• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    The invention relates to color image sensors. In order to benefit from both a good luminance resolution and a colorimetry which is not too much degraded by the sensitivity of silicon to near-infrared radiation, the invention proposes to make a mosaic of pixels comprising colored pixels (R ), (G), (B), coated with color filters, distributed in the matrix, with white pixels (T) not coated with color filters and distributed in the matrix. The colored pixels comprise photodiodes constituted differently from the photodiodes of the white pixels, the different constitution being such that the photodiodes of the colored pixels have a lower sensitivity to infrared radiation than the photodiodes of the white pixels.
    公开号:FR3030885A1
    申请号:FR1463121
    申请日:2014-12-22
    公开日:2016-06-24
    发明作者:Pierre Fereyre；Frederic Mayer；Pascal Douine；Thierry Ligozat；Vincent Presvost；Bruno Diasparra
    申请人:e2v Semiconductors SAS；
    IPC主号:
    专利说明:

    [0001] The invention relates to color electronic image sensors comprising a matrix of photosensitive pixels measuring the amount of light they receive. The most widespread sensors were at the origin of sensors in CCD technology and are more and more often active pixel sensors in MOS technology. The invention will be described with regard to active pixels in MOS technology. A pixel active in MOS technology generally comprises a silicon photodiode and several transistors for detecting a quantity of electric charges generated in the photodiode under the effect of illumination. The detected signal level in the pixel is transferred outside the matrix to read circuits, for example a reading circuit per column of pixels, the reading being carried out line by line in parallel on the different columns. For taking a color image with a single matrix of pixels, it is generally expected to deposit, above the pixels, on the incoming side of the light, a mosaic of different color filters, most often a so-called Bayer mosaic comprising a regular arrangement of groups of four primary color filters which are a red filter, a blue filter and two green filters. Arrangements with complementary cyan, magenta, and yellow secondary color filters are also possible. The filters are made from pigments colored in organic matter. But they have the defect to greatly reduce the amount of light received by the photodiodes of the sensor; a pixel covered with a red filter receives at most only 75% of the light received in the red wavelengths, and in the same way a pixel covered with a blue or green filter does not uniformly convert all the light emitted in blue or green; this results from the fact that the transmission curve of the filter as a function of the frequency is not rectangular but rather Gaussian and that the peak of transmission is not 100%. As a result, with a low level of light, the pixels are no longer sensitive enough and the resulting image can not be used by the eye.
    [0002] It has been proposed in the prior art to make matrices which comprise at the same time pixels coated with colored filters, which will be called hereinafter colored pixels (for example red pixel, blue pixel, green pixel) to remind the presence of the colored filter, and pixels not coated with colored filters, which will be referred to hereinafter as white pixels. Thus, in case of low light, white pixels will collect all of the light and provide the main luminance information. The colored pixels will collect less light and provide chrominance information. This chrominance information will have a lower resolution than with a Bayer filter matrix, and a signal level lower than the white pixel level since it comes from pixels receiving less light, but it is a secondary disadvantage because the luminance information in low light is more important than the colorimetric accuracy. The more white pixels, the more luminance information faithfully represents the luminance of the image. Another problem conventionally encountered with color filter mosaics, especially the least expensive filters that are made from organic pigments, is the fact that these filters are all very transparent to the near infrared rays, and especially in the band of 750 nanometers at 1100 nanometers in wavelength. Silicon is sensitive to these wavelengths and generates electrical charges in proportion to the near-infrared radiation it receives. These charges generated in the red, blue or green pixels are not generated by the real red or blue or green radiation that the pixel receives. They therefore generate in these pixels a portion of the signal that does not represent the actual colorimetry of the scene observed by the sensor. In order that the sensor does not receive this disturbing infrared radiation, it has been proposed in the prior art to place in front of the sensor a band-stop optical filter, which passes light for visible wavelengths but which cuts the near infrared -red. Such a filter is expensive because it must frankly cut the infrared while allowing particular pass the red wavelength band. It has also been proposed to distribute in a conventional mosaic of red, green and blue colored pixels, additional pixels that may be called black pixels, covered with a filter completely cutting visible wavelengths but allowing the infrared to pass through. . Such a filter can be produced using organic pigments such as other filters, for example by superimposing, at the location of the pixels concerned, a red filter and a blue filter. If a Bayer mosaic is modified, then an individual pixel group includes a red pixel, a blue pixel, two green pixels, and a black pixel; these pixels are close to each other and receive substantially the same dose of infrared rays from the scene observed. Subsequently, the signals produced by the red, green, blue pixels, which are affected by the infrared radiation, are subtracted