Distributed feedback laser biosensor noise reduction

Label-free biosensors based upon detection of shifts in the emission wavelength, which occur when biological analytes are adsorbed on the surface of a solid-state laser, represent an important new class of sensors that can simultaneously provide high sensitivity and high resolution. We report on a signal processing approach that enables detection of lasing wavelength shifts as small as $\Delta\lambda\sim 1.5~{\rm pm}$ from a distributed feedback laser biosensor (DFBLB) fabricated upon a flexible plastic substrate and incorporated into standard-format microplates. Because the DFBLB can be optically pumped in a pulsed mode at a repetition rate that is substantially faster than the rate of biomolecular binding interactions, noise may be reduced through the ability to average multiple independent readings through integration of many lasing spectra within a single spectral integration period, and to subsequently use boxcar averaging to collapse multiple readings in a time sequence to a single time point. We have observe that the DFBLB occasionally produces nonclassical output spectra that can lead to an increase in wavelength shift noise, and therefore we implement a statistical metric to automatically determine whether an acquired spectrum should be discarded from analysis. The combined approaches are used to demonstrate the ability to detect a protein-protein interaction with an analyte concentration that would not otherwise be observable over background noise.