JP2011254090A - Semiconductor image sensor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To further improve resolution of a semiconductor image sensor and to simplify an operational expression of a chrominance signal.SOLUTION: A semiconductor image sensor includes a plurality of pixels having: photoelectric conversion means; and MOS transistors selectively reading out the photoelectric conversion means, including: a first pixel group in which the plural pixels 103A are arranged in a horizontal direction and in a vertical direction by respective predetermined pitches in a two-dimensional manner; and a second pixel group in which, in the state that it is shifted relative to the first pixel group in the horizontal direction and in the vertical direction by a substantially 1/2 pitch relative to said pitch, the plural pixels 103B are arranged in the two-dimensional manner. As an on-chip color filter corresponding to the first pixel group, a complementary color filter is used.

Description

  The present invention relates to a semiconductor image sensor used for a video camera, a digital still camera, or the like.

  The semiconductor image sensor is a semiconductor device that includes a plurality of pixels including photoelectric conversion means and a MOS transistor, and selects and reads signals from the pixels by the MOS transistor. In recent years, this semiconductor image sensor, in particular, a so-called CMOS image sensor manufactured by a CMOS (complementary MOS) process has a low voltage, low power consumption, multiple functions, and an SOC (system-on) that can be integrated into one chip with peripheral circuits. Chip). Therefore, it is attracting attention and used as an imaging device for a camera for a mobile phone, a digital still camera, a digital video camera, or the like.

  FIG. 17 shows an example of the configuration (equivalent circuit) of the semiconductor image sensor. In the CMOS image sensor 1, a plurality of pixels 3 each including a photodiode 2 for performing photoelectric conversion and a plurality of MOS transistors for selectively reading out the photodiode 2 are two-dimensionally arranged on the same semiconductor substrate. An imaging element forming region 4 is provided. Further, peripheral circuits 5 and 6 for pixel selection and signal output are provided around the imaging region formation area 4 on the same semiconductor substrate. The region including the circuit 5 and the output circuit 6 for pixel selection is called a peripheral circuit formation region. In the imaging element formation region 4, each pixel 3 is composed of one photodiode 2 and a plurality of MOS transistors, in this example, three MOS transistors including a transfer transistor 8, a reset transistor 9, and an amplification transistor 10. Yes. In the peripheral circuit formation region, a pixel selection circuit (vertical scanning circuit) 5 and an output circuit (horizontal scanning / output circuit) 6 are configured using CMOS transistors.

  In FIG. 17, the photodiode 2 is connected to the source of the transfer transistor 8. A transfer wiring 11 is connected to the gate of the transfer transistor 8. The drain of the transfer transistor 8 is connected to the source of the reset transistor 9, and a so-called floating diffusion FD between the drain of the transfer transistor 8 and the source of the reset transistor 9 is connected to the gate of the amplification transistor 10. The gate of the reset transistor 9 is connected to the reset wiring 12. Further, the drain of the reset transistor 9 and the drain of the amplification transistor 10 are connected to a power supply wiring 13 for supplying power. The source of the amplification transistor 10 is connected to the vertical signal line 14.

In the CMOS image sensor 1, photoelectric conversion is performed in the photodiode 2. Photoelectrons (signal charges) of the photodiode are selected by the transfer transistor 8 and transferred to the floating diffusion FD through the transfer transistor 8. Since the floating diffusion FD is connected to the amplification transistor 10, a signal corresponding to the potential of the floating diffusion FD is output to the vertical signal line 14 through the amplification transistor 10.
The reset transistor 9 resets the signal charge of the floating diffusion FD by discarding the signal charge of the floating diffusion FD to the power supply wiring 13. The horizontal wirings 11, 12, and 13 are common to the pixels 3 in the same row, and are controlled by a circuit 5 for pixel selection.

  Conventionally, a general color solid-state imaging device is provided with a color filter for each pixel and detects an optical signal having a specific wavelength transmitted through the color filter. Therefore, the light reaching the semiconductor portion that performs photoelectric conversion becomes a part of the incident light, and the signal output is reduced as compared with the case where no color filter is mounted.

  Further, in Patent Document 1, as “color separation in an active pixel imaging array using a triple well structure”, signals of three primary colors (red, green, and blue) are detected without using a color filter. A photodetector using a pn photodiode having a three-layer structure has been proposed. According to this proposal, the terminals of each photodiode are formed at the semiconductor interface, and each terminal is connected to the gate of a MOS transistor that functions as an amplifier, so that the signal of each photodiode is amplified and read out.

  FIG. 16 shows the configuration of the optical sensor in Patent Document 1. The optical sensor 31 includes an n-type semiconductor layer (well region) 33 on a p-type semiconductor substrate 32, a p-type semiconductor layer (well region) 34 thereon, and an n-type semiconductor layer (well region) 35 on the semiconductor interface thereon. And a combination of the p-type semiconductor substrate 32 and the n-type semiconductor layer 33, a combination of the n-type semiconductor layer 33 and the p-type semiconductor layer 34, and a combination of the p-type semiconductor layer 34 and the n-type semiconductor layer 35. Three photodiodes are formed, a red photodiode, a green photodiode, and a blue photodiode. The red, green, and blue photodiodes are connected to current meters 41, 42, and 43 for detecting signal charges stored in the photodiodes.

Both the CMOS image sensor and the optical sensor described above utilize the fact that the light absorption coefficient of a semiconductor depends on the wavelength of light with respect to color separation.
In the optical sensor 31 of Patent Document 1, in the same semiconductor substrate 32, the short wavelength, that is, the blue light photoelectric conversion portion (photodiode) is the uppermost layer, and the long wavelength, that is, the red light photoelectric conversion portion (photodiode) is the lowermost layer. In addition, color separation is performed by providing a photoelectric conversion portion (photodiode) of the intermediate wavelength, that is, green light in the intermediate layer. According to this, incident light can be used without waste, and the signal output can be increased as compared with a general color filter system. Also, two-dimensionally, since three types of color signals can be extracted from the same location, the color resolution can be increased as compared with a color filter system that extracts one type of color signal from one pixel. However, since the electrodes of each photodiode are shared with the electrodes of photodiodes adjacent in the depth direction, the signal voltage of one photodiode is affected by the signal voltage of another photodiode adjacent in the depth direction. As a result, it is difficult to take out the signal of each photodiode independently. Further, when resetting each photodiode, there is a problem that S / N cannot be obtained because kTC noise due to the capacitance of the photodiode is applied.

  On the other hand, in Patent Document 2, light reception in which magenta filters are stacked using a color filter in which a magenta (Mg) filter as a complementary color filter and a green (G) filter as a primary color filter are alternately arranged in a hydraulic direction and a horizontal direction is used. Red (R) and blue (B) color signals are detected by the first and second semiconductor layers separated in the depth direction of the part, and a green (G) color signal is detected by the light receiving part in which the green filters are stacked. A MOS-type color solid-state imaging device configured as described above has also been proposed. According to this, faithful color reproducibility can be achieved, and high image quality can also be achieved.

JP-T-2002-513145 JP 2004-281773 A

  By the way, in the MOS type semiconductor image sensor as described above, it is desired to further improve the resolution while suppressing color mixing and noise and having high color separation. Further, in order to increase the speed and simplification of signal processing, further simplification of an arithmetic expression for detecting a color signal is desired.

  In view of the above, the present invention provides a semiconductor image sensor in which the resolution is further improved and the arithmetic expression of the color signal is simplified.

  A semiconductor image sensor according to the present invention is a semiconductor image sensor including a plurality of pixels including a photoelectric conversion unit and a MOS transistor for selectively reading out the photoelectric conversion unit. A first pixel group arranged at a predetermined pitch in each of the vertical directions and a plurality of two-dimensionally a plurality of pixels in a state shifted from the first pixel group by about ½ pitch of the pitch in both the horizontal direction and the vertical direction. And a second pixel group in which the pixels are arranged, and a complementary color filter is used as an on-chip color filter corresponding to the first pixel group.

