Patent Application
20110140220
This application claims priority from French Application for Patent No. 09-59060 filed Dec. 16, 2009, the disclosure of which is incorporated by reference.
The invention relates to the field of microelectronics and more particularly to back side illuminated optoelectronic devices as well as to their production.
Silicon-on-insulator (SOI) substrates are replacing conventional silicon substrates in the production of a number of devices, especially in the microwave field and the transistor field. These substrates are the result of a complex process involving an implantation of hydrogen into the substrate followed by a high-temperature treatment to create a zone of weakness in the crystal. The substrate thus treated is next directly bonded to another substrate before cleavage through the zone of weakness created in the substrate. A sandwich is thus obtained consisting of the silicon substrate, silicon oxide and of silicon in which devices may be produced while being insulated from the substrate. The thickness of silicon in which the devices are produced is limited by the depth to which the hydrogen is implanted so as to form the cleavage zone. The quality of the silicon is that of the initial substrate into which the hydrogen is implanted. The quality of the new substrate so bonded is that of the support onto which the implanted substrate is transferred. The production process allows defect-free crystalline layers and a good quality interface with the insulating oxide to be obtained.
After bonding, silicon may be grown epitaxially on the surface of the silicon layer to adjust the thickness of the silicon layer in which devices will later be produced, the oxide layer being employed as insulation and/or as a stop layer for substrate etch steps. Despite their undeniable advantages, such substrates are more costly than conventional substrates.
Furthermore, CMOS microelectronic imaging devices are experiencing continual and rapid growth of their market. The technological challenge is to reduce the pixel size without compromising pixel performance. Two technologies are currently in competition, front side illumination (FSI) technology and back side illumination (BSI) technology. FSI technology consists in illuminating the device from the front side through the back-end layers, i.e. the side on which the photodiodes or phototransistors are situated and which is overlaid with the circuit interconnects. BSI technology consists in illuminating the device from its back side, the radiation being directly absorbed by the silicon substrate.
The two technologies have advantages and drawbacks. BSI technology allows photogeneration effects (cross-talk, quantum yield) to be more easily controlled and parasitic reflections caused by metal lines to be avoided. BSI technology is thus better suited to the production of pixels with small dimensions. Currently, BSI technology is immature and produced on SOI substrates, which were described above. These particularly costly substrates are slowing the technological and commercial development of such microelectronic devices.
According to one manner of implementation and an embodiment, production of a thinned substrate covered with a superficial layer of SiGe allowing production of a CMOS image sensor designed to be back side illuminated, is proposed.
According to another manner of implementation and an embodiment, it is proposed to produce such a sensor from a bulk substrate thinned by successive selective etches leading to a uniform and controlled thickness across the whole of the semiconductor wafer.
According to one aspect, a process for producing a microelectronic device is proposed, comprising production of a first semiconductor substrate comprising formation of a first layer, for example p++-doped silicon (Si), and of a second layer, for example of silicon-germanium (SiGe), between a first side and a second side of the substrate, production of first components and of an interconnecting part on and above the second side, and thinning of the substrate comprising a first selective etch of the first substrate from the first side, stopped on the first layer, followed by a second selective etch, stopped on the second layer.
The thinning of the substrate especially allows back side illumination of an image sensor produced in and on this substrate.
The first selective etch, preferably a chemical etch, allows the variation in thickness of the substrate to be reduced while the second selective etch (for example a chemical or plasma etch), allows the required thickness of silicon to be obtained in a controlled manner.
The production of the first substrate preferably comprises epitaxial growth of the first layer on an initial substrate, for example a p−-doped bulk silicon wafer, epitaxial growth of the second layer on the first layer and epitaxial growth of a final semiconductor layer, for example p−-doped or n−-doped silicon (Si) on the second layer.
The epitaxy of the final semiconductor layer may be carried out until a layer thickness of between 2 μm and 6 μm is reached. This thickness is the thickness of the semiconductor layer through which the image sensor photodiodes will be illuminated. The production of first components may comprise the production of components such as photodiodes and MOS transistors.
