JP2008186830A - Solid-state image sensor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a lamination type solid-state image sensor capable of satisfying light utilization efficiency and color reproducibility. <P>SOLUTION: In the solid-state image sensor 5, an R photoelectric conversion layer 53 for absorbing light at the wavelength region of mainly red color for generating a signal charge accordingly, a G photoelectric conversion layer 51 for absorbing light at the wavelength region of mainly green color for generating a signal charge accordingly, and a B photoelectric conversion layer 52 for absorbing light at the wavelength region of mainly blue color for generating a signal charge accordingly are laminated on a semiconductor substrate. In the solid-state image sensor 5, the G photoelectric conversion layer 51 is laminated on the uppermost layer, and the absorption rate is lower than 1.0 when the absorption rate at the absorption peak wavelength is set to 1.0 at the maximum. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention mainly includes an R photoelectric conversion layer that absorbs light in a red (R) wavelength region (wavelength ranging from about 550 nm to about 700 nm) and generates a signal charge corresponding thereto, and mainly green (G). A G photoelectric conversion layer that absorbs light in a wavelength region (wavelength of about 450 nm to about 610 nm) and generates a signal charge corresponding thereto, and a blue (B) wavelength region (wavelength is about 380 nm to about 380 nm) The present invention relates to a solid-state imaging device in which a B photoelectric conversion layer that absorbs light in a range (up to about 520 nm) and generates a signal charge corresponding thereto is laminated on a semiconductor substrate.

  In a single-plate color solid-state imaging device typified by a CCD type or CMOS type image sensor, three or four types of color filters are arranged in a mosaic pattern on an array of light receiving units that perform photoelectric conversion. Accordingly, color signals corresponding to the color filters are output from the respective light receiving units, and a color image is generated by performing signal processing on these color signals.

  However, in a color solid-state imaging device in which color filters are arranged in a mosaic shape, about 2/3 of incident light is absorbed by the color filter in the case of a primary color filter, light use efficiency is poor and sensitivity is low. There's a problem. Further, since only one color signal can be obtained at each light receiving unit, the resolution is poor, and in particular, there is a problem that false colors are conspicuous.

  Therefore, in order to overcome such a problem, a stacked solid-state imaging device having a structure in which three photoelectric conversion layers are stacked on a semiconductor substrate on which a signal readout circuit is formed has been researched and developed (for example, Patent Documents 1, 2). This stacked solid-state imaging device has, for example, a light receiving unit structure in which photoelectric conversion layers that generate signal charges (electrons, holes) are sequentially stacked on B, G, and R light sequentially from the light incident surface. A signal readout circuit that can independently read out signal charges generated in each photoelectric conversion layer is provided for each light receiving unit.

  In the case of an image pickup device having such a structure, incident light is almost photoelectrically converted and read out, the utilization efficiency of visible light is close to 100%, and color signals of three colors of R, G, and B are obtained in each light receiving unit. Therefore, it is possible to generate a good image with high sensitivity and high resolution (false color is not noticeable).

Special Table 2002-502120 JP 2002-83946 A

  For color films and single-plate solid-state image sensors used in silver-salt cameras, there is a wealth of knowledge regarding spectral sensitivity design to ensure good color reproducibility. There was no knowledge about the ideal spectral sensitivity that can satisfy the color reproducibility while utilizing the high light utilization efficiency.

  The present invention has been made in view of the above circumstances, and an object thereof is to provide a stacked solid-state imaging device capable of satisfying light utilization efficiency and color reproducibility.

  The solid-state imaging device of the present invention mainly absorbs light in the red wavelength region and generates a signal charge corresponding thereto, and absorbs light in the green wavelength region and signals corresponding thereto A solid-state imaging device in which a G photoelectric conversion layer that generates electric charges and a B photoelectric conversion layer that absorbs light mainly in the blue wavelength region and generates signal charges corresponding thereto are stacked on a semiconductor substrate. When the G photoelectric conversion layer is laminated on the top and the absorption rate at the absorption peak wavelength is 1.0, the lower is lower than that.

  In the solid-state imaging device of the present invention, the absorptance at the absorption peak wavelength of the G photoelectric conversion layer is in the range of 0.5 to 0.8.

