A research group at Eindhoven University of Technology (TU/e) has developed a new near-infrared sensor that is easy to manufacture, comparable in size to sensors in smartphones, and suitable for agricultural and industrial process monitoring.

His innovation has been described in Nature Communications magazine, with co-first author Kaylee Hakkel defending her doctoral degree. Paper on 1 14.

The human eye is an amazing sensor. Using three different types of photosensitive cone cells to convert visible light into signals of various colors, the eye provides important information about our surrounding environment.

When our brain puts the signals together, it predicts the meaning of the signals based on our experience. For example, red strawberries are sweet, but green strawberries are not.

Kaylee Hakkel, co-first author and PhD researcher in the Photonics and Semiconductor Nanophysics group, Department of Applied Physics, Eindhoven University of Technology

Size matters

Although the human eye is excellent, it cannot compare with state-of-the-art natural light sensors.

Mantis shrimp eyes have 16 different types of cells that are sensitive to ultraviolet, visible, and near-infrared (NIR) light. Measuring infrared spectra is most interesting for industrial and agricultural applications, but there is a major problem—current near-infrared spectrometers are too large and too expensive.

Kaylee Hakkel, co-first author of the study and PhD researcher in the Photonics and Semiconductor Nanophysics group, Department of Applied Physics, Eindhoven University of Technology

Hakkel and her team solved this problem by creating near-infrared sensors that can be placed on small chips. Like the mantis shrimp's eye, it contains 16 different sensors—but they are all sensitive to near-infrared light.

Achieving miniaturization of sensors while maintaining low cost is a significant challenge. Therefore, we designed a new wafer-level manufacturing process to achieve this goal. Its cost is low because we can produce multiple sensors simultaneously, and it is now ready for real-world practical applications. The sensor chip is so small that it can even be embedded in future smartphones.

Kaylee Hakkel, co-first author of the study and PhD researcher in the Photonics and Semiconductor Nanophysics group, Department of Applied Physics, Eindhoven University of Technology

Solving spectral sensing

Andrea Fiore, Research Leader in the Department of Applied Physics and the Hendrik Casimir Institute at Eindhoven, expressed satisfaction with the efforts of his research team. "For many years, we have been studying this technology. Now we have successfully integrated spectral sensors onto chips, while also solving another key issue—the effective utilization of data."

Typically, when a sensor quantifies light, the resulting signal is used to reconstruct the material's spectrum or optical fingerprint. Sensing algorithms are then applied to examine the data.

In this new method, scientists demonstrated that the spectral reconstruction step is unnecessary. In other words, the signal generated by the sensor can be directly transmitted to analysis algorithms. "This significantly simplifies the design requirements of the device," said Fiore.

Analyzing milk and plastics

As described by co-first author Maurangelo Petruzzella, after the sensor was ready for use, the team subsequently tested the sensor in several experiments; he is also employed by the startup MantiSpectra.

"We used the sensor to measure the nutritional properties of many materials, including milk. Our sensor provided accuracy comparable to traditional spectrometers in predicting the fat content in milk. We then used the sensor to classify different types of plastics."

The nutritional properties of milk determine its economic value, and the sensor has been validated to correctly quantify these properties. Furthermore, these measurements can be used to track the overall health status of dairy cows. Classifying plastic types using the sensor helps strengthen the waste sorting process.

In addition to these applications, we anticipate that the sensor can also be used for personalized healthcare, precision agriculture (such as monitoring the ripeness of fruits and vegetables), process control, and lab-on-a-chip testing. We now have a complete development kit based on this technology, SpectraPod™, which companies and research institutions are using to build their applications.

Maurangelo Petruzzella, co-first author of the study, Department of Applied Physics, Eindhoven University of Technology

"The best part is that this sensor could even become widespread in future smartphones, meaning people could use it at home to check their food quality or monitor their health," added Petruzzella.

Mantispectra Future

For Hakkel, things are just beginning to get exciting. She will defend her thesis at TU/e on 1 14. Afterward, she will join Petruzzella at the startup MantiSpectra, where they will work on advancing sensors for more real-world applications.

"I am excited to start the next phase of sensor development with MantiSpectra. This sensor can contribute to a cleaner environment and address food waste, applications that are important to everyone."