Active pixel sensor with punch-through reset and cross-talk suppression

<u style="single">FIELD OF INVENTION</u>

The invention relates generally to the field of solid state sensors, and in particular to sensors referred to as active pixel sensors.

<u style="single">BACKGROUND OF THE INVENTION</u>

APS are solid state imagers where each pixel contains a photo-sensing means, charge to voltage conversion means, reset means, and all or part of an amplifier. They have some advantages over charge coupled device imagers (CCD), including single 5V supply operation, x-y addressability, and high level of integration of on-chip signal processing.

APS devices are operated in a manner where each pixel is repeatedly reset and read out. The reset operation is accomplished by resetting the photodetector or charge to voltage conversion means by removing the signal charge that is residing there. This is accomplished by incorporation of a reset transistor in each pixel. Turning the reset transistor on allows signal charge to flow into the drain of the reset transistor, thus being removed from the photodetector region or charge to voltage conversion region. Since a reset transistor is included in each pixel, it occupies area that could otherwise be used for the photodetector, thus reducing the fill factor and sensitivity of the device.

A prior APS pixel is shown in <figref idref="f0001">Fig. 1</figref>. The pixel 10 comprises a photodetector 12, which can conceivably be either a photodiode or photogate, transfer gate 14, floating diffusion 16, reset transistor 17, with a reset gate 18 and reset drain 19, row select transistor (ROWST) 8, with a row select gate (RSG) 9 and signal transistor (SIG) 6. Pixels are arranged in an array (X-columns and Y-rows), to form an image sensor. Device operation occurs in the following manner. A given row is reset by turning on transfer gate 14 and reset gate 18. Any electrons that are in the photodetector 12 or floating diffusion 16 are removed via the reset drain 19. The transfer gate 14 and the reset gate 18 are then turned off and incident light creates electrons in the photodetector 12 for a determined time (integration time). The reset gate is then turned on, removing any electrons that may have accumulated in the floating diffusion region. The reset gate is then turned off and the reset signal level is then read out one column at a time for that row (the details of this operation are not relevant to this invention). Transfer gate 14 is then turned on and these electrons are transferred onto the floating diffusion 16 which is connected to the gate of SIG 6. This signal level is then read out a column at a time for that particular row. A CDS amplifier in each column is used to remove reset noise and noise due to pixel offset. This operation is then repeated for the remaining rows, with the integration time being of constant duration for each row, but integrated during a different time period. As discussed earlier, the incorporation of a reset transistor in each pixel reduces the fill factor and sensitivity of the device. Additionally, the row by row reset integrate and read operation can produce image artifacts.

There are two basic types of charge to voltage conversion means, the floating diffusion and the floating gate. The floating diffusion method provides very good sensitivity, (i.e. small capacitance), but results in an incomplete reset which leads to reset noise. The FG method provides a complete reset and consequently no reset noise, but has poor sensitivity. The floating diffusion approach has been used predominantly in APS and charge coupled devices, (CCD) to obtain good sensitivity. The reset noise is dealt with by performing correlated double sampling (CDS).

In order to perform CDS and eliminate the reset noise the reset operation must occur prior to the read operation. Because of this need to have a reset signal level prior to a read signal level, APS devices have been operated in a manner where each line or row of the imager is reset, integrated and read out at a different time interval than each of the remaining lines or rows. Hence if one were reading out the entire imager, each line would have captured the scene at a different point in time. Since illumination conditions can and do vary temporally, and since objects in the scene may also be moving, this method of read out can produce line artifacts in the resulting representation of the image. This limits the usefulness of APS devices in applications where high quality motion or still images are required. This problem can be overcome by reading the signal level then the reset level, but as stated above this would not eliminate the reset noise.

<patcit id="pcit0001" dnum="US5298778A"><text>U.S. patent 5,298,778</text></patcit> discloses an application solid-state imaging device which includes a plurality of picture elements arranged in a two-dimensional matrix. A sensor region is surrounded by substrate and a gate region is positioned laterally substantially around the sensor region. A sensor region is formed through one surface of the substrate and aligned vertically with the sensor region, while a drain is formed at an opposing surface of the substrate and is likewise aligned with the sensor region. The sensor region and gate region together defined a channel through which source-drain current flows. Light incident on the substrate passes therethrough to the sensor region where charge accumulates photoelectrically for producing an image signal by controlling the source-drain current in proportion to the magnitude of the photoelectrically accumulated charge. The device is reset after reading by removing charge accumulated in the sensor region through the gate region.

In order to solve the problems of APS devices described above, it is desirable to provide a reset means that does not reduce the fill factor of the pixel. It is also desirable to provide a reset means that is complete so that the APS device could be operated in a manner where it is read then reset without introduction of reset noise.

