SUBSTRATE FOR FRONT-FACING IMAGE SENSOR AND METHOD FOR MANUFACTURING SUCH A SUBSTRATE
A double BOX structure with a semiconductor intermediate layer and two separation layers addresses the dislocation issue in SiGe layers, significantly enhancing infrared photon absorption in image sensors.
Patent Information
- Application Number
- FR2018055540
- Authority / Receiving Office
- FR · FR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2018-06-21
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2038-06-21
AI Technical Summary
The critical thickness of SiGe layers in image sensors limits their absorption capacity and introduces dislocations, necessitating a compromise between thickness and germanium concentration to avoid relaxation and enhance infrared photon capture.
A double BOX structure with a semiconductor intermediate layer and two separation layers, combined with a seed layer and epitaxial growth, to optimize photon absorption and minimize dislocations, using a single or dual layer transfer process.
Enhances infrared photon absorption by up to 87% compared to single-layer structures, improving the performance of image sensors.
Smart Images

Figure 00000019_0000 
Figure 00000019_0001 
Figure 00000019_0002
Abstract
Description
certain thickness, called critical thickness. This relaxation results in the formation of dislocations within the SiGe layer. Such dislocations would make the SiGe layer unsuitable for the function of the active layer, particularly for an image sensor, and must therefore be avoided. However, the critical thickness is all the lower as the germanium concentration is high. The thickness of the epitaxial layer and the germanium concentration of said layer therefore result from a compromise between: - on the one hand, a thickness large enough to capture a maximum of photons in the near infrared wavelengths, - on the other hand, a sufficient concentration of germanium to increase the photon absorption capacity of the active layer, particularly in the near infrared, and - finally, a limited thickness (dependent on the concentration), less than the critical thickness, to avoid relaxation of the silicon-germanium and the resulting creation of crystalline defects (dislocations). Typically, the aim is to maximize the thickness and germanium concentration of the epitaxial layer to achieve the best possible absorption in the infrared. Preferably, the germanium content of the active layer is less than or equal to 10%. Indeed, the critical thickness of a Si0,gGe0,i layer is of the order of a few micrometers, which is suitable for the active layer of a "front face" type image sensor. The intermediate layer can be made of a semiconductor material or an electrically conductive material. Indeed, the function of this intermediate layer is to allow polarization of the active layer from the rear, in other words via an area arranged between the support substrate and the second electrically insulating layer. Such polarization allows the application of a potential difference between the active layer and the buried intermediate layer. The intermediate layer can be monocrystalline but this is not essential because we are not looking for a capacity to ensure conduction of electrons through this layer, nor any electronic property as is commonly considered for applications other than that of image sensors, but only a capacity to modify the electrical potential of the active layer at the periphery of the first separation layer. The intermediate layer can thus be polycrystalline and / or amorphous, which makes it less expensive to manufacture, and / or metallic. This layer can be more or less doped so as to ensure a capacity to be polarized. A semiconducting intermediate layer is advantageously made of silicon. Said layer then typically has a thickness of between 20 and 150 nm. The first separating layer, electrically insulating, which is interposed between the active layer and the intermediate layer, fulfills the function of the dielectric of a capacitor, and thus allows the active layer to be polarized at the periphery of the electrically insulating material. For this purpose, the first separation layer is chosen to be thin enough to minimize the potential difference to be applied between the intermediate layer and the active layer. Typically, the thickness of the first separation layer is between 10 and 100 nm. The thickness of the first separation layer, however, is too thin to reflect all the photons passing through the active layer, particularly photons with wavelengths in the near infrared range. Consequently, photons passing through the active layer are likely to pass through the first separation layer and the intermediate layer. The second separation layer has the function of inducing a reflection of photons, in