  In the semiconductor image sensor of the present invention, the first pixel group in which a plurality of pixels are arranged at predetermined pitches in the horizontal and vertical directions, and the first pixel group are shifted by approximately ½ pitch in the horizontal and vertical directions. By providing the second pixel group in which a plurality of pixels are arranged in the above state, the number of recorded pixels is doubled as the number of effective pixels, and the resolution can be improved. By using a complementary color filter for the on-chip color filter corresponding to the first pixel group, it is possible to simplify an arithmetic expression for forming a color signal in so-called diagonal pixel shift.

  A semiconductor image sensor according to the present invention is a semiconductor image sensor including a plurality of pixels including a photoelectric conversion unit and a MOS transistor for selectively reading out the photoelectric conversion unit. A first pixel group arranged at a predetermined pitch in each of the vertical directions and a plurality of two-dimensionally a plurality of pixels in a state shifted from the first pixel group by about ½ pitch of the pitch in both the horizontal direction and the vertical direction. And a second pixel group in which the pixels are arranged, the photoelectric conversion means of the first pixel group has a plurality of stacked structures, and two photoelectric conversion signals having different peak wavelengths are extracted from the same pixel. It is characterized by comprising.

  In the semiconductor image sensor of the present invention, the first pixel group in which a plurality of pixels are arranged at predetermined pitches in the horizontal and vertical directions, and the first pixel group are shifted by approximately ½ pitch in the horizontal and vertical directions. By providing the second pixel group in which a plurality of pixels are arranged in the above state, the number of recorded pixels is doubled as the number of effective pixels, and the resolution can be improved. Since the photoelectric conversion means of the first pixel group has a plurality of stacked structures and takes out two photoelectric conversion signals having different peak wavelengths from the same pixel, an arithmetic expression for forming a color signal in so-called oblique pixel shift Can be simplified.

  A semiconductor image sensor according to the present invention is a semiconductor image sensor including a plurality of pixels including a photoelectric conversion unit and a MOS transistor for selectively reading out the photoelectric conversion unit. A first pixel group arranged at a predetermined pitch in each of the vertical directions and a plurality of two-dimensionally a plurality of pixels in a state shifted from the first pixel group by about ½ pitch of the pitch in both the horizontal direction and the vertical direction. Each of the first pixels in the first pixel group is configured to detect the first and second color component signals of different primary colors separately, Each second pixel of the two pixel groups is configured to detect a third color component signal of a primary color different from the primary color, and an area extending between the adjacent first pixel and the second pixel is a virtual pixel, The virtual pixel And detecting a color signal a length less than one pitch of the pixels of the pixel group in the center of the first pixel and the color component signals of the second pixels included within the circle having the radius and processing.

  In the semiconductor image sensor of the present invention, the first pixel group in which a plurality of pixels are arranged at predetermined pitches in the horizontal and vertical directions, and the first pixel group are shifted by approximately ½ pitch in the horizontal and vertical directions. By providing the second pixel group in which a plurality of pixels are arranged in the above state, the number of recorded pixels is doubled as the number of effective pixels, and the resolution can be improved. The first pixel is configured to detect the first and second color component signals of different primary colors, the second pixel is configured to be able to detect the third color component signal of the primary color, and the adjacent first and second pixels A region extending over the virtual pixel is used as a virtual pixel, and calculation processing is performed from the color component signals of the first pixel and the second pixel included in a circle having a radius of less than one pitch of the pixel of the pixel group around the virtual pixel. By detecting the color signal, the arithmetic expression for forming the color signal can be simplified.

  A semiconductor image sensor according to the present invention is a semiconductor image sensor including a plurality of pixels including a photoelectric conversion unit and a MOS transistor for selectively reading out the photoelectric conversion unit. A first pixel group arranged at a predetermined pitch in each of the vertical directions and a plurality of two-dimensionally a plurality of pixels in a state shifted from the first pixel group by about ½ pitch of the pitch in both the horizontal direction and the vertical direction. And a second pixel group in which the first and second pixels are arranged, and each first pixel of the first pixel group is configured to detect the first and second signal components of different primary colors separately, Each of the second pixels in the pixel group is configured to detect a third color component signal of a primary color different from the primary color, and a region extending between the adjacent first pixel and second pixel is a virtual pixel. The virtual pixel A color signal is detected by performing arithmetic processing from the color component signals of the first pixel and the second pixel included in the circle, including a pixel in contact with a circle whose radius is the length of one pitch of the pixels of the pixel group in the center. It is characterized by that.

  In the semiconductor image sensor of the present invention, the first pixel group in which a plurality of pixels are arranged at predetermined pitches in the horizontal and vertical directions, and the first pixel group are shifted by approximately ½ pitch in the horizontal and vertical directions. By providing the second pixel group in which a plurality of pixels are arranged in the above state, the number of recorded pixels is doubled as the number of effective pixels, and the resolution can be improved. The first pixel is configured to detect the first and second color component signals of different primary colors, the second pixel is configured to be able to detect the third color component signal of the primary color, and the adjacent first and second pixels A first pixel and a second pixel included in the circle including a pixel in contact with a circle whose center is the virtual pixel and whose radius is one pitch length of the pixel group. The arithmetic expression for forming the color signal can be simplified by calculating the color signal and detecting the color signal.

  According to the semiconductor image sensor of the present invention, since the resolution can be further improved, information in the horizontal and vertical directions can be reproduced more precisely than a general square pixel arrangement, and a clearer image can be reproduced. Can do. In addition, since the arithmetic expression of the color signal can be simplified, it becomes easy to speed up and simplify the subsequent signal processing.

1 is a schematic configuration diagram showing a first embodiment of a semiconductor image sensor according to the present invention. It is sectional drawing which shows the semiconductor sectional structure of the pixel (G pixel) of the 2nd pixel group of 1st Embodiment of the semiconductor image sensor which concerns on this invention. It is sectional drawing which shows the semiconductor sectional structure of the pixel (R / B pixel) of the 1st pixel group of 1st Embodiment of the semiconductor image sensor which concerns on this invention. It is a schematic block diagram which shows 2nd Embodiment of the semiconductor image sensor which concerns on this invention. It is sectional drawing which shows the semiconductor sectional structure of the pixel (G pixel) of the 2nd pixel group of 2nd Embodiment of the semiconductor image sensor which concerns on this invention. It is sectional drawing which shows the semiconductor sectional structure of the pixel (Mg pixel) of the 1st pixel group of 2nd Embodiment of the semiconductor image sensor which concerns on this invention. FIGS. 4A and 4B are an explanatory diagram for explaining an arithmetic expression of color signals of a semiconductor image sensor according to the present invention and an explanatory diagram for explaining a pitch of pixels. It is explanatory drawing with which it uses for description of the calculation formula of the color signal of the semiconductor image sensor which concerns on this invention. It is explanatory drawing with which it uses for description of an example of the computing equation which calculates the color signal of the semiconductor image sensor which concerns on this invention. It is explanatory drawing with which it uses for description of the other example of the computing equation which calculates the color signal of the semiconductor image sensor which concerns on this invention. It is explanatory drawing with which it uses for description of the other example of the computing equation which calculates the color signal of the semiconductor image sensor which concerns on this invention. It is explanatory drawing with which it uses for description of the other example of the computing equation which calculates the color signal of the semiconductor image sensor which concerns on this invention. It is an equivalent circuit diagram which shows other embodiment of the semiconductor image sensor which concerns on this invention. It is an equivalent circuit diagram which shows other embodiment of the semiconductor image sensor which concerns on this invention. It is a principle figure of a Bayer type color filter used for explanation of calculation of a color signal of a semiconductor image sensor which has a Bayer type color filter. It is a schematic block diagram of the conventional image sensor without a filter. It is a schematic equivalent circuit diagram of a CMOS image sensor.