from the black pixel immediately adjacent and receiving only the infrared radiation. This subtraction performed for all the pixels eliminates the harmful influence of the infrared radiation on these pixels. The invention is based on the remark that the matrices comprising colored pixels and white pixels can be treated differently because the luminance produced by the infrared rays and received by white pixels when they exist is not necessarily troublesome . Indeed, the radiation received by the white pixels does not distort the colorimetry, and on the contrary contributes to the representation of the overall luminance of a scene, especially in case of low light where a higher signal level is desirable for the quality of the electronic image. Consequently, the invention proposes a color image sensor comprising a matrix of N active photosensitive pixels, a pixel comprising a photodiode for providing an electrical signal which is a function of the electric charges generated in the photodiode by the light, the matrix being covered with a mosaic of colored filters arranged in correspondence with the pixels of the matrix to form pixels called colored pixels, the filters being of K different types corresponding to K different colors and the pixels of each color being distributed in the matrix, with P pixels , P <N, said white pixels, not covered with a color filter and distributed in the matrix, characterized in that the photodiodes of the colored pixels are constituted differently from the photodiodes of the white pixels, the photodiodes of the colored pixels having, for the infrared wavelengths, a conversion coefficient of photons into significantly lower electrical charges to the conversion coefficient of the photodiodes of the white pixels for these same wavelengths. Thus, without necessarily having black pixels and without the need to use a global infrared filter, which would unnecessarily reduce the luminance information given by the white pixels and which would be expensive, an image having a good luminance information is produced. quality, particularly in low light level, and chrominance information weakly degraded by infrared radiation.
    [0003] There are several possible ways to render the photodiodes of the colored pixels less sensitive to infrared radiation than the photodiodes of the white pixels. In general, we know that the shortest wavelengths are the ones that penetrate the least in silicon and therefore generate electrical charges at a shallow depth (100 to 300 nanometers for blue wavelengths) ; green and red wavelengths penetrate deep into the silicon and generate charges a little deeper (up to a micrometer typically); the infrared wavelengths are the ones that penetrate the deepest silicon and generate charges to a depth that can be significantly higher (several micrometers for the near infrared). Means are therefore provided to prevent the light from producing payloads (i.e., charges recoverable by the photodiode to be read) beyond a certain depth from which only infrared rays are practically penetrate. For that, one can either limit the depth of the space charge zone created in the silicon by the inverse polarization of the photodiode, or create in depth a zone for collecting and evacuating the electrons generated in depth. In the first case, the electrons created in depth out of the space charge area of the photodiode of a pixel can not be collected by the pixel and therefore do not constitute payloads for the pixel; these electrons recombine in fact elsewhere than in the space charge zone and do not migrate towards the cathode of the photodiode. In the second case, the charges are collected and discharged to a drain and not to the photodiode. In one embodiment, the photodiodes are formed by an NP junction between an N-type doped localized region and a P-doped active layer and brought to a reference potential, the active layer having a depth H below the junction. . The photodiodes of the colored pixels comprise a P + type localized region, more doped than the P-type doped active layer, this more doped localized region reducing to a value H 'less than H the P-type active layer depth located directly above below the junction and limiting at this depth H 'the space charge area produced in operation by the inverse polarization of the photodiode, and the white pixels not comprising such a localized region of P + type so that their charging zone space can extend to a depth greater than H '.
    [0004] In another embodiment, with the same type of photodiode for the white pixels, it is expected that the photodiodes of the pixels covered with color filters having a buried localized region of N + type, located at a depth H "below the junction, a depth such that the active layer located below the buried localized region receives mainly infrared rays but not or almost no visible radiation, the buried localized region being brought to a potential for draining the charges generated by the infrared radiation; below this region.
    [0005] In this second embodiment, it may be desirable for the pixels covered with color filters to be juxtaposed so that they touch each other on one side or, strictly speaking, by a corner to allow the continuity of the buried regions N + of the pixels to be formed. colored, continuity up to an edge of the matrix. Thus, the buried layers of N + type of all these pixels can be connected to each other and it is then easier to connect the N + buried layer to a positive potential from the edge of the matrix. This avoids providing link contacts within the pixel. It is also possible, however, even with white pixels separating the colored pixels, to create a N + buried layer continuity by giving it a grid shape surrounding the white pixels without coming under the white pixels and thus without significantly modifying the constitution and the behavior of the photodiode of these white pixels.
