Click to learn about ReadCube Desktop
  • Save
  •  
  • Zoom In
  • Zoom Out
  • Fit Width
  •  
  • Set Text Cursor
  • Set Hand Cursor
  • Set Snapshot Cursor
  •  
  • Print
  • Save
    • Save Enhanced PDF

      With free ReadCube Desktop

  •  
  • Closex
  • Find Previous
  • Find Next

Efficient colour splitters for high-pixel-density image sensors

Seiji Nishiwaki,Tatsuya Nakamura,Masao Hiramoto,Toshiya Fujii,Masa-aki Suzuki

DOI: 10.1038/nphoton.2012.345

When the pixel size of image sensors shrinks to the wavelength of light, this results in low signal levels for a given photon flux per pixel as a result of scaling laws. Because many image sensors require colour filters, it becomes crucial for small-pixel sensors to have an efficient filtering method that can capture all incident photons without absorbing them. Here, we propose a new method to split colours by using a microscale plate-like structure with a transparent medium that has a higher refractive index than the surrounding material. We experimentally demonstrate that this principle of colour splitting based on near-field deflection can generate colour images with minimal signal loss. From comparisons of the sum of the total integrated values for the colour channels, we confirm the amount of light received is 1.85 times that of the conventional colour filter method of the Bayer array, while maintaining the same level of resolution.

To view this page in ReadCube Web Reader ensure that Adobe Flash Player version 10.0.0 or greater is installed.

1 Nature Photonics 2013 Vol: 7(3):240-246. DOI: 10.1038/nphoton.2012.345

Efficient colour splitters for high-pixel-density image sensors

Seiji Nishiwaki; Tatsuya Nakamura; Masao Hiramoto; Toshiya Fujii; Masa-aki Suzuki

When the pixel size of image sensors shrinks to the wavelength of light, this results in low signal levels for a given photon flux per pixel as a result of scaling laws. Because many image sensors require colour filters, it becomes crucial for small-pixel sensors to have an efficient filtering method that can capture all incident photons without absorbing them. Here, we propose a new method to split colours by using a microscale plate-like structure with a transparent medium that has a higher refractive index than the surrounding material. We experimentally demonstrate that this principle of colour splitting based on near-field deflection can generate colour images with minimal signal loss. From comparisons of the sum of the total integrated values for the colour channels, we confirm the amount of light received is 1.85 times that of the conventional colour filter method of the Bayer array, while maintaining the same level of resolution.