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

  FIG. 1 shows a schematic configuration of a semiconductor image sensor according to a first embodiment of the present invention, a so-called CMOS image sensor. The semiconductor image sensor of this embodiment is an example of color separation without using an on-chip color filter. The semiconductor image sensor 61 according to the present embodiment is formed on the surface of the same semiconductor substrate 62. The imaging region 64 is a light receiving region in which a plurality of pixels 63 are two-dimensionally arranged, and outside the imaging region 64. Peripheral circuits 65 and 66 for selecting the arranged pixels 63 and outputting signals are provided. One peripheral circuit 65 is configured by a vertical scanning circuit (so-called vertical register circuit) located on the side of the imaging region 64. The other peripheral circuit 66 is a horizontal scanning circuit (so-called horizontal register circuit) and an output circuit (including a signal amplification circuit, an A / D conversion circuit, a synchronization signal generation circuit, etc.) located below the imaging region 64. Composed.

  In the imaging region 64, a plurality of pixels are so-called diagonally arranged. That is, a first pixel group in which a plurality of pixels 63A are two-dimensionally arranged in a substantially lattice pattern with a predetermined pitch W1 in the horizontal direction and the vertical direction, and the pitch W1 in both the horizontal direction and the vertical direction with respect to the first pixel group. And a second pixel group in which a plurality of pixels 63B are two-dimensionally arranged in a state shifted by approximately a half pitch, and the pixels 63A and 63B are arranged in a square lattice pattern that is shifted diagonally. Yes. In this example, the pixels 63B are arranged in the odd rows, and the pixels 63A are arranged in the even rows with a ½ pitch shift.

  Each pixel (so-called first pixel) 63A in the first pixel group has two color component signals of different primary colors from the same pixel, in this example, red component signals corresponding to red (R) and blue (B) colors. And a pixel structure for separately detecting the blue component signal. FIG. 3 shows a semiconductor cross-sectional structure of the first pixel 63A. A second conductivity type semiconductor well region, in this example, a p-type semiconductor well region 72, is formed on a first conductivity type semiconductor substrate, in this example, an n-type silicon semiconductor substrate 71, and the surface of this p-type semiconductor well region 72 is formed. Then, an n-type semiconductor region 73 is formed as a blue light photoelectric conversion region having a required depth h1 in the depth direction, in this example, a width h1 of 0.1 μm to 0.3 μm. Further, in the deep part in the p-type semiconductor well region 72, photoelectric conversion of red light having a required width h2 in the depth direction, in this example, a depth h2 of 0.8 μm to 2.5 μm in depth from the surface. An n-type semiconductor region 74 to be a region is formed. Further, a p-type accumulation region 75 for suppressing dark current is formed so as to extend over the surface of the n-type semiconductor well region 73 and the surface of the region of the n-type semiconductor region 74 that extends partially toward the substrate surface. .

  The n-type semiconductor region 73, the p + accumulation region 75 and the p-type semiconductor well region 74 form a photodiode PDb which is a blue photoelectric conversion means. The n-type semiconductor region 74, the p-type accumulation region 75 and the p-type accumulation region 75 A photodiode PDr which is a red photoelectric conversion means is formed by the type semiconductor well region 72. These photodiodes PDb and PDr are configured as a so-called HAD sensor (Hole Accumulation Diode) in which a p-type accumulation region 75 is formed at the interface of the substrate surface.

  In the p-type semiconductor well region 72, the n-type semiconductor regions 77 and 78 serving as the floating diffusion FD are formed on both sides of the photodiodes PDb and PDr, that is, on the substrate surface side. Then, between the n-type semiconductor region 73 constituting the blue photodiode PDb on the substrate surface and the n-type semiconductor region 77 serving as one floating diffusion, that is, continuously to the n-type semiconductor region 73 on the substrate surface. A gate electrode 80 is formed on the facing channel region 79 via a gate insulating film, and a transfer transistor Tr1b is formed. Further, between the n-type semiconductor region 74 constituting the red photodiode PDr on the substrate surface and the n-type semiconductor region 78 serving as the other floating diffusion, that is, continuous to the n-type semiconductor region 74 on the substrate surface. A gate electrode 82 is formed on the facing channel region 81 via a gate insulating film, and a transfer transistor Tr1r is formed. The pixel 63A is separated from adjacent pixels by an element isolation region 84 formed in the p-type semiconductor well region 72, for example, an insulating isolation region by a selective oxide layer or a trench isolation region.

  The pixel 63A has a structure in which two photodiodes PDb and PDr that separate and photoelectrically convert blue light and red light are stacked in one pixel structure. According to this pixel 63A, the blue light having a short wavelength is photoelectrically converted by the photodiode PDb formed in the shallow region, and signal charges are accumulated in the shallow n-type semiconductor region 73. At the time of signal reading, the transfer transistor Tr1b is turned on, and the signal charge is transferred through the channel region 79 under one gate electrode 80 to the n-type semiconductor region 77 which becomes one floating diffusion. Red light having a long wavelength is photoelectrically converted by the photodiode PDr formed in a deep region, and signal charges are accumulated in the n-type semiconductor region 74 at a deep position. At the time of signal reading, the transfer transistor Tr1r is turned on, and the signal charge is transferred through the channel region 81 below the other gate electrode 82 to the n-type semiconductor region 78 serving as the other floating diffusion.

  Each pixel (so-called second pixel) 63B of the second pixel group has a pixel structure in which a color component signal of a different primary color, in this example, a green component signal corresponding to green (G) color is detected. The FIG. 2 shows a semiconductor cross-sectional structure of the second pixel 63B. A second conductivity type p-type semiconductor well region 72 is formed in the same n-type silicon semiconductor substrate 71 as described above. Corresponding to the required depth h3 of the p-type semiconductor well region 72, that is, the intermediate position between the n-type semiconductor region 73 and the n-type semiconductor region 74 described in FIG. In other words, an n-type semiconductor region 86 serving as a photoelectric conversion region for green light having a width h3 in the depth direction of 0.3 μm to 0/8 μm is formed. A p-type accumulation region 87 for suppressing dark current is formed in a region from the n-type semiconductor region 86 to the substrate surface.

  From the n-type semiconductor region 86, the p-type accumulation region 87, and the p-type semiconductor well regions 72 and 5, a photodiode PDg which is a green photoelectric conversion means is formed. This photodiode PDg is configured as a so-called HAD sensor in which a p-type accumulation region 87 is formed at the interface of the substrate surface.

  In the p-type semiconductor well region 72, an n-type semiconductor region 88 to be a floating diffusion FD is formed. Then, between the n-type semiconductor region 86 constituting the green photodiode PDg and the n-type semiconductor region 88 that becomes a floating diffusion on the substrate surface, that is, the n-type semiconductor region 86 continues to face the substrate surface. A gate electrode 90 is formed on the region 89 via a gate insulating film, and a transfer transistor Tr1g is formed. The pixel 63B is separated from adjacent pixels by the element isolation region 84 similar to the above.

  According to the second pixel 63B, the green light is photoelectrically converted by the photodiode PDg formed in the intermediate depth region, and signal charges are accumulated in the n-type semiconductor region 86. At the time of signal reading, the transfer transistor Tr1g is turned on and the signal charge is transferred through the channel region 89 under the gate electrode 90 to the n-type semiconductor region 88 that becomes floating diffusion.

