Cambridge Team Builds $10 Chip-Sized Spectrometer Aimed at Wearable Health Sensing

Researchers at the University of Cambridge and GlitterinTech, a start-up founded by the same research group, have developed a miniature optical spectrometer that they say combines laboratory-grade accuracy with a size of just centimeters across and a cost of around $10.

Spectrometers analyze light to reveal information about the substances it has interacted with, making them essential in medicine, manufacturing, chemical testing and environmental research. But shrinking them has usually meant trade-offs. Compact versions often have limited measurement ranges, lower resolution or reduced accuracy, which makes them better at identifying substances than measuring them precisely.

The Cambridge team's design, called a convolutional spectrometer, takes a different route. Traditional instruments typically split incoming light into its component wavelengths or rely on complex calculations to reconstruct a spectrum. The new device instead performs a mathematical operation called convolution directly on the light, drawing on a principle known as the convolution theorem.

To do this, the researchers arranged optical components with repeating spectral responses in sequence. These can include unbalanced Mach-Zehnder interferometers or micro-ring resonators. By adjusting them in proportion to one another, the system shifts its response to different wavelengths in a controlled, linear way. Fast Fourier transforms are then used to recover the light spectrum.

"The key insight was to go back to the mathematics and ask whether there was a fundamentally cleaner way to retrieve spectra," said Dr. Chunhui Yao, a lead author of the study. Yao said the approach gives high precision, strong noise tolerance and very low computational overhead.

The team built the device on a silicon nitride photonic platform and integrated it with onboard electronics. It operates across near-infrared wavelengths from 1200 to 1700 nanometers and can collect and process measurements in less than a second. Because the design relies on repeating optical patterns, the researchers say its spectral bandwidth could theoretically be extended almost without limit, and adding more components in sequence could increase resolution exponentially.

In tests, the spectrometer achieved a 100% classification success rate when identifying plastics, pharmaceuticals, coffee, flour and tea, according to the reported results. It also measured concentrations in water-based and organic solutions with accuracies of approximately 0.01%, exceeding the performance of commercial benchtop spectrometers in those experiments.

The health demonstrations drew particular attention. Under conditions designed to reflect real physiological environments, the device accurately measured several human biomarkers: skin moisture, blood alcohol, blood lactate and blood glucose. The researchers also showed noninvasive glucose tracking over an extended period in one participant. That is a single-person demonstration, and the researchers note that further testing would be needed before the technology could become a practical medical product.

The sensor also kept working between -20°C and 80°C, an unusually wide range for a miniature spectrometer. Prof. Qixiang Cheng, who led the project, said the work stands out for its technical readiness as well as its lab performance, describing a packaged system that operates reliably across temperature extremes and real-world conditions.

The design may also ease the computing burden. Many miniature spectrometers depend on complicated calibration and demanding algorithms to reconstruct spectra, methods that can become unreliable when measurements are noisy. Because the convolutional approach follows a linear mathematical relationship, the researchers say it can recover spectra rapidly and consistently with relatively little processing power. Prof. Richard Penty, who contributed to the photonic system design and integration, called the architecture "simple, scalable, and manufacturable."

The researchers envision inexpensive spectrometers installed throughout factories to check material composition and product quality, or used to monitor food freshness and environmental conditions. In healthcare, they suggest future wearables could measure hydration, detect signs of alcohol intoxication, track fitness-related biomarkers or continuously monitor glucose for people with diabetes. "Our vision is to make spectrometry as ubiquitous as temperature or motion sensing," Cheng said.

Several caveats apply. The accuracy figures come from the team's own reported tests on a limited set of materials, and the source material does not detail sample sizes. The glucose result involved one participant. Claims about unlimited bandwidth are theoretical, and the healthcare uses described are potential future applications rather than approved products.

Still, the work suggests that rethinking the underlying mathematics, not only shrinking the hardware, could help bring precise optical sensing out of the laboratory and into everyday devices.