<u style="single">SUMMARY OF THE INVENTION</u>

The present invention addresses the foregoing problems by providing a new pixel architecture and reset method that eliminates the need for a reset transistor per pixel, and provides a complete reset thus eliminating reset noise and the need to do correlated double sampling (CDS). The present invention deals with an APS image sensor according to claim 1. In an active pixel sensor having a plurality of pixels, each of the pixels having a photodetector for accumulating charge from incident light, a transfer gate for removing charge from the photodetector, a floating diffusion that acts as a sense node to an amplifier input, and a drain the improvement comprising providing reset of each pixel by means of application of a potential adjacent the floating diffusion such that the area between the floating diffusion and the drain becomes depleted. In a similar manner the photodetector can be reset in the same manner by placing the reset drain next to the photodetector.

The present invention provides a means for reducing the components required for the reset operation, and further provides a complete reset so that the device can be operated in a manner where the entire imager is integrated at the same time (rather than a row at a time), without suffering from reset noise. This new architecture will be referred to as an APS with punch-through reset. The floating diffusion is reset by completely depleting the region between the reset drain and floating diffusion, so that one has "punch-through" from the reset drain to the floating diffusion. By doing so, all of the charge stored on the floating diffusion will be taken out through the reset drain. Two physical embodiments of the new pixel architectures are shown in <figref idref="f0001">Figs. 2A and 2B</figref> (only the features relevant to this invention are shown). Other specific physical embodiments are realizable. These two are chosen for illustration.

In the first physical embodiment (shown in <figref idref="f0001">Fig. 2A</figref>, the pixel comprises a photodetector, transfer gate, floating diffusion, SIG, RSG, ROWST and reset drain. It does not contain a separate reset gate or reset transistor. In this architecture, the reset operation is accomplished by applying the supply voltage, VDD, or some other appropriate voltage to the desired reset drains (OV, or other appropriate voltage, to the remaining reset drains), so that the p-type region in-between the reset drain and the floating diffusion is depleted. By depleting the p-type region, all of the electrons stored on the floating diffusion will be spilled into the reset drain, providing a complete reset of the floating diffusion. Pixel offset cancellation can still be done by a read then reset clamp and sample. The area that was previously used for RES and RG can now be used for the photodetector, thus improving the fill factor and sensitivity of the pixel.

The buried reset drains shown in <figref idref="f0001">Fig. 2A</figref> are connected together within a row by a buried layer. The buried layer and buried reset drain can be constructed by several means known within the art of integrated circuit fabrication (e.g. high speed BICMOS processes).

The pixel in <figref idref="f0001">Fig. 2B</figref> is similar to that of <figref idref="f0001">Fig. 2A</figref> except the reset drain 29 is formed by a deep n-diffusion from the surface, rather than a buried reset drain 27. By application of an appropriate positive voltage (e.g. VDD) the p-type region is depleted and the floating diffusion is completely reset.

These and other aspects, objects, features and advantages of the present invention will be more clearly understood and appreciated from a review of the following detailed description of the preferred embodiments and appended claims, and by reference to the accompanying drawings.

<u style="single">Advantageous Effect Of The Invention</u>

Fill factor is improved by the reset means requiring no pixel surface area. A complete reset is achieved with a high sensitivity floating diffusion structure, so reset noise is eliminated. Pixel cross-talk is suppressed.

<u style="single">BRIEF DESCRIPTION OF THE DRAWINGS</u>

<ul id="ul0001" list-style="none"><li><figref idref="f0001">FIG. 1</figref> is a prior art pixel;</li><li><figref idref="f0001">FIG. 2A</figref> is a first embodiment of a pixel of the present invention.</li><li><figref idref="f0001">FIG. 2B</figref> is a second embodiment of a pixel of the present invention.</li><li><figref idref="f0002">FIG. 2C</figref> shows a row of pixels from <figref idref="f0001">FIG. 2A</figref> where the reset drains are connected together by a patterned buried layer, and</li><li><figref idref="f0002">FIG. 2D</figref> shows a row of pixels from <figref idref="f0001">FIG. 2b</figref> where the reset drains are connected together by a patterned buried layer.</li><li>To facilitate understanding, identical reference numerals have been used, where possible, to designate identical elements that are common to the figures.</li></ul>

<u style="single">DETAILED DESCRIPTION OF THE INVENTION</u>

This invention provides a means for reducing the components and area required for the reset operation of APS devices, providing higher fill factor and sensitivity over prior art devices. It also provides a complete reset so that the device can be operated in a manner where the entire imager is integrated at the same time (rather than a row at a time), without suffering from reset noise.

This new architecture will be referred to as an APS with punch-through reset. Punch through reset can be accomplished either by resetting the photodetector portion, which is not the object of the present invention, or by resetting the floating diffusion, which is the subject matter of the present invention. In the case of the photodetector, a drain is positioned adjacent the photodetector. The floating diffusion is reset by completely depleting the region between the reset drain and floating diffusion, so that one has "punch-through" from the reset drain to the floating diffusion. By doing so, all of the charge stored on the floating diffusion will be taken out through the reset drain. Two physical embodiments of the new pixel architectures are shown in <figref idref="f0001">Figs. 2A and 2B</figref>. Other specific physical embodiments are realizable. These two are chosen for illustration.

