CdSe/ZnS quantum dot fluorescence spectra shape-based thermometry via neural network reconstruction

摘要

As a system of interest gets small, due to the influence of the sensor mass and heat leaks through thesensor contacts, thermal characterization by means of contact temperature measurements becomescumbersome. Non-contact temperature measurement offers a suitable alternative, provided a reliablerelationship between the temperature and the detected signal is available. In this work, exploiting thetemperature dependence of their fluorescence spectrum, the use of quantum dots as thermomarkers onthe surface of a fiber of interest is demonstrated. The performance is assessed of a series of neural networksthat use different spectral shape characteristics as inputs (peak-based—peak intensity, peakwavelength; shape-based—integrated intensity, their ratio, full-width half maximum, peak normalizedintensity at certain wavelengths, and summation of intensity over several spectral bands) and that yieldat their output the fiber temperature in the optically probed area on a spider silk fiber. Starting fromneural networks trained on fluorescence spectra acquired in steady state temperature conditions, numericalsimulations are performed to assess the quality of the reconstruction of dynamical temperaturechanges that are photothermally induced by illuminating the fiber with periodically intensitymodulatedlight. Comparison of the five neural networks investigated to multiple types of curve fitsshowed that using neural networks trained on a combination of the spectral characteristics improvesthe accuracy over use of a single independent input, with the greatest accuracy observed for inputs thatincluded both intensity-based measurements (peak intensity) and shape-based measurements (normalizedintensity at multiple wavelengths), with an ultimate accuracy of 0.29K via numerical simulationbased on experimental observations. The implications are that quantum dots can be used as a more stableand accurate fluorescence thermometer for solid materials and that use of neural networks for temperaturereconstruction improves the accuracy of the measurement.