chip works at room temperature, consumes about 10 milliwatts of power, and can be mass-produced with existing fab tools. That trade-off makes it attractive for portable and industrial applications.
Luna: What's the catch? I always feel like there's a hidden catch when a paper sounds this perfect.
Lucas: The main challenge is the diamond growth itself. Chemical vapor deposition of high-quality diamond on an iridium buffer layer is still a specialized process — not yet a standard foundry service. The yield of NV centers with good spin properties also varies. And the photodiode collection efficiency is around 30 percent, which limits the overall sensitivity. They're working on integrating photonic waveguides to boost that.
Luna: So we're probably a few years away from a commercial product. But the fact that it's cmos compatible at all is huge. The foundries won't have to retool completely.
Lucas: Exactly. The team is now exploring bonding the diamond layer directly onto CMOS circuits — a heterogeneous integration approach. They've already demonstrated operation with standard 180-nanometer CMOS, and they're moving to 65-nanometer nodes for better signal processing.
Luna: If this gets cheap enough, every smartphone could have a quantum sensor. Imagine pointing your phone at a wall and detecting live wiring behind it, or scanning your brain for early signs of epilepsy.
Lucas: Right now, the cost is still prohibitive — each diamond wafer costs thousands of dollars to produce. But as the process matures, scale will bring that down. The same thing happened with MEMS accelerometers: they were military-grade expensive in the 1990s, and now they're in every phone.
Luna: Alright, so we went from diamond defects to brain helmets. I'm sold. If our listeners want to dive deeper, where should they look?
Lucas: The paper is in Nature Electronics, April 2026, by a group led by Professor Dirk Englund at MIT. Search for 'on-chip diamond NV magnetometer' and you'll find it. Also, if today's hardware deep-dive gave you something useful — a new perspective on quantum sensors or the thrill of seeing diamond on silicon — you should know that this show stays ad-free thanks to listener support. You can find us at buy me a coffee dot com slash fexingo. It's a simple way to keep these conversations going without sponsors.
Luna: Yeah, it really does make a difference. Even a small contribution helps us invest in better research and guest experts. So if you enjoyed this, consider it.
Lucas: And speaking of integrating new materials, the next frontier could be adding diamond NV sensors directly onto a processor die — imagine a chip that can both compute and sense magnetic fields at the atomic scale. That would open up real-time magnetic microscopy for debugging circuits.
Luna: Fault localization without destroying the chip — that's a game-changer for failure analysis. I'll be watching that space.
Lucas: Not a bad place to leave it. Thanks for listening, and we'll see you next time on The Hardware Podcast.

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.