  According to the semiconductor image sensor 61 of the first embodiment, a first pixel group in which a plurality of first pixels 63A are arranged in a substantially lattice shape at a predetermined pitch W1 in the horizontal and vertical directions, and the first pixel group In other words, a so-called diagonal pixel shift arrangement is used, which includes a second pixel group in which a plurality of second pixels 63B are arranged two-dimensionally in a state where the pitch W1 is shifted by about 1/2 of the pitch W1 in both the horizontal and vertical directions. Therefore, the number of recorded pixels (corresponding to the number of virtual pixels described later) can be doubled as the number of effective pixels (corresponding to the number of so-called physical pixels). That is, since the pixel size per pixel can be increased, the resolution can be improved by increasing the number of recording pixels without lowering the saturation charge amount (dynamic range) and sensitivity.

  The human retina is considered to be sensitive in the vertical and horizontal directions and inferior to the vertical and horizontal directions in the oblique direction. If the number of pixels is the same, the diagonal pixel arrangement can increase the number of recorded pixels in the vertical and horizontal directions. Therefore, information in the vertical and horizontal directions can be reproduced more precisely than a general square pixel arrangement, and a clearer image can be reproduced.

In addition, since the photodiodes PDb, PDr, and PDg constituting the pixels 63A and 63B have a HAD structure, generation of noise can be remarkably suppressed. In particular, even if the reset operation is performed in the pixel 63A, noise does not occur in the n-type semiconductor regions 74 and 73 that are red (R) and blue (B) charge storage regions. Therefore, S / N is improved. In the pixels 63A and 63B, the positions in the depth direction of the n-type semiconductor regions 73, 74, and 86 that contribute to the photoelectric conversion of light of different colors are different depending on the wavelength. It is suppressed and the color separation property is increased.
Even in a configuration without an on-chip color filter, color separation is possible, cost can be reduced, and sensitivity can be improved.

  Further, since the first pixel 63A for red / blue and the second pixel 63B for green are obliquely arranged, calculation for calculating R, G, B color signals from pixels to be imaged as will be described later. Since the equation can be simplified, it is easy to speed up and simplify the subsequent signal processing.

  FIG. 4 shows a schematic configuration of a semiconductor image sensor according to the second embodiment of the present invention, a so-called CMOS image sensor. The semiconductor image sensor of this embodiment is an example in which color separation is performed using an on-chip color filter composed of complementary colors and primary colors. As the on-chip color filter of this example, two colors of on-chip color filters, a magenta (Mg) filter that is a complementary color filter and a green (G) filter that is a primary color filter, are used. In the second embodiment, portions corresponding to those in the first embodiment in FIGS. 1 to 3 are described with the same reference numerals.

  The semiconductor image sensor 100 according to the present embodiment is configured in the same manner as in FIG. 1 except for the on-chip color filter. That is, the imaging region 64 formed on the surface of the same semiconductor substrate 62 as a light receiving region in which a plurality of pixels 103 are two-dimensionally arranged, and the selection and signal output of the pixels 103 arranged outside the imaging region 64 Peripheral circuits 65 and 66 are provided. One peripheral circuit 65 is configured by a vertical scanning circuit (so-called vertical register circuit) located on the side of the imaging region 64. The other peripheral circuit 66 is a horizontal scanning circuit (so-called horizontal register circuit) and an output circuit (including a signal amplification circuit, an A / D conversion circuit, a synchronization signal generation circuit, etc.) located below the imaging region 64. Composed.

  In the imaging region 64, the plurality of pixels 103 are so-called diagonally arranged. That is, a first pixel group in which a plurality of pixels 103A are two-dimensionally arranged in a substantially grid pattern with a predetermined pitch W1 in the horizontal direction and the vertical direction, and the pitch W1 in both the horizontal direction and the vertical direction with respect to the first pixel group. And a second pixel group in which a plurality of pixels 103B are arranged two-dimensionally in a state shifted by approximately a half pitch, and the pixels 1063A and 103B are arranged in a square lattice pattern that is shifted diagonally. Yes. On the imaging region 64, a magenta (Mg) filter that is a complementary color filter corresponds to each pixel 103A of the first pixel group, and a green (G) filter that is a primary color filter corresponds to each pixel 103B of the second pixel group. Correspondingly, two-color on-chip filters of Mg filter and G filter are arranged. In this example, the pixels 103B are arranged in the odd rows, and the pixels 103A are arranged in the even rows with a ½ pitch shift.

  Each pixel (A so-called first pixel: hereinafter referred to as Mg pixel) 103A in the first pixel group has two color component signals of different primary colors from the same pixel, that is, in this example, red (R) and blue (B) colors. And a pixel structure for separately detecting and detecting the red component signal and the blue component signal corresponding to. FIG. 6 shows a semiconductor cross-sectional structure of the Mg pixel 103A. Similar to the configuration of FIG. 3 described above, a second conductivity type semiconductor well region, in this example p-type semiconductor well region 72, is formed in a first conductivity type semiconductor substrate, in this example n-type silicon semiconductor substrate 71. On the surface of the p-type semiconductor well region 72, an n-type semiconductor region 73 is formed as a blue light photoelectric conversion region having a required width h1 in the depth direction, in this example, a width h1 of 0.1 μm to 0.8 μm. Is done. Further, in the deep part in the p-type semiconductor well region 72, photoelectric conversion of red light having a required width h2 in the depth direction, in this example, a depth h2 of 0.8 μm to 2.5 μm in depth from the surface. An n-type semiconductor region 74 to be a region is formed. Further, a p-type accumulation region 75 for suppressing dark current is formed so as to extend over the surface of the n-type semiconductor well region 73 and the surface of the region of the n-type semiconductor region 74 that extends partially toward the substrate surface. .

  The n-type semiconductor region 73, the p + accumulation region 75, and the p-type semiconductor well region 74 form a photodiode PDb that serves as a blue photoelectric conversion means. The n-type semiconductor region 74, the p-type accumulation region 75, and the p-type accumulation region 75 A photodiode PDr serving as a red photoelectric conversion means is formed by the type semiconductor well region 72. These photodiodes PDb and PDr are configured as so-called HAD sensors in which a p-type accumulation region 75 is formed at the interface of the substrate surface.

  In the p-type semiconductor well region 72, the n-type semiconductor regions 77 and 78 serving as the floating diffusion FD are formed on both sides of the photodiodes PDb and PDr, that is, on the substrate surface side. Then, between the n-type semiconductor region 73 constituting the blue photodiode PDb on the substrate surface and the n-type semiconductor region 77 serving as one floating diffusion, that is, continuously to the n-type semiconductor region 73 on the substrate surface. A gate electrode 80 is formed on the facing channel region 79 via a gate insulating film, and a transfer transistor Tr1b is formed. Further, between the n-type semiconductor region 74 constituting the red photodiode PDr on the substrate surface and the n-type semiconductor region 78 serving as the other floating diffusion, that is, continuous to the n-type semiconductor region 74 on the substrate surface. A gate electrode 82 is formed on the facing channel region 81 via a gate insulating film, and a transfer transistor Tr1r is formed. The Mg pixel 103 </ b> A is separated from adjacent pixels by an element isolation region 84 formed by, for example, a selective oxide layer or a trench isolation region formed in the p-type semiconductor well region 72.

  Then, a magenta (Mg) filter 105 of the on-chip color filters is formed on each Mg pixel 102A via a planarizing film.