  In the solid-state imaging device of the present invention, the G photoelectric conversion layer has an absorption peak wavelength in the range of 533 nm to 547 nm.

  In the solid-state imaging device of the present invention, the absorption peak wavelengths of the R photoelectric conversion layer and the B photoelectric conversion layer are shifted to the absorption peak wavelength side of the G photoelectric conversion layer from the ideal wavelength based on sRGB.

  In the solid-state imaging device of the present invention, the R photoelectric conversion layer has an absorption peak wavelength in the range of 575 nm to 605 nm, and the B photoelectric conversion layer has an absorption peak wavelength in the range of 445 nm to 478 nm.

  In the solid-state imaging device of the present invention, the half width of the absorption spectrum of the R photoelectric conversion layer is in the range of 65 nm to 100 nm.

  In the solid-state imaging device of the present invention, the half width of the absorption spectrum of the B photoelectric conversion layer is in the range of 50 nm to 110 nm.

  In the solid-state imaging device of the present invention, the B photoelectric conversion layer, the R photoelectric conversion layer, and the G photoelectric conversion layer are stacked in this order on the semiconductor substrate.

  According to the present invention, it is possible to provide a stacked solid-state imaging device capable of satisfying light utilization efficiency and color reproducibility.

  Embodiments of the present invention will be described below with reference to the drawings.

The inventor has found the characteristics of each photoelectric conversion layer of a stacked solid-state imaging device capable of satisfying light use efficiency and color reproducibility by the following simulation model.
FIG. 1 is a diagram illustrating a configuration of an imaging system used for a simulation model.
The imaging system shown in FIG. 1 includes a Macbeth chart 1 as a subject, a light source 2 that illuminates the Macbeth chart 1 with D65 light, a solid-state imaging device 5 for photographing the Macbeth chart 1, and a front surface of the solid-state imaging device 5. The provided IR cut filter 4 that cuts light in the infrared region, the imaging optical system 3 disposed between the IR cut filter 4 and the color chart 1, and the white balance of the imaging signal output from the solid-state imaging device 5 A WB circuit 6 that performs arithmetic processing to adjust the output, an MTX circuit 7 that corrects an output signal of the WB circuit 6, a monitor 8 that supports sRGB for displaying an image based on the signal corrected by the MTX circuit 7, A color difference calculation unit 9 for calculating a color difference between the Macbeth chart image displayed on the monitor 8 and the Macbeth chart 1, and a Macbeth chart image displayed on the monitor 8 It consists S / N calculating section 10 for calculating the luminance S / N.

FIG. 2 is a diagram showing a cross-sectional configuration for one pixel of the solid-state imaging device 5 shown in FIG. It is assumed that the solid-state imaging device 5 includes at least 24 pixels shown in FIG. 2 arranged in a two-dimensional manner in order to capture a 24-color Macbeth chart.
As shown in FIG. 2, the solid-state imaging device 5 includes a B photoelectric conversion layer (hereinafter referred to as B layer) 53 stacked on a semiconductor substrate (not shown), and an R photoelectric conversion layer (hereinafter referred to as B layer 53). , R layer) 52 and a G photoelectric conversion layer (hereinafter referred to as G layer) 51 laminated on the R layer 52. The G light of the reflected light from the number n chart of the Macbeth chart 1 incident from above the IR cut filter 4 is absorbed by the G layer 51, where a G signal charge corresponding to the G light is generated, and the G layer R light out of the light transmitted through 51 is absorbed by the R layer 52, where an R signal charge corresponding to the R light is generated, and B light out of the light transmitted through the R layer 52 is absorbed by the B layer 53. Here, a B signal charge corresponding to the B light is generated. An imaging signal Srn corresponding to the R signal charge, an imaging signal Sgn corresponding to the G signal charge, and an imaging signal Sbn corresponding to the B signal charge are output to the outside.

  The spectral sensitivities Sr (λ), Sg (λ), and Sb (λ) of the R layer 52, the G layer 51, and the B layer 53 that constitute one pixel of the solid-state imaging device 5 are expressed by the following equation (1). Is done.

