At the University of Cambridge, a team of physicists has accomplished something that seemed unlikely just years ago: they've crammed laboratory-grade precision into a device small enough to slip onto your wrist, all for around $10.
The breakthrough is a new type of optical spectrometer developed by researchers at Cambridge and GlitterinTech, a startup founded by the same research group. For decades, miniaturizing spectrometers meant accepting painful compromises—smaller devices meant sacrificing accuracy, resolution, or the range of wavelengths they could measure. This new convolutional spectrometer, detailed in Nature Photonics, eliminates those trade-offs by rethinking how light itself is measured.
Rather than dispersing light or using algorithms to reconstruct spectra, the device physically performs a mathematical operation called convolution directly on incoming light using a simple cascade of optical components with periodic responses. The elegance lies in the mathematics: by using the convolution theorem in the optical domain, the team solved a problem that had constrained miniaturized spectrometers for generations. "By using the convolution theorem directly in the optical domain, we avoid many of the limitations that have held miniaturized spectrometers back," said Dr. Chunhui Yao, a lead author of the study. "This gives us high precision, strong noise tolerance, and very low computational overhead, all in a compact and low-cost system."
Built on a silicon nitride photonic integration platform, the spectrometer operates across an ultra-wide near-infrared range of 1,200–1,700 nanometers with sub-second sampling times. Its periodic design allows bandwidth to expand almost indefinitely without changing the hardware, while resolution can be scaled by simply cascading additional components.
The practical demonstrations reveal why this matters beyond the lab. In materials and food analysis, the device classified plastics, pharmaceuticals, coffee, flour and tea with perfect accuracy. When measuring chemical concentrations in solutions, it achieved precision around 0.01%—outperforming commercial benchtop spectrometers that cost thousands of dollars. Perhaps most striking were the biomedical applications: the team demonstrated noninvasive sensing of human biomarkers including skin moisture, blood alcohol, blood lactate, and blood glucose. Glucose tracking over extended periods showed high accuracy, suggesting a future where wearable devices could continuously monitor metabolic health.
The system proved remarkably robust, remaining stable across temperatures from minus 20 degrees Celsius to 80 degrees—a level of durability rarely seen in miniaturized spectrometers and essential for devices meant to survive in pockets, on skin, or outdoors. Professor Qixiang Cheng, who led the project, emphasized that the breakthrough wasn't just lab performance. "What makes this work stand out is not just performance in the lab, but technical readiness," he said. "That combination is what opens the door to practical deployment."
What sets this approach apart is also its simplicity. High-performance miniaturized spectrometers typically require complex calibration and computationally intensive algorithms that can be sensitive to noise. The convolutional spectrometer's linear nature enables fast, stable spectrum recovery with minimal processing power—crucial for wearable devices running on battery power.
These characteristics point to a near future where real-time chemical sensing becomes practical in everyday devices. Industrial quality control, environmental monitoring, and personalized health tracking all lie within reach. For the first time, the tools of laboratory science are becoming accessible at the scale and cost of consumer electronics.