Modelling of diffraction grating based optical filters for fluorescence detection of biomolecules.

The detection of biomolecules based on fluorescence measurements is a powerful diagnostic tool for the acquisition of genetic, proteomic and cellular information. One key performance limiting factor remains the integrated optical filter, which is designed to reject strong excitation light while transmitting weak emission (fluorescent) light to the photodetector. Conventional filters have several disadvantages. For instance absorbing filters, like those made from amorphous silicon carbide, exhibit low rejection ratios, especially in the case of small Stokes' shift fluorophores (e.g. green fluorescent protein GFP with λ exc = 480 nm and λ em = 510 nm), whereas interference filters comprising many layers require complex fabrication. This paper describes an alternative solution based on dielectric diffraction gratings. These filters are not only highly efficient but require a smaller number of manufacturing steps. Using FEM-based optical modelling as a design optimization tool, three filtering concepts are explored: (i) a diffraction grating fabricated on the surface of an absorbing filter, (ii) a diffraction grating embedded in a host material with a low refractive index, and (iii) a combination of an embedded grating and an absorbing filter. Both concepts involving an embedded grating show high rejection ratios (over 100,000) for the case of GFP, but also high sensitivity to manufacturing errors and variations in the incident angle of the excitation light. Despite this, simulations show that a 60 times improvement in the rejection ratio relative to a conventional flat absorbing filter can be obtained using an optimized embedded diffraction grating fabricated on top of an absorbing filter.