particular photons whose wavelength is in the near infrared range, towards the pixel formed in the active layer through the stack comprising: the second separation layer, the intermediate layer and the first separation layer. For this purpose, this second separation layer has a thickness large enough to have a high reflectivity (or optical reflection coefficient) in particular in the near infrared range. Typically, the thickness of the second separation layer, made for example of silicon oxide, is between 100 and 300 nm. Figure 2 is a sectional view of a substrate for a front-facing image sensor according to one embodiment of the invention. Said substrate successively comprises, from its rear face to its front face: - a support substrate 1, preferably semiconductor, - the second electrically insulating separation layer, 2b, - the intermediate semiconductor layer 4, - the first electrically insulating separation layer 2a, and - the 3a semi-conducting, monocrystalline seed layer. Examples of manufacturing processes for the substrate illustrated in Figure 2 will now be described. According to a first embodiment, illustrated in Figures 3A-3E, the substrate manufacturing method comprises two successive layer transfer steps, for example by implementing the Smart Cut™ method twice. On the one hand, with reference to FIG. 3A, a first donor substrate 40 is provided comprising a semiconductor material intended to constitute the intermediate layer 4. On the other hand, with reference to FIG. 3B, the support substrate 1 is provided, and the donor substrate is bonded to the support substrate, the second separation layer 2b being at the bonding interface. As shown in FIG. 3A, said layer 2b is for example previously formed on the surface of the first donor substrate 40 before the bonding. Alternatively, layer 2b could be formed on the support substrate 1, or even consist of the assembly of a layer formed on the first donor substrate and a layer formed on the support substrate. The first donor substrate is then thinned so as to transfer a layer 4 of the semiconductor material onto the receiving substrate. This thinning can be carried out by polishing or etching the semiconductor material via the face opposite the bonding interface. However, advantageously, before the bonding step, a weakening zone 41 is formed in the semiconductor material so as to delimit a surface layer 4 to be transferred; said weakening zone can be formed by implantation of atomic species such as hydrogen and / or helium (said implantation being shown diagrammatically by the arrows in FIG. 3A). After the bonding step, the thinning consists of detaching the first donor substrate 40 along the weakening zone 41, which leads to the transfer of the intermediate layer 4 onto the support substrate 1 (see FIG. 3C).Typically, the thickness of the transferred layer 4 is less than or equal to 300 nm. Optionally, a finishing treatment is carried out on the free surface of the transferred layer in order to promote the implementation of a new layer transfer step, this treatment being able to lead to thinning and reducing the roughness of the transferred layer. With reference to FIG. 3D, a second donor substrate 30 is also provided comprising a monocrystalline material suitable for the epitaxial growth of the active layer and intended to constitute the seed layer 3a. With reference to Figure 3E, this second donor substrate is bonded to the intermediate layer 4 previously transferred to the support substrate 1, the first separation layer 2a being at the bonding interface. As shown in Figure 3D, said layer 2a is for example previously formed on the surface of the second donor substrate 30 before bonding. Alternatively, the layer 2a could be formed on the intermediate layer 4 after its transfer to the support substrate 1, or even constituted by the assembly of a layer formed on the second donor substrate and a layer formed on the transferred intermediate layer. The second donor substrate is then thinned so as to transfer a layer 3a of the semiconductor material onto the receiving substrate, which makes it possible to obtain the substrate shown in FIG. 2. This thinning can be carried out by polishing or etching the semiconductor material so as to obtain the desired thickness and surface condition for the epitaxy of the active layer. However, advantageously, before the bonding step, a weakening zone 31 is formed in the monocrystalline semiconductor material so as to delimit the seed layer 3a to be transferred. After the bonding step, the thinning consists of detaching the second donor substrate 30 along the weakening zone 31, which leads to the transfer of the seed layer 3a onto the structure consisting of the support substrate 1, the second separation layer 2b and the intermediate