This application claims priority from French patent application No. 07/56447, filed Jul. 12, 2007, which is incorporated herein by reference.
1. Technical Field
An embodiment of the present invention relates to an image sensor made in monolithic form and intended to be used in shooting devices such as for example cameras, camcorders, digital microscopes, digital photographic devices, or endoscopes. More specifically, an embodiment of the present invention relates to a semiconductor-based image sensor.
2. Discussion of the Related Art
There are two large families of image sensors. Image sensors made according to the CCD technology (CCD standing for Charge-Coupled Device) and image sensors made according to the CMOS technology (CMOS standing for Complementary Metal Oxide Semiconductor). In all cases, the light is converted in a layer of a semiconductor material, for example, silicon, into exploitable electric signals.
Conventionally, an image sensor made in monolithic form comprises photosensitive cells, or pixels, each comprising charge storage means formed in a single-crystal silicon substrate, the storage means generally corresponding to a photodiode for a CMOS transistor and to a MOS capacitance for a CCD sensor. A stacking of insulating and conductive layers covers the substrate, and metal tracks and vias are formed in these layers to connect the storage means to other components. Microlenses are distributed on the upper surface of the stacking of insulating and conductive layers, each lens being associated with a pixel and focusing the light rays reaching the upper surface of the image sensor towards the storage means of the associated pixel. In the case of a color image sensor, color filters may be added under the focusing microlenses. Such an image sensor structure is called a front-lit image sensor.
According to another conventional image sensor example, the charge storage means are formed in a substrate corresponding to a thin single-crystal silicon layer which is lit on its rear surface. The sensor thus formed is called a back-lit image sensor. An advantage of a back-lit image sensor over a front-lit image sensor is that the focusing of the light rays towards the storage means of the pixels is eased since the light rays do not have to cross the stacking of insulating and conductive layers.
When the image sensor is lit, the incident photons are absorbed in the silicon substrate and cause the forming of electron/hole pairs, the electrons being captured by the storage means of the pixels. Generally, the silicon substrate thickness necessary for a majority of the photons of a given wavelength to be absorbed increases along with the wavelength. To ensure the absorption of a sufficient amount of photons of large wavelengths of the visible spectrum (mainly corresponding to red), it is generally necessary for the silicon substrate thickness to be greater than 3 μm. However, even with such thicknesses, it is possible for the absorption rate of photons of large wavelengths not to be maximum so that the image sensor sensitivity is different according to the wavelength of the incident photons. It may then be necessary to provide a compensation processing of the signals provided by the image sensor to take into account this non-uniform sensitivity.
The electrons resulting from the absorption of photons focused by a microlens are captured by the associated storage means. However, it may be observed that some electrons resulting from the absorption of photons are not captured by the storage means for which they are intended, but by neighboring storage means. This translates as a disturbance in the amount of charges stored by the storage means and thus as a disturbance of the signals provided by the sensor. This phenomenon is called “crosstalk” and is all the stronger as the electron-forming location is remote from the photodiodes.
For a front-lit image sensor, the crosstalk mainly concerns the photons of the visible spectrum of large wavelengths (mainly corresponding to color red) which may cause the forming of electrons distant by several micrometers from the storage means. For a back-lit image sensor, the crosstalk is particularly significant for photons of small wavelengths (mainly corresponding to colors blue and green) for which most of the electrons form down to a small substrate depth, smaller than one micrometer, from the rear surface of the sensor, that is, on the substrate side opposite to the storage means. Indeed, for such wavelengths, the distance that the electrons must cover to reach the storage means for which they are intended is typically greater than one micrometer and the risk for electrons to move towards other more distant storage means is not negligible.
It may be devised, to decrease crosstalk phenomena, to decrease the substrate thickness, which would bring the storage means closer to the electron/hole pair forming areas. However, this may not be envisaged since the image sensor sensitivity might then be too strongly altered for large wavelengths.
Another possibility would be to form, across the entire substrate thickness, insulation areas separating the pixels and for example corresponding to regions of an insulating material or to heavily-doped regions. It is however difficult to form insulation areas which are both deep and narrow on a substrate with a thickness of several micrometers, without greatly increasing the total sensor surface area.
An embodiment of the present invention aims at providing an image sensor enabling decreasing the crosstalk phenomenon between adjacent pixels while improving the sensor sensitivity.
Thus, an embodiment of the present invention provides an image sensor comprising photosensitive cells, each photosensitive cell comprising at least one charge storage means formed at least partly in a substrate of a semiconductor material. The substrate comprises, for at least one first photosensitive cell, a portion of a first silicon and germanium alloy having a first germanium concentration, possibly zero, and for at least one second photosensitive cell, a portion of a second silicon and germanium alloy having a second germanium concentration, non-zero, greater than the first germanium concentration.