  The Mg pixel 103A has a structure in which two photodiodes PDb and PDr that separate and photoelectrically convert blue light and red light incident through the Mg filter 105 are stacked in one pixel structure. ing. According to the Mg pixel 103A, the blue light having a short wavelength transmitted through the Mg filter 105 is photoelectrically converted by the photodiode PDb formed in the shallow region, and signal charges are accumulated in the shallow n-type semiconductor region 73. At the time of signal reading, the transfer transistor Tr1b is turned on, and the signal charge is transferred through the channel region 79 under one gate electrode 80 to the n-type semiconductor region 77 which becomes one floating diffusion. Further, red light having a long wavelength transmitted through the Mg filter 105 is photoelectrically converted by the photodiode PDr formed in a deep region, and signal charges are accumulated in the n-type semiconductor region 74 at a deep position. At the time of signal reading, the transfer transistor Tr1r is turned on, and the signal charge is transferred through the channel region 81 below the other gate electrode 82 to the n-type semiconductor region 78 serving as the other floating diffusion.

  Each pixel in the second pixel group (so-called second pixel: hereinafter referred to as G pixel) 103B has a pixel structure for detecting a color component signal of a different primary color, that is, a green component signal corresponding to green (G) in this example. It is comprised. FIG. 5 shows a semiconductor cross-sectional structure of the second pixel 63B. Similar to the configuration of FIG. 2 described above, a second conductivity type p-type semiconductor well region 72 is formed in the same n-type silicon semiconductor substrate 71. Corresponding to the required depth h3 of the p-type semiconductor well region 72, that is, the intermediate position between the n-type semiconductor region 73 and the n-type semiconductor region 74 described in FIG. In other words, an n-type semiconductor region 86 serving as a photoelectric conversion region for green light having a width h3 in the depth direction of 0.1 μm to 2.5 μm is formed. A p-type accumulation region 87 for suppressing dark current is formed in a region from the n-type semiconductor region 86 to the substrate surface.

  From the n-type semiconductor region 86, the p-type accumulation region 87, and the p-type semiconductor well regions 72 and 5, a photodiode PDg serving as a green photoelectric conversion means is formed. This photodiode PDg is configured as a so-called HAD sensor in which a p-type accumulation region 87 is formed at the interface of the substrate surface.

  In the p-type semiconductor well region 72, an n-type semiconductor region 88 to be a floating diffusion FD is formed. Then, between the n-type semiconductor region 86 constituting the green photodiode PDg and the n-type semiconductor region 88 that becomes a floating diffusion on the substrate surface, that is, the n-type semiconductor region 86 continues to face the substrate surface. A gate electrode 90 is formed on the region 89 via a gate insulating film, and a transfer transistor Tr1g is formed. The G pixel 103B is separated from adjacent pixels by the same element isolation region 84 as described above.

  Then, a green (G) filter 106 of the on-chip color filters is formed on each G pixel 103B via a planarizing film.

  According to the G pixel 103B, the green light transmitted through the G filter 106 is photoelectrically converted by the photodiode PDg formed in the intermediate depth region, and signal charges are accumulated in the n-type semiconductor region 86. At the time of signal reading, the transfer transistor Tr1g is turned on and the signal charge is transferred through the channel region 89 under the gate electrode 90 to the n-type semiconductor region 88 that becomes floating diffusion.

  According to the semiconductor image sensor 100 of the second embodiment, the same effects as those of the first embodiment described above can be obtained. That is, a first pixel group in which a plurality of Mg pixels 103A are arranged in a substantially lattice shape at a predetermined pitch W1 in the horizontal and vertical directions, and about 1/2 of the pitch W1 in the horizontal and vertical directions with respect to the first pixel group. Since the so-called diagonal pixel shift arrangement is used, which includes a second pixel group in which a plurality of G pixels 103B are arranged two-dimensionally in a state where the pitch is shifted, the number of recorded pixels (corresponding to the number of virtual pixels described later) The number of effective pixels (corresponding to the number of so-called physical pixels) can be doubled. That is, since the pixel size per pixel can be increased, the resolution can be improved by increasing the number of recording pixels without lowering the saturation charge amount (dynamic range) and sensitivity. Therefore, information in the vertical and horizontal directions can be reproduced more precisely than a general square pixel arrangement, and a clearer image can be reproduced.

  In addition, since the photodiodes PDb, PDr, and PDg constituting the Mg pixel 103A and the G pixel 103B have the HAD structure, generation of noise can be remarkably suppressed. In particular, in the Mg pixel 103A, even if the reset operation is performed, noise does not occur in the n-type semiconductor regions 74 and 73 that are red (R) and blue (B) charge storage regions. Therefore, S / N is improved. In the Mg pixel 103A and the G pixel 103B, the positions in the depth direction of the n-type semiconductor regions 73, 74, and 86 that contribute to photoelectric conversion of light of different colors are different depending on the wavelength. Color mixing is suppressed, and color separation is improved.

  Two color on-chip color filters, magenta (Mg), which is a complementary color filter, and green (G), which is a primary color filter, are used. G filters are alternately arranged in odd rows, and Mg filters are alternately arranged in even rows. . By using this on-chip color filter, color mixing can be reduced. Furthermore, by using a multilayer photodiode using a magenta (Mg) complementary color filter as a blue and red detection sensor, it is possible to suppress light from being absorbed by the filter than in the case of a photodiode using a primary color filter. Because it can, sensitivity does not decrease. The cross-sectional structure of the sensor to be detected, that is, the photodiode, is substantially the same as that of the first embodiment without the on-chip color filter. However, since the on-chip color filter is used, the width of the n-type semiconductor region is increased in width. Roughly designed. For example, an n-type semiconductor region serving as a photoelectric conversion region of a green (G) photodiode can be formed with a width of 0.1 μm to 2.5 μm, and an n-type semiconductor region serving as a photoelectric conversion region of a blue (B) photodiode is The n-type semiconductor region which can be formed with a width of 0.1 μm to 0.8 μm and becomes the photoelectric conversion region of the red (R) photodiode can be formed with a width of 0.8 μm to 2.5 μm.

  Further, since the Mg pixel 103A and the G pixel 103B are obliquely arranged, the arithmetic expression for calculating the R, G, B color signals from the pixel to be imaged can be simplified as will be described later. It is easy to speed up and simplify the processing.

  Next, an arithmetic expression for calculating R, G, B color signals applied to the semiconductor image sensor according to the first embodiment and the second embodiment will be described with reference to FIGS. . Although this arithmetic expression is applied to the semiconductor image sensor of the second embodiment using Mg pixels and G pixels, it can be similarly applied to the first embodiment.

Prior to the description of the arithmetic expression according to the present invention, in order to facilitate understanding of the arithmetic expression of the present invention, the arithmetic expression in the case of the Bayer array that has been widely used conventionally will be described with reference to FIG. .
FIG. 15 shows a main part of a semiconductor image sensor using a Bayer type color filter. The principle of this Bayer type color filter is shown. The basic layout of the color filter starts with a red filter at the top left of the image sensor. For the color information, a virtual pixel 52 is provided in the middle surrounded by four pixels 51 [51R, 51Ga, 51Gb, 51B]. The four filtered pixels 51 around the virtual pixel 52 are two pixels (51Ga, 51Gb) for green (G), one for red (R) pixel (51R), and one for blue (B) pixel (51B). Consists of. It can be seen that the proportion of green is large. This is because humans have good visibility against green and the filter is also balanced by giving extra light.

Attention is paid to the central virtual pixel Px which is the virtual pixel 52 in FIG. The color information PxG, PxR, PxB of the virtual pixel Px is obtained by matrix calculation from the color information of neighboring pixels. Green (G) is divided by 2 by adding the green pixel G7 and pixel _{G 10} in the vicinity. Red (R) is also considered the red information around it in preference to the red pixel R _11. Blue (B) is the same as red. In this way, the color of an arbitrary cell is specified. The calculation formula is as follows.