    [0006] Different mosaic patterns are possible, an advantageous configuration being an arrangement of pixels with periodic groups of pixels consisting of a blue pixel, a red pixel, two green filter pixels and X unfiltered pixels between two color pixels of the same line or the same column of the matrix. X is greater than or equal to let of equal to 1, 2 or 3. Other features and advantages of the invention will appear on reading the detailed description which follows and which is given with reference to the appended drawings in which: FIG. 1 schematically represents the organization of a color electronic image sensor comprising a matrix of silicon-based pixels, coated with a mosaic of color filters with unfiltered pixels distributed in the matrix; FIG. 2 represents an example of possible spectral responses of the pixels coated with color filters; FIG. 3 shows an example of a mosaic of the prior art making it possible to compensate the influence of radiation in the near infrared; FIG. 4 represents the organization of a sensor according to the invention; FIG. 5 represents an embodiment of photodiodes with attenuation of the influence of the infrared on the colored pixels; FIG. 6 represents another embodiment according to the invention; FIG. 7 represents a way of connecting the N + regions under the photodiodes of non-adjacent colored pixels; FIG. 8 represents several interesting configurations of the mosaic of pixels according to the invention; FIG. 9 represents configurations that also incorporate black pixels sensitive only to infrared rays.
    [0007] In Figure 1, there is shown in top view the principle of a matrix sensor of the prior art. It comprises a matrix of lines and columns of photosensitive pixels. This matrix is covered with a mosaic of colored filters. In this example, there are both pixels called "colored" because they are covered with a filter of a given color and so-called white pixels that are not covered with a color filter. By way of example, there is shown a regular distribution of groups of four pixels which are respectively a red pixel (R) covered by a filter allowing the red light to pass but not or not the other colors, a blue pixel (B) covered a filter allowing the blue light to pass but not the other colors, a green pixel (G) covered by a filter allowing the green light to pass but not or not the other colors, and a white pixel (T) not covered of a filter and thus letting all the colors pass. Other color distributions in the mosaic exist in the prior art, including distributions with complementary color filters cyan, magenta, yellow, instead of primary colors red, blue, green. The sensor is a silicon-based sensor, which converts the light it receives into electrical charges into a wavelength band of about 300 nanometers to about 1100 nanometers. Each pixel is constituted by a photodiode and a few transistors providing an electrical signal representing the illumination received by the pixel. The filters are generally filters comprising colored organic pigments that mainly pass through the wavelengths corresponding to the color in question, namely: about 300 to 530 nanometers for blue, 480 to 620 nanometers for green, and about 580 at 650 nanometers for red. FIG. 2 represents a typical sensitivity diagram of a silicon sensor covered with organic color filters. Five different curves are represented on this diagram.
    [0008] The scale on the abscissa is graduated in wavelengths; the ordinate scale is a sensitivity scale in arbitrary units, representing the output signal of a pixel for a given amount of photons at the wavelength considered. The RNF curve is the response curve of a white pixel, that is to say devoid of filter. It represents the intrinsic intrinsic sensitivity of silicon. It has a maximum in the red and it decreases on both sides of this maximum. The other curves RB, RG, RR are the sensitivity curves of the blue, green, and red pixels respectively, taking into account therefore the presence of a color filter. These three curves show a strong rise in sensitivity in the near-infrared range: beyond about 820 nanometers, the filters are all practically transparent, they do not eliminate these near-infrared wavelengths. The near infrared thus distorts the colorimetry of the image by suggesting for example that a blue pixel provides a signal representing the luminance for the blue wavelengths, while the signal represents a sum of a blue luminance and of an infrared luminance received on the pixel. An infrared filter placed in front of the sensor makes it possible to eliminate this disadvantage, but it is expensive.
    [0009] In the prior art, mosaics of differently constituted pixels have already been proposed, with a blue pixel (B), a green pixel (G), a red pixel (R) and a so-called "black" pixel (IR) which pass no color and let only infrared radiation. A black pixel can be made by simply superimposing a red filter and a blue filter; its response curve in Figure 2 would be the RIr curve. Figure 3 shows the organization of such a sensor. Each colored pixel provides a signal representing the sum of the luminance received in the color under consideration and the luminance received in the near infrared. The black pixel provides a signal representing only the luminance received in the near infrared. Subtraction between the signal from a colored pixel and the signal from the black pixel most eliminates the infrared-related component in the signal of the color pixel. FIG. 4 represents the general organization of a sensor according to the invention, which is based on a different principle making it possible to avoid the use of an infrared filter in front of the sensor and making it possible to use a mosaic of filters also comprising white pixels in a way that maximizes the luminance of white pixels. The mosaic of filters which covers the sensor may be the same as that of FIG. 1, that is to say that it uses (R), (G), (B) so-called "colored" pixels because they are covered with colored filters, and pixels (T) called "white" which are not covered with colored filters. It will be considered in the following that the colors of the filters are the primary colors red green blue but it could alternatively provide secondary color filters magenta, cyan, yellow.