According to an embodiment, the substrate comprises, for at least one third photosensitive cell, a portion of a third silicon and germanium alloy having a third germanium concentration, non-zero, greater than the second germanium concentration.
According to an embodiment, the substrate has a thickness smaller than 1 micrometer, for example, smaller than 500 nanometers.
According to an embodiment, the substrate comprises, at least for the second photosensitive cell, a single-crystal silicon portion adjacent to the second silicon and germanium alloy portion, the single-crystal silicon portion containing the charge storage means associated with the second photosensitive cell.
According to an embodiment, the image sensor comprises an insulation area separating the first and second photosensitive cells and extending across the entire thickness of the substrate.
According to an embodiment, the first photosensitive cell comprises a first filter capable of letting through light rays having first wavelengths and the second photosensitive cell comprises a second filter capable of letting through light rays having second wavelengths greater than the first wavelengths.
An embodiment of the present invention provides a method for manufacturing an image sensor comprising photosensitive cells, comprising the steps of:
(a) providing a single-crystal silicon layer;
(b) forming, at least partly in the layer, for at least one first photosensitive cell, a portion of a first silicon and germanium alloy having a first germanium concentration, possibly zero, and, for at least one second photosensitive cell, a portion of a second silicon and germanium alloy having a second non-zero germanium concentration, greater than the first germanium concentration; and
(c) forming, for each photosensitive cell, a charge storage means at least partly in the layer.
According to an embodiment, step (b) comprises at least one step of germanium ion implantation in the layer.
According to an embodiment, step (b) comprises the steps of forming on the layer, at least at the level of the second photosensitive cell, a portion of a fourth silicon and germanium alloy; and growing, by thermal oxidation, a silicon oxide portion which extends into the portion of the fourth silicon and germanium alloy, which results in the migration of the germanium into said layer to form the second silicon and germanium alloy portion.
According to an embodiment, step (b) comprises the steps of forming on the layer, at least at the level of the second photosensitive cell, a portion of a fifth silicon and germanium alloy; insulating the second photosensitive cell with areas of an insulating material; and melting, at least partially, the fifth silicon and germanium alloy portion, which results, by interdiffusion, in the forming of the second silicon and germanium alloy portion.
The foregoing and other features and advantages of one or more embodiments of the present invention will be discussed in detail in the following non-limiting description in connection with the accompanying drawings.
For clarity, the same elements have been designated with the same reference numerals in the different drawings and, further, as usual in the representation of integrated circuits, the various drawings are not to scale. In the following description, color image sensors comprising three types of pixels respectively associated with colors blue, green, and red will be considered.
To limit the crosstalk phenomenon in an image sensor, it is provided to decrease the substrate thickness so that the electron generation occurs as close as possible to the charge storage means. This further eases the forming of insulation areas separating the pixels and extending across the entire substrate thickness. However, to avoid altering the sensor sensitivity, in particular to large wavelengths, it is provided, at least for the pixels associated with photons of large wavelengths, that is, the red pixels in the present example, to form the substrate portion associated with the pixel with a material different from single-crystal silicon, thus enabling more efficient absorption of photons of large wavelengths. An additional constraint is that the used material is compatible with conventional CMOS and CCD sensor manufacturing processes. By many tests, it has been shown that a silicon and germanium alloy fulfills such requirements. Indeed, the greater the germanium concentration in a silicon-germanium alloy, the more the capacity to absorb photons of large wavelengths in the visible spectrum increases.
An embodiment of the present invention will now be described for a CMOS image sensor. It should however be clear that an embodiment of the present invention also applies to a CCD image sensor. Further, in the following description, a back-lit image sensor is considered. It should however be clear that an embodiment of the present invention also applies to a front-lit image sensor.