PxG = (G ₇ + G ₁₀ ) / 2
PxR = (9R ₁₁ + 3R ₃ + 3R ₉ + R ₁ ) / 16
PxB = (9B ₆ + 3B ₈ + 3B ₁₄ + B ₁₆ ) / 16

Here, the details of the above calculation of PxR and PxB are as follows.
For PxR, it is taken into consideration with the red pixel _{R 11,} blue pixels _{B 6,} red information of the red pixels around each of the green pixel _{G 7, G 10.}
R ₁₁
Ra (red information around the pixel B6) = (R ₁ + R ₃ + R ₉ + R ₁₁ ) / 4
Rb (red information around the pixel _{G 7) = (R 3 +} R 11) / 2
Rc (red information around the pixel _{G 10) = (R 9 +} R 11/2
∴PxR = Ra + Rb + Rc + R ₁₁ ) / 4
= [{(R ₁ + R ₃ + R ₉ + R ₁₁ ) / 4}
+ {(R ₃ + R ₁₁ ) / 2}
+ {(R ₉ + R ₁₁ ) / 2}
+ R ₁₁ ] / 4
= (R ₁ + R ₃ + R ₉ + R ₁₁ + 2R ₃ + 2R ₁₁ + 2R ₉ + 2R ₁₁
+ 4R ₁₁ ) / 16
= (9R ₁₁ + 3R ₃ + 3R ₉ + R ₁ ) / 16

For PxB, blue information of the blue pixel B ₆ and the blue pixels around each of the green pixels G ₇ and G ₁₀ is added.
B ₆
Ba (blue information around the pixel _{R 11) = (B 6 +} B 8 + B 14 + B 16) / 4
Bb (blue information about the pixel _{G 7) = (B 6 +} B 8) / 2
Bc (blue information around the pixel _{G 10) = (B 6 +} B 14) / 2
∴PxB = (Ba + Bb + Bc + B ₆ ) / 4
= [{(B ₆ + B ₈ + B ₁₃ + B ₁₆ ) / 4}
_{+ {(2B 6 + 2B 8} /2} / 4
+ {(2B ₆ + 2B ₁₄ ) / 4
+ 4B ₆ ] / 16
= (B ₆ + B ₈ + B ₁₄ + B ₁₆ + 2B ₆ + 2B ₈ + 2B ₆ + 2B ₁₄
+ 4B ₆ ) / 16
= (9B ₆ + 3B ₈ + 3B ₁₄ + B ₁₆ ) / 16
This calculation method is an epoch-making method that allows easy coloring.

  However, in the case of oblique pixel shifting, a two-step calculation is performed in which a virtual pixel is first virtualized at the intersection of the pixels, and the final virtual pixel is calculated at the center of four pixels using the virtual pixel. At this time, the R, G, and B pixels are composed of columns in which the G column and RB are alternately arranged, and the calculation formula differs depending on whether the virtual pixel touches the R pixel or the B pixel. It becomes complicated.

Next, an arithmetic expression according to the present embodiment will be described.
As shown in FIG. 9 to FIG. 12, when the pixel is shifted diagonally by the Mg pixel [pixel Mg11, Mg12, Mg21, Mg22] 103A and the G pixel [pixel G11, G12, G21, G22] 103B, the adjacent Mg pixel 103A And G pixels 103B are defined as virtual pixels (corresponding to recording pixels) P1 to P9. Now, the green (G), red (R), and blue (B) color signals G _P5 , R _P5 , and B _P5 of the virtual pixel _P5 are calculated. G ₁₁ ~G ₂₂ green component signal of each pixel _G11~G22, Mg 11R ~Mg _22R red component signal of each pixel _Mg11~Mg22, Mg 11B ~Mg _22B are respectively a blue component signal of the pixel Mg11～Mg22.

  9A and 9B, as shown in FIGS. 7A and 7B, respectively, when the length of one side of one square pixel (physical pixel) 103 is set to one pitch p1, the center of the virtual pixel P5 (pixel This is an example of calculating a signal of a pixel included in a circle C1 having a radius R1 less than 1 pitch p1 from the boundary between Mg11 and the pixel G22.

FIG. 9 shows a first embodiment of the arithmetic expression. The green (G), red (R), and blue (B) color signals G _P5 , R _P5 , and B _P5 of the virtual pixel P5 are obtained by an arithmetic expression represented by the following equation (1).
[Equation 1]
G _P5 = (G ₂₂ + Mg _11G ) / 2
Mg _11G = (G ₂₁ + G ₁₂ ) / 2
G _P5 = {G ₂₂ + (G ₂₁ + G ₁₂ ) / 2} / 2
∴G _P5 = (G ₂₁ + G ₁₂ + 2G ₂₂ ) / 4

R _P5 = (Mg _11R + G _22R ) / 2
G _22R = (Mg _21R + Mg _12R ) / 2
R _P5 = {g _11R + (Mg _21R + Mg _12R ) / 2} / 2
∴R _P5 = (2Mg _11R + Mg _21R + Mg _12R ) / 4

B _P5 = (Mg _11B + G _22B ) / 2
G _22B = (Mg _21B + Mg _12B ) / 2
B _P5 = {Mg _11B + (Mg _21B + Mg _12B ) / 2} / 2
∴B _P5 = (2Mg _11B + Mg _21B + Mg _12B ) / 4

FIG. 10 shows a second embodiment of the arithmetic expression. In order to perform the calculation of the virtual pixel P5, first, the color component signals of the virtual pixels K1, K2 shifted diagonally are obtained, and the color signals G _P5 , R _P5 , B _P5 of the virtual pixel P5 are obtained from the virtual pixels K1, K2. Ask. That is, it is obtained by an arithmetic expression shown by the following formula 2. G _K1 , R _K1 , and B _K1 are the green component signal, red component signal, and blue component signal of the virtual pixel K1, respectively. G _K2 , R _K2 , and B _K2 are the green component signal, red component signal, and blue component of the virtual pixel K2, respectively. Signal.

[Equation 2]
G _K1 = (G ₂₁ + G ₂₂ ) / 2
G _K2 = (G ₁₂ + G ₂₂ ) / 2
∴G _P5 = (G _K1 + G _K2 ) / 2

R _K1 = (Mg _11R + Mg _21R ) / 2
R _K2 = (Mg _11R + Mg _12R ) / 2
∴R _P5 = (R _K1 + R _K2 ) / 2

B _K1 = (Mg _11B + Mg _21B ) / 2
B _K2 = (Mg _11B + Mg _12B ) / 2
∴B _P5 = (B _K1 + B _K2 ) / 2

Equation 1 can be calculated by Equations G _P5 , R _P5 , and B _P5 of Equation 2.
G _P5 = (G _K1 + G _K2 ) / 2
= [{(G ₂₁ + G ₂₂ ) / 2} + {(G ₁₂ + G ₂₂ ) / 2}] / 2
_{= (G 21 + G 12 +} 2G 22) / 4
R _P5 = (R _K1 + R _K2 ) / 2
= [{( _Mg11R + _Mg21R ) / 2} + {( _Mg11R + _Mg12R ) / 2}] / 2
_{= (2Mg 11R + Mg 21R +} Mg 12R) / 4
B _P5 = (B _K1 + B _K2 ) / 2
= [{( _Mg11B + _Mg21B ) / 2} + {( _Mg11B + _Mg12B ) / 2}] / 2
_{= (2Mg 11B + Mg 21B +} Mg 12B) / 4

  As described above, in this embodiment, since the R signal and the B signal can be extracted from the same pixel, the arithmetic expressions of the R, G, and B color signals can be simplified as shown in Equations 1 and 2. Can do. The arithmetic expression of Expression 2 can further simplify the calculation than Expression 1.