    [0010] The particularity of the invention is the fact that the colored pixels are made in silicon from photodiodes which have a constitution different from the photodiodes of the white pixels. White pixels have photodiodes that are quite strongly sensitive to infrared radiation as in the prior art, but the colored pixels are made with different photodiodes that are significantly less sensitive to infrared radiation. The photodiodes of the colored pixels therefore have, for the infrared wavelengths, a conversion coefficient (or more generally a response level as a function of the wavelength) that is significantly lower than the photodiodes of the white pixels. The colored pixels R, G, B are shown in FIG. 4 with a dotted-surface disk which symbolizes the fact that the photodiode of these pixels is different from the photodiode of the white pixels. White pixels are represented without this dotted-surface disk and have a stronger response in the infrared range.
    [0011] The photodiodes of the usual sensors are generally formed by an NP junction in an active layer of P type monocrystalline silicon. To give a conversion coefficient in the colored pixels smaller than the conversion coefficient in the white pixels, it is possible, for example to limit the depth of the space charge zone created by the inverse polarization of the photodiode, or it is possible to create in the active layer an intermediate zone for collecting and evacuating the electrons generated in depth in the active layer. The space charge zone, also called depleted zone or deserted zone, is the region free of free carriers which appears in an N-P junction, in particular when it is reverse biased. It extends all the way further, on either side of the junction, the layer is less doped and the reverse bias voltage is higher. This zone, in which a non-zero electric field develops, allows a collection efficiency of very high generated charges. The NP junction is formed between an N-type doped localized region and the P type doped active layer. is brought to a reference potential. All types of conductivity could be reversed without changing the nature of the invention and it will be considered that the definition given for a conductivity type applies identically to the inverse type without departing from the scope of the invention. Figure 5 shows in section in the silicon of the sensor, the photodiodes of three adjacent pixels. The pixels are two colored pixels red (R) and green (G), and a white pixel (T). Only the photodiodes are represented to simplify the diagram, the transistors of the pixels not being represented. The active layer 10 of monocrystalline silicon in which the photodiodes are formed is a P-type active layer. It may be the upper layer of a monocrystalline silicon substrate or the upper epitaxial layer of an SOI ("Silicon On lnsulator") substrate. that is, silicon on insulator). The substrate 12 is simply represented by hatching in FIG. 5; its thickness is much greater than the thickness of the active layer 10, which may have a thickness of 3 microns to 30 microns. The photodiodes are formed from N-type individual regions 14 diffused in the active layer 10, which regions form an NP junction with the underlying silicon. Conventionally but not necessarily, each region N 14 may be covered with a surface area 16 of P + type, which is brought to a reference potential which is the potential applied to the active layer 10. The pixels are active pixels and comprise transistors for reading the charges accumulated in the regions 14 during operation. The pixels are separated from one another by insulating regions consisting of silicon of the same type as the active layer 10 but more doped, the insulation being optionally reinforced by the presence of surface trenches of silicon oxide (STI insulation, of English "Shallow Trench Isolation"). These areas are not represented; they prevent or limit the displacement of electric charges generated by the light of a pixel to a neighboring pixel. Finally, an insulating layer, for example made of silicon oxide 5iO 2, can cover the entire surface of the photodiodes.
    [0012] The photodiode of the green pixel is covered with a green filter GF and the photodiode of the red pixel is covered with a red filter GR. It can be seen that there is no filter above the photodiode of the white pixel T. What will be called photodiode in this description is the set of the region N 14 (with its region 16 if any) and the semiconductor regions located below the region 14. In the embodiment of FIG. 5, the difference in constituting the photodiodes of the colored pixels and the white pixels is a difference which causes a difference in the possibilities of extension in depth of the charging zone. space created in the active layer by the inverse polarization of the photodiode. The possibilities of extending the space charge area (represented by a dotted line) are more limited in the colored pixels, thanks to the particular constitution of the photodiode of these pixels, than in the white pixels. This extension depends on the doping of the semiconductor below the junction and depends on the reverse bias voltage of the photodiode. But we establish an extension limit in the colored pixels, lower than in the white pixels, thanks to a deeper layer of P + type more doped than the active layer.