The epitaxy of the second layer may comprise an epitaxy of SiGe. The SiGe layer, at least partially preserved after thinning, serves as a channel for electrons so as to reduce, or even eliminate, the dark current of the image sensor.
The epitaxy of the first layer may comprise an epitaxy of p++-doped Si.
An intermediate epitaxy of p−-silicon may be added between the p++-silicon layer and the SiGe layer leading to an increase in the etch rate selectivity between the p-doped silicon layers and the SiGe.
The process may comprise, before the thinning of the substrate, an encapsulation of the first components and of the interconnecting part with at least one insulating material, and the attachment of a second substrate to the encapsulating insulator.
The process may comprise, after the second etch, deposition of a passivation layer on the second layer. This passivation layer thus forms the back side of an image sensor.
The method may comprise production of second components on the passivation layer, for example lenses and color filters.
The first layer may have an etch rate selectivity with respect to the second layer (4) of between 30 and 500, preferably between 60 and 100.
According to another aspect, a microelectronic device is proposed comprising a semiconductor layer, a superficial layer formed on a first side of the semiconductor layer, first components and an interconnecting part that are placed on and above a second side of the semiconductor layer.
The semiconductor layer may have a thickness of between 2 μm and 6 μm.
The superficial layer may comprise an SiGe layer.
The first components may comprise optoelectronic components, for example photodiodes, designed to be illuminated through the semiconductor layer and the superficial layer.
The first components and the interconnecting part may be encapsulated in at least one insulating material, a second substrate being attached to the insulating encapsulation.
The microelectronic device may comprise a passivation layer on the superficial layer.
The microelectronic device may comprise second components on the passivation layer.
The second components may comprise optical components, for example microlenses and/or color filters.
According to another aspect, an image sensor is proposed comprising a microelectronic device such as that defined hereinabove designed to be illuminated from the back side.
According to one embodiment, such an image sensor has the advantage of being produced in two steps. The first step allows the thickness variation of the CMOS portion to be reduced, by grinding the bulk silicon substrate until there remains a thickness beneath the CMOS of 5 to 10 μm, and then by selective (preferably chemical) etching of the p−-doped silicon, stopping on the p++-doped silicon layer. The second step allows the required thickness of silicon beneath the CMOS to be obtained in a controlled manner by highly selective, especially chemical or plasma, etching of the p++-doped silicon layer, stopping on the SiGe layer.
According to one embodiment, such an image sensor also has the advantage of preserving a SiGe layer after the thinning step, which provides a channel for electrons so as to reduce, or even eliminate, the dark current.
Other objects, features and advantages will become clear on reading the following description, given merely by way of non-limiting example, and with reference to the appended drawing in which:
In
To obtain this first substrate 1, a first layer 5 is produced on a silicon substrate 6, for example a p+-doped silicon substrate. Next, a second layer 4 is produced on the first layer 5 followed by a final semiconductor layer 3 on the second layer 4.
The production of the first layer 5, of the second layer 4, and of the final semiconductor layer 3 may result from an epitaxy.
The epitaxial thicknesses are dependent on the etch rate selectivities and tolerances during etching. The higher the etch rate selectivity, the thinner the layer relative to the thickness of the etched layer may be. Particular mention must be made concerning the epitaxy of the final semiconductor layer. This is because the components of the device are intended to be produced on both sides of the final semiconductor layer after the thinning step. The thickness of the final semiconductor layer thus determines the thickness of the silicon in the device on completion of the production process. Generally, an epitaxial layer of between 2 μm and 6 μm is produced.
Likewise, the materials of the first layer 5 and of the second layer 4 are chosen so that there exists a difference between the etch rate of the first layer 5 and the silicon substrate 6 and between the first layer 5 and the second layer 4. The differences in etch rates are generally called etch rate selectivities or just selectivities.
By the choice of materials, it is sought, during the second step, to etch the silicon substrate 6 faster than the first layer 5, the etch rate of which is itself faster than that of the second layer 4. For example, to obtain such selectivities, silicon, highly p++-doped to around 5×1019 cm−3 to 2×1020 cm−3, is chosen as the material of the first layer 5 and SiGe as the material of the second layer 4. Generally an etch selectivity between the first layer and the second layer of between 30 and 500, preferably between 60 and 100, is chosen.