Assuming that the absorption coefficient spectra of the R layer 52, the G layer 51, and the B layer 53 are αr (λ), αg (λ), and αb (λ), respectively, and assuming that each has a Gaussian distribution, As shown, the absorption spectrum A subscript (λ) and the transmission spectrum T subscript (λ) of each layer shown in Equation 1 are the absorption peak wavelength λ 0 of the absorption coefficient spectrum α (λ) of the layer shown by the subscript. And the layer thickness t of the absorption coefficient spectrum α (λ) of the layer indicated by the subscript and the half width HW of the absorption coefficient spectrum α (λ) of the layer indicated by the subscript can be described as parameters. That is, the absorption spectrum A (λ) of each layer can be adjusted by adjusting λ0, t, HW of each layer.

  The signal output from the solid-state imaging device 5 is expressed by the following equation (3).

  In this simulation, only random noise generated in the solid-state imaging device 5 is considered, noise included in the output signal Srn is defined as Nrn, noise included in the output signal Sgn is defined as Ngn, and noise included in the output signal Sbn is determined. Nbn.

  In the WB circuit 6, the output signal Srn is multiplied by the gain Gr to obtain the output signal Swrn, the output signal Sgn is multiplied by the gain Gg, the output signal Swgn is obtained, and the output signal Sbn is multiplied by the gain Gb. Performs processing to obtain Swbn. In the WB circuit 6, the WB chains Gr, Gg, and Gb are set so that the output signals Swrn, Swgn, and Swbn are set to Swrn = Swgn = Swbn = constant value for complete diffusion white plate imaging. As a result, the noise output from the WB circuit 6 is given by the equation (4).

  The MTX circuit 7 corrects the output signals Swrn, Swgn, Swbn by performing the calculation shown in the following equation 5 using a 3 × 3 matrix coefficient, and outputs the corrected output signals SRn, SGn, SBn to the monitor 8 and S / N Output to the arithmetic unit 10. In this simulation, the RMS value (square root of the sum of squares) of the color difference ΔE * ab of each color between the Macbeth chart 1 itself and the Macbeth chart image obtained when the Macbeth chart 1 is imaged by the solid-state imaging device 5 is minimized. Determine the matrix coefficients.

  As a result, the noise included in the output signal of the MTX circuit 7 is given by the equation (6).

  When the output signals SRn, SGn, BGn are input to the monitor 8, the chromaticity points XYZ of the Macbeth chart image displayed on the monitor 8 are given by the following equation (7).

  The color difference calculation unit 9 obtains a color difference between the XYZ chromaticity coordinates of the Macbeth chart image displayed on the monitor 8 and the XYZ chromaticity coordinates of the Macbeth chart 1 itself in the CIELAB color space. The color difference ΔE * ab is obtained by the following equation (8). The color difference ΔE * ab is obtained for each of the 24 colors of the Macbeth chart 1 by the calculation in the color difference calculation unit 9.

  The S / N calculator 10 calculates the luminance S / N in the sYCC space. The conversion from the output signals SRn, SGn, SBn of the MTX circuit 7 to the sYCC space is given by the equation (9).

Therefore, the conversion from the noise signals Nwrno, Nwgno, Nwbno included in the output signal of the MTX circuit 7 to the sYCC space is given by the equation (10).

For this reason, the luminance S / N when the number n of the Macbeth chart 1 is photographed is obtained by Y / Ynoise.

  In the simulation model having such a configuration, among the three parameters λ0, HW, t that determine the absorption spectrum A (λ) of each of the R layer 52, the G layer 51, and the B layer 53, “half-value width HW and layer thickness” When t is a fixed value and the peak wavelength λ0 is changed by a small amount, the matrix coefficient of the MTX circuit 7 is optimized so that the RMS value (= average color difference ΔE) of 24 color differences ΔE * ab is minimized. Repeated the work. The result is shown in FIG.

  FIG. 3A shows an absorption peak wavelength (horizontal axis) of the absorption spectrum Ag (λ) when set to a certain peak wavelength λ0, and an average color difference ΔE (optimized by a matrix coefficient at the peak wavelength λ0). It is the figure which showed the relationship with the vertical axis | shaft. FIG. 3 (b) shows the absorption peak wavelength (horizontal axis) of the absorption spectrum Ar (λ) when set to a certain peak wavelength λ0, and the average color difference ΔE () obtained by optimizing the matrix coefficient at the peak wavelength λ0. It is the figure which showed the relationship with the vertical axis | shaft. FIG. 3 (c) shows the absorption peak wavelength (horizontal axis) of the absorption spectrum Ab (λ) when set to a certain peak wavelength λ0, and the average color difference ΔE (optimized by the matrix coefficient at the peak wavelength λ0). It is the figure which showed the relationship with the vertical axis | shaft.