    [0013] In the example given in FIG. 5, it is considered that the depth H of the active layer is approximately 3 to 10 micrometers and that the space charge zone occupies the entire depth or practically the entire depth of the active layer at the same time. below white pixels. However, one could have an active layer depth H different from the depth Z of the space charge zone, in particular when the depth of the active layer is greater than 10 or 20 microns or more. A localized region 20 of the same conductivity type as the active but more doped layer, limiting at a depth Z 'below the NP junction, the extension of the charging zone space produced in operation by the inverse polarization of the photodiode of the pixel. The depth Z 'is less than Z and in practice the top of the localized region P + may be located at a depth H' = Z 'below the junction, or a little above this depth if the localized region P + is much more doped than the active layer. Indeed, the space charge area extends very little in a heavily doped region. The extension of the space charge area is proportional to the square root of the inverse of the doping atom concentration. The radiation of visible wavelengths blue, green, red, penetrate the silicon and create electrical charges in the depth Z '. They do not penetrate deeper or much deeper. Infrared rays penetrate deeper into the P + region 20 and beyond. If they create electric charges in the P + region, these charges recombine rapidly and the electrons are not attracted to the N region of the photodiode. If they create charges between the P + region 20 and the substrate 12, these charges can possibly be dispersed towards the neighboring white pixels, but not towards the colored pixels because of the P + 20 regions which hinder their passage towards the regions 14. photodiodes of white pixels do not have this P + region. Visible or near-infrared wavelength radiation creates electrical charges throughout the Z-depth of the space charge area, here practically throughout the depth of the active layer. The electrons drained by the electric field present in the space charge zone can be collected by the N region 14 of the white pixel.
    [0014] All the luminance received contributes, whatever the wavelength, to the signal produced by the white pixel. This gives a luminance indication the best possible. The risk of receiving electrons from infrared rays coming from neighboring colored pixels exists, generating a risk of loss of resolution of the luminance, but this risk is even lower than the ratio between the number of colored pixels and the number white pixels is lower. We will see later configurations in which, to obtain a good luminance resolution, there are many more white pixels than colored pixels. Thus, it has been realized as desired a matrix in which the conversion coefficient of photons into a useful signal for the infrared wavelengths is lower for the photodiodes of the pixels coated with color filters than for the photodiodes of the white pixels. Typically, the depth Z 'can be between 1 to 5 micrometers, preferably between 1 and 3 micrometers, the difference in height ZZ' being preferably greater than 2 micrometers and can even be 10 to 20 micrometers in the case where the active layer is simply constituted by the upper part of a monocrystalline silicon substrate. The depth H 'may be equal to Z'. The depth H is at least equal to Z but can be much greater in some cases.
    [0015] The doping of the P + type localized region (typically 1 to 1016 atoms / cm3, corresponding to a resistivity of a few ohms-cm) is preferably at least 10 times and preferably at least 100 times greater than the doping of the layer. P-type active (typically from 1 to 10 times 1014 atoms / cm 3, corresponding to a resistivity of a few tens of ohm-cm) so that the space charge zone substantially stops at the depth H 'in top of the P + 20 region penetrating very little in the P + region. The P + doped region may be carried out by deep implantation of impurities of the same type as the impurities which doped the active layer, preferably boron. This implantation takes place before the implantation of the regions 14 and 16 of the photodiodes. FIG. 6 represents another exemplary embodiment resulting in the same desired result, namely a response of the photodiodes to an infrared wavelength (the response being the quantity of electric charges generated by a quantity of photons at a wavelength given and actually retrieved by the photodiode to provide a useful signal for that wavelength) is lower for the colored pixels than for the white pixels, and this in the same way for all the colored pixels. In this embodiment, in the photodiodes, colored pixels are provided not a P + region which limits the depth of the space charge zone but a localized region of the N + type located at a depth H "and brought to a potential This N + region is positive to drain the electrons generated by the light to a depth in which it is essentially the infrared rays and not the visible radiation that generates electron-hole pairs. be the general power supply Vdd or a lower potential but higher than the reference potential of the active layer The electrons are discharged towards this fixed potential 3030 885 14 The photons resulting from the visible light are absorbed in the silicon at a depth weaker than the depth H "of the N + region and create electrons that are drained as in the photodiodes of the white pixels towards the N region of the photodiode to provide a useful signal. Infrared photons absorbed near the N + region or in the N + region or below the N + region are removed and do not provide a useful signal. The photodiodes of the white pixels do not include this region N + 30. This again results in an electric response for the infrared wavelengths lower in the colored pixels than in the white pixels. To produce the deep N + regions, it is possible to start from a substrate comprising a P type epitaxial active layer having a thickness HH "which is smaller than the desired final thickness, that is to say smaller than the sum of the depth H and the depth of