Progress in a-Si:H based multispectral sensor technology and material recognition

Abstract

In this paper we describe the development of four different amorphous silicon based pixn multispectral photodetectors and discuss their optical characteristics as a result of extensive bandgap engineering. Upcoming from a sensor structure providing narrow band absorption peaks between 450nm and 540nm with a maximum applied bias of −12V, we developed bias optimized detectors with increased bandwidth by changing the composition and thickness of the absorbing material. By applying just −2.5V, one sensor obtains almost a Gaussian spectral response with peaks ranging from 420nm to 580nm. We present a specific algorithm to simulate color recognition probabilities for 20 different whitish powders by using two similar detectors. For the sensor providing sensitivity maxima reaching from 450nm to 600nm with sampling peaks in the range between 400nm and 670nm, the simulation discloses enhanced recognition probabilities of more than 70.2%, requiring a readout time of at least 15.5ms. As assumed, the competetive sensor structure providing just a sampling bandwidth between 420nm and 630nm achieves recognition probabilities of 62.5% with a reduced readout time of only 6.1ms. Possible sensor applications may exist in fields of fluorescence and spectrophotometric measurements, in chemical analysis, medical diagnostics or in colorimetric and multispectral imagery.