  11 and 12, respectively, as shown in FIG. 8, the signals of pixels in contact with a circle C2 having a radius R2 of 1 pitch p1 from the center of the virtual pixel P5 (on the boundary line between the pixels Mg11 and G22), That is, this is an example of calculating the signal of the pixels in the circle C2 including the pixels in contact with the circle C2 having the radius R2.

FIG. 11 shows a third embodiment of the arithmetic expression. The green (G), red (R), and blue (B) color signals G _P5 , R _P5 , and B _P5 of the virtual pixel P5 are obtained by an arithmetic expression represented by the following equation (3).
[Equation 3]
G _P5 = {G ₂₂ + Mg _11G + (G ₁₂ + G ₂₂ ) / 2 + (G ₂₁ + G ₂₂ ) / 2} / 4
Mg _11G = (G ₁₁ + G ₂₁ + G ₂₁ + G ₂₂ ) / 4
G _P5 = [G ₂₂ + {(G ₁₁ + G ₂₁ + G ₂₁ + G ₂₂ ) / 4}
+ {(2G ₁₂ + 2G ₂₂ ) / 4} + {(2 ₂₁ + 2G ₂₂ ) / 4}] / 4
∴G _P5 = (G ₁₁ + 3G ₂₁ + 3G ₁₂ + 9G ₂₂ ) / 16

R _P5 = (Mg _11R + G _22R + G _12R + G _21R ) / 4
G _22R = (Mg _11R + Mg _21R + Mg _12R + Mg _22R ) / 4
G _12R = (Mg _11R + Mg _12R ) / 2
G _21R = (Mg _11R + Mg _21R ) / 2
_{R P5 = [Mg11R + {(} Mg 11R + Mg 21R + Mg 12R + Mg 22R) / 4}
+ {(Mg _11R + Mg _12R ) / 2}
+ {(Mg _11R + Mg _21R ) / 2}] / 4
∴R _P5 = (9Mg _11R + 3Mg _21R + 3Mg _12R + Mg _22R ) / 16

_BP5 = ( _Mg11B + _G22B + _G12B + _G21B ) / 4
G _22B = (Mg _11B + Mg _21B + Mg _12B + Mg _22B ) / 4
G _12B = (Mg _11B + Mg _12B ) / 2
G _21B (Mg _11B + Mg _21B ) / 2
_{B P5 = [Mg11B + {(} Mg 11B + Mg 21B + Mg 12B + Mg 22B) / 4}
+ {(Mg _11B + Mg _12B ) / 2}
+ {(Mg _11B + Mg _21B ) / 2}] / 4
∴B _P5 = (9Mg _11B + 3Mg _21B + 3Mg _12B + Mg _22B ) / 16

FIG. 12 shows a fourth embodiment of the arithmetic expression. In order to perform the calculation of the virtual pixel P5, first, color component signals of the virtual pixels K1, K2 shifted diagonally are obtained, and the color signal G _{P5 of the} virtual pixel P5 is obtained from the virtual pixels K1, K2, the pixel Mg11, and the pixel G22. , R _P5 and B _P5 are obtained. That is, it is obtained by an arithmetic expression shown by the following expression 4. Mg _11G is a green component signal obtained from pixels G _{11 to} G ₂₂ in contact with the pixel Mg ₁₁ as a virtual pixel, and G _22R and G _22B are pixels when the pixel G ₂₂ is a virtual pixel. It is a red component signal and a blue component signal obtained from the pixels Mg _{11 to} Mg ₂₂ in contact therewith.

[Equation 4]
G _K1 = (G ₂₁ + G ₂₂ ) / 2
G _K2 = (G ₁₂ + G ₂₂ ) / 2
Mg _11G = (G ₁₁ + G ₁₂ + G ₂₁ + G ₂₂ ) / 4
G ₂₂
∴G _P5 = (G _K1 + G _K2 + Mg _11G + G ₂₂ ) / 4

R _K1 = (Mg _11R + Mg _21R ) / 2
R _K2 = (Mg _11R + Mg _12R ) / 2
G _22R = (Mg _11R + Mg _12R + Mg _21R + Mg _22R ) / 4
Mg _11R
∴R _P5 = (R _K1 + R _K2 + Mg _11R + G _22R ) / 4

B _K1 = (Mg _11B + Mg _21B ) / 2
B _K2 = (Mg _11B + Mg _12B ) / 2
G _22B = (Mg _11B + Mg _12B + Mg _21B + Mg _22B ) / 4
Mg _11B
∴B _P5 = (B _K1 + B _K2 + Mg _11B + G _22B ) / 4

Equation 3 can be calculated using Equations G _P5 , R _P5 , and B _P5 of Equation 4.
G _P5 = (G _K1 + G _K2 + Mg _11G + G ₂₂ ) / 4
= [{(G ₂₁ + G ₂₂ ) / 2} + {(G ₁₂ + G ₂₂ ) / 2}
_{+ {(G 11 + 12 +} G 21 + G 22) / 4} + G 22 ] / 4
= (G ₁₁ + 3G ₂₁ + 3G ₁₂ + 9G ₂₂ ) / 16
R _P5 = (R _K1 + R _K2 + Mg _11R + G _22R ) / 4
= [{(Mg _11R + Mg _21R ) / 2} + {(Mg _11R + Mg _12R ) / 2}
+ Mg _11R + {(Mg _11R + Mg _12R + Mg _21R + Mg _22R ) / 4}] / 4
_{= (9Mg 11R + 3Mg 21R +} 3Mg 12R + Mg 22R) / 16
B _P5 = (B _K1 + B _K2 + Mg _11B + G _22B ) / 4
= [{(Mg _11B + Mg _21B ) / 2] + {(Mg _11B + Mg _12B ) / 2}
+ Mg _11B + {(Mg _11B + Mg _12B + Mg _21B + Mg _22B ) / 4}
_{= (9Mg 11B + 3Mg 21B +} 3Mg 12B + Mg 22B) / 16

  As described above, in this embodiment, since the R signal and the B signal can be extracted from the same pixel, the arithmetic expressions of the R, G, and B color signals can be simplified as shown in Equations 3 and 4. Can do. The calculation formula of Formula 4 can further simplify the calculation from Formula 3.

  On the other hand, the semiconductor image sensor according to the present invention can share the MOS transistors constituting the pixels in the plurality of pixels of the first pixel group and the second pixel group. That is, in the case of four MOS transistors, for example, the selection transistor, the reset transistor, the amplification transistor, or all of these transistors can be shared. In the case of three MOS transistors, each of the reset transistor and the amplification transistor, or all of these transistors can be shared.

The embodiment is shown in FIG. 13 and FIG. FIG. 13 shows an embodiment in which an amplification transistor, a reset transistor, and a selection transistor are shared by two pixels, the Mg pixel and the G pixel. In the present embodiment, the sources of the transfer transistors Tr1g, Tr1r, Tr1b corresponding to the green photodiode PDg by the G pixel and the red photodiode PDr and the blue photodiode PDb by the Mg pixel are respectively connected. The Transfer wirings 111 (G), 111 (MgR), and 111 (MgB) are connected to the gates of the transfer transistors Tr1g, Tr1r, and Tr1b. The drains of the transfer transistors Tr1g, Tr1r, Tr1b are connected in common and connected to one reset transistor Tr _RES , and a so-called floating diffusion FD between the drain of the transfer transistor and the source of the reset transistor is amplified by one. Connected to the gate of the transistor Tr _AMP . The drain of the reset transistor Tr _{RES and} the drain of the amplification transistor Tr _AMP are connected to the power supply wiring 112 (Vcc). The gate of the reset transistor Tr _RES is connected to the reset wiring 113. Further, the source of the amplification transistor Tr _AMP is connected to the drain of one selection transistor Tr _SEL . The source of the selection transistor Tr _SEL is connected to the vertical signal line 115 and the selection wiring 114 is connected to the gate thereof.