the zones 14 and 16. A mask is produced which protects the surfaces corresponding to the locations of the photodiodes of the white pixels and the desired N + zones are implanted on the surface of this partial layer under the pixels which will be covered with then the epitaxial growth of the entire desired active layer height is continued before forming the N and P + regions of the photodiodes To avoid making a connection contact of the N + layer in each color pixel (for a potential bond fixed drainage), it is preferably provided that all N + regions 30 are connected to one another, and contact is made only on one or more outer edges of the die This is entirely possible in the case of FIG. 4, by making the N + region of a colored pixel extend to touch the region of an adjacent color pixel. It is also possible to do this in all cases where the pixels touch one side, or even only one corner by touching the N + regions on two adjacent sides or two adjacent corners. In addition, even if the colored pixels do not form a continuity of edges or corners up to an edge of the matrix, it is still possible to make a continuity of N + regions in the form of a grid whose openings are located at the place of the white pixels only, as represented by way of example in FIG. 7. In this example (usable also in the embodiment of FIG. 5), the pixels are organized by elementary patterns of sixteen pixels comprising a red pixel, a blue pixel, two green pixels, and twelve white pixels. The colored pixels are all separated from each other by a white pixel and their corners do not touch. We can then make a grid of N + regions as shown in Figure 7 below the mosaic represented; this grid has continuous areas of N + areas below the colored pixels G, R, B, open surfaces below the white pixels T, and links around the white pixels, sufficiently narrow links that do not substantially disturb the pixels white.
    [0016] Figure 8 shows various other arrangements of colored and white pixels corresponding to different proportions of colored pixels (50%, 25% and 6% respectively). The more color pixels are better colorimetry. The more white pixels there are, the better the luminance image resolution. Mosaic 8A includes a periodic repetition of 16-pixel squares with 8 white pixels, 4 green pixels, 2 red pixels, and 2 blue pixels. The mosaic 8B comprises a periodic pattern of squares of sixteen pixels with twelve white pixels and four colored pixels (two green, one red, one blue, aligned along a diagonal of the square). The mosaic 8C includes periodical squares of 25 pixels including four colored pixels (two green, one red and one blue) and 21 white pixels, offering a very good luminance resolution. In an alternative embodiment, there is provided in the mosaic of color filters a distribution of black pixels each covered with a filter allowing the near infrared and not passing the visible light. These black pixels are made with photodiodes identical to the photodiodes of the colored pixels, that is to say, attenuating the influence of the near-infrared rays in the same way as for the colored pixels. The signal coming from each colored pixel and the signal coming from the nearest black pixel (or interpolation of signals coming from the nearest black pixels) are collected, and this signal is subtracted from the signal coming from the colored pixel. But we do not subtract between the signal from the white pixels and the signals from the black pixels. The colorimetry of the color pixels is thus improved without deteriorating the sensitivity of the white pixels.
    [0017] FIG. 9 represents two exemplary embodiments of this variant. The first example (9A) is drawn from that of FIG. 8C, in which the white central pixel has been replaced by a black IR pixel. From the point of view of the present invention, the black pixel is considered as a colored pixel, coated with an infrared filter which can be achieved by the combination (superposition or mixture of pigments) of a red filter and a blue filter . The second example (9B) comprises a greater proportion of black pixels, the elementary pattern repeated in the mosaic comprising eight white pixels, four colored pixels, and four black pixels, each black pixel being equidistant from four colored pixels.
    [0018] To end this description, it is indicated that N and P are called conductivity types usually corresponding to donor or electron acceptor impurity atoms, but since the conductivity types can all be reversed without changing anything. the principles of the invention, it is considered for the purposes of this description and claims that the N and P names are purely conventional and may designate doping contrary to the usual use. The recovered electrical charges are then no longer holes and the directions of the potential differences must be reversed. This notation avoids making the language heavier by expressions such as "first type of conductivity", "second type of conductivity", "positive electrical charges", "negative electrical charges", etc.
    权利要求:
    Claims (10)
    [0001]
    REVENDICATIONS1. Color image sensor, comprising a matrix of N active photosensitive pixels, a pixel comprising a photodiode for providing an electrical signal according to the electric charges generated in the photodiode by the light, the matrix being covered with a mosaic of colored filters arranged in correspondence with the matrix of pixels to form pixels called colored pixels (R, G, B) coated with a color filter, the filters being of K different types corresponding to K different colors and the pixels of each color being distributed in the matrix, the matrix further comprising P pixels, P <N, said white pixels (T), not covered by a color filter and distributed in the matrix, characterized in that the photodiodes of the colored pixels are constituted differently from the photodiodes of the white pixels, photodiodes of colored pixels having, for infrared wavelengths, a conversion coefficient of photons into electrical charges It is significantly lower than the conversion coefficient of photodiodes of white pixels for these same wavelengths.