layer 4. Typically, the thickness of the transferred seed layer is less than or equal to 300 nm. Optionally, a finishing treatment is carried out on the free surface of the transferred seed layer in order to promote the implementation of epitaxy, this treatment being able to lead to thinning the transferred layer and / or to reducing its roughness. According to a second embodiment, illustrated in Figures 4A-4D, the method of manufacturing the substrate comprises a step of depositing the intermediate layer (instead of transferring said layer from a donor substrate) and a single layer transfer step, for the formation of the seed layer. This second embodiment of the method takes advantage of the fact that the intermediate semiconductor layer does not have an optical or electronic function and can therefore be made of a material which is not monocrystalline but polycrystalline and / or amorphous. Thus, the intermediate layer can be formed by deposition on the second underlying electrically insulating layer 2b. With reference to Figure 4A, the support substrate 1 is provided covered with the second separation layer 2b. Said layer 2b is typically formed by thermal oxidation of the support substrate 1 if the latter is made of silicon. This can also be formed by chemical vapor deposition (CVD technique, acronym for the English term “Chemical Vapor Deposition”) which may then require or not a treatment aimed at reducing its roughness. With reference to Figure 4B, the intermediate layer 4 is deposited, for example in polycrystalline and / or amorphous silicon. This deposition can be carried out by CVD or by epitaxy at different temperatures (ranging from 300°C to more than 800°C depending on the techniques used). This deposition can be followed by polishing or a smoothing treatment, for example of the plasma type, of the layer 4 in order to obtain a surface state suitable for bonding and then transferring the seed layer. With reference to Figure 4C, a donor substrate 30 is provided comprising a monocrystalline material suitable for the epitaxial growth of the active layer and intended to constitute the seed layer 3a. According to one embodiment, the seed layer 3a is delimited by a weakening zone 31 formed by implantation of atomic species, such as hydrogen and / or helium. With reference to Figure 4D, the donor substrate 30 is bonded to the intermediate layer 4 previously deposited on the support substrate 1, the first separation layer 2a being at the bonding interface. As shown in Figure 4C, said layer 2a is for example previously formed on the surface of the donor substrate 30 before bonding. Alternatively, the layer 2a could be formed on the intermediate layer 4 after its deposition on the support substrate 1, or even made up of the assembly of a layer formed on the donor substrate and a layer formed on the deposited intermediate layer. The donor substrate 30 is then thinned so as to transfer the layer 3a onto the intermediate layer 4, which makes it possible to obtain the substrate shown in FIG. 2. Advantageously, the thinning consists of detaching the donor substrate 30 along the weakening zone 31. Alternatively, the thinning could be carried out by polishing or etching the donor substrate from the face opposite the bonding interface so as to obtain the desired thickness and surface condition for the epitaxy of the active layer. Typically, the thickness of the transferred seed layer is less than or equal to 300 nm. Optionally, a finishing treatment is carried out on the free surface of the transferred seed layer in order to promote the implementation of the epitaxy, this treatment being able to lead to thinning of the transferred layer and / or to reducing its roughness. This second form of execution of the process is particularly advantageous in that it is less expensive since it involves a single layer transfer step instead of two. Whatever the manufacturing process of the structure illustrated in Figure 2, the epitaxial growth of a layer 3b of silicon-germanium or silicon is then carried out on the transferred seed layer 3a until the desired thickness for the active layer is obtained (see Figure 5), i.e. typically greater than or equal to 1 pm. The epitaxial layer 3b can be lightly doped. The seed layer 3a and the epitaxial layer 3b together form the active layer 3. The thickness of the epitaxial layer 3b being significantly greater than that of the seed layer 3a, it is considered that the optical properties of the active layer are essentially imposed by the epitaxial layer 3b, even if the layers 3a and 3b are made of different materials. Thus, for example, if the epitaxial layer is SiGe but the seed layer is not SiGe, for example when it is silicon, the silicon layer is sufficiently thin (with a thickness less