Thus, by sharing the reset transistor Tr _RES , the amplification transistor Tr _AMP and the selection transistor Tr _SEL with respect to the two pixels of the Mg pixel and the G pixel, the Mg pixel and the G pixel described above are arranged obliquely. In addition to the effect, the area of the MOS transistor of the pixel can be reduced. Accordingly, since the photodiode area (opening area of the light receiving portion) can be further increased, the saturation charge amount and sensitivity can be further improved.

FIG. 14 shows an embodiment in which an amplification transistor, a reset transistor, and a selection transistor are shared by a total of four pixels including two Mg pixels and two G pixels. In the present embodiment, transfer transistors respectively corresponding to green photodiodes PDg1 and PDg2 using two G pixels, and red photodiodes PDr1 and PDr2 and blue photodiodes PDb1 and PDb2 using two Mg pixels, respectively. The sources of Tr1g, Tr2g, Tr1r, Tr2r, Tr1b, Tr2b are connected. In this case, there are six transfer transistors. Transfer wirings 111G1, 111G2, 111MgR1, 111MgR2, 111MgB1, 111MgB2 are connected to the gates of the transfer transistors Tr1g, Tr2g, Tr1r, Tr2r, Tr1b, Tr2b. The drains of the transfer transistors Tr1g, Tr2g, Tr1r, Tr2r, Tr1b, Tr2b are connected in common and connected to one reset transistor Tr _SEL , and so-called floating between the drain of the transfer transistor and the source of the reset transistor Tr _RES. The diffusion FD is connected to the gate of one amplification transistor Tr _AMP . The drain of the reset transistor Tr _{RES and} the drain of the amplification transistor Tr _AMP are connected to the power supply wiring 112 (Vcc). The gate of the reset transistor Tr _RES is connected to the reset wiring 113. Further, the source of the amplification transistor Tr _AMP is connected to the drain of one selection transistor Tr _SEL . The source of the selection transistor Tr _SEL is connected to the vertical signal line 115 and the selection wiring 114 is connected to the gate thereof.

In this way, by sharing the reset transistor Tr _RES , the amplification transistor Tr _AMP, and the selection transistor Tr _SEL with respect to a total of four pixels, that is, two Mg pixels and two G pixels, the above-described Mg pixel and G pixel are slanted. In addition to the effect of the shifted arrangement, the area of the MOS transistor of the pixel can be further reduced. Accordingly, since the photodiode area (opening area of the light receiving portion) can be further increased, the saturation charge amount and sensitivity can be further improved.

  13 and 14 show the case where four pixel transistors are provided as the pixel transistors, but the present invention can also be applied to the case where the pixel transistors shown in FIG. 17 have three MOS transistors. In that case, the reset transistor, the amplification transistor, and the like can be shared.

  As described above, according to the present embodiment, a 2 × 2 spectral structure (that is, a structure in which the G detection sensor and the B and R detection sensors are separated and B and R can be detected from the same pixel) is used. In addition, since the oblique pixel shift is used, higher sensitivity and higher resolution can be achieved. B and R can be detected using the HAD structure, and noise can be reduced. By sharing a MOS transistor for reading a signal for a plurality of pixels, the aperture ratio of the light receiving portion can be improved. An arithmetic expression for obtaining the R, G, B color signals of the virtual pixel can be simplified. Even if the on-chip color filter of the first embodiment is not provided, color separation is possible, the cost can be reduced, and the sensitivity can be improved.

  In the above example, the green (G) filter is used as the primary color filter and the magenta (Mg) filter is used as the complementary color filter. However, the present invention can be applied to other combinations of primary color filters and complementary color filters.

  61, 100... Semiconductor image sensor, 63 [63A, 63B], 103 [103A, 103B] .. Pixel, 64, 65 .. Peripheral circuit, 71... N-type semiconductor substrate, 72. 73, 74, 86,... N-type semiconductor region, PDb, blue photodiode, PDr, red photodiode, PDg, green photodiode, 75, 87, p-type accumulation region, 105・ ・ Magenta (Mg) filter, 106 ・ ・ Green (G) filter

Claims (7)

A semiconductor image sensor comprising a plurality of pixels including a photoelectric conversion means and a MOS transistor for selectively reading out the photoelectric conversion means,
A first pixel group in which a plurality of pixels are two-dimensionally arranged at a predetermined pitch in each of a horizontal direction and a vertical direction;
A second pixel group in which a plurality of pixels are two-dimensionally arranged in a state where the pitch is shifted by approximately ½ pitch of the pitch in both the horizontal direction and the vertical direction with respect to the first pixel group,
A semiconductor image sensor using a complementary color filter as an on-chip color filter corresponding to the first pixel group. A semiconductor image sensor comprising a plurality of pixels including a photoelectric conversion means and a MOS transistor for selectively reading out the photoelectric conversion means,
A first pixel group in which a plurality of pixels are two-dimensionally arranged at a predetermined pitch in each of a horizontal direction and a vertical direction;
A second pixel group in which a plurality of pixels are two-dimensionally arranged in a state where the pitch is shifted by approximately ½ pitch of the pitch in both the horizontal direction and the vertical direction with respect to the first pixel group,
The photoelectric conversion means of the first pixel group has a plurality of stacked structures, and is configured to take out two photoelectric conversion signals having different peak wavelengths from the same pixel. A semiconductor image sensor comprising a plurality of pixels including a photoelectric conversion means and a MOS transistor for selectively reading out the photoelectric conversion means,
A first pixel group in which a plurality of pixels are two-dimensionally arranged at a predetermined pitch in each of a horizontal direction and a vertical direction;
A second pixel group in which a plurality of pixels are two-dimensionally arranged in a state where the pitch is shifted by approximately ½ pitch of the pitch in both the horizontal direction and the vertical direction with respect to the first pixel group,
Each first pixel of the first pixel group is configured such that first and second color component signals of different primary colors are detected separately, and each second pixel of the second pixel group includes the primary color and A third color component signal of a different primary color is detected;
A region extending between the adjacent first pixel and the second pixel is defined as a virtual pixel, and the first pixel includes a circle having a radius of less than one pitch of the pixels of the pixel group with the virtual pixel as a center. A semiconductor image sensor that detects color signals by performing arithmetic processing from color component signals of pixels and second pixels. A semiconductor image sensor comprising a plurality of pixels including a photoelectric conversion means and a MOS transistor for selectively reading out the photoelectric conversion means,
A first pixel group in which a plurality of pixels are two-dimensionally arranged at a predetermined pitch in each of a horizontal direction and a vertical direction;
A second pixel group in which a plurality of pixels are two-dimensionally arranged in a state where the pitch is shifted by approximately ½ pitch of the pitch in both the horizontal direction and the vertical direction with respect to the first pixel group,
Each first pixel of the first pixel group is configured such that first and second color component signals of different primary colors are detected separately, and each second pixel of the second pixel group includes the primary color and A third color component signal of a different primary color is detected;
Including a pixel in contact with a circle having, as a virtual pixel, a region extending between the adjacent first pixel and the second pixel, the radius of which is a pitch of one pixel of the pixel group with the virtual pixel as a center A semiconductor image sensor that detects a color signal by performing arithmetic processing from the color component signals of the first pixel and the second pixel included in the circle. 5. The semiconductor image sensor according to claim 1, wherein a plurality of pixels of the first pixel group and the second pixel group share an amplification transistor of the MOS transistors. 5. The semiconductor image sensor according to claim 1, wherein a plurality of pixels of the first pixel group and the second pixel group share a reset transistor among the MOS transistors. 5. The semiconductor image sensor according to claim 1, wherein a plurality of pixels of the first pixel group and the second pixel group share a selection transistor among the MOS transistors.

Images