    [0002]
    2. Color image sensor according to claim 1, characterized in that the photodiodes are formed by an NP junction between an N-type doped localized region (14) and an active layer (12) doped with a P type and carried at a reference potential, the active layer having a depth H below the junction, the photodiodes of the colored pixels comprising a localized region (20) of P + type, more doped than the active layer, this localized region more doped reducing to a value H 'less than H the P-type active layer depth located directly below the junction and limiting at this depth H' the space charge area produced in operation by the inverse polarization of the photodiode, and white pixels not having such a P + type localized region so that their space charge area can extend to a depth greater than H '.
    [0003]
    3. Color image sensor according to claim 2, characterized in that the depth H 'is between 2 and 3 micrometers and the depth difference H-H' is greater than 2 micrometers.
    [0004]
    4. Color image sensor according to one of claims 2 and 3, characterized in that the doping of the P + type localized region is at least 10 times and preferably at least 100 times greater than the doping of the active layer. of type P.
    [0005]
    5. Color image sensor according to claim 1, characterized in that the photodiodes are formed by an NP junction between a localized region (14) doped N type and a active layer (12) doped type P and brought to a reference potential, the active layer having a depth H below the junction, the photodiodes of the pixels covered with color filters having a buried localized region (30) of N + type, located at a depth H "below the junction, a depth such that the active layer located below the buried localized region receives mainly infrared rays but practically no visible radiation, the buried localized region being brought to a potential enabling the charges generated by the infrared radiation to be drained at below this region, in the second photodiode, and the photodiodes of the pixels not covered with color filters not including such a region ocialized buried type N.
    [0006]
    6. color image sensor according to claim 5, characterized in that the depth H "of the localized buried region of N + type is at least 3 microns.
    [0007]
    7. Color image sensor according to one of claims 5 and 6, characterized in that the colored pixels are arranged in series so that they touch by a corner or by a side and localized buried N + regions form a doping continuity N + up to an edge of the matrix.
    [0008]
    8. Color image sensor according to one of the preceding claims, characterized in that the matrix comprises an arrangement of pixels with filters of three different colors, including a periodic arrangement of groups of pixels comprising a blue filter, a filter red, two green filters and X (X> = 1) white pixels between two colored pixels of the same row or column of the matrix.
    [0009]
    9. Color image sensor according to claim 8, characterized in that X is equal to 1 or 2 or 3.
    [0010]
    10. A color image sensor according to one of the preceding claims, characterized in that the pixel array comprises so-called black pixels passing infrared rays but not visible light, distributed in the matrix, the photodiodes of these pixels. being formed as the photodiodes of the colored pixels, for signal subtraction between a colored pixel and a nearest black pixel or a combination of the nearest black pixels.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3238254B1|2021-06-30|Colour image sensor with white pixels and colour pixels
    EP3098858B1|2017-10-25|Photodetector with high quantum efficiency
    EP2786412B1|2021-04-14|Optical detector unit
    FR2930676A1|2009-10-30|IMAGE SENSOR WITH VERY LOW DIMENSIONS
    WO2013064753A1|2013-05-10|Monolithic multispectral visible-and-infrared imager
    EP2064742A2|2009-06-03|Colour image sensor with improved optical crosstalk
    FR2974237A1|2012-10-19|REAR-SIDED IMAGE SENSOR WITH TRANSPARENT ELECTRODE
    BE1024783B1|2018-07-02|Detector device with majority current and isolation means
    EP3016141A1|2016-05-04|Image sensor with vertical electrodes
    EP2452361B1|2013-05-08|Photo-detector device and method of producing a photo-detector device
    FR2954854A1|2011-07-01|PHOTODETECTOR WITH PLASMON STRUCTURE
    FR2906082A1|2008-03-21|PHOTODIODE AT AVALANCHE
    EP3155661B1|2019-07-31|Method for acquiring images using a cmos image sensor
    WO2014118308A1|2014-08-07|Photodiode array having a charge-absorbing doped region
    EP2840616B1|2017-09-06|SPAD photodiode with high quantum efficiency
    FR3009890A1|2015-02-27|BOD PHOTODIODE WITH HIGH QUANTUM EFFICIENCY
    EP2846357A1|2015-03-11|Photodetector device with semiconductor regions divided by a potential barrier
    EP1061732A1|2000-12-20|Method for biasing the photodiodes of a matrix sensor through their associated diodes/photodiodes
    EP3240030B1|2021-08-18|Photo-detection device with overdoped inter-diode array and manufacturing method
    WO2014060479A1|2014-04-24|Image sensor having improved quantum efficiency at large wavelengths
    FR2968877A1|2012-06-15|Image sensor for detecting color on surface of substrate semiconductor, has pixels arranged in rows adjacent to each other based on pixel width, where pixels of adjacent rows are being offset relative to each other than half of pixel length
    FR2969390A1|2012-06-22|IMAGING DEVICE WITH INFRARED RADIATION FILTRATION.
