Fig. 1. The strategy for fabricating the QD spectrometer. The absorption spectra of the QDs are tuned via their size and composition, and the QD filter array is fabricated by the electrohydrodynamic jet printing technique. The substrate with the QD filter array is assembled on top of a CMOS image sensor to fabricate the spectrometer. A new reconstruction algorithm is proposed to enhance the noise tolerance performance.

Fig. 2. (a) A picture of some typical QD solutions. (b) Absorption spectra of selected QDs. (c) Plots of QD absorption peak as a function of growth time in the seeded growth process for the QDs with different compositions.

Fig. 3. (a) Redshift of the absorption peak of QDs with the increase of the MCH ligand. (b) A picture of the QD filter array, in which the scale bar is 0.50 mm. (c) Uniformity and surface profile of the filter units. (d) Transmittance spectra of some filter units.

Fig. 4. Spectral reconstruction results of the TKVA algorithm for (a) monochromatic light, (b) a white LED spectrum, and (c) the xenon lamp spectrum. For comparison, the results of the Ridge algorithm are presented.

Fig. 5. Spectral reconstruction results of different algorithms at noise levels of (a) 10 dB, (b) 0 dB, and (c) −3dB.

Fig. 6. Spectral reconstruction results of the (a) TKVA and (b) Ridge algorithms at different interpolation factors. Reconstruction results of room light (c) without and (d) with 10× interpolation using the TKVA and Ridge algorithms. (e) Reconstruction results of two monochromatic lights with peaks 0.1 nm apart. (f) Spectral resolution as a function of the number of QD filters.