In a field long shaped by the trade-off between the size of a measuring instrument and the accuracy of what it reads, a team at the University of Cambridge has offered a quiet but consequential answer: a spectrometer that fits on a chip and costs about the price of a lunch. Working with GlitterinTech, the start-up that grew out of their own research group, the engineers replaced the old habit of reconstructing light after the fact with a design that performs the mathematics in the light itself. The promise is laboratory-grade precision moving out of the laboratory and into everyday hands, from
Cambridge's $10 spectrometer chip identifies food, plastics and drugs, eyes wearable health monitoring
In a field long shaped by the trade-off between the size of a measuring instrument and the accuracy of what it reads, a team at the University of Cambridge has offered a quiet but consequential answer: a spectrometer that fits on a chip and costs about the price of a lunch. Working with GlitterinTech, the start-up that grew out of their own research group, the engineers replaced the old habit of reconstructing light after the fact with a design that performs the mathematics in the light itself. The promise is laboratory-grade precision moving out of the laboratory and into everyday hands, from food and plastics to the body itself, though whether that promise survives beyond a single participant remains an open question.
A spectrometer that costs about $10 and fits on a single chip has emerged from the University of Cambridge, and its makers say it matches the precision of instruments that normally fill laboratory benches. The work came from the university's Department of Engineering and GlitterinTech, a start-up founded by the same research group.
The central idea is what the team calls a convolutional spectrometer. Most instruments either spread light out physically or rely on software to rebuild a spectrum afterward. This design carries out the convolution on the light itself, using cascaded optical components whose tuning shifts the response in a predictable, linear way, so the spectrum can be recovered quickly with fast Fourier transforms. Dr Chunhui Yao, a lead author, argues that this sidesteps the usual shrinking penalty, where small instruments lose bandwidth, resolution or accuracy and need fragile calibration.
Built on a silicon nitride platform, the chip covers near-infrared wavelengths and processes readings in under a second. It remained stable from minus 20°C to 80°C, which the team sees as a sign it can work in wearable, industrial and outdoor settings. In tests it classified plastics, pharmaceuticals, coffee, flour and tea without error and measured solution concentrations to around 0.01%, which the researchers say beat commercial benchtop models.
The health results drew the most excitement. The chip sensed skin moisture, blood alcohol, lactate and glucose non-invasively under realistic conditions. Glucose, however, was followed over extended periods in only one person. Professor Qixiang Cheng praised Dr Yao for turning a mathematical concept into a robust, packaged product, but the next chapter belongs to validation: whether the wearable promise holds across larger groups of people.
A spectrometer that costs about $10 and fits on a chip has come out of the University of Cambridge, and its makers say it matches the precision of laboratory instruments. The device was built by researchers in the university's Department of Engineering together with GlitterinTech, a start-up founded by the same research group.
The design rests on what the team calls a convolutional spectrometer. Conventional instruments either spread light out physically or rely on algorithms to reconstruct a spectrum after the fact. This one performs a convolution operation on the incoming light itself. It uses cascaded optical components with periodic spectral responses, such as unbalanced Mach-Zehnder interferometers or micro-ring resonators. Tuning those components proportionally shifts the system's spectral response in a linear way, which lets the spectrum be recovered with fast Fourier transforms.
Dr Chunhui Yao, a lead author of the study, said that applying the convolution theorem directly in the optical domain avoids many of the limitations that have held miniaturised spectrometers back. In his account, the result is high precision, strong noise tolerance and very low computational overhead in a compact, low-cost system. Smaller spectrometers have typically given up bandwidth, resolution or accuracy as they shrink, and often need complex calibration and noise-sensitive reconstruction software.
The chip is built on a silicon nitride photonic integration platform and packaged with on-board electronics. It covers a near-infrared range of 1,200 to 1,700 nanometres, with sampling and processing in under a second. Because the design is periodic, the bandwidth can be widened without changing the hardware, and resolution can be scaled exponentially by cascading more components. The team also reported that it stayed stable from minus 20°C to 80°C, which they say supports use in wearable, industrial and outdoor settings.
In tests on materials and food, the device classified plastics, pharmaceuticals, coffee, flour and tea with a 100% success rate. It measured concentrations in aqueous and organic solutions to around 0.01% accuracy, outperforming commercial benchtop spectrometers according to the researchers. They point to materials analysis, food testing and industrial quality control as natural uses.
The health results are what the team calls particularly exciting. The chip sensed human biomarkers non-invasively under realistic physiological conditions, measuring skin moisture, blood alcohol, blood lactate and blood glucose with high accuracy. Glucose was tracked over extended periods, but in a single participant.
Project leader Professor Qixiang Cheng said the work stands out not only for lab performance but for technical readiness. He credited Dr Yao with turning a mathematical concept into a fully packaged, robust system that works reliably across temperature extremes and real-world conditions, which he said opens the door to practical deployment.
What to watch next is validation. The glucose demonstration involved one person, so tests across larger groups will show whether the wearable promise holds.
Quer a matéria completa?
Leia o original em The Times of India ↗
Technology