    FR3008230A1|2015-01-09|MATRIX COLOR IMAGE SENSOR, WITHOUT COLOR FILTERS
    EP3000132A1|2016-03-30|Photodiode array having adjustable charge-absorption
    同族专利:
    公开号 | 公开日
    WO2016102267A1|2016-06-30|
    US20180006078A1|2018-01-04|
    EP3238254A1|2017-11-01|
    EP3238254B1|2021-06-30|
    FR3030885B1|2017-12-22|
    TW201635509A|2016-10-01|
    CN107210309B|2021-05-14|
    US10128298B2|2018-11-13|
    CN107210309A|2017-09-26|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    EP0605898A1|1993-01-01|1994-07-13|Canon Kabushiki Kaisha|Solid-state image pickup device|
    EP1630871A1|2003-11-10|2006-03-01|Matsushita Electric Industrial Co., Ltd.|An imaging device and an imaging method|
    US20050173742A1|2004-01-20|2005-08-11|Kabushiki Kaisha Toshiba|Solid state image sensing device and method of manufacturing the same|EP3582494A4|2017-02-10|2020-01-08|Hangzhou Hikvision Digital Technology Co., Ltd.|Multi-spectrum-based image fusion apparatus and method, and image sensor|JP5262180B2|2008-02-26|2013-08-14|ソニー株式会社|Solid-state imaging device and camera|
    US7910961B2|2008-10-08|2011-03-22|Omnivision Technologies, Inc.|Image sensor with low crosstalk and high red sensitivity|JP6953297B2|2017-12-08|2021-10-27|キヤノン株式会社|Imaging device and imaging system|
    CN108305883A|2018-01-30|2018-07-20|德淮半导体有限公司|Imaging sensor|
    KR20200098032A|2019-02-11|2020-08-20|삼성전자주식회사|Pixel array included in image sensor and image sensor including the same|
    CN113192993A|2019-03-26|2021-07-30|福州鑫图光电有限公司|Imaging method of image sensor|
    KR102230609B1|2019-11-26|2021-03-19|가천대학교 산학협력단|Deep Learning-Based Sparse Color Sensor Image Processing Apparatus, Image Processing Method and Computer-readable Medium|
    法律状态:
    2015-11-23| PLFP| Fee payment|Year of fee payment: 2 |
    2016-06-24| PLSC| Publication of the preliminary search report|Effective date: 20160624 |
    2016-11-28| PLFP| Fee payment|Year of fee payment: 3 |
    2017-11-27| PLFP| Fee payment|Year of fee payment: 4 |
    2018-06-22| CD| Change of name or company name|Owner name: TELEDYNE E2V SEMICONDUCTORS SAS, FR Effective date: 20180523 |
    2019-12-26| PLFP| Fee payment|Year of fee payment: 6 |
    2020-12-27| PLFP| Fee payment|Year of fee payment: 7 |
    2021-12-27| PLFP| Fee payment|Year of fee payment: 8 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1463121A|FR3030885B1|2014-12-22|2014-12-22|COLOR IMAGE SENSOR WITH WHITE PIXELS AND COLOR PIXELS|FR1463121A| FR3030885B1|2014-12-22|2014-12-22|COLOR IMAGE SENSOR WITH WHITE PIXELS AND COLOR PIXELS|
    CN201580070000.1A| CN107210309B|2014-12-22|2015-12-16|Color image sensor with white pixels and color pixels|
    EP15817203.1A| EP3238254B1|2014-12-22|2015-12-16|Colour image sensor with white pixels and colour pixels|
    PCT/EP2015/080004| WO2016102267A1|2014-12-22|2015-12-16|Colour image sensor with white pixels and colour pixels|
    US15/538,294| US10128298B2|2014-12-22|2015-12-16|Colour image sensor with white pixels and colour pixels|
    TW104142744A| TW201635509A|2014-12-22|2015-12-18|Colour-image sensor with white pixels and coloured pixels|
    [返回顶部]