Mentions
Figures
Figure 1: Operational principle of a symmetric deflector. a, The symmetric deflector has a plate-like structure and is composed of a transparent medium that has a higher refractive index than the surrounding material. When light enters a symmetric deflector of width w and length l, a phase difference δ develops between the light propagated through the deflector and the light propagated through the surrounding volume. If δ is an even multiple of π, the transmitted light is undeflected (arrow U), and if δ is an odd multiple of π, the transmitted light is deflected at equal ± angles (arrows +D and −D). The phase difference δ rises and falls as a function of wavelength λ, so the ratio between the amount of undeflected light and the amount of ± deflected light varies depending on λ. b, Plot of the cross-sectional refractive index distribution for a symmetric deflector. Shape and position of the deflector: w = 0.28 µm, l = 1.20 µm, d = 2.00 µm. Refractive indices of the deflector medium (SiN) and the surrounding medium (SiO2): n = 2.03 and 1.46. c,d, Plots of cross-sectional light intensity distribution for a symmetric deflector with λ = 430 nm (c) and λ = 600 nm (d). The Gaussian beam has a waist diameter D (full width at 1/e2 maximum) of 1.43 µm. As the wavelength increases, deflected light becomes undeflected. e,f, Plots of light intensity distribution at the detector surface for a symmetric deflector for λ = 430 nm (e) and λ = 600 nm (f). Three 1.43 µm × 1.43 µm detectors (R, C and L) are placed next to one another on the detecting surface. As the wavelength increases, the spots on the detector move towards the centre. Figure 2: Operational principle of an asymmetric deflector. a, The asymmetric deflector has a structure in which the side surface has a step; this distinguishes it in shape from the symmetric deflector. When light enters an asymmetric deflector with widths w and w1 and lengths l and l1, phase differences δ and δ1 develop between the light propagated through the deflector and the light propagated through its surroundings. These phase differences closely approximate a slanted wavefront. As the phase differences δ and δ1 rise and fall as a function of wavelength λ, their polarity may change. In short, the direction of propagation of the deflected light varies as a function of wavelength like arrows U, +D and −D. b, Plot of cross-sectional refractive index distribution for an asymmetric deflector. Shape and position of the deflector: w = 0.32 µm, w1 = 0.16 µm, l = 1.20 µm, l1 = 0.60 µm, d = 2.00 µm. Refractive indices of the deflector medium (SiN) and the surrounding medium (SiO2): n = 2.03 and 1.46. c,d, Plots of cross-sectional light intensity distribution for an asymmetric deflector with λ = 420 nm (c) and λ = 700 nm (d). The Gaussian beam has a waist diameter D (full width at 1/e2 maximum) of 1.43 µm with a 0.20 µm shift along the x-axis. Diffracted light moves to the right with lengthening wavelength. e,f, Plots of light intensity distribution at the detector surface for an asymmetric deflector at λ = 420 nm (e) and λ = 700 nm (f). Three 1.43 µm × 1.43 µm detectors (R, C and L) are placed next to one another on the detecting surface. When the wavelength increases, the spots on the detector move to the right. Figure 3: Spectra from symmetric and asymmetric deflectors. For the symmetric deflector, the solid line plotting the amount of light obtained by the C-detector peaks at λ = 590 nm and troughs at λ = 410 nm. The solid line plotting the sum of the amount of light obtained by the R- and L-detectors peaks at λ = 430 nm and troughs at λ = 620 nm. For the asymmetric deflector, the broken lines plotting the amount of light obtained by the R-, C- and L-detectors have their maxima at λ = +700 nm, 520 nm and 420 nm, respectively. Figure 4: The two-deflector method and demonstration of colour splitting. a, Location of R-deflectors and detectors in the two-deflector method. R-deflectors split colour to form the colours W + R in areas between neighbouring R-deflectors, and W − R just beneath the R-deflectors. b, Location of B-deflectors and detectors in the two-deflector method. B-deflectors split colour to form W + B colours in areas between neighbouring B-deflectors, and W − B just beneath the B-deflectors. c, Colour detection by detectors. The colours W + R, W − R, W + B and W − B are detected by the four detectors. d, Cross-sectional SEM image of an R-deflector immediately after etching the silicon nitride layer. The R-deflectors have a depth of 1.68 µm, and a trapezoidal cross-section that is 0.33 µm wide at the top and 0.43 µm at the bottom, with a height of 1.14 µm. e, Cross-sectional SEM image of a B-deflector immediately after etching the silicon nitride layer. The B-deflectors have a depth of 1.68 µm, and a trapezoidal cross-section that is 0.30 µm wide at the top and 0.34 µm at the bottom, with a height of 0.42 µm. f, Cross-sectional configuration diagram of the evaluation sample. R-deflector: d = 4.97 µm; B-deflector: d = 5.02 µm. g, Microscope image showing the transmitted light of an R-deflector. Light transmitted through the hole in the silicon substrate of the R-deflectors is observed at the undersurface of the silicon oxide using a microscope. Colour splitting of W + R and W − R is demonstrated. h, Microscope image showing the transmitted light of a B-deflector. Light transmitted through the hole in the silicon substrate of the B-deflectors is observed at the undersurface of the silicon oxide using a microscope. Colour splitting of W + B and W − B is demonstrated. Figure 5: Cross-sectional SEM images