In the first preferred embodiment (shown in <figref idref="f0001">Fig. 2A</figref>), the pixel 20 comprises a Photodetector 22, Transfer Gate 24, Floating diffusion 26, SIG 6, ROWST 8, RSG 9 and reset drain 27. It does not contain a separate reset gate or reset transistor. In this architecture, the reset operation is accomplished by applying the supply voltage VDD, or some other appropriate voltage to the desired reset drain (OV, or appropriate voltage, to the remaining reset drains), so that the p-type region in between the reset drain 27 and floating diffusion 26 is depleted. By depleting the p-type region, all of the electrons stored on the floating diffusion 26 will be spilled into the reset drain 27, providing a complete reset of the floating diffusion 26. Pixel offset cancellation can still be done by differencing the read and reset signals. The reset drain 27 is now buried underneath the floating diffusion 26, and the reset drains in that row are connected together by an n-type buried layer 28, so that none of the pixel surface area is used for the reset operation, thus improving the fill factor and sensitivity of the pixel.

The buried reset drains are connected together within a row by a buried layer 28. The buried layer 28 and buried reset drain 27 can be constructed by several means. In one case both the buried layer 28 and buried drain 27 can be constructed by 2 separate n-type implants prior to epitaxial deposition. The implant doses would be designed to provide the appropriate up diffusion during the rest of the thermal processing of the device. The buried layer 28 would be patterned underneath at least a portion of the photodetector 22 with the appropriate implant dose and/or species to provide cross-talk suppression but still have adequate red response. The buried reset drain implant would be patterned under at least a portion of the floating diffusion 26 with the appropriate implant dose and/or species to provide a p-type region left between the floating diffusion 26 and reset drain 27 that could be completely depleted with a desirable reset potential.

This could also be accomplished by forming the buried layer 28 with an n-type implant prior to epitaxial deposition, and forming the buried reset drain 27 with a high energy implant after epitaxial deposition.

The pixel in <figref idref="f0001">FIG. 2B</figref> is similar to that of <figref idref="f0001">FIG. 2A</figref>, comprising like elements except that reset drain 29 is formed by a deep n-diffusion from the surface, rather than a buried reset drain 27 with an n-type buried layer 28 as used in <figref idref="f0001">FIG. 2A</figref>. Any application of an appropriate positive voltage, (e.g. VDD), results in the p-type region in between the reset drain 28 and the floating diffusion 26 being depleted, and resulting in the floating diffusion 26 being completely reset. This approach has the disadvantage of requiring pixel surface area compared to the buried reset drain approach, but takes less surface area than the separate transistor used in prior art.

In this case a buried layer is not required, and the reset drain region could be formed by a separate implant that is done prior to the N-well drive or field oxidation (i.e. the largest thermal process step). However, a buried layer could be utilized to connect to the reset drain region so that a separate reset buss in an interconnect layer is not required. The buried layer would be formed and patterned as described above.

<figref idref="f0002">FIG. 2C</figref> shows a row of the pixels as illustrated in <figref idref="f0001">FIG. 2A</figref> wherein the reset drains 27 are connected together by a patterned buried layer 28. <figref idref="f0002">FIG. 2D</figref> shows a row of pixels as illustrated in <figref idref="f0001">FIG. 2B</figref> wherein the reset drains 29 are connected together by a patterned buried layer 28. In both <figref idref="f0002">FIG. 2C and FIG. 2D</figref> the SIG 6, RSG 9 and ROWST 8 have been omitted for simplicity of drawing.

<u style="single">Parts List</u>

<dl id="dl0001" compact="compact"><dt>6.</dt><dd>signal transistor</dd><dt>8.</dt><dd>row select transistor</dd><dt>9.</dt><dd>a row select gate</dd><dt>10.</dt><dd>pixel</dd></dl><dl id="dl0002" compact="compact"><dt>12.</dt><dd>photodetector</dd><dt>14.</dt><dd>transfer gate</dd><dt>16.</dt><dd>floating diffusion</dd><dt>17.</dt><dd>reset transistor</dd><dt>18.</dt><dd>reset gate</dd><dt>19.</dt><dd>reset drain</dd></dl><dl id="dl0003" compact="compact"><dt>20.</dt><dd>pixel</dd><dt>22.</dt><dd>photodetector</dd><dt>24.</dt><dd>transfer gate</dd><dt>26.</dt><dd>floating diffusion</dd><dt>27.</dt><dd>buried reset drain</dd><dt>28.</dt><dd>buried layer</dd><dt>29.</dt><dd>reset drain</dd></dl>