  Furthermore, among the three parameters λ0, HW, t that determine the absorption spectrum A (λ) of each of the R layer 52, the G layer 51, and the B layer 53, “the peak wavelength λ0 and the layer thickness t are fixed values, and the half value The work of “optimizing the matrix coefficient of the MTX circuit 7 so that the RMS value (= average color difference ΔE) of 24 color differences ΔE * ab is minimized by changing the width HW by a small amount” was repeated. The result is shown in FIG.

  FIG. 4 (a) shows the half width (horizontal axis) of the absorption spectrum Ag (λ) when set to a certain half width HW, and the average color difference ΔE (vertical length) obtained by optimizing the matrix coefficient with the half width HW. It is the figure which showed the relationship with an axis | shaft. FIG. 4B shows the half-value width (horizontal axis) of the absorption spectrum Ar (λ) when set to a certain half-value width HW and the average color difference ΔE (vertical length) obtained by optimizing the matrix coefficient with the half-value width HW. It is the figure which showed the relationship with an axis | shaft. FIG. 4 (c) shows the half width (horizontal axis) of the absorption spectrum Ab (λ) when set to a certain half width HW, and the average color difference ΔE (vertical length) obtained by optimizing the matrix coefficient with the half width HW. It is the figure which showed the relationship with an axis | shaft.

  Furthermore, among the three parameters λ0, HW, t that determine the absorption spectrum Ag (λ) of the G layer 51, “the peak wavelength λ0 and the half-value width HW are fixed values, and the layer thickness t is changed by a minute amount. The operation of “optimizing the matrix coefficient of the MTX circuit 7 so that the RMS value of 24 color differences ΔE * ab (= average color difference ΔE) is minimized” was repeated. The result is shown in FIG.

  FIG. 5 (a) is obtained by optimizing the matrix coefficient with the absorption rate (abscissa) at the absorption peak wavelength of the absorption spectrum Ag (λ) when the G layer is set to a certain layer thickness t and the layer thickness t. It is the figure which showed the relationship with the obtained average color difference (DELTA) E (vertical axis). Fig. 5 (b) shows the absorption coefficient (abscissa) at the absorption peak wavelength of the absorption spectrum Ag (λ) when the G layer is set to a certain layer thickness t, and the matrix coefficient optimized by the layer thickness t. S / N of the output signal of the MTX circuit 7 (vertical axis, more precisely, the relative S / N based on the S / N of the output signal from the single-plate image sensor having ideal imaging characteristics) It is the figure which showed the relationship.

  According to the simulation results of FIGS. 3 to 5, it can be seen that the average color difference ΔE can be minimized when the absorption rate at the absorption peak wavelength of the absorption spectrum Ag (λ) is changed. As shown in FIG. 5, when the absorption rate at the absorption peak wavelength of the absorption spectrum Ag (λ) is about 0.6, the average color difference ΔE takes the minimum value, but as the absorption rate decreases, G Since the amount of signal obtained from the layer 51 decreases, assuming that the noise generated in the G layer 51 does not depend on the amount of absorbed light, the relative S / N deteriorates as the absorption rate decreases. . In order to make the difference almost indistinguishable when the Macbeth chart image displayed on the monitor 5 and the Macbeth chart 1 itself are visually compared, the average color difference ΔE is preferably about 1.0 or less. In order to achieve a certain degree of sensitivity, the relative S / N is preferably in the range of -4 to -1. For this reason, in order to achieve both high sensitivity and good color reproducibility, the absorbance at the absorption peak wavelength of the absorption spectrum Ag (λ) is about 0.5 to about 0. It can be said that being in the range of 8 is the most important requirement.