In an imaging application, the thicknesses and the doping of the p++- and SiGe epitaxial layers are determined so as to ensure a lateral electric field (or drift field) beneath the photodiode.
By way of indication, the thickness of the p++-layer is between 1 and 2 μm, whereas the thickness of the SiGe layer is between 20 and 50 nm.
The first semiconductor substrate 1 thus obtained comprises a first side 7 which is the same as the lower surface of the silicon substrate 6 and a second side 2 which is the same as the upper surface of the final semiconductor layer 3.
During a second step (
The first components 10 and the interconnect area are then encapsulated with an insulator 11. This material may be a dielectric used in microelectronics, for example a silicon oxide or nitride.
After a planarization step not illustrated in
After coarse grinding of the silicon 6, a first etch, stopped on the first layer 5, is performed which allows the complete removal of the silicon substrate 6, as well as a part of the first layer 5. The first layer 5 may, without distinction, be partially or totally removed during this first etch. The result is illustrated by
The first etch may, without distinction, be associated with a wet etch step (for example a dilute tetramethylammonium hydroxide (TMAH) solution) and/or a fluorine plasma dry etch. The aim of this etch is to reduce the thickness of the substrate to a thickness close to the value required for the device. The most important characteristic of this first etch is that it has a high etch rate allowing the targeted thickness to be rapidly obtained. The thickness variation of the silicon in which the CMOS components were fabricated is reduced relative to the grinding step. The total thickness variation (TTV) is between 0.5 and 2 μm depending on the method used. However, the result obtained is not sufficient to guarantee a uniform behavior of the imagers over the entire wafer. A selectivity of approximately 10 between the etch of the silicon substrate 6 and of the p++-doped silicon first layer 5 allows a fast etch to be obtained.
The process then continues with a second etch to remove the material left over by the first etch. The second layer 4 may, without distinction, be left unchanged or partially or totally removed during the second etch. Optionally, a layer of p−-silicon may be inserted between the layers 4 and 5 in order to separate said layers and to further increase the overall selectivity of the process. In this case, the first etch will be selective with respect to the p−-doped silicon and an additional etch, selective with respect to the layer 4, will be added to selectively remove the inserted p−-doped silicon layer. The additional etch is preferably a wet etch of HNO3+HF type.
This being so, in an imager application, this second layer 4 of SiGe is advantageously at least partially preserved.
The second etch is, without distinction, a wet etch and/or a remote plasma dry etch that uses a gas mixture, for example a CF4+N2+O2+CH2F2 mixture. The aim of this etch is to complete the reduction in thickness of the substrate to the thickness required for the device. The most important characteristic of this second etch is a very high selectivity between the etching of the first silicon layer 5 and of the second SiGe layer 4. The selectivity is greater than 30. This enables a very low silicon TTV, between 0.05 and 0.3 μm, to be obtained. The result of the second etch is illustrated by
The process continues with the deposition of a passivation layer 13, which may be silicon oxide with a thickness of between 0.05 and 0.3 μm, on the second layer 4 after the second etch. The result is illustrated by
The process concludes with production, in a conventional manner, of the second portion 14 of device components on the passivation layer 13. The second portion 14 of components may comprise through-vias, anti-reflection coatings, filters and lenses. The result is illustrated by
Thus, the device of
The upper side of the passivation layer, i.e. the side on which the optical components 14, such as color filters and/or microlenses, are produced, constitutes the back side of the image sensor device. This image sensor may thus be back side illuminated. In this respect, the semiconductor layer 3 of crystalline quality and small thickness (typically between 2 μm and 6 μm) allows, the CMOS components to function and, at the same time, photon capture and electron transfer as close as possible to the component.
Furthermore, by the diffusion of p-dopants from the p++-layer, a doping profile is obtained beneath the photodiodes allowing the setup of a non-retrograde lateral electric field (drift field) in the silicon facilitating electron transfer.
| Number | Date | Country | Kind |
|---|---|---|---|
| 0959060 | Dec 2009 | FR | national |
