Development of a compound-eye spectrometer for calibration-transfer-free spectroscopy

Area of Science:

• Spectroscopy
• Analytical Chemistry
• Instrumentation

Background:

• Near-infrared (NIR) spectroscopy offers rapid, nondestructive, and cost-effective analysis.
• Wider application of NIR spectroscopy is hindered by spectral data variations due to instrument differences.
• Complex calibration transfer (CT) procedures are currently required to address these variations, being time-consuming and labor-intensive.

Purpose of the Study:

• To investigate a novel compound-eye spectrometer concept to reduce instrument differences in NIR spectroscopy.
• To eliminate the need for complex calibration transfer (CT) procedures.
• To improve sensitivity and enable CT-free spectroscopy by averaging signals from multiple sensors.

Main Methods:

• A custom-built compound-eye spectrometer was developed, integrating 16 micro-electro-mechanical system (MEMS) spectral sensors in a circular array.
• The sensors functioned as a virtual spectrometer, averaging signals to minimize individual sensor variations.
• Calibration models for sucrose solutions (0-30 wt%) were developed and compared against three commercial reference spectrometers.

Main Results:

• Virtual spectrometers composed of five MEMS sensors showed superior performance in predicting sucrose concentrations compared to reference instruments.
• Significantly smaller absolute bias and root mean squared error of prediction were observed with the compound-eye spectrometer.
• Lower variation in absorbance at Brix-correlated wavelengths contributed to the improved performance.

Conclusions:

• The compound-eye spectrometer concept shows potential for developing calibration transfer-free (CT-free) spectroscopy.
• Averaging signals from multiple MEMS sensors effectively reduces spectral data variations.
• Further investigation is needed for limitations like measurement noise and performance variations, with future work focusing on fruit and vegetable samples.