An element forming a support 5, for example, a semiconductor wafer, is covered with a stacking of insulating and conductive layers 7 and with a substrate 9. Substrate 9 is formed of adjacent portions of lightly P-type doped silicon-germanium alloys 10B, 10G, 10R (P31). Respectively call XB, XG, and XR are the germanium concentrations in the silicon-germanium alloy of portions 10B, 10G, and 10R. The germanium concentration is substantially uniform across each portion 10B, 10G, and 10R. The thickness of substrate 9, which is substantially constant, is designated with reference τf. As an example, thickness τf is smaller than or equal to 1 μm, for example, smaller than or equal to 500 nm. Photodiodes 11B, 11G, 11R (not shown in detail) and MOS transistor source and drain regions (not shown) are formed in substrate 9 on the side of stacking 7. The insulation between adjacent pixels is provided by P-type areas 16 which are more heavily doped (P+) than portions 10B, 10G, and 10R and which extend across the entire thickness of substrate 9. The transistor gates are formed at the surface of substrate 9 on the side of stacking 7. Metal tracks and vias enabling connection between the different system components (photodiodes, transistors . . . ) are formed in stacking 7. On the surface of substrate 9 opposite to stacking 7 are formed, at the level of each pixel B, G, and R, color filters 17B, 17G, and 17R and microlenses 19B, 19G, 19R enabling focusing the light rays towards photodiodes 11B, 11G, and 11R. Color filter 17B, associated with pixel B, lets through, in the visible spectrum, the light rays having wavelengths close to color blue, color filter 17G, associated with pixel G, lets through, in the visible spectrum, the light rays having wavelengths close to color green, and color filter 17R, associated with pixel R, lets through, in the visible spectrum, the light rays having wavelengths close to color red.
When image sensor 1 is illuminated, the incident photons are absorbed in substrate 9 by causing the forming of electron/hole pairs, the electrons being captured by photodiodes 11B, 11G, and 11R of pixels B, G, and R.
Absorption rate A of photons in each portion 10B, 10G, and 10R of substrate 9 is provided by the following relation:
where k is the extinction coefficient of the material forming the considered portion 10B, 10G, or 10R of substrate 9 and λ is the wavelength of the incident photons.
Concentrations XB, XG, and XR are selected so that, for a same thickness τf of substrate 9, the absorption rates of portions 10B, 10G, and 10R of pixels B, G, and R are substantially sufficient for the wavelengths of the light rays associated with these pixels. For example, the absorption rates of portions 10B, 10G, and 10R are maximum and identical. For this purpose in this example, concentration XR is greater than concentration XG which is itself greater than concentration XB. As an example, for a concentration XB equal to 0%, concentration XG may be on the order of 15% and concentration XR may be on the order of 30%. The general sensitivity of the image sensor may thus be substantially uniform for all wavelengths in the visible spectrum. Further, since thickness τf may be smaller than 1 μm, the electron generation location is close to the photodiodes, which enables decreasing the crosstalk. Further, insulation areas 16 may easily be formed across the entire thickness τf of substrate 9. This enables further limiting the crosstalk.
For the image sensor shown in
According to a variation of the previously described embodiments, silicon-germanium portions 10R, 10G are provided only at the level of each red and green pixel, the substrate portion associated with each blue pixel corresponding to single-crystal silicon. Since most of the photons corresponding to color blue are absorbed in a thickness on the order of 500 nm in the case of a single-crystal silicon substrate, a reduced thickness may be maintained for substrate 9.
According to another variation of the previously described embodiments, a portion only of silicon-germanium 10R is provided at the level of each red pixel, the substrate portions associated with each blue and green pixel corresponding to single-crystal silicon. Since most of the photons corresponding to color green are absorbed in a thickness on the order of 1 μm in the case of a single-crystal silicon substrate, a reduced thickness may be maintained for substrate 9.
According to another variation of the previously described embodiments, insulation areas 16 may correspond to regions of an insulating material, for example, silicon oxide, and may be formed by a method of shallow trench insulation type (STI).
According to another variation of the previously-described embodiments, for each pixel B, G, and R, a P-type dopant concentration gradient may be provided in silicon-germanium portions 10B, 10G, and 10R according to the stacking direction of sensor layers, that is, the dopant concentration decreases from the rear surface of the sensor. Such a gradient causes, for each pixel R, G, and B, the forming of an electric field in silicon-germanium portion 10B, 10G, and 10R oriented substantially perpendicularly to the rear sensor surface and which tends to direct the electrons photogenerated in the silicon-germanium portion 10B, 10G, and 10R towards the corresponding photodiode 11B, 11G, 11R.
An example of a method for manufacturing the image sensor of
forming over the entire layer 25 a layer of the silicon-germanium alloy with concentration X0 and thickness τiB, for example, by epitaxial growth under a gas flow, for example, of silane and germane;
forming a first mask on the silicon surface layer covering the silicon surface layer only at the level of the blue pixels;
anisotropically etching the portions of the silicon surface layer and of the silicon-germanium layer unprotected by the first mask to form silicon-germanium portion 30B and silicon portion 32B;
removing the first mask;
depositing a protection layer over the entire obtained structure;
forming a second mask which covers the protection layer only at the level of the green and red pixels;
anisotropically etching the portions of the protection layer unprotected by the second mask to expose silicon portion 32B at the level of each blue pixel and form protection layer 34; and
removing of the second mask.
