Photonics Research, Volume. 7, Issue 9, 961(2019)

Compact CMOS spectral sensor for the visible spectrum Editors' Pick

Yibo Zhu1,4、†,*, Xin Lei1、†, Ken Xingze Wang2,5、*, and Zongfu Yu3
Author Affiliations
  • 1Coherent AI LLC, Redwood City, California 94065, USA
  • 2School of Physics, Huazhong University of Science and Technology, Wuhan 430079, China
  • 3Department of Electrical and Computer Engineering, University of Wisconsin–Madison, Madison, Wisconsin 53705, USA
  • 4e-mail: yibo@coherent.ai
  • 5e-mail: wxz@hust.edu.cn
  • show less
    Figures & Tables(4)
    Sensor operating principle and device images. (a) Sensor operating principle and schematics. The spectral sensor consists of an array of PC slabs atop a CMOS image sensor. The transmission spectra of the PC slabs differ significantly from each other and form a sampling basis T. For an incident light, the pixel values of the sensor and the sampling basis T are used to infer the unknown spectrum of the light. (b) Device photograph. The sensor consists of an array of PC slabs fabricated on a SiNx/quartz substrate, placed on a CMOS image sensor. (c) Optical micrograph of a PC slab array.
    Transmission of PC slabs fabricated using different materials. (a) The real parts and imaginary parts of refractive indices for α-Si, SiNx, SiC, and SiO2 deposited using chemical vapor deposition. (b) Representative transmission spectra for SiNx, SiO2, and SiC by numerical simulation. (c) Mapping of correlation coefficients between each pair of the 42 simulated transmission spectra in different wavelength ranges. Blue indicates weak or negative correlation; red indicates strong positive correlation. For each mapping plot, the value on the secondary diagonal is the mean value of this matrix.
    Simulated spectra reconstruction results. (a) Reconstruction of a three-peak spectrum in different spectral ranges using three different materials. SiC-based PC exhibits large error in the 400–500-nm range. (b) and (c) Reconstruction of a more complex spectrum using fewer PC slabs. The SiNx PC slab achieves significantly higher reconstruction accuracy than SiO2 and SiC when the PC number is reduced to 12. The reconstruction error presented here is calculated using ‖I0−IR‖1/‖I0‖1, where I0 is the simulated spectrum, IR is the reconstructed spectrum, and ‖x‖1 indicates the L-1 norm of x. In (a) and (b), the solid lines indicate original simulated spectra, while circles represent recovered spectra.
    Experimental results. (a) Image captured by the PC-slab-mounted CMOS image sensor at an illuminating wavelength of 450 nm. (b) Scanning electron microscopy image of a PC slab. (c) Normalized transmission spectrum of the PC in (b) as measured using the CMOS image sensor. The as-measured data are divided by the measured light power at each wavelength to achieve the corrected data for spectral reconstruction. (d) Sensing of monochromatic light using the single-shot sensor. (e)–(h) Spectral sensing using 42 and 9 PC slabs. (e) and (f) Sensing of narrowband (bandwidth ≈10 nm) spectra at 450 and 600 nm, respectively. (g) and (h) Sensing of broader band (bandwidth ≈40 nm) spectra at 500 nm. The spectrum of (g) is achieved using a bandpass filter, while the spectrum of (h) is from a stack of colorful plastic films.
    Tools

    Get Citation

    Copy Citation Text

    Yibo Zhu, Xin Lei, Ken Xingze Wang, Zongfu Yu. Compact CMOS spectral sensor for the visible spectrum[J]. Photonics Research, 2019, 7(9): 961

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Nanophotonics and Photonic Crystals

    Received: May. 13, 2019

    Accepted: Jul. 3, 2019

    Published Online: Aug. 2, 2019

    The Author Email: Yibo Zhu (yibo@coherent.ai), Ken Xingze Wang (wxz@hust.edu.cn)

    DOI:10.1364/PRJ.7.000961

    Topics