From the simulation results of FIGS. 3 and 4, it was found that the color reproducibility can be further improved by adding the following additional requirements 1 to 6.
(Additional requirement 1)
The upper limit of the allowable range of the average color difference ΔE is around ΔE = 2.0. Therefore, from the result of FIG. 3A, the preferable range of the absorption peak wavelength of the absorption spectrum Ag (λ) is a range in which the average color difference ΔE is 2.0 or less, that is, 533 nm to 547 nm.
(Additional requirement 2)
FIG. 6 is a diagram showing the sRGB ideal imaging characteristics. From this sRGB ideal imaging characteristic, it can be seen that the absorption peak wavelength of the absorption spectrum Ar (λ) has the best color reproducibility when its value is 605 nm. Therefore, when viewing FIG. 3B, the range of wavelengths in which the value can be made smaller than ΔE at 605 nm, which is an ideal value based on the sRGB ideal imaging characteristics (wavelength range of 575 nm to 605 nm). Is present. In other words, the additional requirement 2 is that the absorption peak wavelength of the absorption spectrum Ar (λ) is in the wavelength range of 575 nm to 605 nm. By doing so, the color reproducibility is improved more than the sRGB ideal imaging characteristics. Can do.

(Additional requirement 3)
From the sRGB ideal imaging characteristics shown in FIG. 6, it can be seen that the absorption peak wavelength of the absorption spectrum Ab (λ) has the best color reproducibility when its value is set to 445 nm. Therefore, when viewing FIG. 3 (c), the wavelength range in which the value can be made smaller than ΔE at 445 nm, which is an ideal value based on the sRGB ideal imaging characteristics (wavelength range of 445 nm to 478 nm). Is present. That is, the additional requirement 3 is to make the absorption peak wavelength of the absorption spectrum Ab (λ) in the wavelength range of 445 nm to 478 nm. By doing so, the color reproducibility is improved more than the sRGB ideal imaging characteristics. Can do.

  It should be noted from the additional requirements 3 and 4 that the absorption peak wavelength of the absorption spectrum Ar (λ) and the absorption peak wavelength of the absorption spectrum Ab (λ) are absorbed more than ideal values based on the sRGB ideal imaging characteristics, respectively. By shifting to the absorption peak wavelength side of the spectrum Ag (λ), good color reproducibility can be secured.

(Additional requirements 4-6)
As shown in FIG. 4, a preferable range in consideration of color reproducibility can also be set for the half width of the absorption spectrum A (λ) of each photoelectric conversion layer. In each of FIGS. 4 (a) to 4 (c), when an optimum half-value width is defined in a range where the color reproducibility can be tolerated (a range where the average color difference ΔE increases from the minimum value to a value increased by 50% from the minimum value). The half width of the absorption spectrum Ag (λ) is preferably in the range of 70 nm to 120 nm (additional requirement 4), and the half width of the absorption spectrum Ar (λ) is preferably in the range of 65 nm to 100 nm (additional requirement). 5) The half width of the absorption spectrum Ab (λ) is preferably in the range of 50 nm to 110 nm (additional requirement 6).

  Incidentally, these additional requirements 1 to 6 can realize better color reproducibility by combining at least one of these with the important requirements as compared with the case where only the important requirements are used.

  In this simulation, the simulation is performed using the solid-state imaging device 5 having a configuration in which the B layer 53, the R layer 52, and the G layer 51 are stacked in this order. However, when the positions of the B layer 53 and the R layer 52 are switched. However, it can be applied without changing the above requirements.

Examples of the material of each layer that can realize the various requirements described above include the following.
R layer 52: squarylium compound, croconium compound, phthalocyanine compound, merocyanine compound G layer 51: quinacridone compound, merocyanine compound B layer 53: merocyanine compound, porphyrin compound

  Next, as a result of changing the parameters λ0, HW, t and trying the combinations of the absorption spectra A (λ) of the layers of the solid-state imaging device 5 in every pattern, the result of the pattern having the smallest average color difference ΔE is shown in FIG. 7 shows.

  FIG. 7A shows the absorption spectrum A (λ) and transmission spectrum T (λ) of each layer. In FIG. 7A, B-Abs is the absorption spectrum of the B layer 53, G-Abs is the absorption spectrum of the G layer 51, R-Abs is the absorption spectrum of the R layer 52, and G-Trans is the transmission spectrum of the G layer 51. , R-Trans is the transmission spectrum of the R layer 52. FIG. 7B is a diagram showing the imaging characteristics of the entire system when each layer has the characteristics shown in FIG. FIG. 7 (c) is a diagram showing a value for each chart number of the Macbeth chart 1 of the color difference ΔE * ab when the characteristics of FIG. 7 (a) are set.