than or equal to 300 nm) compared to the thickness of the SiGe layer so as not to significantly affect the properties of the active layer in terms of absorption in the infrared. However, it is possible to modify the nature of the seed layer, for example by means of a process called "thermal mixing" according to English terminology. In a manner known per se, said process comprises an oxidation of the epitaxial SiGe layer on a silicon layer, said oxidation having the effect of consuming only the silicon (to form silicon oxide) and of causing the germanium to migrate towards the face opposite the free surface of the SiGe layer. A layer of SiO2 is then obtained on the surface which can be removed by etching. With reference to Figure 6, a plurality of electrically insulating trenches 5 are formed in the active layer 3 which extend to the first electrically insulating layer 2a. These trenches are known in the field of image sensors under the term CDTI, an acronym for the English term “Capacitive Deep Trench Isolation”, i.e. deep isolation trench. Each region of the active layer delimited by such trenches is intended to form a pixel of the image sensor. For this purpose, a subsequent step in the method of manufacturing the image sensor is to form a photodiode (not shown) in said region. The methods of manufacturing the trenches and the photodiodes are known to those skilled in the art and will therefore not be described in detail in the present text. According to an optional but advantageous embodiment, with reference to FIG. 7A, the active layer 3 in which the isolation trenches 5 have been formed is covered with an optical confinement layer 6 having an optical reflection coefficient from the front face to the active layer greater than the reflection coefficient from the active layer to the front face. Said optical confinement layer 6 consists of a stack of layers which ensure such selectivity of the reflectivity as a function of the direction of the incident photon. According to a preferred embodiment, said optical confinement layer 6 comprises a layer of titanium nitride between two layers of silicon oxide having different thicknesses. An advantage of such a stack is that it is compatible with the methods used in microelectronics; the formation of the optical confinement layer can therefore be easily integrated into the manufacturing method of the image sensor.For example, the optical confinement layer 6 comprises, from the front face to the back face, a 100 nm thick SiO2 layer, a 10 nm thick TiN layer and a 200 nm thick SiO2 layer. The reflectivity of such a stack from the front face of the sensor to the active layer is 0.5%, while its reflectivity from the active layer to the front face is 37%. Said optical confinement layer 6 allows the incident radiation to pass through the surface of the image sensor substantially without reflecting it, but instead reflects the photons present in the active layer and reflects by the double BOX structure, which has the effect of trapping them in the active layer and increasing the length of their path in the active layer. Said optical confinement layer thus makes it possible to increase the optical absorption of the active layer. According to an embodiment illustrated in Figure 7B, the electrically insulating trenches 5 also extend into the optical confinement layer 6. This configuration advantageously makes it possible to electrically isolate, with respect to each other, two adjacent pixels (or two image sensors) even at the level of the confinement layer, in particular to avoid parasitic or shadowing effects. According to an embodiment illustrated in Figure 8, it is possible to polarize each pixel entirely (i.e. over its entire thickness). For this purpose, each trench 5 is formed of a semiconductor via 5a, for example made of silicon, or electrically conductive, extending to the intermediate layer 4 between walls 5b made of an electrically insulating material. This arrangement is particularly advantageous in that it allows, with a single contact, to polarize the entire pixel since the semiconductor layers 5a and 4 are electrically connected. Finally, according to an embodiment illustrated in Figure 9, it is possible to polarize each pixel entirely and independently. Indeed, by adjusting the depth and thicknesses of the inner and outer walls 5b of each trench, each pixel can be polarized independently of the adjacent pixel. For example, each pixel can be delimited on one side (right side for the central pixel of Figure 9) by a wall 5b made of a relatively thin electrically insulating material which extends to the intermediate layer 4, and on another side (left side for the central pixel of Figure 9) by a wall 5b made of a relatively thick electrically insulating material which extends