of a conventional CCD and MiCS-IS. a, Cross-sectional SEM image of a conventional CCD. b, Cross-sectional SEM image of an R-deflector. The R-deflectors have a depth of 1.43 µm and a trapezoidal cross-section that is 0.14 µm wide at the top and 0.26 µm at the bottom, with a height of 1.02 µm. The etched lenses have a cuboid configuration of 0.81 µm × 0.81 µm with a height of 0.26 µm. The collecting plates have a cuboid configuration of 1.02 µm × 1.02 µm and height of 0.19 µm. c, Cross-sectional SEM image of a B-deflector. The B-deflectors have a depth of 1.43 µm and a trapezoidal cross-section that is 0.19 µm wide at the top and 0.33 µm at the bottom, with a height of 1.21 µm. The upper end position of the B-deflectors is aligned with that of the R-deflectors and the bottom of the etched lenses. The head position of the collecting plates is aligned with the bottom of the B-deflectors. Figure 6: Imaging characteristics of a MiCS-IS. a, Photographic image from the MiCS-IS. b, Photographic image from the colour filter method. c, Experimental spectroscopic characteristics of MiCS-IS and those of the colour filter method. The illumination spectrum is corrected so as to remain constant over the whole visible range, and the detected light value of the vertical axis is defined as the energy absorbed at the detectors (that is, a silicon layer) in measurement units of millivolts. The amount of light received is 1.85 times that obtained using the colour filter method using a Bayer array. d, Simulated spectroscopic characteristics of MiCS-IS and those of the colour filter method. The detected light value of the vertical axis is defined as the energy absorbed at the detectors (assuming that the silicon layer is 3.2 µm thick) using normalized units (a value of 1.0 indicates the incident light value). The analysis curves closely match the measured curves, but there are some differences, which are caused by simulation errors in the analytical method (B-BPM) or by measurement errors in the physical dimensions. Measurement errors, for example in relation to the optical constants of silicon and the colour filters, are likely to have a major influence. e,f, Resolution chart for MiCS-IS (e) and the colour filter method (f). The MiCS-IS image has a resolution that is nearly the same as that obtained using the colour filter method.
Altmetric
References
  1. Fossum, E. R. CMOS image sensors: electronic camera-on-a-chip IEEE Trans. Electron. Dev. 44, 1689-1698 (1997) .
      • . . . The pixel size in state-of-the-art image sensors is now approaching 1 µm, and this is placing a practical limit with respect to the signal-to-noise ratio of detected signals1, 2, 3 . . .
    • Abe, H. Device technologies for high quality and smaller pixel CCD and CMOS image sensors , 989-992 (2004) .
      • . . . The pixel size in state-of-the-art image sensors is now approaching 1 µm, and this is placing a practical limit with respect to the signal-to-noise ratio of detected signals1, 2, 3 . . .
    • Huo, Y.; Fresemaier, C. C.; Catrysse, P. B. Microlens performance limits in sub-2 µm pixel CMOS image sensors Opt. Express 18, 5861-5872 (2010) .
        • . . . The pixel size in state-of-the-art image sensors is now approaching 1 µm, and this is placing a practical limit with respect to the signal-to-noise ratio of detected signals1, 2, 3 . . .
      • Zhang, F.; Zhang, J.; Yang, C. Performance simulation and architecture optimization for CMOS image sensor pixels scaling down to 1.0 µm IEEE Trans. Electron. Dev. 57, 788-794 (2010) .
          • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
        • Jin, X. Measurement of high sensitivity and low crosstalk of zero-space microlens for 2.8-µm -pitch active pixel sensor IEEE Trans. Electron. Dev. 57, 415-421 (2010) .
            • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
          • Fesenmaier, C. C.; Huo, Y.; Catrysse, P. B. Optical confinement methods for continued scaling of CMOS image sensor pixels Opt. Express 16, 20457-20470 (2008) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Watanabe, H. A 1.4 µm front-side illuminated image sensor with a novel light guiding structure consisting of stacked lightpipes , 179-182 (2011) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Agranov, G. Pixel continues to shrink.... Pixel development for novel CMOS image sensors International Image Sensor Workshop , 69-72 (2009) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Pain, B. A back-illuminated megapixel CMOS image sensor Proc. 2005 IEEE Workshop on CCD and Advanced Image Sensors , 35-38 (2005) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Pain, B. Fabrication and initial results for a back-illuminated monolithic APS in a mixed SOI/bulk CMOS technology Proc. 2005 IEEE Workshop on CCD and Advanced Image Sensors , 102-104 (2005) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Iwabuchi, S. A back-illuminated high sensitivity small-pixel color CMOS image sensor with flexible layout of metal wiring Tech. Dig. ISSCC 49, 302-303 (2006) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Lee, K. SNR performance comparison of 1.4-mm pixel: FSI, light-guide, and BSI International Image Sensor Workshop , 9-11 (2011) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Michelot, J. Back illuminated vertically pinned photodiode with in depth charge storage International Image Sensor Workshop , 24-27 (2010) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Rhodes, H. The mass production of second generation 65 nm BSI CMOS image sensors International Image Sensor Workshop , 28-33 (2010) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Lee, Y. A highly manufacturable backside illumination technology for CMOS image sensor International Image Sensor Workshop , 46-49 (2010) .