X
B
=X
0(τiB/τf) (2)
removing protection portion 34 and silicon oxide portion 36B;
forming over the entire layer 25 a layer of the silicon-germanium alloy with concentration X0 and thickness τiG, for example, by epitaxial growth under a gas flow, for example, of silane and germane;
forming a single-crystal silicon surface layer on the silicon-germanium alloy;
forming a first mask on the silicon surface layer covering the silicon surface layer at the level of the green pixels only;
anisotropically etching the portions of the silicon surface layer and of the silicon-germanium layer unprotected by the first mask to form silicon-germanium layer 30G and silicon portion 32G;
removing the first mask;
depositing a protection layer over the entire obtained structure;
forming a second mask which covers the protection layer at the level of the blue and red pixels only;
anisotropically etching the portions of the protection layer unprotected by the second mask to expose silicon portion 32G at the level of each green pixel and form protection portion 40; and
removing the second mask.
The forming of silicon-germanium portion 10R at the level of each red pixel may be carried out by steps similar to what has been described previously for the forming of portion 10G. Heavily-doped P-type insulation areas 16 are finally formed to delimit the pixels.
The image sensor manufacturing method carries on with the steps of:
to obtain the structure shown in
forming the stacking of insulating and conductive layers 7 on the obtained structure;
attaching support 5, for example, by gluing, to the stacking of insulating and conductive layers 7;
thinning down the structure by removal of support 27 and of insulating layer 26 to expose the rear sensor surface;
forming filters 17B, 17G, and 17R and microlenses 19B, 19G, and 19R on the rear surface side of the sensor.
To obtain a front-lit image sensor, the thinning step does not take place and filters 17B, 17G, and 17R and microlenses 19B, 19G, and 19R are formed on the stacking of insulating and conductive layers 7.
In the previously-described manufacturing method example, different germanium concentrations XB, XG, and XR for silicon-germanium portions 10B, 10G, and 10R associated with the blue, green, and red pixels are obtained by providing different initial thicknesses τiB, τiG, and τiR for silicon-germanium portions 30B, 30G, and 30R for which the initial germanium concentration is X0.
According to a variation of the previously-described manufacturing method, silicon-germanium portions 30B, 30G, and 30R are formed before the oxidation step, which then results in the simultaneous forming of silicon-germanium portions 10B, 10G, and 10R. Since portions 30B, 30G, and 30R then have the same thickness, different germanium concentrations XB, XG, and XR for portions 10B, 10G, and 10R are obtained by providing different initial germanium concentrations for portions 30B, 30G, and 30R.
An example of a method for manufacturing a variation of the image sensor of
The initial structure corresponds to that shown in
forming over the entire layer 25 a layer of the silicon-germanium alloy with concentration X0B and thickness τi, for example, by epitaxial growth under a gas flow, for example, of silane and germane;
forming a mask on the silicon-germanium layer covering the silicon layer at the level of the blue pixels only;
anisotropically etching the portions of the silicon-germanium layer unprotected by the mask to form silicon-germanium portion 50B;
removing the mask;
repeating the preceding steps to form silicon-germanium portions 50G and 50R; and
depositing silicon layer 52.
In this example of a manufacturing method, the different concentrations XB, XG, and XR of silicon-germanium portions 10B, 10G, and 10R are obtained by varying initial concentrations X0B, X0G, and X0R.
The next steps of the present example of a manufacturing method may be identical to what has been described for the example of a manufacturing method illustrated in
Another example of an image sensor manufacturing method of
The initial structure corresponds to a structure similar to that shown in
The forming of silicon-germanium portion 10R associated with each red pixel R may be performed by steps similar to those previously described for the forming of portion 10G. A final step of forming of insulation areas 16 of the pixels, which may correspond to heavily-doped P-type regions, is provided.
The next steps of the present example of a manufacturing method may be identical to what has been described for the manufacturing method example illustrated in
Specific embodiments of the present invention have been described. In particular, although the previously-described embodiments relate to color image sensors comprising pixels associated with colors blue, green, and red, it should be clear that an embodiment of the present invention may also apply to color sensors for which the pixels are associated with other colors, for example, the three primary colors (blue, yellow, red), the colors complementary to the primary colors (green, orange, purple), etc.
Such alterations, modifications, and improvements are intended to be part of this disclosure, and are intended to be within the spirit and the scope of the present invention. Accordingly, the foregoing description is by way of example only and is not intended to be limiting.
Number | Date | Country | Kind |
---|---|---|---|
0756447 | Jul 2007 | FR | national |