at least partly into the second separation layer 2b.The portion 4a of the intermediate layer 4 located under the pixel is electrically connected to the semiconductor layer 5a located only on one side of the pixel (right side for the central pixel in figure 9) and is electrically isolated from the rest of the intermediate layer 4. Thus, each pixel can advantageously be addressed independently. Although not shown in Figures 8 and 9, the optical confinement layer could be present on the active layer and be crossed or not by the trenches 5 as shown in Figures 7B and 7A. Examples Numerical simulations of optical absorption were carried out for different substrates, according to the state of the art (with a single silicon oxide layer between the support substrate and the active layer, as illustrated in Figure 1) and according to the invention (with a double BOX structure between the support substrate and the active layer, as illustrated in Figure 5). Said double BOX structure consists of the following stack, from the front face to the back face of the substrate: - first separation layer 2a: SiO2, 40 nm - intermediate semiconductor layer 4: polycrystalline silicon, 100 nm - second separation layer 2b: SiO2, 150 nm. The reflectivity of such a stack is of the order of 72% for an incident wavelength of 940 nm. In these simulations, some substrates are covered with an optical confinement layer having a reflectivity from the front face to the active layer greater than the reflectivity from the active layer to the front face. Said optical confinement layer consists of the following stack, from the front face to the back face of the substrate: SiO2, 100 nm / TiN, 10 nm / SiO2, 200 nm. The active layer consists of either a 6 pm thick silicon layer or a 2 pm thick SiGe layer with a germanium concentration of 10%. The table below shows the optical absorption coefficient in the active layer, for normal incident radiation (perpendicular to the front face of the substrate) and having a wavelength of 940 nm. The simulation does not take into account the influence of diffraction or refraction on the trenches separating the different pixels. Optical confinement layer No No No Yes Yes Yes Support substrate-active layer interface SiO2 30 nm Double BOX Double BOX SiO2 30 nm Double BOX Double BOX Active layer Si 6 pm Si 6 pm Si0,gGe0,i 2 pm Si 6 pm Si 6 pm Si0,gGe0,i 2 pm Optical absorption coefficient 33% 47% 49% 47% 68% 87% A significant gain in absorption is observed when a double BOX structure is used instead of a single SiO2 layer. Optical absorption is further improved by making the active layer of SiGe instead of silicon, and / or by adding the optical confinement layer which confines the photons in the active layer. REFERENCES US 2016 / 0118431
Claims
CLAIMS 1. Front face type image sensor, characterized in that it successively comprises, from the rear face to the front face of said sensor: - a semiconductor support substrate (1), - a first electrically insulating separation layer (2a), and - a monocrystalline semiconductor layer (3), called the active layer, comprising a matrix network of photodiodes, said image sensor being characterized in that it further comprises, between the support substrate (1) and the first electrically insulating layer (2a): - a second electrically insulating separation layer (2b) and - a second electrically conductive or semi-conductive layer (4), called an intermediate layer, arranged between the second separation layer (2b) and the first separation layer (2a), the second separation layer (2b) being thicker than the first separation layer (2a).
2. Image sensor according to claim 1, wherein the first separation layer (2a) has a thickness of between 10 and 100 nm.
3. Image sensor according to one of claims 1 or 2, in which the second separation layer (2b) has a thickness of between 100 and 300 nm.
4. Image sensor according to one of claims 1 to 3, in which the intermediate layer (4) is made of a doped amorphous or polycrystalline material.
5. Image sensor according to one of claims 1 to 4, in which the intermediate layer (4) is made of doped silicon.
6. Image sensor according to one of claims 1 to 4, in which the intermediate layer (4) is made of a metallic material.
7. Image sensor t according to one of claims 1 to 6, in which the intermediate layer (4) has a thickness of between 20 and 150 nm.
8. Image sensor according to one of claims 1 to 7, in which the active layer (3) comprises a seed layer (3a) of silicon.
9. Image sensor according to one of claims 1 to 7, in which the active layer (3) comprises a seed layer (3a) of silicon-germanium.
10. Image sensor according to one of claims 8 or 9, wherein the active layer (3) comprises a monocrystalline layer (3b) of silicon-germanium on the seed layer (3a).