              • . . . The first is to retain the maximum amount of light received at the pixels by optimizing the microlenses4, 5 or by using a light-guiding structure6, 7, 8 or a back-illuminated structure (back side illuminated (BSI) image sensor)9, 10, 11, 12, 13, 14, 15 . . .
            • Dammann, H. Color separation gratings Appl. Opt. 17, 2273-2279 (1978) .
                • . . . The second is to use colour-splitter structures16, 17, 18, 19, 20 rather than colour filters, as the latter tend to be based on absorption and introduce loss. . . .
                • . . . Conventional diffractive gratings16 are based on the principle of interference by phase-matching of the periodic wave surfaces of diffracted light, so they split colours in the far field . . .
              • Suzuki, Y. Proposal of micro-trichroic structures for high-sensitivity color image sensors Jpn J. Appl. Phys. 50, 042201 (2011) .
                  • . . . The second is to use colour-splitter structures16, 17, 18, 19, 20 rather than colour filters, as the latter tend to be based on absorption and introduce loss. . . .
                • Xu, T.; Wu, Y.; Luo, X.; Guo, L. Plasmonic nanoresonators for high-resolution colour filtering and spectral imaging Nature Commun. 1, 59 (2010) .
                  • . . . The second is to use colour-splitter structures16, 17, 18, 19, 20 rather than colour filters, as the latter tend to be based on absorption and introduce loss. . . .
                • Nguyen-Hua, N.; Loa, Y.; Chen, Y. Color filters featuring high transmission efficiency and broad bandwidth based on resonant waveguide–metallic grating Opt. Commun. 284, 2473-2479 (2010) .
                    • . . . The second is to use colour-splitter structures16, 17, 18, 19, 20 rather than colour filters, as the latter tend to be based on absorption and introduce loss. . . .
                  • Kanamori, Y.; Shimono, M.; Hane, K. Fabrication of transmission color filters using silicon subwavelength gratings on quartz substrates IEEE Photon. Technol. Lett. 18, 2126-2128 (2006) .
                      • . . . The second is to use colour-splitter structures16, 17, 18, 19, 20 rather than colour filters, as the latter tend to be based on absorption and introduce loss. . . .
                    • Nishiwaki, S. Imaging photodetection device , (2010) .
                      • . . . Here, we propose a new method—a micro colour splitter (MiCS)21, 22—to split colours using a miniature plate-like structure with a high refractive index to induce a spectral-dependent deflection . . .
                    • Nishiwaki, S.; Suzuki, M.; Wakabayashi, S.; Yonemoto, K.; Hiramoto, M. Imaging photodetection device , (2011) .
                      • . . . Here, we propose a new method—a micro colour splitter (MiCS)21, 22—to split colours using a miniature plate-like structure with a high refractive index to induce a spectral-dependent deflection . . .
                    • Longhurst, R. S. Geometrical and Physical Optics , (1967) .
                      • . . . In addition, although the diffractive angle of gratings is proportional to the wavelength divided by the periodic interval23, the deflection angle of a symmetric or asymmetric deflector, because of its unitary construction, is proportional to wavelength and also to the geometric or optical parameters (for example, width, length, refractive index and polarization ), but not a periodic interval. (Further information on the deflection angle can be found in Supplementary Section SC.) . . .
                    • Feit, M. D.; Freeck, J. A. Jr. Light propagation in graded-index optical fibers Appl. Opt. 17, 3990-3998 (1978) .
                        • . . . In our wave-optics analysis of Figs 1–3, we used an improved fast Fourier transformation beam propagation method (FFT-BPM24, 25), termed Babinet-BPM (B-BPM; see Methods for more details) . . .
                      • Thylen, L. The beam propagation method: an analysis of its applicability Opt. Quantum Electron. 15, 433-439 (1983) .
                          • . . . In our wave-optics analysis of Figs 1–3, we used an improved fast Fourier transformation beam propagation method (FFT-BPM24, 25), termed Babinet-BPM (B-BPM; see Methods for more details) . . .
                        • Yee, K. S. Numerical solution of initial boundary value problems involving Maxwell's equations in isotropic media IEEE Trans. Antennas Propag. AP-14, 302-307 (1966) .
                          • . . . Because the analytical accuracy of B-BPM is close to that of the finite-difference time-domain (FDTD) method26 for s-polarization (see Supplementary Section SA for details) and provides faster computing performance than FDTD, we used B-BPM to design and practically optimize the following MiCS image sensor (MiCS-IS). . . .
                        • Hiramoto, M. Solid-state imaging device , (2010) .
                          • . . . In this Article, we apply a method27 to reproduce colours by combining the outputs of the R- and B-deflectors using a two-dimensional charged-coupled device (CCD) . . .
                        • Bayer, B. E. Color imaging array , (1976) .
                          • . . . Based on the above two-deflector method, in place of colour filters with a Bayer array28, an R-deflector and a B-deflector were formed on conventional CCDs with a 1.43 µm × 1.43 µm square pixel in which inner lenses had been fabricated . . .
                        Expand