Time- and frequency-domain analyses of fluorescence lifetime for temperature sensing

Fluorescence lifetime detection is widely used for sensing physical and chemical quantities. The measurement accuracy of fluorescence lifetime-based sensing systems, either in time or frequency domain relies on their capability of detection and analysis of low level signal superimposed to noise. In this work a quantitative assessment of several data processing and analysis methods for the estimation of the fluorescence lifetime was carried out by using an experimental arrangement based on a fiber optic temperature sensor system. A comparison between the various methods was performed using actual signals from an optical sensing medium. The basic principles of time- and frequency-domain lifetime measurements were also reviewed and discussed in order to point out the limit of the cw frequency-domain approach and to suggest a way to overcome it. The investigated lifetime interval was from 200 to about 2200μs, corresponding to a temperature span of the sensor of about 300 °C. The results showed that in time domain (such as with Marquardt, integration, and log-fit algorithms) a good agreement, with relative differences from 0.2% to 0.5%, can be reached. Frequency-domain results based on an N-point fast Fourier transform (FFT) compare favorably with the previous ones in the long lifetime region (resulting in relative differences lower than 0.2%) with larger differences for short lifetimes. For each data processing method, the uncertainty associated with lifetime estimation was evaluated. Sampling and harmonics effects on the estimation accuracy for N-point FFTs were also investigated to trade-off between speed and accuracy of the algorithm in view of its application in real-time detection systems.