11. Image sensor according to claim 10, wherein the germanium content of the silicon-germanium layer (3b) is less than or equal to 10%.
12. Image sensor according to one of claims 10 or 11, in which the thickness of the layer (3b) of silicon-germanium is less than a critical thickness defined as being a thickness beyond which relaxation of the silicon-germanium occurs.
13. Image sensor according to claim 8, wherein the active layer (3) comprises a monocrystalline layer of silicon on the seed layer.
14. Image sensor according to one of claims 1 to 13, further comprising, on the active layer (3), a layer (6) called optical confinement layer having an optical reflection coefficient from the front face towards the active layer greater than the reflection coefficient from the active layer towards the front face.
15. Image sensor according to claim 14, wherein said optical confinement layer (6) comprises a layer of titanium nitride between two layers of silicon oxide.
16. Image sensor according to one of claims 1 to 15, wherein each photodiode is separated from an adjacent photodiode by at least one electrically insulating trench (5) extending to the first electrically insulating layer (2a).
17. Image sensor according to claim 16, wherein said trench comprises a semiconducting or electrically conductive via (5a) extending to the intermediate layer (4) between walls (5b) of an electrically insulating material.
18. Image sensor according to one of claims 16 or 17 in its dependent relationship with claim 14, wherein said at least one trench (5) extends through the optical confinement layer (6).
19. Image sensor according to claim 17, wherein each trench (5) comprises a first wall (5b) extending to the intermediate layer (4) and a second wall (5b) extending at least partly into the second separation layer (2b) so as to electrically isolate a portion of the intermediate layer (4), the semiconducting or electrically conductive via (5a) being electrically connected to said portion of the intermediate layer (4).
20. Method for manufacturing a front-face type image sensor according to one of claims 1 to 19, comprising the following steps: - providing a first donor substrate (40), - formation of a weakening zone (41) in said first donor substrate, so as to delimit a first semiconductor layer (4), - transfer of said first layer (4) onto a semiconductor support substrate (1), an electrically insulating layer (2b) being at the interface between the donor substrate (40) and the support substrate (1) so as to form a structure comprising the support substrate (1), the electrically insulating layer (2b) forming the second separation layer, and the transferred layer (4), - providing a second donor substrate (30), - formation of a weakening zone (31) in said second donor substrate, so as to delimit a monocrystalline semiconductor layer (3a), - transfer of said monocrystalline semiconductor layer (3a) onto the structure, an electrically insulating layer (2a) forming the first separation layer being at the interface between the second donor substrate (30) and the structure, - epitaxial growth of a monocrystalline semiconductor layer (3b) on the transferred monocrystalline semiconductor layer (3a), said monocrystalline semiconductor epitaxial layer (3b) forming, with the transferred monocrystalline semiconductor layer (3a), an active layer (3) of the image sensor.
21. Method for manufacturing a front-face type image sensor according to one of claims 1 to 19, comprising the following steps: - formation of a structure by depositing a semi-conductive or electrically conductive layer (4) on a support substrate (1) covered with an electrically insulating layer (2b) forming the second separation layer, - providing a donor substrate (30), - formation of a weakening zone (31) in said donor substrate (30), so as to delimit a monocrystalline semiconductor layer (3a), - transfer of said monocrystalline semiconductor layer (3a) onto the structure, an electrically insulating layer (2a) forming the first separation layer being at the interface between the second donor substrate (30) and the structure, - epitaxial growth of a monocrystalline semiconductor layer (3b) on the transferred monocrystalline semiconductor layer (3a), said monocrystalline semiconductor epitaxial layer (3b) forming, with the transferred monocrystalline semiconductor layer (3a), an active layer (3) of the image sensor. 10 22. Method according to one of claims 20 or 21, further comprising the formation of a so-called optical confinement layer (6) on the active layer (3), said optical confinement layer (6) having an optical reflection coefficient from the front face towards the active layer greater than the reflection coefficient from the active layer 15 towards the front face.