  When the average color difference ΔE was determined from the result of FIG. 7C, the value was 0.510. This value is a level at which almost no color difference can be confirmed, and is not inferior to that of an optimized single-plate image sensor. As can be seen from FIG. 7A, the absorption rate at the absorption peak wavelength of the G layer 51 is about 0.6, which satisfies the above-mentioned important requirement. Further, it can be seen that the absorption peak wavelength of the G layer 51 is around 540 nm and satisfies the additional requirement 1. Further, it can be seen that the absorption peak wavelength of the R layer 52 is around 590 nm and satisfies the additional requirement 2. Further, it can be seen that the absorption peak wavelength of the B layer 53 is around 460 nm and satisfies the additional requirement 3. Further, it can be seen that the half width of the G layer 51 is about 95 nm and satisfies the additional requirement 4. Further, the half width of the R layer 52 is about 95 nm, and it can be seen that the additional requirement 5 is satisfied. In addition, the half width of the B layer 53 is about 80 nm, and it can be seen that the additional requirement 6 is satisfied.

  Thus, it was found that the optimum color reproducibility can be secured by combining the important requirements and the additional requirements 1 to 6.

Diagram showing the configuration of the imaging system used in the simulation model The figure which showed the cross-sectional structure for 1 pixel of the solid-state image sensor 5 shown in FIG. Figure showing the change in average color difference ΔE when the absorption peak wavelength of each layer is changed The figure which showed the change of average color difference ΔE when changing the half width of each layer Figure showing the change in average color difference ΔE when the absorptance at the absorption peak wavelength of the G layer is changed Diagram showing ideal imaging characteristics of sRGB Diagram showing the characteristics of each layer when the average color difference ΔE is minimized in the simulation model

Explanation of symbols

51 G photoelectric conversion layer 52 R photoelectric conversion layer 53 B photoelectric conversion layer

Claims (8)

An R photoelectric conversion layer that mainly absorbs light in the red wavelength region and generates a signal charge corresponding thereto, and a G photoelectric conversion layer that absorbs light in the green wavelength region and generates signal charge corresponding thereto And a solid-state imaging device in which a B photoelectric conversion layer that mainly absorbs light in a blue wavelength range and generates a signal charge corresponding thereto is laminated on a semiconductor substrate,
A solid-state imaging device in which the G photoelectric conversion layer is stacked on the top and the absorptance at the absorption peak wavelength is 1.0, which is lower than that. The solid-state imaging device according to claim 1,
The solid-state image sensor whose absorptivity in the absorption peak wavelength of the said G photoelectric converting layer exists in the range of 0.5-0.8. The solid-state imaging device according to claim 1 or 2,
A solid-state imaging device in which an absorption peak wavelength of the G photoelectric conversion layer is in a range of 533 nm to 547 nm. The solid-state image sensor according to any one of claims 1 to 3,
A solid-state imaging device in which absorption peak wavelengths of the R photoelectric conversion layer and the B photoelectric conversion layer are shifted to an absorption peak wavelength side of the G photoelectric conversion layer from an ideal wavelength based on sRGB. The solid-state imaging device according to claim 4,
The absorption peak wavelength of the R photoelectric conversion layer is in the range of 575 nm to 605 nm,
A solid-state imaging device in which an absorption peak wavelength of the B photoelectric conversion layer is in a range of 445 nm to 478 nm. The solid-state imaging device according to claim 4 or 5,
A solid-state imaging device having a half-value width of an absorption spectrum of the R photoelectric conversion layer in a range of 65 nm to 100 nm. The solid-state imaging device according to claim 6,
A solid-state imaging device having a half-value width of an absorption spectrum of the B photoelectric conversion layer in a range of 50 nm to 110 nm. The solid-state imaging device according to any one of claims 1 to 7,
A solid-state imaging device in which the B photoelectric conversion layer, the R photoelectric conversion layer, and the G photoelectric conversion layer are stacked in this order on the semiconductor substrate.

Images