Miniature and durable spectrometer for wearable applications

Researchers have developed an ultra-thin chip-scale spectrometer suitable for wearable applications. The powerful gallium nitride chip-on-chip lab instrument can also withstand harsh environments with harsh radiation, such as space exploration or those with high temperatures—and can be tuned to perform a blood analysis just by shining a light on the skin. It measures only 0.16mm².

“As one wearable devicewe will be able to place our device on a flexible substrate, such as a sheet or cloth—or possibly on skin,” said researcher Dr. Tuba Sarwar.

One spectrograph uses spectroscopy to gather a variety of information about a substance based on how the substance absorbs and reflects light. Spectroscopy, which measures the spectrum of an optical signal, is one of the most common forms of spectroscopy. Specifically, it measures the relative spectral power density of the spectrum at different wavelengths.

The prototype spectrometer created by Sarwar and other members of Professor PC Ku’s team was originally developed with a specific purpose: to measure athletes’ perspiration in a wearable device in the form of a pad. skin stickers. This app was identified by UM’s Exercise and Sports Science Initiative (ESSI). Equipment does not need to be used for a long time; In fact, disposable is an advantage.

Creating such a device would require extreme miniaturization of devices currently on the market, which are close to 12 cubic inches in size and cost hundreds to thousands of dollars. It will also require creating a device that can operate in real time under changing conditions.

“Commercially, there are many spectrometers that can outperform ours,” says Sarwar. “But our focus is on miniaturization, especially the thickness of the device. These two features can help us analyze an active sample, such as sweat. We don’t have to go into the room. experiment to use complex instruments to get very precise measurements.”

Sarwar’s mention of a laboratory underscores the fact that most commercially available spectrometers require a bench-top system consisting of several separate optical components: 1) a source light to stimulate the sample; 2) spectral encoder to translate light from sample into readable output; and 3) photodetector to translate the encoded signal into spectrum.

Of these three elements, the spectrophotometer encoder usually takes up the most space or thickness, making it difficult for a thin sheet of spectrometers to be suitable for epidermal applications.

Ku, Qu and their team took on the challenge of creating a device that could achieve ESSI’s goals and create an integrated, miniaturized, low-power device that operates in visible light , specifically in the 400–645 nm wavelength range.

The group’s spectrometer consisted of only 16 photodetectors, each responding to a unique spectrum of light. This low number of photodetectors is made possible by two main techniques.

First, the team used a strain technique on a gallium nitride (GaN)-based spectroscopic encoder. Strain engineering is a technique used, for example, in semiconductor manufacturing where the material is stressed or deformed. If done properly, it can lead to new material properties that are more suitable for specific applications. GaN semiconductors were chosen as base materials because of their excellent optical properties on the visible spectrum.

The desired result of a significant reduction in the angle dependence of the light has been achieved, which eliminates the need for precise positioning of the spectrometer and associated optics. It also allows photodetectors to reside with spectral encoders on the same chip.

Second, the team incorporated machine learning into the operation of the device to decode the signal emitted by the detector. PhD researcher Can Yaras used a simple non-negative least squares (NNLS) algorithm to enable an efficient computational algorithm to recover spectral information from the detector’s signal.

In terms of performance, the device has high accuracy in determining peak wavelengths (with standard deviation is 0.97%), but is less accurate in measuring intensity ratios across different peak positions (with a standard deviation of 21.1% or 10.4% after removing one outlier) .

The team hopes that readings of intensity ratios can be improved by increasing the number of photodetectors and further developing machine learning algorithms, such as by applying deep learning techniques. They are also working on several other improvements to the prototype spectrometer.

“Our goal is not to build the best spectrometer in the world in terms of resolution,” says Ku, “but to focus on other aspects that are just as interesting if not more: size, thickness, power consumption and ease of operation.”

The work was published in the journal The word nano.

As for future applications, Sarwar says the miniaturized spectrometer could be integrated into a skin patch for health monitoring and diagnosis. Its advantage over existing devices is that the excitation light source can also be integrated easily. The radiative stiffness of the GaN semiconductor also makes the device potentially suitable for space exploration.